HVWXGLRV VREUH OD HFRQRPLD HVSDÔROD

Holger Bonin Joan Gil Concepció Patxot

EEE 38

http://www.fedea.es/hojas/publicado.html

Holger Bonin University of Freiburg Institute of Public Finance D-79098 Freiburg (Germany) holbonin@vwl.uni-freiburg.de

Joan Gil Universitat de Barcelona Facultat d’Economia Av. Diagonal 690 08034 Barcelona (Spain) jgil@eco.ub.es

Concepció Patxot Universitat de Barcelona Facultat d’Economia Av. Diagonal 690 08034 Barcelona (Spain) patxot@eco.ub.es

April 1999

Abstract

The paper investigates the intergenerational impact of the Spanish public pension system after the 1997 Pension Reform Act. Using a Generational Accounting framework we find that maintaining the new legal setting could leave future generations with liabilities as high as 176 percent of base year GDP. As the recent reform measures have been insufficient to achieve sustainability of the current pension system, we also analyse the impact of alternative reform strategies. Within the current pay-as-you-go setting, a further improvement of the tax-benefit linkage in the original spirit of the Toledo Agreement is shown to yield the intergenerationally most balanced outcome, compared to an increase in retirement age or an expansion of public subsidies financed through indirect taxes. Finally, we investigate the generational impact of a move toward a partially funded pension system which could restore the intergenerational balance.

J.E.L Classification: E62, H55. Key words: Spanish pension reform, intergenerational redistribution, generational accounting

We wish to thank J.A. Fernández Cordón for providing us their last updated Spanish population projections. Joan Gil and Concepció Patxot acknowledge the CICYT financial help under the project SEC98-0314.

1. Introduction

Social Security is on the verge of severe crisis in many countries with public pension systems that predominantly work on a pay-as-you-go (paygo) scheme. In the course of the next decades progressive population ageing will hit most western societies, as a consequence of sharply declined fertility combined with ever increasing life expectancy. With a diminished ratio of contributors to pensioners the financial base of paygo-systems is expected to deteriorate. Rising concerns that pension schemes could hence impose severe pressure on future public budgets have induced an ongoing debate on Social Security reform to prevent substantial future deficits.

In Spain, population ageing will be particularly severe, due to fertility rates which rank among the world’s lowest since almost two decades. Old-age dependency is projected to double in the next fifty years, even if fertility rates would recover.¹ Social Security provision seems not well prepared for this unfavourable demographic trend: without reforms, the present contributive pension system could run deficits as high as 2-3 percent of GDP in the decades to come.² The Spanish government has recently enacted a set of reform measures to improve the fiscal sustainability of the Social Security system. The 1997 Pension Reform Act was partly inspired by the so-called Toledo Agreement which, after a series of round-table talks among political parties, social forces and economic experts, was approved in Parliament in April 1995.³ The ultimate intention behind the recent legal amendments is to strengthen the correspondence of lifetime payroll contributions and individual pension receipts, i.e., to stress the contributive character of a pension system which exhibits a strongly redistributive nature at present. This policy in particular demands to lower the average replacement rate currently provided by public pensions.

According to the Instituto de Demografía (1994) the old-age dependency ratio (measured as the number of people aged 65 and over per 100 people in the working age 15-64) increases from 22.5 in 1996 to 33.4 in 2026 and 50.5 in 2050. Fernández Cordón (1998) arrives at similar results.

2 Cf. MTSS (1995), FBBV (1997), Piñera (1996) and Herce and Alonso (1998)

3 Initially the Toledo Agreement was a mere declaration of purposes. A more concrete agreement (Acuerdo de Consolidación y Racionalización del Sistema de Seguridad Social) was signed only in October 1996 by the new government and the two main unions. The 1997 Pension Reform Act (Ley 24/1997, de 15 de Julio, de Consolidación y Racionalización del Sistema de Seguridad Social) culminates the process to reach broad consensus on pension reform.

Herce and Alonso (1998) have demonstrated that the 1997 Pension Reform Act has failed to improve significantly the long-run viability of the Spanish pension system. They estimate the reduction in the annual deficit of the pension system to be lower than 0.1 percent of GDP in 2050. In light of these results, more thorough reform measures appear necessary to stabilise Social Security finance. In this paper, we investigate the long-run budgetary impact of various pension reform policies in Spain. As annual budgets are an ill-defined concept from a neoclassical viewpoint, we employ the intergenerational budgeting concept of Generational accounting suggested by Auerbach et al (1991, 1992, 1994) which takes the intertemporal budget constraint of the pension system as a starting point to calculate cohort specific individual lifetime fiscal burdens, obviously relevant in a life-cycle framework.⁴ This method, first introduced to Spain by Berenguer et al (1997, 1999) for the entire public sector, not only allows us to compute the implicit liabilities accumulated by the present pension insurance system, and the influence of reform measures on its sustainability. It also permits us to analyse the redistributive impact of the reforms across generations.

The first concern of this paper is to judge the intergenerational impact of the Spanish pension system after the new 1997 pension legislation. We find that the measures taken are insufficient by far to reach intertemporal sustainability of the system. We therefore also consider a set of more radical reform proposals. In line with the original spirit of the Toledo Agreement, we compute the intergenerational impact of a higher reduction in average replacement rates than is implied by the 1997 reform. Further, we investigate measures which are directed at the revenue side of the system instead, like the partial substitution of payroll finance by public subsidies financed out of indirect taxes. Finally, as the before-mentioned reform strategies still generate sizeable intertemporal redistribution, we analyse the consequences of a semi-private partial funding system that would restore intergenerational balance.

The paper is organised as follows. Section 2 briefly introduces the methodological framework of generational accounting. Section 3 discusses the utilised data and provides basic generational accounts for the legal status quo. Section 4 analyses the intergenerational impact of various reforms within the current paygo framework. Section 5 proposes a partial funding of future pensions which ensures intergenerational sustainability. We end with our main conclusions in section 6.

4 For critical discussions of the generational accounting approach, cf. Havemann (1994), Diamond (1994), Raffelhüschen and Risa (1997) and Buiter (1997).

2. Generational Accounting for Social Insurance

Conventionally, generational accounting is based on the intertemporal budget constraint of the overall public sector. As was shown by Boll et al (1994) the method can be easily modified to address the intergenerational sustainability of Social Security systems. Our computations follow their approach, but incorporate the most recent methodological refinements suggested by Bonin et al (1997) and Raffelhüschen (1999).

