After decades of expansion, reform of the social security system in Sweden has become the focus of attention as a result of the country’s overall budget crisis since the early 1990s. Although the need for pension reform had long been acknowledged, the pressing need to reform other parts of the system was not generally recognized until the adverse budgetary effects of the most recent economic downturn became apparent. Reforms have been inspired greatly by concern over the economic costs of the system as currently structured. The most obvious long-term threat to the system’s viability is the prospective aging of the Swedish population which, as in other OECD countries, will place considerable pressure on public finances in the decades ahead. Equally important are the adverse effects of the expansion of the social welfare system on the economic growth potential of the Swedish economy.1

After decades of expansion, reform of the social security system in Sweden has become the focus of attention as a result of the country’s overall budget crisis since the early 1990s. Although the need for pension reform had long been acknowledged, the pressing need to reform other parts of the system was not generally recognized until the adverse budgetary effects of the most recent economic downturn became apparent. Reforms have been inspired greatly by concern over the economic costs of the system as currently structured. The most obvious long-term threat to the system’s viability is the prospective aging of the Swedish population which, as in other OECD countries, will place considerable pressure on public finances in the decades ahead. Equally important are the adverse effects of the expansion of the social welfare system on the economic growth potential of the Swedish economy.1

This section reviews the development of and prospects for social security in Sweden. It briefly examines the growth of public spending in Sweden in comparison with other countries; analyses the non-pension components of social security in Sweden, noting especially the ways in which they have been or could be reformed; discusses the major reform of the pension system that is currently being introduced; and provides some concluding remarks.

Historical and International Perspectives

Growth of General Government Spending

The past quarter of a century has witnessed a considerable expansion of the public sector throughout the OECD countries, reflected in large part in the persistent growth of general government spending as a share of GDP (Chart 5-1). Having started from an initially higher share of general government spending in GDP than in most other European Union countries, the higher rate of increase of government spending in Sweden during this period resulted in a widened gap between Sweden and the other countries shown. By 1993, public spending had reached 72 percent of GDP in Sweden, compared with 51 percent of GDP in the European Union. Although a small part of the recent increase is attributable to cyclical conditions, most of the past growth of public spending reflects structural factors.

Chart 5-1.
Chart 5-1.

General Government Spending in Sweden and Selected OECD Countries

(In percent of GDP)

Source: Organization for Economic Cooperation and Development.

A widely observed source of spending growth has been the expansion of the welfare state. As can be seen in the central panel of Chart 5-1, the share of transfers in GDP increased generally more rapidly in Sweden than in the European Union as a whole, as well as compared with the other countries shown. By 1993, transfer spending reached 29 percent of GDP in Sweden, or roughly 40 percent greater than the European Union average. Government consumption spending also increased during the period as a whole, remaining significantly above the average level observed in the European Union.

International comparisons are often complicated by differences of definition. Even generally comparable data, such as those shown in Chart 5-1, mask potentially important distinctions between Sweden and the other members of the European Union. For instance, health care services in Sweden are delivered directly through government production, whereas in many other countries public medical care is provided via transfers to private producers. Another important distinction is in the tax treatment of benefits. In contrast to many European Union countries, most transfers and social security benefits in Sweden (80 percent of gross payments) are subject to income taxation.2 Finally, the share of GDP spent on social protection is attributable in part to the higher old-age dependency ratio in Sweden than elsewhere in the European Union.3

Notwithstanding these considerations, the important policy issue is whether the economic efficiency losses associated with the marginal tax rates needed to finance social security and other outlays are too high.4 Although it is often argued that what matters for efficiency is the “net” tax (taxes less transfers), it is important to recognize that most transfers are contingent. To the extent that individuals may be able to affect the amount of transfers they receive without affecting the cost to them, moral hazard can drive up benefit claim rates beyond the levels that might otherwise be observed. As emphasized below, there is increasing evidence that the interplay of social security benefits and taxation in Sweden has resulted in a high degree of benefit dependence.

