De Economist (2005) 153:331–347 DOI 10.1007/s10645-005-1991-y
© Springer 2005
DE ECONOMIST 153, NO. 3, 2005
NOTES AND COMMUNICATIONS
REFORM OF OCCUPATIONAL PENSIONS IN THE NETHERLANDS
1 INTRODUCTION
With population ageing looming, pension reform is on top of the agenda. Also in the Netherlands, which used to be acclaimed for its sound funded pension system, significant reforms have been necessary in recent years. These reforms particularly concern the second pillar of occupational pensions; the first pillar, which provides a basic public pension to all people aged 65 or older, has been remarkably steady. Triggered by the plunging reserves during the creeping stock market collapse, many pension funds adjusted their pension contracts, often moving from pensions based on a fixed percentage of final earnings to more flexible systems based on average earnings. Also the linkage of pensions to price and wage increases has been loosened. Many pension funds have introduced schemes which explicitly make the indexing of entitlements conditional on the fund’s financial position. Finally, also the supervisory rules for the pension funds are in a process of change. The new supervisory regime is foreseen to become effective in 2006. This article gives an overview of the major reforms in the system of supplementary pensions, and discusses their relevance for the Dutch economy. We focus on old age pensions, although there is a clearly tendency for early retirement schemes to become integrated in old age pensions. In fact, most pension schemes by now have a flexible date of retirement together with actuarial fair recalculation of pensions. We start with a brief description of the system, then we discuss the major reforms, and finally, we analyse the economic impact on the Dutch labour market on the basis of a general equilibrium OLG model for the Netherlands. 2 THE DUTCH PENSION SYSTEM
The pension system in the Netherlands consists of the well-known three pillars. The first pillar comprises the public pension. This is a PAYG-financed lump-sum benefit for all individuals that is linked to the minimum wage. The second pillar includes the occupational pensions. These are organised by employers and employees on the company of industry level. These pensions are mandatory for all workers concerned, funded, and largely defined benefit. The role of the government in the second pillar is limited; it mainly provides the fiscal and legal framework. The third pillar features strictly individual
332
DE ECONOMIST 153, NO. 3, 2005
retirement provisions. Part of those receive a favourable tax treatment, in particular as far as these savings are meant to supplement a deficit in the individual’s occupational pension. Mandatory pensions in the Netherlands are high by international standards. First and second pillar pensions together aim at a level of 70% of final earnings for all income classes (Table 1) at the pension age of 62. At the – statutory – age of 65 pensions can grow up to 100% of the final earnings. The ambitious level of the mandatory pensions in the first and second pillar, is reflected in little savings in the third pillar (Table 2). In the Netherlands, the first pillar accounts for 50% of retirement income, the second pillar for 40%, whereas the third pillar only for the remaining 10%, of which about 60% is tax favoured. In the future, when the occupational system becomes more mature, the share of second pillar pensions is expected to increase up to some 60% of total retirement income. The dominant role of supplementary pensions is reflected in the composition of household wealth (Table 3). Total pension wealth, i.e. the present value of pensions to be expected, constitutes a substantial part of total household portfolio, comfortably exceeding financial wealth (including home ownership) for all age cohorts. The weight of supplementary pensions increases with the age; since entitlements are built up over time, the share is small for the younger cohorts, but for older cohorts it grows up to some three quarters of public pension wealth. As the coverage of supplementary pensions
TABLE 1 – THE AMBITION LEVEL OF MANDATORY PENSIONS (1ST AND 2ND PILLAR) Pension (percentage of the final wage) by income group
Netherlands Germany France Italy Spain Sweden Switzerland U.K. U.S.
Fraction of workers covered %
0.5× average wage
average wage
1.5× average wage
2.0× average wage
2.5× average wage
70 50 80 58 88 93 63 51 57
70 38 72 58 88 69 58 35 45
70 38 65 58 88 66 46 30 39
70 32 54 58 76 65 34 22 33
70 26 48 58 61 65 28 18 29
Source: Whitehouse (2002).