Since a paygo pension system has not accumulated any assets in the past, the present value of all prospective pension expenditure must equal the present value of aggregate payroll contributions made by either presently living or future cohorts. Let denote the lifetime net contributions, i.e. Social Security contributions net of pension receipts, made by all individuals born in year k and discounted back to a base year t. Assuming that pensions and contributions constitute the only payment flows between Social Security and the personal sector, the intertemporal budget constraint of the Social Security system can be written as

\[\sum_ {k = t - D} ^ {t} N _ {t, k} + \sum_ {k = t + 1} ^ {\infty} N _ {t, k} = 0\tag{1}\]

In equation (1), D denotes the maximum life span of individuals. The first sum aggregates the net contributions of all individuals alive in the base year, while the second represents the aggregate net contributions of all generations not yet born. Generational accounting regards as sustainable or intergenerationally balanced any specific tax and transfer policy that, if maintained unchanged forever, does not offend against the intertemporal budget constraint. If a policy is unsustainable, it requires to adjust the fiscal burden of either present or future living cohorts, in order to meet the intertemporal budget constraint.

To analyse the sustainability of base year pension policy, the method in a first step determines the remaining lifetime net contributions to Social Security for all cohorts, assuming that the initial contribution and pension level could be maintained infinitely. For each cohort k the aggregate lifetime net contribution is computed as

5 For reasons of expositional clarity, the presentation assumes migration absent. The methodological modifications necessary to cope with migration have been fully incorporated in the computations [cf. Bonin et al (1997)].

\[N _ {t, k} = \sum_ {s = \max \{t, k \}} ^ {k + D} P _ {s, k} T _ {t, t - (s - k)} \left(\frac {1 + g}{1 + r}\right) ^ {s - t} \quad k = t - D, \dots , \infty .\tag{2}\]

In equation (2), denotes the base year net contributions of a representative individual born in year t-(s-k). stands for the number of individuals born in year k who survive through year s. Further, g represents the annual growth rate of labour productivity, assumed constant, and r denotes the time invariant interest rate used to discount future payment streams. Equation (2) allocates to any future individual of age s-k the average net contributions of a s-k-year-old in the base year, uprated however for prospective wage growth. The application of the discount factor brings all future payments back to the base year.

The aggregate of all present and future cohorts’ net contributions according to (2) provides a direct test, if base year pension policy satisfies the intertemporal budget constraint. Unless the aggregate is zero, the pension system would accumulate intertemporal debt (wealth), and the net contributions of at least one cohort would have to be increased (reduced) in the future. The size of thus defined implicit liabilities is a sufficient indicator for the long term sustainability of the base year status quo. However, generational accounting is also concerned about the intergenerational redistributive impact of fiscal policy, and hence translates intertemporal liabilities into per capita lifetime burdens.

To do so, the method draws a clear distinction between cohorts currently alive and generations not yet born: For any living generation, the division of its aggregate remaining lifetime payments according to (2) by the initial cohort size yields its so-called generational account. The generational accounts measure the per capita rest-of-life contributions of the living by age, holding to a specific policy. The induced intertemporal liabilities are allocated exclusively to future generations. By convention the adjustment required to meet the intertemporal budget constraint is spread evenly among all prospective cohorts. Considering income growth, all future newborns are assumed to experience an equal proportionate change in those lifetime contributions that would result if the policy under investigation could be maintained. The implied generational account for a representative newborn of year t+1 which is representative for all cohorts to follow yields an alternative measure for intergenerational sustainability. A comparison with the generational account of a base year newborn allows to judge the extent of intergenerational redistribution which is imposed by a given policy on the individuals’ level. Any policy that induces distinct fiscal burdens over a complete life-cycle is regarded as intergenerationally imbalanced by generational accounting, as it shifts fiscal burdens over time.

3. The Collapse of the Present Pension System

In this section we are concerned with the intergenerational viability of the current state of Social Security insurance in Spain, considering the reform measures that were introduced by the 1997 Pension Reform Act for a baseline. The intergenerational impact of the reform is analysed through the construction of generational accounts as outlined in the previous section. We begin by summarising the main assumptions of our computations, before turning to the accounting results.

3.1 Baseline Data and Parameter Values

The computation of isolated generational accounts for the Spanish pension insurance requires a long-run demographic projection and data on the average age-specific payroll taxes and pension benefits in a given base year. Our analysis starts from 1996 as the most recent year for which the required data could be assembled.

For the population forecast we extend the projections provided by Fernández-Cordón (1996, 1998). A medium mortality hypothesis is adopted which assumes a continuous moderate decline of mortality rates. This appears compatible with the secular trend in Spain and evidence from other countries. Life expectancy at birth reaches 77.2 years for males and 84.0 years for females in 2025 which means a gain of respectively 2.9 and 2.6 years from the base year. Since future fertility and migration are rather difficult to predict, we distinguish two demographic scenarios, adopting identical mortality patterns. In the more optimistic upper scenario, the total fertility rate rises linearly from 1.16 in 1996 to 1.8 in 2025. Immigration is determined endogenously to keep constant the absolute size of the workforce. The age structure of the population at working age changes nevertheless, because the age composition of immigrants differs from that of the initial labour force. The alternative scenario adopts a rather more pessimistic view. The fertility increase is assumed to come to an end at 1.6 in 2025, and net immigration is kept fixed at 30,000 people per annum. In either scenario, all demographic parameters are maintained constant after 2026 until the final year of the long term projection, which is 2200.

Age- and gender-specific profiles on average individual payroll contributions were derived from the 1996 wave of the Continuous Household Expenditures Survey, a rotary panel run since 1985. Incidence of Social Security contributions is assumed to fall exclusively on labour income earners.⁶ Relative profiles for Social Security receipts by pension type were taken from Social Security administration data [cf. INSS (1995, p. 281)]. Using this data source, we can distinguish between old-age (retirement) pensions that will undergo reform, and other pensions like disability, orphan and widows pensions, which are kept constant in real terms over the entire projection period. Following generational accounting conventions, all original micro profiles were revaluated to yield the corresponding macroeconomic aggregate in the base year, taking into account the observed population structure. We assign 3.67 billions of pesetas to old-age pension and 3.04 billions of pesetas to other pensions [cf. IGAE (1997, p. 74)]. As social insurance in Spain follows a single cash approach, specific contributions to Social Security cannot be isolated. We therefore initially equate payroll revenue to overall pension expenditure, supposing that the pension system actually worked as a pure paygo scheme in the base year.