Social Spending in Sweden and the European Union

The greater share of transfers in Sweden than in other European Union countries reflects the heavier emphasis placed in Sweden on social protection. Social protection is also effected through the direct provision of services such as health care. Although comparative data are limited, some are available from EUROSTAT (1994). These are reported in Table 5-1, which shows the percentage of spending on social protection in Sweden, the European Union, and several other countries.5 Although social spending in Sweden has been affected since 1989 by the sharp economic downturn, the structural excess of Swedish social spending relative to that in the other European Union countries is evident. In fact, social spending in Sweden is higher than in every country in each year shown.6 Table 5-2 provides a crosscountry comparison of social spending by category for 1991 (the most recent year for which comparable data are available). Although interpretation is problematic because of difficulties of classification, it is striking that social spending in Sweden in practically all categories substantially exceeds the average level in the European Union. The substantial gap in sickness-related spending has been reduced since 1991 as a result of the measures that have been adopted.7 However, as the statutory provisions of sick pay remain seemingly favorable in Sweden (see below), the gap is unlikely to have been closed.

Table 5-1.

Selected European Union Countries: Expenditure on Social Protection1

(In percent of GDP)

article image
Source: EUROSTAT (1994).

Includes health care, sickness and invalidity benefits, old-age pensions, maternity benefits, family allowances, unemployment compensation, and housing assistance.

Table 5-2.

Selected European Union Countries: Social Protection Spending by Category, 1991

(In percent of GOP)

article image
Source: EUROSTAT (1994).

Includes old-age and survivors’ pensions, as well as invalidity payments.

Included in invalidity, which is in turn included in pensions.

Social Security and Benefit Dependence


Sweden has one of the most elaborate, comprehensive, and complex social security systems in the world. The overall composition of social security benefits is summarized in Table 5-3. The bulk of these benefits is financed by payroll contributions and interest earnings on accumulated surpluses (Table 5-4). In addition, transfers from the central government are provided. Currently, the payroll tax is equal to 33 percent of wages; self-employed persons contribute about 30 percent of active income, and roughly 18 percent of passive income. Included in the Government’s recently announced savings package is a 2 percentage point increase in the employee’s contribution in 1995 (1 percentage point for sickness and 1 percentage point for unemployment insurance), with further increases of 1 percentage point in each of the years 1996-98.

Table 5-3.

Main Benefits of Social Insurance

article image
Source: National Social Insurance Board, Social Insurance in Sweden: Fact Sheets on Sweden, December 1993.
Table 5-4.

Financial Operations of the Principal Social Insurance Programs, 19921

article image
Source: National Social Insurance Board, Social Insurance Statistics: Facts 1993.

Excluding training allowance, housing allowance, special allowance for single parents, car allowance, voluntary pension, compensations for military conscripts, occupational health care, disease carrier compensation, adoption allowance, special allowance for certain adopted children, insurance for owners of small firms, and so on.

Including rehabilitation allowance and cash benefit for closely related persons.

Including cash labor market assistance.

Profit tax of SKr 600 million has been deducted and transferred to the owners of small firms fund.

Refers to persons liable for the maintenance.

Despite the apparent contributory basis of the social insurance system, two aspects of the system’s financing suggest that the payroll deduction acts more like a tax than an insurance premium. First, the bulk of the contribution is paid by the employer, and therefore the link between benefits and contributions is weak, blurring the cost to insured persons. Second, while the payroll tax is levied on the whole of the wage bill, pension benefits are limited to a maximum of 7.5 base amounts (basbelopp).8 Contributions above this ceiling are purely taxes.

A large part of the growth in spending on social security programs since 1970 is attributable to benefit enhancements and to rising take-up rates (Chart 5-2). As a consequence, social insurance benefits and other transfers have become important components of disposable income for a large proportion of households. This is illustrated vividly in Chart 5-3. The chart depicts the incomes of four family types relative to the social assistance threshold, which is the minimum income prescribed by the National Board of Health and Welfare as being necessary to achieve a socially acceptable standard of living in each community.9 As the diagrams demonstrate, income transfers and social security benefits are important sources of income for many households whose market incomes are inadequate to reach the social assistance threshold. As many as 30 percent of couples with two children had market income below the social assistance threshold in 1991; among couples with three children, the percentage increases to over 50 percent. Also noticeable is the fact that a large proportion of households with children have incomes that do not exceed greatly the social assistance threshold, and are thus vulnerable to falling below the threshold as a result of even small losses of income.10

Chart 5-2.
Chart 5-2.