91 46 – 5 – 90 – 46 45
NOTES AND COMMUNICATIONS
333
TABLE 2 – SIZE OF THE PENSION PILLARS (SHARE IN PENSION INCOME, TOTAL = 100)
Share in percent 1st pillar 2nd pillar Other
Netherlands
Germany
France
Italy
Spain
Switzerland
U.K
U.S.
50 40 10
85 5 10
79 6 15
74 1 25
92 4 4
42 32 26
65 25 10
45 13 42
¨ Source: Borsch-Supan (2004).
TABLE 3 – COMPOSITION OF HOUSEHOLD WEALTH, CAPITAL BY AGE COHORT (2000) Cohort by age
Financial wealth, including house ownership Human Capital Public pension (1st pillar) Supplementary pensions (2nd pillar) Total
INCLUDING
HUMAN
% Total 25–34
35–44
45–54
55–64
3.7 82.0 12.8 1.5 100
10.5 70.4 15.0 4.1 100
20.9 51.0 19.3 8.8 100
31.7 27.4 25.1 15.8 100
65+ 41.9 0.0 32.4 25.7 100
Source: Westerhout et al. (2004).
is greater for the younger cohorts it can be expected that the value of supplementary pensions will exceed public pensions in the future. The share of second pillar pensions in the Netherlands is much larger than in other countries. Assets of pension funds have grown from approximately 70% of GDP in 1991 to 120% of GDP in 1999, at the top of the stock market boom. Since then they have declined to somewhat below 100% of GDP. Several factors account for the sharp increase in pension fund assets in the 1990s. First, more employees gained access to collective pension schemes. Also, the maturation of pension funds leads to a growth in assets. Lastly, in recent years the pay-as-you-go early retirement schemes have been gradually replaced by funded pre-pension schemes. Also surge in stock market prices in the second half of the 1990s had a significant effect on the value of pension funds assets. Upon the relaxation of regulation in the beginning of the 1990s, Dutch pension funds increased their share in equities substantially from about 15% to 50% of their portfolios (Figure 1). This resulting risk exposure was comparable to that of pension funds in the UK and the US, but much higher than in most other countries, as Canada, Germany, France and Italy (Davis, 2004). This helped
334
DE ECONOMIST 153, NO. 3, 2005
Figure 1 – Asset allocation of Dutch pension funds, 1990–2002. Source: Westerhout et al. (2004)
pension funds to benefit from the stock market boom in the 1990s, but also made them extra vulnerable to the collapse in prices after 2000. 3 REFORM OF THE PENSION CONTRACT
In recent years, pension funds changed their policy in two major ways. First, the typical defined benefit pension based on final earnings was changed into a pension based on average earnings by most pension funds. Table 4 compares the distribution of pension contracts in 2000 and 2004. It is evident that a number of funds, which are some 900 in total, switched from a final earnings system to a average earnings pension. The share of pension funds with a final earnings pension decreased from 59.4% to 46.2%. Measured by the number of participants the change is far more dramatic, as in the share of active participants with a final pay wage pension dropped from over 50% in 2001 to only a sheer 12.5% in 2005. This remarkable difference is due to the fact that the two largest pension funds, ABP (for civil servants) and PGGM (for workers in care sectors), changed their contract into an average pay pension. Still defined benefit pensions are by far dominant. Only 8½% of the pension funds provides a defined contribution pension, comprising only some 3% of all active participants. The second major reform concerns the way pensions are coupled to price and wage increases. Up to the recent reforms it was more or less standard practice that pensions where fully indexed to prices or wages. In the typical
NOTES AND COMMUNICATIONS
335
TABLE 4 – PENSION FUNDS ACCORDING TO TYPE OF PENSION CONTRACT, 2000, 2004 %Total Number of funds, 2000 Defined benefit final earnings average earnings Other Defined contribution Other Total
59.4 16.5 12.9 8.0 3.2 100
Number of active participants, 2000
59.3 31.3 8.1 1.1 0.2 100
Number of funds, 2004
46.2 25.6 18.1 8.4 1.7 100
Number of active participants, 2004
12.5 74.2 10.0 3.2 0.1 100
Source: PVK, Pensioenmonitor 2004.