For a baseline, all prospective contribution and pension profiles are submitted to annual labour productivity growth which is set constant at an annual rate of 1.5 percent reflecting the long term growth trend in Spain.⁸ With respect to pension payment profiles, we deviate from generational accounting’s standard practice of growth uprating since the Spanish pension system insures against inflation risk only. Therefore, we assume that the primary insurance amount of consecutive cohorts entering retirement increases at rate g in line with their lifetime income. Once retired, the cohort-specific pension benefits upon death are considered to remain constant in real terms. Differentiating cohort-specific pension profiles, we also take into account the final maturing of old-age pension system which at present still grants significantly lower average payments to the oldest males, as compared to the more recently retired. The initially observed variations in average pensions are maintained upon death of all base year pensioners. Finally, we deviate from standard growth uprating to model the long term effects of the 1997 Pension Reform Act which are introduced in the following section. The discount rate used to take all future contributions and pension payments back to the base year is set to 4 percent, mirroring the past trend of real pre-tax interest rates in Spain.

6 This standard incidence assumption is shown empirically valid for Spain by Argimón and González-Páramo (1987) and Escobedo (1991). As the survey collects net income only, we translated net income into gross terms applying effective income tax rates. Contributions are assumed proportionate to gross income.

7 This approach might roll over an actual deficit of the pension system that is obscured due to the single cash approach to other parts of the social insurance system. If there was a deficit in the base year, the implicit liabilities of the pension system would increase.

8 MTSS (1995), FBBV (1997), Herce and Alonso (1998) opt for similar values which range from 1.1 to 2.5 percent.

3.2 Modelling the Pension Reform Act

In July 1997 the Spanish government enacted a Pension Reform Act that was influenced by proposals to reform pension insurance generally known as the Toledo Agreement. The measures taken aim to rationalise the pension insurance system, in order to consolidate Social Security finance in face of the demographic challenges ahead. The guideline of the reform suggestions was to strengthen the actuarial fairness of the pension system by improving the correspondence of individuals’ lifetime contributions and pension receipts. To investigate the intergenerational sustainability of Spanish pension insurance, the new legal status quo should be considered as the baseline.

Nevertheless, of the entire set of reform measures enacted in 1997, we have only taken into account those which are expected to reduce the implicit indebtedness of the pension system, as they lower the average pension received for prospective retirees. The majority of the omitted legal amendments tends to increase intertemporal liabilities [cf. Herce and Alonso (1998)], but their financial effects are difficult to quantify accurately.¹⁰ To avoid ambiguities, we rather prefer to analyse a lower bound of prospective liabilities.

In detail, we adjust the standard productivity growth uprating of base year contribution and pension profiles for two measures enacted in 1997. First, the reform will lower the average pension of future retirees, as it reduces the minimum replacement rate promised by the Social Security system for labourers with short earnings history. Before the reform, 15 years of contributions granted 60 percent of the full pension, with each additional year of contributions until a maximum of 35 adding two percentage points. Under the new regime, 15 years of contributions promise only 50 percent of the full pension, while each year of contributions between the 16th and the 25th adds three, and each year between the 26th and 35th two percentage points of the full pension. Secondly, the 1997 Pension Reform Act changes the pension formula, gradually prolonging the earnings history that enters the calculation of the primary insurance amount from 8 to 15 years between 1997 and 2002. From 1997 onward one contribution year is added until 2002, when two additional years are taken into account. Considering the typical lifetime earnings profile, this measure leads to a decline in assessed past wages, although wages are still adjusted for inflation in the pension formula, except for the two final years. Both measures combined might affect the entry level of retirement pensions for prospective pensioner cohorts. The projection of the base year old-age pension profile needs to be adjusted accordingly.

9 This real interest rate is compatible with a 6 percent nominal long term interest rate, adjusted by a 2 percent expected inflation rate.

10 In particular, our analysis leaves out a) the progressive elimination of wage contribution limits for certain professional categories, b) the partial equalisation of contribution payments between the general scheme and categorical schemes, c) the pension increase for some orphans and widows, d) the envisaged more severe control of disability benefits.

Following empirical evidence,¹¹ we assume that the earnings history of a representative wage income earner lasts 30 years and that he or she retires at age 63. In addition, we set past real wage growth to 1.5 percent per annum and suppose annual inflation equals 2 percent. For this scenario we find that the representative pensioner when retiring will experience a gradual decrease in old-age pensions which ranges from 0.74 percent in 1997 to 4.96 percent from year 2002. It is worth noting that the new treatment of workers with short earnings history under the given assumptions does not affect at all the percentage of the regulatory base that the representative individual receives: with 30 years of contributions he or she can still expect 75.6 percent of the regulatory base under the new legislation.¹² The entire reduction of the average pension level can be attributed to the less generous formula which will be used to compute the primary insurance amount.

3.3 Baseline Generational Accounts

Figure 1 plots the generational accounts for each cohort alive in 1996 and a representative future generation, given the above outlined baseline and employing the upper demographic scenario. Basically, the generational accounts for the base year living exhibit a familiar lifecycle pattern.¹³ The present value of lifetime net contributions, i.e., contributions net of pension receipts, to the Social Security system is positive for a base year newborn. Lifetime contributions thus exceed lifetime pension receipts by 1,371 thousand pesetas in present value terms, suggesting that the internal rate of return is likely to be rather low. During childhood and youth, the generational accounts gradually increase with age, because the period of active contribution payments and pension receipts is less heavily discounted for these cohorts. Remaining lifetime tax payments reach their maximum with entry to the labour force. The generation aged 20 in the base year faces a net contribution burden of about 2,171 thousand pesetas in present value.

¹¹ Cf. MTAS (1997) and Monasterio et al (1996).

¹² Since the average earnings history lacks five years to the necessary career of 35 years, full replacement is first reduced by 10 percent. Additional 8 percent, or 7.2 percentage points, are deducted for each year of early retirement.

13 For a comparison, cf. the generational accounts for the isolated German social insurance system found in Bonin et al (1999).