Transfers to Households

(In percent of GDP)

Source: Ministry of Finance.
Chart 5-3.
Chart 5-3.

Income as a Percent of Social Assistance Threshold, 1991

Source: Expert Group on Public Finance (1994).

The social assistance allowance is seen by many analysts as playing a crucial, if indirect, role in raising outlays on other social benefits.11 In effect, the social assistance threshold sets a floor below which other forms of compensation—including combined social security benefits—cannot fall. The Expert Group on Public Finance (1994) reports reservation wages that result from the interplay of the social assistance allowance and the tax system.12 These are shown in Table 5-5. As the Expert Group has emphasized, the higher the social assistance norm, the greater must be other forms of income and social insurance benefits in order to prevent households from turning to social assistance. To the extent that the social assistance norm plays so crucial a role in the de-termination of the willingness to work, careful consideration should be given to reducing its level, while of course preserving the safety net function it is intended to serve.

Table 5-5.

Before-Tax Market Income Required to Exceed Social Assistance Threshold, 1994 Rules

(In kronor a month)

article image
Source: Expert Group on Public Finance (1994).Note: The minimum requirement before tax—without benefits column shows the earnings requirement if no child allowance, housing benefit, or maintenance advance (calculated as 66 percent of the advance) are paid out. This earnings requirement is further increased by the fact that income from work, but not in the form of benefits, reduces the rebate on day-care charges. The earnings requirement for couples applies to each adult, on condition that both have equally high incomes. Trade union dues, housing benefit, and day-care charges are indicated for the income at which the household ceases to be entitled to social assistance.

Nonpension Benefits

Nonpension social security benefits in Sweden can be categorized either as those relating to sickness, work injury, and disability or those relating to family allowances.

Sickness Benefits

After pensions, sickness-related benefits are the most important component of social security in Sweden. During the past several decades, program enhancements have led to substantially increased total sickness-related spending. Four income replacement programs account for the majority of sickness related expenditures of the social insurance system: sick pay; work injury insurance; rehabilitation payments; and disability pensions.13 During the second half of the 1980s, the incapacity rate, defined as the total number of days for which income compensation is paid under all of these programs relative to the number of insured persons, increased dramatically.14 In order to reverse this trend, sick pay and work injury compensation have been reviewed intensively, and significant reforms have been adopted.

Sick pay. In the early stages of the social security system, Sweden’s sick pay provisions were less generous than those in other OECD countries.15 Sickness benefits were paid at a low and declining replacement rate, for up to a maximum of two years, and only after an initial three-day waiting period. However, subsequent program changes dramatically altered this situation. The limit on the duration of benefit payments was eliminated in 1963, followed by the reduction of the waiting period to two days in 1967, before its total elimination in 1987. In 1974, sick pay benefits were raised, bringing the replacement rate for most workers to 90 percent. As illustrated in Chart 5-4, these changes contributed to the maintenance of a strikingly high number of compensated days of sick leave throughout the 1970s and 1980s.

Chart 5-4.
Chart 5-4.

Sick Leave and Disability

Source: Ministry of Finance.

Beginning in March 1991, the 90 percent compensation rate for sickness benefit was paid only after three months of absence. During the first three days only 65 percent was paid, with 80 percent compensation from the fourth to the ninetieth day. In January 1992, the cost of the first 14 days of sick pay was shifted to employers, and a rehabilitation allowance was introduced. The objective of this policy was to strengthen the incentive for employers to monitor more closely the benefit claims of their workers. More recently, a one-day waiting period was reintroduced in April 1993, and the replacement rate has been reduced further, to 65 percent for the second and third days, 80 percent for the period after the three hundred and sixty-fifth day, and 70 percent thereafter.