final earnings contract workers build up entitlements each year as a fraction of final wage. For example, by adding entitlements of 1.75% of the final wage each year an employee can build up a pension of 70% in 40 years. In an average earnings system participants build up entitlements as a percentage of their current wage rather than their final wage. By subsequent indexation to prices and wages these entitlements – through ‘back service’ – can grow up to a certain percentage of average wages upon retirement. While full indexation used to be more or less standard in the past, it was not longer possible to adhere to this practice after 2001. Several pension funds introduced indexing schemes which link the degree of indexation (varying from e.g. 0% to 100%) on the financial position, for example measured by the coverage ratio that measures total assets as a percentage of the funds’ pension liabilities. Since 2002 indexation was on average significantly below 100%. Such a flexible system of conditional indexation is especially effective in an average earnings system, as indexation pertains both to the pensions of the inactive members and the entitlements built up by the participants who are still active in the labour force. This is in contrast with a final wage system where only those who are already retired or otherwise inactive suffer from incomplete indexation. 4 BACKGROUND AND REGULATORY REFORM
These reforms in the pension contract were motivated by three factors: (1) the sharp drop in reserves ensuing from the stock market crash, (2) the growing difficulty of passing through shocks in contribution rates to the active members, and finally, (3) the new accounting standards and regulatory framework. We discuss each of these factors briefly. Figure 2 shows the evolution of the coverage ratio of pension funds, i.e. the ratio between assets and total
336
DE ECONOMIST 153, NO. 3, 2005
pension liabilities. The solid curve represents the coverage ratio when liabilities are measured according to the ‘fair value’ method. The intermitted curve represents the traditional actuarial measure of the coverage ratio which used a fixed interest rate of 4% to determine the present value of pension entitlements. Up to the mid 1990s, this method clearly overestimated the true value of pension liabilities providing a prudential buffer in pension funding requirements. Due to the falling market rate of interest this buffer eroded rapidly during this period. This gave rise to a paradoxical situation. While according to traditional measures pension wealth surged following the boom in the financial markets the true coverage rate showed a falling trend, due to the concomitant decline in interest rates. So while in reality the financial position already eroded since the beginning of the 1990s, the general sentiment – misguided by the traditional accounting methods – was one of great optimism about alleged excess reserves, leading to cuts in contribution rates for employees and premium holidays for companies. When stock market prices began to fall after 2000 the coverage ratios worsened sharply according to all standards. With the prudential buffer implied in the actuarial method already gone astray, the situation became critical for many pension funds when their value of assets dropped by 25% on average. The increased risk position that had been advantageous before 2000, now turned into its opposite. In the years 2000–2002 pension funds recorded unprecedented losses causing their coverage ratio to plunge. For some funds, the coverage ratio even fell below the critical 100% of nominal entitlements. This development prompted the supervisor PVK (‘Pensioen- en Verzekeringskamer’)1 to tighten its regulations and to demand recovery to at least a coverage ratio of 105 by the end of 2003, and to form additional buffers over a longer time horizon. The second factor leading to pension reform was the growing awareness that it becomes increasingly difficult to accommodate asset shocks by passing them through to the active participants via the higher of lower contribution rates. Due to demographic change and the maturing of pension funds the burden of pension liabilities grows rapidly relative to the wage sum which serves as the basis for raising contributions. Pension funds become top heavy relative to the basis of active participants. The ratio between pension liabilities and wage sum in the Netherlands increases from about 200% in 1990 to 250% in 2001, and is expected to increase to above 350% in 2030 (Figure 3). If the pension contract is not changed this would lead to a sharp rise in the leverage by which asset shocks of pension funds translate into shocks in contribution
1 As from January 1st 2005, the PVK merged with De Nederlandsche Bank (the Dutch central bank) (DNB) and is now operating under the latter name.