After reaching the peak, the generational accounts fall continuously for cohorts initially in the labour force. A relatively shorter contribution period and less discounted benefits offset the effects of rising income (and thus contributions) with working age. The rest-of-life net tax payments turn negative for a 35-year-old. Over his or her remaining life-cycle, he or she will experience a net transfer from the Social Security system. The maximum transfer is reached for the base year 65-year-old, the cohort that reaches legal retirement age in the base year. For all older cohorts, remaining lifetime pension receipts gradually decrease in line with their shorter life expectancy.

As generational accounting is a purely forward looking concept, the remaining lifetime net contributions of cohorts at different stages of their life-cycle cannot be compared. Nevertheless, our findings help making apparent a distinctive feature of the Spanish pension system. The fact that the maximum net payment to the system (at age 20) amounts to only about 15 percent of the maximum receipts suggests significant income redistribution across generations to the favour of retirees. Our findings for the base year living thus support recent empirical evidence by Gil and López-Casasnovas (1998).

Turning to the question of intergenerational sustainability of the present Social Security system, we find that it could leave future generations with an overwhelming bill, despite the measures enacted with the 1997 pension reform. Maintaining the current state of pension insurance (including the envisaged pension cut) upon infinity would accumulate liabilities as high as 175.7 percent of base year GDP. If this debt burden is rolled over entirely to generations not yet born, the pension system must raise lifetime net tax contributions of 7,143 thousand pesetas from a representative member. The lifetime net payment to Social Security for future cohorts would be almost five times higher than for a base year newborn. The mammoth degree of intergenerational redistribution, mainly induced by the rapid ageing of the Spanish population, fundamentally questions the long-run viability of the pension system. In face of the severe demographic transition, the current vast redistribution by Social Security across generations, i.e. from younger and richer cohorts to older and poorer cohorts, appears no longer sustainable.

Our findings also indicate that the 1997 Pension Reform Act has achieved little to shield Social Security against the demographic challenges ahead. In fact, if we withdraw the envisaged pension cut from the baseline, the implicit liabilities accumulated by the pension system are only insignificantly higher at 187.6 percent of 1996 GDP. This result is hardly surprising, given that the reform has imposed only a very moderate burden on the present living: according to the generational accounts the enacted pension cut will raise the lifetime net contributions of, e.g., a base year newborn by a mere 6.1 percent. Further, policy has left all transfers to base year pensioners completely untouched, as the reform guarantees existing pensions.

The finding of this section that maintaining the current state of Spanish pension insurance will accumulate immense debt to finance all current pension claims in a progressively ageing society does hold for a wide range of scenarios. Variations of productivity growth and interest rates within an economically reasonable range do not change the basic result. For example, supposing an annual growth of 2 percent and a discount rate of 3 percent yields intertemporal liabilities as high as 528.8 percent of 1996 GDP. As compared to the baseline, this economic scenario gives higher weight to payments in a more distant future. The higher present value of distant pension payments in an aged society increases the intertemporal deficit. At the opposite end, employing a growth rate of 1 percent and an interest rate of 5 percent leads to intertemporal liabilities of only 86.2 percent. Since distant pension payments are more heavily discounted, the present favourable demographic situation with a large labour force gains more weight. Nevertheless, for reasonable combinations of interest and growth rates, the qualitative finding of a severe intergenerational imbalance remains sufficiently robust.

Also the choice of the demographic setting does not seriously alter our basic findings. Employing the more pessimistic demographic scenario outlined in section 3.1 leads to even higher intergenerational redistribution. The additional contribution burden of future cohorts amounts to 8,658 thousand pesetas, as compared to 5,772 thousand pesetas in the baseline. Of course, assuming an even more optimistic fertility development than in the already optimistic baseline would show an improved sustainability of the pension system. However, in line with other generational accounting studies, we find that the overall impact of future fertility on the tax burden of future generations remains insignificant.

3.4 A Paygo Perspective

The huge intergenerational debt computed in the previous section is the result of a thought experiment: The base year contribution and pension profiles were projected adapting a simple productivity rule. This proceeding implies to keep the initial legal parameters of the pension system, in particular contribution and average replacement rates, constant in all future, even if that would contradict the institutional arrangements of a paygo system which cannot run deficits. The above generational accounting results therefore reflect an extreme scenario which gives up the non-deficit requirement of a paygo scheme right from the base year.

As the opposite extreme, one could preserve a pure paygo system upon infinity. To achieve an annually balanced budget, payroll contributions and/or pension transfers need to be continuously adjusted. As the consequence no intertemporal liabilities are accumulated, and hence intergenerational balance of generational accounts is achieved by definition.¹⁴ This approach translates the intertemporal debt that was accumulated in section 3.3 into possible time paths of contribution or replacement rates. It allocates the burden from population ageing among all present generations and future cohorts, if at varying degree. Obviously, any ‘real world’ development would be found between the two extremes of rigidly maintained paygo practice and a policy switch to the accumulation of debt that is rolled over to future generations.

We investigate two benchmark scenarios which maintain the paygo system forever. The first is a defined benefit plan. Contribution rates are adjusted annually to hold a zero deficit, while keeping the replacement rate of pensions constant. The alternative is a defined contribution plan, which takes the pension level as endogenous, and fixes base year contributions rates. Figure 2 plots the development of contribution rates and pension levels for either scenario, indexing the respective base year level with 100. As a direct reflection of the demographic development, endogenous adjustment of contributions or transfers mainly takes place in the three decades after 2015, before the system converges to a stable state. The necessary adjustments to maintain paygo finance are sizeable – pensions need to be reduced by about 50 percent at the maximum, or, alternatively, contributions have to be raised by 100 percent. In light of these figures, which imply a considerable intergenerational distribution conflict, the viability of the Spanish pension system is clearly in doubt. However, the still rather favourable demographic environment during the next two decades might seduce short sighted politicians to postpone the necessary reforms.

14 In fact, there might occur slight differences in the generational accounts of the base year newborn and their immediate successors. The life-cycle of two consecutive cohorts is shifted by one year, and hence both experience a marginally different demographic environment.

Both scenarios leave cohorts at the beginning of their working life or younger with the highest burdens in present value terms. These cohorts will live entirely through the period of highest contribution rates or lowest pension replacement. Other cohorts are affected too, but at lesser degree. Obviously, in the endogenous contributions scenario, base year pensioners are not hit at all, in contrast to the endogenous pensions scenario, where they share their part of the fiscal burden from population ageing.