The recent changes to the sick pay system were clearly warranted in the face of mounting outlays, some of which could be attributed to abuse and lack of self-policing by employers due to the problem of asymmetric information among the interested parties. Moreover, it was more widely recognized that sick pay provisions can reduce labor supply.16 However, the changes made still leave the Swedish sick pay schemes more generous than those in most countries in the European Union, despite the fact that in many countries employers are required to provide sickness benefits which, in turn, increases labor costs. First, a three-day waiting period before compensation is paid is common in the European Union. Second, the period during which the employer is required to provide compensation is often considerably longer: 30 days in Belgium and six weeks in Germany, for instance. Third, no member of the European Union allows compensation to be received for an unlimited duration as in Sweden, although, unlike in Sweden, some countries allow automatic eligibility for a disability pension when sickness benefits are extended.17

The relative generosity of the Swedish sick pay provisions is further underlined by a cross-country comparison using an effective replacement rate; that is, the percentage of pre-episode after-tax income that is replaced for typical workers. Hansen (1994) attempts to make such comparisons for five OECD countries, using as a common beneficiary the OECD Average Production Worker (APW). Based on benefit rules applicable in 1992, for one week of illness, an APW would have received a replacement rate of 81 percent in Sweden, 63 percent in Denmark, 100 percent in Germany, 42 percent in the Netherlands, and 8 percent in the United Kingdom.18

Bolder reforms to the Swedish sick pay system than have been adopted thus far could include extending the waiting period to three days, as well as lengthening the period of coverage for which employers are responsible. An alternative could be to provide only catastrophic coverage through the national social insurance system and leave to workers and employers the responsibility for insuring against normal lesser levels of risk.

Work injury. Compensation for income lost owing to illnesses and injuries in connection with employment is provided by the Work Injury Insurance System. As in the case of nonwork related sick pay, the workman’s compensation system experienced considerable growth in outlays during much of the 1980s because of generous benefits and increased claim rates. In 1980, there were 190,000 reported industrial injuries; by 1988 the number had increased to 260,000. Concomitantly, outlays on benefits rose from 0.1 percent of GDP in 1970 to 0.8 percent of GDP in 1992.

The rapidly increasing cost of the work injury program has made it a target of reform. Until January 1992, benefits under the Work Injury Insurance System were equivalent to those provided under the sickness insurance system during the first 90 days, after which an annuity was paid if the illness or injury was considered work related. Beginning in January 1992, the period was extended to 180 days, and statutory eligibility was tightened the following year by raising the qualifying standard from a “probability” to a “high probability” of a causal link between the work environment and an injury or illness. Notwithstanding these changes, however, expenditures have remained high, at 0.7 percent GDP in 1993.

A number of aspects of the work injury system remain problematic, and warrant reform. For instance, as suggested by Lindbeck and others (1994), as opposed to general sickness insurance, compensation for work-related injuries could be restricted to a specific list of injuries rather than for diffuse symptoms. Also, with a view to reducing the moral hazard that the existing system creates, a closer linkage of the premium to costs could be achieved by differentiating contribution rates on the basis of each enterprise’s claim experience. Finally, a declining replacement rate over the course of an illness episode, combined with rehabilitation, could help to provide an incentive to return to work in suitable types of employment.

Disability benefits. When a person’s work capacity has been durably reduced by at least one quarter, the worker is entitled to a temporary or permanent disability pension. The pensions are payable at full or partial rates depending on the degree of incapacity. Most recipients receive a full pension. The pension is paid at the same rate as the standard old-age pension. As in many countries, outlays on disability pensions have risen continuously during the past 20 years, reflecting in large part the increasing number of recipients. In 1970, 188,000 persons or 2.3 percent of the population received a disability pension. In 1993, the number of recipients had increased to an estimated 402,000, or 4.6 percent of the population (Chart 5-4). The real value of the disability pension has also increased greatly over time. In reflection of these developments, there has been an increase in the weight of disability pension outlays, from 1.2 percent of GDP in 1975 to 2.1 percent of GDP in 1993.