NOTES AND COMMUNICATIONS
337
rates. This increased leverage in effect makes the instrument of adjusting contribution rates blunt. Finally, the new regulatory framework plays an important role in the pension reforms. Following international accounting standards, the new framework 250%
200%
150%
100%
50%
0% 1988
1989
1990
1991
1992
1993
1994
1995
Fair value coverage ratio
1996
1997
1998
1999
2000
4% coverage ratio
Figure 2 – Pension funds coverage ratio (assets/ liabilities), 1986–2003 1400
1200
1000
800
600
400
200
0 1990
2001 Liabilities
2030 Wages
Figure 3 – Pension liabilities and the sum of wages, 1990–2030
2001
2002
338
DE ECONOMIST 153, NO. 3, 2005
adopts the principle of market value or ‘fair value’ for the valuation of assets and liabilities. This has a pervasive impact on the way of thinking about pensions, funding and investment strategies of pension funds. One crucial implication is that realistic valuation based on actual market rates of interest will make the value of liabilities and the coverage ratio of pension funds far more volatile in comparison with the traditional actuarial methods based on a fixed discount rate. This is especially so, if investment strategies aim at maximising expected returns rather than at minimising mismatch of risks between liabilities and assets, as is the case for most important pension funds in the Netherlands. A second major implication is that companies become more reluctant towards pension risks as they have to account for the true value of liabilities and their risks on the balance sheet. The fear for excessive volatility of the value of liabilities has lead to a compromise with pension funds, as they will get the possibility to smooth interest rate volatility by taking an average over the last 10 years. The principle of fair value will be incorporated in the new regulatory framework to be introduced in 2006. In March 2004 the Dutch parliament agreed with the final draft of this framework, which will be incorporated in the new Pension Act planned for 2006. Anticipating the new framework the supervisor (DNB/PVK) can now already take the new principles into account. With regard to the funding requirements the new pension act makes a somewhat artificial distinction between ‘hard’ and ‘soft’ pension entitlements. The ‘hard’ entitlements form the basis for the solvency requirements. Hard entitlements are considered to be those that are clearly promised by pension funds. For most pension funds, these consist of the nominal entitlements that participants build up each year. In contrast, the future indexation to prices and wages is generally considered as a ‘soft’ entitlement, thereby being exempted from solvency requirements, provided that pension funds emphasise that indexation is conditional on future decisions by the management of the fund. That is, pension funds should adopt a disclaimer stating that pension indexation is conditional upon future management decisions, that there is no legal right to indexation, and that it is also uncertain for the long term whether or not pension benefits will be indexed. Then indexation does not have to be included in the official liabilities, and therefore also does not have to be funded either. Even despite the fact that the expected indexation may be strictly positive on average. This has two important consequences: first that it discourages pension funds to be transparent about the true value of pension entitlements of its participants. Second, it creates a system where funding is based on the nominal entitlements rather than on a realistic measure of actual pension liabilities. It should be noticed that there is a huge gap between the nominal entitlement and what is being considered as a decent pension to be expected
NOTES AND COMMUNICATIONS
339
upon retirement. For a participant aged 45 the nominal entitlements make up only some 50% of the true expected pension in case of full indexation to a – modest – nominal wage growth of 3%. Consequently, pension funds supervision threatens to be focused on the rather meaningless minimum for Dutch pensions rather than on some more realistic reflection of pension liabilities. This nominal bias is mitigated – but not taken away – by the system of solvency requirements that require a buffer on top of the funding of the nominal entitlements. The solvency buffer should be high enough to honour the ‘hard’ pension liabilities over a period of one year with a confidence level of 97.5%. For an average pension fund with a fifty–fifty mix of equities and bonds, this implies a required coverage ratio of around 130% of the nominal entitlements. Pension funds with a riskier portfolio should hold a larger solvency buffer. For pension funds with less risk, as smaller buffer may suffice. The new rules thus produce an incentive to hold less risk portfolios. At present it is unclear how pension funds are going to react on the new rules. This new supervisory framework has four major draw-backs: First, the focus on the nominal minimum for pensions is at variance with the general preference