In the consequence, the two strategies yield sustainable lifetime net contributions that are significantly different. With endogenous pensions, the generational account of a base year newborn would amount to 2,791 thousand pesetas, as compared to 1,371 thousand pesetas in the generational accounting scenario. Allocating parts of the fiscal burden to present pensioners about doubles the lifetime contribution rate of a current newborn to the pension insurance, to the favour of future generations which are left with a significantly lower burden (2,784 thousand pesetas instead of 7,143 thousand pesetas). Since a policy of endogenous contributions exempts current pensioner generations, the sacrifice required from the remaining cohorts is necessarily greater. Moreover, the contribution adjustments hit at an earlier and hence less discounted part of the life-cycle. The sustainable generational account of a base year newborn therefore almost doubles in comparison to the policy of endogenous pensions. He or she faces a lifetime tax burden of 4,675 thousand pesetas.

From the viewpoint of generational accounting, there is no way to choose between these two extreme paygo policies that are hardly politically feasible. Either paygo strategy restores intergenerational sustainability of the pension system which is the essential normative criterion of the method. Policy recommendations beyond the pure summation of payment streams would require a welfare analysis which only a general equilibrium model could provide. It is likely that an efficient policy requires some combination of the two policies. There are positive reasons that could move the policy choice in the direction of reduced pensions: first, the lower life-cycle contribution rate implied might reduce the labour market distortions of pension insurance. Second, it might be reasonable to punish future pensioner cohorts for their fertility behaviour which is partially responsible for the demographic destabilisation of the system.¹⁵

4. Reforming the Paygo System

Irrespective whether the generational accounting approach or a pure paygo perspective is adopted, the lesson from the above analysis pints at the same direction: The recent reform attempt to prepare Spanish pension insurance for the demographic challenge ahead was thoroughly insufficient to set the Spanish pension system on an intergenerationally balanced time path. There is a clear need for further reform. In this section, we investigate the intergenerational impact of alternative policy measures that might become part of the political agenda to restore intergenerational sustainability. Any reform experiment is judged against the baseline scenario established by the 1997 Pension Reform Act.

4.1 Strengthening the Tax Benefit Linkage

As we have noted above, one intention of the 1997 Pension Reform Act was to improve the tax-benefit linkage of the Spanish pension insurance, in order to reduce its considerable redistributive impact. Politicians have been very timid however to translate the suggestions of the Toledo Agreement into law. In a first experiment, we therefore investigate the impact of a more rigid attempt to strengthen the tax-benefit linkage by lowering the entry pension level of future generations of retirees. In line with the strategies analysed by Monasterio et al. (1996) and FBBV (1997), we opt for a moderate phasing in of more severe measures to reduce future pension levels which are assumed to mature after 20 years only. Obviously, a faster pension cut could be fiscally more effective, but would have difficulties to gather political support, due to sizeable adjustment costs.

Our analysis accounts for three reform proposals. First, we assume that the maximum number of contribution years required to receive a full pension is linearly increased from 35 to 40 within five years from the base year, thus approaching a standard set by many OECD countries [Cf. World Bank (1994)]. At the same time, strict proportionality is introduced with respect to the weight of each contribution year, rather than giving a higher weight to the first 15 years. We assume that from year 2001 each year of contributions gains the same percentage: 2.5 percent of the full pension. If an average earnings career lasts 30 years, this policy reduces the pension a contributor with average income would obtain by 15 percentage points, compared to the baseline legal setting.

15 Patxot (1994) argues that cohorts which enter retirement in an ageing population have partially opted out the investment in children. Their consequently higher personal lifetime consumption possibilities might overcome the lower pension received from a paygo insurance with fixes contribution level. Sinn (1997) offers reform proposals, including a pension cut, for German Social Security based on this idea.

Secondly, in gradual steps we take into account the full earnings history in the computation of the primary insurance amount. From 1997 to 2002 one additional year enters the regulatory base formula. From 2003 onward, two additional years are considered, until a maximum of 40 years is reached in 2015. Finally, the reform strengthens the incentives to refrain from early retirement. For each year of early retirement, the loss in pensions is increased from 8 to 9 percent. All three measures combined lead to a significant reduction of future pensioners’ primary insurance amount. For a representative individual with 30 years of average contributions who retires at age 63, the pension cut amounts to about 30 percent from 2010.

The intergenerational impact of this reform is summarised in the second column of Table 1 which collects the basic generational account indicators for the various policy experiments of this section. Improving the correspondence of lifetime contribution payments and pensions received clearly enhances the sustainability of the Social Security system. Implicit liabilities of the pension insurance are greatly reduced by almost 60 percentage points to 116.5 percent of base year GDP. The fiscal imbalance between current and future newborns decreases accordingly. It falls from 5,772 thousand pesetas in the baseline to 3,826 thousand pesetas in a scenario which opts for a more rigid realisation of the Toledo Agreement. Future newborns are relieved at the expense of the present living. For example, over the entire lifetime of a base year newborn, net pension contributions rise by about 400 thousand pesetas. St ill, the reform is far away from achieving intergenerational balance, mainly due to its smooth introduction which delays its full effect, and the exemption of current pensioner cohorts whose expected lifetime transfer are left untouched. Hence, even if a policy toward more actuarial fairness marks a step in the right direction, there is a clear need to complement it with other reform measures.

4.2 Payroll Contributions vs. Consumption Taxes

In Spain, indirect taxes, and value-added taxes (VAT) in particular, are well below the average of the other EU members at present. As harmonisation of indirect tax rates within the EU is still a major political objective, there appears to be ample scope to raise VAT. In this section we investigate the intergenerational impact of a base year increase in the average VAT tax rate. The additional revenue from indirect taxation serves to reduce payroll contributions to Social Security by the same amount. The initial budget of pension insurance thus remains in balance. The idea to substitute payroll contributions to Social Security by public subsidies that are financed out of consumption based taxes is appealing. In general, consumption is more evenly spread over the life-cycle than labour income. Relying on consumption taxation to finance social security could give pensioners a share in financing pensions (reducing the net transfer to the old), and hence make the system less prone to demographic ageing.