An important factor underlying the growth of the beneficiary population has been the loosening of eligibility criteria. In particular, while eligibility was originally limited to health-related reasons, rules were changed in the early 1970s extending eligibility to labor market considerations. Unemployed persons close to retirement age could qualify for an early “disability” pension on nonmedical grounds. These opportunities were closed off only in 1991. Notwithstanding this policy change, however, there has been a significant increase in the number of new claimants of disability pensions since then. Whereas the number of new claimants ranged between 45,000 and 52,000 a year during the period before 1991, it increased to over 58,000 in 1992 and to 62,000 in 1993; the majority of new claimants are in the 30-59 year age group.19

There is a clear need to tighten eligibility for disability pensions among those persons who have not yet reached formal retirement age. In a similar fashion, as suggested above in the case of work injury insurance, the incentives to return to gainful employment—albeit perhaps in a job that permits reduced work capacity—could be sharpened by reducing the replacement rate over the course of a number of months during which rehabilitation would have to be provided.

Family Benefits

Family benefits in Sweden are among the most elaborate and the most generous in the world. These include pregnancy benefits, parental insurance, temporary parental leave, child allowances, and maintenance advances. Pregnancy benefits provide for income replacement if an expectant mother is unable to continue normal work. The benefit is paid at the same rate as sickness benefits, for up to 50 days during the last two months of pregnancy.

Parental insurance benefits. These benefits consist of income compensation for up to 450 days, until a child reaches eight years of age. The allowed days can be split between the parents as they wish. During the first 360 days, the benefit equals approximately 90 percent of the parent’s normal income. During the final 90 days, a fixed amount of SKr 60 a day is paid. In addition, the system provides for temporary parental benefits, ensuring compensation for up to 60 (occasionally 120) days for each child each year in the event a parent is required to stay home to care for a sick child. Income is replaced at a rate of 80 percent during the first 14 days, and increases to 90 percent for the remaining qualifying days.20

While a few countries provide some leave for parenting, in no country are the benefits as generous as in Sweden.21 In turn, parental insurance has become one of the most cherished transfers in Sweden, and is thought by some to have had a positive impact on the birthrate, while also contributing to a higher rate of female labor force participation rate. At the same time, evidence also suggests that transfers to families have reduced male labor force participation.22

Outlays on parental leave have increased continuously, rising from 0.3 percent of GDP in 1970 to 1.3 percent of GDP in 1993, while the average number of days taken for parental leave approximately tripled over the same period.

Child allowances. All families with children under 16 domiciled in Sweden receive relatively generous child allowances (Table 5-6). Extended allowances are provided to children over age 16 if attending compulsory school, and persons with three or more children receive additional allowances. In 1992, the last year for which detailed data are available, 1.7 million child allowances were provided at a cost of SKr 20.2 billion, which averaged SKr 11,800 per child.23

Table 5-6.

Selected Countries: Increase in Disposable Income Owing to Family Allowances, 1992

(OECD Average Production Worker)

article image
Source: Hansen (1994).

Outlays on child allowances could be reduced in several ways, depending on the objectives of the allowances. For instance, if the allowance were to be designed to operate as an anti-poverty measure, it could be means-tested, as suggested by both Lindbeck and others (1994) and the Expert Group on Public Finance (1994). Alternatively, if the objective were to increase the average number of children per family, the allowances could be restricted to families with three or more children.24

Budgetary Impacts of Reforms

Estimating the budgetary savings from changes in any particular social insurance program is enormously complicated in Sweden due to the taxation of benefits and the interdependence of certain benefits. Quantifying potential budgetary savings from reforming the social security system requires knowing, among other things, the number of recipients of each benefit at each income level and the size and composition of each family. For this reason, in analyzing possible budget savings from reforming the social security system, the Lindbeck Commission (Lindbeck and others (1994)) relied on the household tax-transfer model of the Ministry of Finance. In order to provide a sense of the magnitude of the budgetary savings that could be achieved from various measures, some of which were noted above, the Lindbeck Commission’s estimates are reported in Table 5-7. It is interesting to note that while the reforms could be expected to result in substantial efficiency gains, the net budgetary savings from these proposed reforms would be relatively small in view of the tax revenue losses that would accompany reductions in transfer payments. The total net saving from combined measures even under the Lindbeck Commission’s most radical savings package would amount to around 2 percentage points of GDP.