for real rather than nominal guarantees to pensions. Second, the new framework lacks any incentive for pension funds to increase transparency. On the contrary, it essentially puts a high penalty on pension funds when they are too clear about their indexation policies. As a result, the information to participants is limited to their nominal minimum pension only. Second, as funding requirements are based on the nominal liabilities, they are generally insufficient to cover the value of total liabilities which correspond to actual expected pensions. This may lead to systematic underfunding of pension funds, and thereby shifting the burden of present pensions to future generations. In case of full indexation to prices or wages, it can be obtained that total funding should be 50–100% higher than funding of the nominal liabilities only (Westerhout et al. (2004)). This additional funding requirement is substantially higher than the solvency buffer of 30% mentioned above. Third, by linking funding requirements to the nominal liabilities the new supervisory regime fails to encourage matching in pension fund investments. Although the solvency requirements entail some incentive to reduce volatility in the coverage ratio, this is directed to the match between assets and the nominal liabilities only. Furthermore, for many pension funds the solvency requirement will not be binding in long run equilibrium. As mentioned above, return to full funding of indexed pensions will require substantially more additional capital than the solvency buffer for an average pension fund. In these cases, where the solvency constraint is not binding, the regulatory framework produces no incentive for matching strategies at all. Current investment policies of most pension funds are best described as maximising the expected return under the constraint that the solvency restrictions are satisfied, although there are some recent indications that some funds consider to give more weight to
340
DE ECONOMIST 153, NO. 3, 2005
matching strategies. This is motivated by accounting standards more than by the regulatory framework. Moreover, recent history teaches us that pension funds are not always able to deal smoothly with shocks in their assets and liabilities. 5 LABOUR MARKET CONSEQUENCES OF PENSION REFORM
Thanks to the adjustments in the pension contract the consequences of the pension shock to the Dutch economy have been kept within bounds. The transition to an average pay system together with partial indexation to prices and wages greatly helped to accommodate for the pension shock ensuing form the creeping stock market crash. Even despite these reforms the impact of the pension shock on the economy has been considerable. The rise in pension contribution rates by 4 to 5 percentage points in reaction the weakening of pension funds financial position, caused severe job losses in the labour market. This rise in pension premiums is in fact one of the principal factors in the persisting weakness of the Dutch labour market, featuring rising unemployment rates from 3.3% in 2001 to more than 6% in 2005. Without the reforms of the pension contract the effects on the labour market would have been far more dramatic. The influence of the pension sector on the labour market can be analysed using the dynamic general equilibrium model GAMMA for the Dutch economy (Draper et al. (2005)). This model integrates comprehensive generation accounts in a CGE-OLG model featuring optimising behaviour of households. This model incorporates endogenous life-cycle behaviour with regard to savings and consumption. It also allows for endogenous labour supply depending on government taxes and on the implicit tax comprised in mandatory pension contributions. The model incorporates a detailed representation of the government budget as well as a comprehensive model for the second pillar of occupational pension funds, including the pension funds behaviour with regard to funding and contribution rates. On the government side the model assumes tax smoothing based on perfect foresight of future expenditure and revenues. At any time, the tariff for indirect taxes (consumption tax) is adjusted to ensure sustainability of government finances. It does so only from the perspective of the government; it does not adjust government tax rate to the variations in the implicit tax of pension funds. In other respects the model is necessarily kept simple, so that the results have yet to be treated with care. First, it is based on perfect foresight and perfect capital markets. As a result there is perfect substitution between – mandatory – pension savings and private savings. Second, the production side of the model is of utmost simplicity assuming perfect international mobility of physical capital. Also there is only one
NOTES AND COMMUNICATIONS
341