The normal VAT tax rate in Spain is 16 percent at present, as opposed to 19.4 percent for the average of all EU member states, and 18 percent in the four major economies, i.e. France, Germany, Italy and the UK [cf. Alonso et al (1997)]. To analyse the redistributive effects of switching tax bases for financing pension schemes, one might assume that a three percentage point increase of the normal VAT rate is feasible in light of the European harmonisation effort. Taking the average Spanish VAT rate of 13.25 percent [cf. MEC (1995)] as a starting point, we get that a one percentage point increase in the VAT rate collects 289.6 thousand million pesetas of additional revenue which would allow to decrease the implicit contribution rate to Social Security by 1.22 percentage points.¹⁶

Given the above assumptions, we find that switching to a consumption oriented revenue base to finance public pensions does not enhance the sustainability of the pension system. As reported in the third column of Table 1, the extent of intergenerational redistribution indeed hardly changes at all. The life-cycle effects of the reform almost cancel out in present value terms. The generational account of a present newborn rises by a mere one thousand pesetas. Only present pensioners face a burden as compared to the baseline. They can expect slightly reduced rest-of-life net transfers. Therefore, the intertemporal liabilities of the pension system fall by about 4 percentage points, to 172 percent of base year GDP. Accordingly, the tax raise for future generations is only marginally diminished: their tax burden falls by only 125 thousand pesetas.

16 This reasoning admittedly neglects general equilibrium considerations, or even partial equilibrium results for the labour market. A move toward higher consumption taxation would in general induce favourable macroeconomic repercussions, as it eliminates labour market distortions and raises incentives to save [cf. Lopez-García (1996)]. Generational accounting omits this dynamic welfare effects, and is limited to the first-round income effects. For Spain Vilches and Salas (1994) and Nadal (1994) have shown that the possible reduction in payroll contribution rates actually might be higher. Our findings therefore provide only a lower bound of the actual improvements of sustainability.

This somewhat unexpected result mainly reflects the age-specific profile of average VAT payments in Spain. In contrast to many other countries, indirect tax payments are spread very similar to payroll contributions among age groups. It is beyond the scope of this paper to decide whether this is due to a peculiarity of the Spanish tax system (or consumption behaviour), or due to deficiencies in the microeconomic data. If the former is true, pension policy cannot expect much from a switch of the tax base toward consumption.

4.3 Increasing Retirement Age

The 1997 Pension Reform Act has not established a concrete measure to raise retirement age, although this proposal appears as an obvious measure to curb the effects of reduced mortality which has prolonged and will prolong even further the average life span in retirement. Measures to increase legal retirement age have already been enacted in a number of countries, and have been shown to be effective to improve the long-run viability of pension systems. 17

In the retirement age experiment for Spain, we assume that policy manages to increase the effective average retirement age by two years. One may think that this is achieved by a raise in official retirement age, or by further removing incentives for early retirement [cf. Gómez-Sala (1993)]. We assume that if effective retirement age increases, the entire relative age profile of old-age pension receipts initially observed is shifted forward by two years. This process takes place gradually over a period of five years from the base year. As the incentives on active labour force participation are rather difficult to predict, the relative age profile of payroll contributions has been left unchanged in the computations. This procedure neglects the potential revenue effects of a prolonged working life. As we do not consider the level effects of a longer stay in the labour force, we implicitly assume the system as actuarially fair, at least as far as the final years at the end of the working life are concerned.

As it could be expected from this type of reform, an increase in retirement age significantly improves the sustainability of the pension system, as can be seen from the final column of Table 1. The measure increases the generational accounts of all living cohorts younger than 65, because of the prolonged working period and reduced benefits during retirement. In the consequence, the intertemporal liabilities accumulated by the system fall by about 30 percentage points to 146.5 percent of the 1996 GDP. Correspondingly the system moves closer to intergenerational balance. While present newborns experience net lifetime contributions of 1,590 thousand pesetas, future generations are exposed to a 4,813 thousand pesetas higher burden, contributing 6,403 thousand pesetas net.

¹⁷ Cf., for example, Crettez et al (1999) for the French case.

Although the increase in retirement age alone could not restore intergenerational balance, it helps, applied together with other measures, to re-address the long-term viability of the pension system. Combined with the harder realisation of the Toledo Agreement discussed in section 4.1, it brings implicit pension liabilities down to around 80 percent of GDP. While still not sufficient to restore full sustainability, there is hence a non-negligible potential for reforms within the existing system. Later retirement might also show a positive effect not considered in this framework. It could ease the possible pressure on the labour market when the baby-boomers retire.

5 Restoring Sustainability: A Move Toward Partial Funding

Section 4 focused on reforms that are directed at restoring intergenerational balance within the existing paygo framework, neither of which was sufficient however to ensure intergenerational sustainability of the pension system. As was shown in section 3.4, intergenerational balance would require fully endogenous taxes and/or transfers. Yet, the associated fiscal burdens which are spread unevenly among cohorts are unlikely to be tolerated by the generations affected.

To reduce the demographic pressure on paygo Social Security, several countries have recently made a step forward to (partially) funded systems [cf. CBO (1999)]. In a funded system the rapidly increasing demand of old-age pensions in an ageing population is met by the accumulation of a capital stock in good time. The transition to a funded system necessarily puts a double fiscal burden on cohorts living during the transition period. In order to accumulate an asset that can be handed over to prospective retirees, they have to forgive part of their current income, but at the same time must still finance the requirements of the expiring paygo system. Nevertheless, the transition to a fully funded system could be pareto-superior, if economic distortions originating from the paygo system, e.g. in the labour market, are eliminated. The resulting dynamic efficiency gains could suffice to compensate the welfare losses of generations living through the transition [cf. Raffelhüschen (1993) and Homburg (1996)].

In the following, we employ generational accounting to derive a partial funding strategy for Social Security in Spain. Following Boll et al (1994) we calculate the immediate once-andfor-all proportionate variation in contribution and/or pension levels that would bring the intertemporal liabilities of the pension insurance down to zero, and thus implies intergenerational balance. Since all adjustments take place immediately, the pension insurance runs surpluses in the first decades, as long as the ratio of contributors and pensioners is comparatively favourable. When the demographic situation worsens, the accumulated assets are employed to partially finance the pension claims of an increased number of retirees.