Table 5-7.

Examples of Savings in the Income Transfer Systems

(In billions of 1993 kronor)

article image
Source: Ministry of Finance, in Lindbeck and others (1994).

Public Pensions

The need to reform the public pension system has been recognized for some time in Sweden. The main impetus for reform has been the concern that the existing system would place considerable pressure on the public finances in the years ahead, especially as a result of the aging of the Swedish population. As elsewhere in the OECD, the baby boom after World War II is expected to result in a substantial, if temporary, increase in the old-age dependency ratio during the first half of the next century, although the projected increase is lower than in Germany or Italy (Chart 5-5). While the approximately 80 percent labor force participation rate for women in Sweden offsets the effects of some of this aging, an expected persistent increase in the number of disability pensioners would operate in the opposite direction.

Chart 5-5.
Chart 5-5.

Old-Age Dependency Rates in Selected Industrial Countries

(Percent of population 65 and older)

Source: World Bank.

Even abstracting from the aging of the population, there are further concerns about the sustainability of the existing system stemming from the lack of linkage of pension costs to the capacity of the economy to pay pension benefits. The lack of actuarial fairness in the system is yet another reason for reforming the system.25 After a decade of analysis and debate, the principles underpinning a far-reaching reform of the public pension system were approved by parliament in June 1994. The reformed system will be phased in over the next several decades beginning in 1995.

The Current System

Widespread support for a universal compulsory public pension system emerged in Sweden during the 1950s. It was not until 1960, however, that the scheme received legislative approval. The system introduced at that time was a two-part defined-benefit scheme financed on a pay-as-you-go basis from employer-paid contributions. The first part, the basic pension (folkpension), is a flat-rate pension paid to all persons who have resided in Sweden for at least 40 years or on whose behalf contributions have been paid for 30 years. The full pension is payable at age 65, but it is reduced by 1/30th for each year of short-fall relative to the 30-year required contribution period. The full old-age basic benefit is equal to 96 percent of the reduced base amount (basbelopp) for a single person, and 157 percent for qualifying retired couples.26 The principal aim of the basic pension is to ensure an income in old age that is independent of pre-retirement income, but not of residence and gainful employment. With contributions paid on all wages, the basic pension provides an element of income redistribution. As the minimum benefit level has been increased considerably since the inception of the program, the redistributional dimension has been also increased.

The second pillar of the current system is the earnings-related supplementary pension, or ATP pension (Allmdntilldggspensionering). Pension credits are earned between the ages of 16 and 64, and at least three years of qualifying income are needed.27 A full pension is payable at age 65 after 30 years of contributions paid by the employer at the current rate of 13 percent on all wages.28 The pension is equal to 60 percent of the average of the highest 15 years’ real wages up to 7.5 base amounts. Since contributions are paid on earnings above the ceiling, the earnings-related pension also provides for some income redistribution.

Pension benefits are comparatively generous in Sweden. Hansen (1994) calculated that in 1992 the net replacement rate of the old-age pension was around 69 percent in Sweden, compared to 59 percent in Denmark, 50 percent in the Netherlands, and 47 percent in the United Kingdom. Only the German pension scheme provided a more generous benefit, calculated to be 73 percent.

Problems with Existing System

Several aspects of the existing system have been especially problematic. An inherent conflict exists in the existing system between, on the one hand, an indexing policy that would ensure that the average replacement rate (that is, the ratio of the average pension to the average wage) remains relatively constant over time and, on the other hand, an indexation policy that makes the system responsive to the capacity of the economy to provide the pensions.29 Moreover, real wage indexation combined with the highest 15 years’ of earnings, makes the long-run sustainability of the system very dependent on a constant high rate of real wage growth (Table 5-8).30 Another problem with the existing system is that the pension scheme does not take into account rising life expectancy. For the same lifetime contributions, retirees receive increasing lifetime benefits due to rising longevity.31

Table 5-8.