representative type of households in each cohort. These simplifications help to focus on the pension system and its consequences for government finance. This model can be used to compare the impact of alternative pension contracts on the recovery of pension funds starting from the low initial coverage ratio in 2002. Table 5 compares the economic effects of the recovery process for the ‘new’ pension contract based on average earnings and conditional indexation with the ‘old’ system with full indexation to contractual wages. First of all, contribution rates are 8.4 lower in the reformed system than under the – hypothetical – old system with full indexation. The drop in contribution rates reflects the much greater weight that is given in cutting back indexation during the recovery period of pension funds. The flipside of this greater emphasis on cuts of indexation is that pension payments are reduced for retirees. The fall in pension payments increases over time as also entitlements of those who are still active are sized down, leading to lower pension later in time. The fall in contribution rates clearly diminishes the distortion of the labour market and boosts employment. The increase in employment is smaller, however, than at first sight might be expected for a premium reduction by more than 8 points given the semi-elasticity of 0.3 used for of labour supply. One should notice, however, that not only the contribution rate goes down, but also the value of expected pensions. Therefore, the reduction in the implicit tax is considerably smaller than the cut in contribution rates. Yet, as also the pensioners pay part of the burden (through the cuts in indexation of their pensions) a positive net impact on employment remains, which is further magnified by the fall in government taxation
TABLE 5 – ECONOMIC EFFECTS OF PENSION REFORM: THE AVERAGE ` PAY PENSION WITH CONDITIONAL INDEXATION VIS-A-VIS AN AVERAGE PAY SYSTEM WITH FULL INDEXATION, 2010, 2015∗ Period Pension contributions (% wage sum) Pension payments (% wage sum) Tax revenue (% GDP) Public debt (% GDP) GDP (market prices) Employment firms (labour years) Consumption Tariff consumption tax %
D D D D % % % D
2010
2015
−8.4 −1.3 0.7 −2.9 0.5 1.0 −0.4 −0.1
−7.4 −2.4 0.6 −6.7 0.7 1.2 −0.2 −0.1
∗ 5-year moving averages over the past 5 years. D indicates an absolute differential vis-`a-vis the baseline scenario, % indicates a relative change. Source: Westerhout et al. (2004).
342
DE ECONOMIST 153, NO. 3, 2005
following the long-term improvement of the government budget. The impact of GDP and the government budget largely mirrors the favourable impact on employment. The distributional consequences are shown in Figure 4 which gives the net benefit from the pension system for each cohort, indicated by date of birth. The net benefit is given by the present value of pensions and contributions, discounted by a real discount rate equal to the rate of return of pension funds portfolio (In this analysis we neglect other possible welfare effects of mandatory pensions, and concentrate on the income flows only). These figures show that the older generations born between 1920 and 1960 are hit by the pension reform. Their pensions are cut, while they benefit little or nothing from the lower contribution rates. Generations born in 1960 and later benefit from the reform. The net effect is of the pension reforms is thus a redistribution from the pensioners to the working population. This is also the key factor behind the improvement in labour supply and employment. The other factor underlying the favourable labour market effect is fact that changes in indexation are less distortionary than changes in contribution rates. A cut in indexation is essentially lump sum, as it applies to ‘old’ stock of entitlements, whereas the implicit tax in pension contributions affects the labour decision at the margin.
5000 4000 3000 2000 1000 0 1900 1920
1940
1960
1980
2000
2020
2040
2060
2080
-1000 -2000 -3000 -4000 -5000 Figure 4 – Intergenerational impact of pension reform: the change net benefit of supplementary pensions by age cohort, year of birth (billions of euros). The horizontal axis features the year of birth, the vertical axis the present value of total net benefits for the cohort in billions of euros in 2001
NOTES AND COMMUNICATIONS
343
6 TOWARDS A MORE ROBUST PENSION SYSTEM
Although reform of the pension contract was primarily motivated from the short-term perspective of the pension shock, it also contributes to a more robust pension system in the future. By putting less weight on contribution rates as a policy instrument, and more on the indexation of existing entitlements, the volatility of the implicit tax on labour can be reduced significantly. Because of the growing unbalance between the huge pension fund assets relative to the wage sum of active participants, it is clear that rebalancing the burden of risk is highly demanded for. Due to this unbalance, the contribution rate is becoming blunt as an instrument to accommodate shocks in the pension funds’ position. Risk should therefore be settled by other instruments, either by adjusting entitlements or by changing investment strategies. Up to present, the main development is the introduction of more flexible policies with regard to indexation of