The first three columns of Table 2 summarise the intergenerational impact of alternative partial funding policies. For a comparison we also report again the findings for the intergenerationally balanced paygo strategies analysed in section 3.4. The immediate and permanent measures required to achieve partial funding are rather severe. This result directly reflects the high implicit liabilities of the pension system in the baseline setting. To accumulate the required assets pension insurance contributions need to be raised from the base year by 43.2 percent, or 6.2 percent of GDP. A 30.1 percent immediate cut in all pensions, which means a reduction in pension spending of 2.8 percent of base year GDP, would fulfil the same task. As a less extreme alternative, we also consider a solution that raises half of the needed funds by contributions, and half by a pension cut. This policy is a simple linear combination of the extreme scenarios: Contributions must be raised by 21.6 percent and pensions cut by 15.5 percent.

The intergenerational impact of the alternative funding strategies varies considerably. Partial funding appears as the preferred strategy of the present young and future generations, as compared to maintenance of the paygo system. The immediate increase of payroll contributions leaves them with a lifetime net contribution of 3,333 thousand pesetas, against a burden of 4,675 thousand pesetas, if the system is continued adjusting contributions annually. An immediate pension cut lets current pensioners participate in the accumulation of temporary Social Security wealth. Consequently, the lifetime tax burden for present young and future generations is even more reduced. A representative newborn is left with 2,328 thousand pesetas, which is again less than in the corresponding scenario with endogenous pensions (2,791 thousand pesetas).

These results notwithstanding the generational accounts do not necessarily allow to argue in favour of partial funding. This is seen from Figure 3 which reports the age-specific deviations from the baseline accounts for all three funding strategies and the paygo scenario with endogenous pensions. It reveals that a policy that continually adjusts pension levels rather than rising contributions once-and-for-all would be preferred by all cohorts younger than 40, at the expense of all older base year cohorts. This result should be approached with some caution however, as it neglects the macroeconomic repercussions of funding strategies. In particular young and future cohorts might benefit from a higher capitalisation of the economy that could result from the accumulation of a capital stock for Social Security. At this stage, the standard generational accounting approach hits a limit that can be overcome only with a general equilibrium approach which lies beyond the scope of this paper.¹⁹

18 Ours is a stylised view on the transition to partial funding. Any immediate adjustment would be difficult to implement. In practice, one would prefer a more gradual transition [cf. Besendorfer et al (1998)], and

In terms of political feasibility, the government might want to opt for a funding strategy that spreads the transitional burdens most evenly among all cohorts. In this respect, Figure 3 suggests that a mixed strategy which leaves equal parts of the adjustment costs to contributors and pensioners must be regarded as comparatively balanced intergenerationally. Less perceptible than the alternatives, this policy could have a higher potential to find support from a majority willing to preserve Social Security insurance, but trying to minimise the individual share in adjustment costs. However, to determine the actual combination of contribution raise and pension cut which could get the strongest backing from Spanish citizens would require a more thorough analysis of political economy.

6. Summary

The intergenerational redistribution impact of the Spanish public pension system after the 1997 Pension Reform Act has been the concern of this paper. Using generational accounts we find that, in order to meet the intertemporal budget constraint of public pension insurance in a rapidly ageing society, the new legal setting would leave future generations with a five times higher lifetime contribution rate than a base year newborn. Stated differently, implicit debt rolled over to future generations amounts to 176 percent of base year GDP. The recent reform effort has apparently done little to improve the long-run sustainability of what remains a generous Social Security system. If a paygo setting is maintained, contribution rates double within the next fifty years, or replacement rates must be cut by one half.

should opt for personalised pension accounts, to put pension wealth out of the reach of greedy politicians.

19 Taking into account macroeconomic repercussions within a generational accounting context, Raffelhüschen and Risa (1997) have demonstrated the possibly ambiguous impact of a funding strategy on individual welfare.

As either development would call the long term political feasibility of the system into question, more vigorous reform measures must be considered. There appears room to improve sustainability while maintaining paygo finance: A further improvement of the tax-benefit linkage, in line with the original intentions of the Toledo Agreement, which gradually reduces average pensions for prospective retirees by 30 percent, would reduce the implicit liabilities of the system by about 60 percentage points. Combined with an increase in retirement age by two years, a more stringent realisation of the Toledo Agreement could bring down pension liabilities even further to about 80 percent of base year GDP.

Nevertheless, reforms within the paygo system are unlikely sufficient to fully restore intergenerational balance. To achieve sustainable pension finances, a complementary transition to a partially funded system could be necessary. Strategies which generate the required funds through either contribution raises or pension cuts distribute transition costs rather unevenly among the present living. Our findings rather support a mixed funding which spreads the funding burden more equally between current contributors and retirees. The latter strategy might provide a useful starting point for the necessary re-thinking of Social Security reform in Spain.

g=0.015, r=0.04 (Figures in thousand pesetas).

Table 1: Intergenerational Impact of Paygo-Reforms

g=0.015, r=0.04 (Figures in thousand pesetas).

Table 2: Intergenerational Impact of Policies with Zero Intertemporal Liabilities

Figure 1: 1996 Social Security Accounts of Present and Future Generations (Baseline, g=0.015 , r=0.04)

Figure 2: Relative Pension and Contribution Change in a Paygo Setting

Figure 3: Strategies with Zero Implicit Liabilities - Deviations from Baseine Accounts