Old-Age Pension Outlays Under Existing System: Alternative Indexation Policies

(In percent of contribution base)

article image
Source: Ministry of Health and Social Affairs.

The Pension Reform

Main Elements of the Reform

In June 1994 parliament approved a bipartisan proposal for long-run pension reform.32 The proposal calls for gradually replacing the basic, ATP, and supplementary pensions with a two-part old-age pension scheme that combines pay-as-you-go with some funding.33 The principal component will consist of a pension that is to be based on lifetime contributions and financed on a pay-as-you-go basis. The lesser element of the system will be an individual capitalization account (a so-called premium reserve account) that is to be fully funded. For future retirees whose lifetime contributions are too low to qualify for an adequate pension, a guaranteed pension will be provided, financed by general revenues.

Of the current total payroll tax, 18.5 percentage points are to be earmarked for the two schemes. The proposal calls for half of this contribution to be paid by employees, but the final split remains to be decided. Only the employer’s contributions would continue to be made on covered incomes above 7.5 base amounts, but these would be earmarked for the central government budget. Thus, there would remain a degree of taxation in the system. Of the 18.5 percentage points, 16.5 percentage points would be earmarked for the principal component, while the remaining 2 percentage points are to be invested in the individual capitalization accounts, which may be privately or publicly managed. The pension benefit from the latter will depend on both the individual’s contributions and the rate of return on investments. The retirement age will be flexible from age 61, although the pension would be reduced or raised for early or delayed retirement, respectively. A full pension will be payable at age 65.

Contributions derived from the 16.5 percent payroll tax will be used to pay ongoing pension obligations, both from the old system during the transition period, and those arising from the new scheme. However, each insured person’s contributions will be registered in an individual pension account at the National Social Insurance Board to form a notional capitalization account. Throughout an individual’s working life, the balance in the notional account is to be adjusted annually using an index that takes into account the change in prices and real wage growth in the economy. This can be thought of as a notional “interest” payment. Upon retirement, the initial pension will be determined by (1) the average life expectancy at retirement age in that year;34 (2) the amount that has been accumulated in the account, including notional interest payments; and (3) an assumption about the rate at which real wages can be expected to grow during the person’s retirement (the so-called norm).35 The norm is, in effect, an assumed real rate of return. A decision has not been reached on what norm to assume, but the Working Group on Pensions recommended a rate of 1.5 percent.

After retirement, the pension will be adjusted annually using an economic adjustment index. The index in each year is to be equal to the divergence of real wage growth from the assumed norm. Pensions in each year will be increased by the inflation rate plus the economic adjustment index. Thus, for real wage growth in excess of the assumed norm, full protection against inflation is assured, plus a sharing of the gains from productivity increases. For real wage growth that is less than the norm, there is only partial inflation protection, and for any positive norm, the replacement rate would decline.36 However, the higher the norm, the higher the initial level of the pension at the outset of retirement.

Impact on Public Finances

As the system will be phased in over a 20-year period, there will be relatively little impact on the public finances in the very near term. However, although pension expenditures under the old system will continue to account for over half of total pension outlays until approximately 2015, the use of the indexation scheme described above for all pensions beginning in the year 2000 will significantly affect the system’s costs from 2000 onward.

The impact of the reform on pension expenditures relative to the old system depends mainly on which indexation scheme—prices or wages—would have been used systematically in the future as regards pensions in the old system. First, as noted earlier, if the ceiling of 7.5 base amounts were to be price indexed during the next several decades, the average replacement rate would decline over time and, hence, so would outlays. If the ceiling were to be indexed by wages, the average replacement rate would be maintained, resulting in a higher level of aggregate pension spending than otherwise. Second, the relative costs of the old and reformed schemes depend on the level of the assumed real growth norm, which affects the adjustments of pensions in the future. The higher the norm, the lower the adjustment of pensions paid under the old system in the future and, hence, the greater the savings from the reform.37