entitlements. As long as there is significant positive (wage) inflation this alternative instrument seems fairly effective, especially since the maximum recovery period has been extended from 8–10 to 15 years. Whether it will be sufficient is difficult to tell at this stage. The choice of the pension contract requires a delicate balance between the benefits of safe pensions and the necessity to diminish the welfare cost of volatility in contributions rates. One way to improve this balance might be to reduce mismatch risk. This would require a change in investment policy from a return maximizing strategy towards strategies that put more emphasis on matching. Also the Dutch government could contribute to such a strategy by supplying longterm bonds and indexed bonds. Another direction to go is to rely more on direct settlement of shocks through adjustment of entitlements, and put less weight on contribution rates as a policy instrument. This does not necessarily mean a complete transition to a pure defined contribution system. In fact, the distinction between defined benefit systems and defined contribution systems is far form clear cut. In practice pension systems form hybrid combinations of both, with very fluent transitions from one to the other. When choosing the balance between the certainty of pensions and the robustness of the system, one should also take account of the labour market consequences. As defined benefit systems generally pass on more of the shocks to participants in the labour market, it is obvious that these systems tend to have higher welfare costs than systems which are closer to defined contribution. This is illustrated in Table 6 which compares the economic consequences of a pension shock for a defined benefit system and a defined contribution system. It is assumed that pension funds are hit by fall in stock market prices by 20% in 2010. In the long run, there is little difference between the two systems. The essential difference concerns the adjustment process after the shock. In a DC system the loss of a shock is settled
344
DE ECONOMIST 153, NO. 3, 2005
TABLE 6 – ECONOMIC EFFECTS OF A PENSION SHOCK DUE TO FALLING STOCK MARKET PRICES BY 20% IN 2010∗
Defined benefit (average earnings, conditional indexation) Pension contributions (% wage sum) Pension payments (% wage sum) Indexation to wage growth Tax revenue (% GDP) Public debt (% GDP) Employment (%) Tariff consumption tax Defined contribution Pension contributions (% wage sum) Pension payments (% wage sum) Indexation to wage growth Tax revenue (% GDP) Public debt (% GDP) Employment (%) Tariff consumption tax
2015
2050
∞
D D D D D D D
0.7 −0.2 −33.3 −0.1 1.5 −0.8 0.2
0.6 −1.6 −1.2 0.0 4.0 −0.3 0.2
0.0 0.0 0.0 0.1 7.6 −0.2 0.2
D D D D D D D
0 −2.2 0 −0.6 3.2 0.0 0.1
0 −0.9 0 0.2 10.0 −0.1 0.1
0 0.0 0 0.1 7.4 −0.1 0.1
∗ 5-year moving averages over the past 5 years. D indicates an absolute differential vis-a-vis the baseline scenario, % indicates a relative change. Source: Westerhout et al. (2004).
immediately by a cut in entitlements. This leads to only a very small impact on the economy. The only effect follows from the decrease in pensions and the consequential loss in tax revenue. In order to restore sustainability the government increases the tax rate on consumption, leading to some distortion in the labour market. Government debt grows a little, but stabilises in the medium term. In contrast to the DC system, the economic impact of the asset shock is substantial under the DB system, even for the case of a flexible DB system considered here. Due to the prolonged cut in indexation and the rise in contribution rates unemployment increases over a considerable period of time. Obviously, pensions are more stable in the short run, which has a beneficial impact on tax revenue and public debt in the short run. In order to maintain sustainability, however, tax rates have to be raised more than under the DC regime, thereby aggravating the distortionary impact on the labour market. The difference between the DB system and the DC system can also be considered from the point of the intergenerational distribution. Figure 5 illustrates the shock is distributed over the three groups of active participants, pensioners and future participants. While the share of the shock borne by active participants is practically the same under both regimes, i.e. roughly three quarters of the shock, the crucial difference concerns the share borne by
NOTES AND COMMUNICATIONS
future generations 16%
345
pensioners 6%
active participants 78% future generations 0% pensioners 25%
active participants 75% Figure 5 – Intergenerational risk sharing through pension funds after a stock market shock in a defined system (upper panel) and a defined contribution system (lower panel)
future generations. Under a DC system this is zero, while under a DB system some 16% of the shock is passed on to future participants. As this represents an implicit tax for these generations it distorts the labour supply decision of future cohorts, leading to welfare losses in the future.