References

1. Alonso, L.M., J. F. Corona and F. Valera (1997) La Armonización Fiscal en la Unión Europea. Cedecs Editorial S.L.
2. Argimón, I. and J.M. González-Páramo (1987) “Traslación e Incidencia de las Cotizaciones Sociales por Niveles de Renta en España: 1980-84”, Documentos de Trabajo de la Fundación FIES, 1.
3. Auerbach, A.J., J. Gokhale and L.J. Kotlikoff (1991) “Generational Accounts: a Meaningful Alternative to Deficit Accounting”, in: D. Bradford (ed.), Tax Policy and the Economy, Vol. 5, Cambridge, MIT Press: 55-110.
4. Auerbach, A.J., J. Gokhale and L.J. Kotlikoff (1992) “Generational Accounting: A New Approach to Understanding the Effects of Fiscal Policy”, Scandinavian Journal of Economics, 94: 303-318.
5. Auerbach, A.J., J. Gokhale and L.J. Kotlikoff (1994) “Generational Accounting: A Meaningful Way to Evaluate Fiscal Policy”, Journal of Economic Perspectives, 8: 73-94.
6. Berenguer, E., B. Raffelhüschen and H. Bonin (1997) “Generational Accounting in Spain: Has Public Sector Grown too Much”, Documents de Treball de la Divisió de Ciències Jurídiques, Econòmiques i Socials. Col.lecció d’Economia. Universitat de Barcelona, E98/30.
7. Berenguer, E., B. Raffelhüschen and H. Bonin (1999) “The Spanish Need for a Broader Tax Base”, in: European Commission (ed.), Generational Accounting in Europe, forth.
8. Besendorfer, D., C. Borgmann and B. Raffelhüschen (1998), “Ein Plädoyer für Intergenerative Ausgewogenheit: Rentenreformvorschläge auf dem Prüfstand, ifo Studien, 44: 209- 231.
9. Boll, S., B. Raffelhüschen and J. Walliser (1994) “Social Security and Intergenerational Redistribution: A Generational Accounting Perspective”, Public Choice, 81: 79-100.
10. Bonin, H., B. Raffelhüschen and J. Walliser (1997) “Can Immigration Alleviate the Demographic Burden – An Assessment with Generational Accounts”, Working Papers in Economics no 7/97, University of Bergen.
11. Bonin, H., B. Raffelhüschen and J. Walliser (1999) “The German Squeeze: Unification and Ageing ”, in: European Commission (ed.), Generational Accounting in Europe, forth.
12. Buiter, W.H. (1997) “Generational Accounts, Aggregate Saving and Intergenerational Redistribution”, Economica, 64: 605-626.
13. CBO – Congressional Budget Office (1999) “Social Security Privatization: Experiences Abroad, CBO Papers, Washington.
14. Crettez, B, K. Feist and B. Raffelhüschen (1999) “Generational Imbalance and Social Insurance Reform in France”, in: European Commission (ed.), Generational Accounting in Europe, forth.
15. Diamond (1994) “Generational Accounts and Generational Balance: An Assessment”, National Tax Journal, 49: 597-607.
16. Escobedo, I. (1991) “Un Análisis Empírico de los Efectos Finales Producidos sobre el Empleo Industrial por el Sistema de Financiación de la Seguridad Social: 1975-1983”, Investigaciones Económicas, 15:169-192.
17. FBBV (1997) Pensiones y Prestaciones por Desempleo, 2ª Edición. Fundación BBV.
18. Fernández-Cordón, J.A. (1996) “Demografía, Actividad y Dependencia en España”, Serie Economía Pública, Fundación BBV.
19. Fernández-Cordón, J.A. (1998) “Proyección de la Población Española 1991-2026. Revisión 1997”, Documento de Trabajo de FEDEA 98-11.
20. Gil, J. and G. López-Casasnovas (1998) “Life-Time Redistribution Effects of the Spanish Public Pension System”, Economics Working Paper 242, Universitat Pompeu Fabra.
21. Gómez-Sala (1993) “Actividad Laboral y Protección Pública a la Vejez: un Análisis Internacional”, Hacienda Pública Española, 126: 51-84.
22. Haveman, R. (1994) “Should Generational Accounts Replace Public Budgets and Deficits?”, Journal of Economic Perspectives, 8: 95-111.
23. Herce, J.A. and J. Alonso (1998) “Los Efectos Económicos de la Ley de Consolidación de la Seguridad Social: Perspectivas Financieras del Sistema de Pensiones tras su Entrada en Vigor”, Documento de Trabajo de FEDEA 98-16.
24. Homburg (1996) “The Efficiency of Unfunded Pension Systems”, Journal of Institutional and Theoretical Economics, 146: 640- 647.
25. IGAE (1997) Cuentas de las Aministraciones Públicas 1996. Intervención General de la Administración del Estado. Ministerio de Economía y Hacienda.
26. INSS (1995) Memoria 1995. Secretaría de Estado de la Seguridad Social, Ministerio de Trabajo y Asuntos Sociales.
27. Instituto de Demografía (1994) Poyección de la Población Española: España 1991-2026. Comunidades Autónomas y Provincias 1991-2006. Madrid.
28. López-García, M.A. (1996) “Consumption and Income as Tax Bases for Social Security”, Public Finance, 1: 54-70.
29. MEC (1995) Memoria de la Administración Tributaria 1995. Ministerio de Economía y Hacienda.
30. Monasterio, C., I. Sánchez and F. Blanco (1996) “Equidad y Estabilidad del Sistema Español de Pensiones”, Serie Economía Pública, Fundación BBV.
31. MTAS (1997) “Anuario de Estadísticas Laborales y Asuntos Sociales”, Ministerio de Trabajo y Asuntos Sociales.
32. MTSS (1995) “La Seguridad Social en el Umbral del Siglo XXI: Estudio Económico Actuarial”, Ministerio de Trabajo y Seguridad Social.
33. Nadal, M. (1994) “Una Nota sobre la Sustitución de Cotizaciones Sociales por IVA”, Cuadernos de Actualidad 4/1994. Instituto de Estudios Fiscales.
34. Patxot, C. (1994) Efectos del Envejecimiento de la Población sobre el Ahorro. Tesis Doctoral. Departamento de Teoría Económica. Facultad de Ciencias Económicas y Empresariales. Universidad de Barcelona.
35. Piñera, M. (1996) Una Propuesta de Reforma del Sistema de Pensiones en España, Círculo de Empresarios.
36. Raffelhüschen, B. (1993) “Funding Social Security Through Pareto-Optimal Conversion Policies”, Journal of Economics, Supplement 7: 105 – 131.
37. Raffelhüschen, B. and A. Risa (1997) “Generational Accounting and Intergenerational Welfare”, Public Choice, 93: 149-163.
38. Raffelhüschen, B. (1999) “Generational Accounting: Method, Data and Limitations”, in: European Commission (ed.), Generational Accounting in Europe, forth.
39. Sinn, H.-W. (1997) “The Value of Children and Immigrants in a Pay-As-You-Go Pension System: A Proposal for a Partial Transition to a Funded System, CES Working Paper 141, University of Munich.
40. Vilches, G. and R. Salas (1994) La Sustitución de Cuotas a la Seguridad Social a Cargo del Empresario por IVA. Instituto de Estudios Fiscales.
41. World Bank (1994) Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth, Washington D. C.