346
DE ECONOMIST 153, NO. 3, 2005
As long as the system is symmetric the opposite holds in case of positive shock. Then a windfall gain in assets will lead to lower contribution rates in a DB system, and thus an implicit subsidy on labour. What matters for the net welfare effect of these shocks in contribution rates is therefore the convexity of the welfare losses due to (implicit) tax rates. Unfortunately, we know little about the size of these losses, like also little is known about the size of the welfare gains due to intergenerational risk sharing (cf. Teulings and De Vries, 2004).
7 CONCLUSION
The ideal pension contract must strike a balance between the wish to provide safe pensions to the old, and the necessity to limit risks passed on to younger and future generations. It is clear that too high volatility in contribution rates leads to undesirable welfare costs through their distortionary impact on the labour market. On the other hand some degree of intergenerational risk sharing can be beneficial as it solves a market failure due to a missing market for contracts with future generations (Bohn (1999)). There are a number of reasons to expect that this balance tends shift in favour of the latter. First, the ageing of population together with the maturing of pension funds makes it increasingly difficult to pass on shocks to younger generations. Pension contribution rates become increasingly blunt as policy instrument to accommodate asset shocks. Second, increasing labour mobility between sectors tends to magnify the distortionary impact of shocks, especially if shocks affect different sectors and different pension funds in an uneven manner. Finally, future generations might be expected to become more critical on implicit taxes they have to pay for older generations. This is not to say that it is better to switch to some form of DC pensions altogether. There is a broad range of hybrid systems and the distinction between DC and DB is opaque in practice. Whatever the system chosen, a conditio sine qua non for any efficient pension system is that it should be transparent with regard to the benefits and cost to each individual participant. Transparency of the relationship between contributions and benefits is necessary to avoid unnecessary distortions of the labour market. Even if this would bring out that some people are net contributors while others are net receivers. A robust and efficient pension system is worth paying for. Casper Van Ewisk∗
∗ CPB Netherlands Bureau for Economic Policy Analysis, Universiteit van Amsterdam & Tinbergen Institute, The Netherlands
NOTES AND COMMUNICATIONS
347
REFERENCES Barr, N. (1992), ‘Economic Theory and the Welfare State: A Survey and Interpretation,’ Journal of Economic Literature, 30, 741–803. Bohn, H. (1999), ‘Social Security and Demographic Uncertainty: The Risk Sharing Properties of Alternative Policies,’ NBER Working paper 7030. ¨ Borsch-Supan, A (2004), ‘Mind the Gap: The Effectiveness of Incentives to Boost Retirement Saving in Europe’, Discussion Paper no. 52–04, Mannheim Research Institute for the Economics of Aging. Disney, R. (2004), ‘Are contributions to public pension programmes a tax on employment?’ Economic Policy, 267–311. Draper, D.A.G. and E.W.T.M. Westerhout (2002), ‘Ageing, sustainability and the interest rate; the GAMMA model’, CPB Report 2002/4, 38–41. Draper, D.A.G., A. Nibbelink, J. Bonenkamp and R. Rosenbrand (2005), Description Gamma Model; version Pension Study, CPB Memorandum 115, CPB, The Hague. Euwals, R. (2000), ‘Do Mandatory Pensions Decrease Household Savings? Evidence for the Netherlands,’ De Economist 148, 38–41. Ewijk, C. van, B. Kuipers, H. ter Rele, M. van de Ven and E. Westerhout (2000) ‘Ageing in the Netherlands’, CPB, The Hague. Teulings, C.N. and C.G. de Vries (2004), ‘Generation Accounting, Solidarity and Pension Losses,’ Discussion Paper 03–94/3, Tinbergen Institute, Rotterdam. Westerhout, E., M. van de Ven, C. van Ewijk en N. Draper (2004), ‘Naar een schokbestendig pensioenstelsel,’ CPB Document 67, CPB, Den Haag. Whitehouse, E. (2002), ‘Pension Systems in 15 Countries Compared: The value of Entitlements,’ Centre for Pensions and Super Annuation, Discussion Paper 02/04, London.