With the norm set at 1.5 percent, if real wage growth were also to average 1.5 percent, the ratio of total pension outlays to the contribution base would be significantly lower than under the existing pension system until the middle of the next century (Chart 5-6). The strong impact on the pension burden of the level of the assumed norm is evident in Chart 5-6, in which actual real growth is assumed to be 2 percent a year, while the norm is assumed to re-main unchanged at 1.5 percent.38 Under such circumstances, the potential savings from the new scheme over the existing scheme would be relatively limited. More generally, any excess of real wage growth over the norm reduces the burden of providing the pension.

Chart 5-6.
Chart 5-6.

Old-Age Pension Costs Under Existing and Reformed Schemes

Source: Ministry of Health and Social Affairs.

Concluding Remarks

Social security benefits account for a substantial share of income in Sweden. There is a tendency, however, for the comparison of social security spending in Sweden to ignore the fact that most social security benefits are subject to income taxation. This tends to overstate the generosity of benefits, which are retained on a net-of-tax basis. At the same time, the scope and level of transfers and exhaustive public spending in Sweden continue to require very high levels of gross taxes. Economists have recognized increasingly that the main element of high marginal tax rates consists of the distortionary effects of the overall tax system. Reforms introduced recently can be expected to help reduce such adverse effects. More reforms are, however, urgently needed.

The pension reform that is currently being introduced in Sweden addresses several of the problems that plague the existing scheme. Among the key changes that the new scheme introduces is linking future pensions to lifetime contributions and to the rate of growth of real wages. By providing less than full pass-through of productivity gains, the new scheme can be expected to make the burden on future workers lower than otherwise.


Some observers, such as Lindbeck and others (1994) and Hansson and Henrekson (1994), consider that the generosity and complexity of the transfer system have contributed significantly to the weakening of economic incentives, which in turn has led to Sweden’s declining rank among OECD countries in terms of per capita GDP.


On a net-of-tax basis, the growth of transfer payments in Sweden since 1970 has been much slower than on a gross-of-tax basis, except for the period after 1990. After-tax transfer payments as a percent of GDP at factor prices remained between 10 percent and 13 percent during the period 1970-90, rising subsequently to close to 18 percent (Expert Group on Public Finance (1994)). This contrasts with the generally constant growth of gross transfers, visible in Chart 5-1. Moreover, adjusted for their taxation, transfers in Sweden are roughly half as large as government consumption as a share of GDP.


In 1990, 17.5 percent of Sweden’s population was older than 65, compared with the OECD average of 13.5 percent. The Organization for Cooperation and Development (1994a) reports that social expenditures could be lowered by 3.8 percent of GDP if Sweden had the same elderly dependency rate as the average for OECD countries.


Sandmo (1991) emphasizes the fact that the “willingness” to pay for transfers has to be adjusted to take into account the fact that real world taxes are not lump-sum and are, therefore, distortionary.


”Social protection” as defined by EUROSTAT includes spending on health care, sickness and disability benefits, old-age pensions, maternity benefits, family allowances, unemployment compensation, and housing assistance.


Again, it is important to recall that the gap on a net-of-tax basis would be smaller than appears in the table.


The substantial difference between sickness outlays in Sweden and elsewhere is partly attributable to Sweden’s public provision of medical services. More than half of the share of GDP shown in Table 5-2 is for “public health.”


The basbelopp is the basic accounting unit of public pensions and much of the rest of the social insurance system. The base amount is typically adjusted annually for changes in consumer prices, and currently equals SKr 35,200.


The sample of households, taken from the 1991 Income Distribution Survey, excludes pensioners and dependents under age 18, as well as a few other groups such as military personnel and families with negative taxable income.


The Expert Group on Public Finance (1994) also reports that empirical analysis supports the view that there is a significantly higher probability of benefit dependence—which the authors de-fine as a situation in which social insurance and other transfers account for at least 66 percent of annual spendable income—among households that receive labor market support or comparatively large amounts of sickness insurance in an early year in working life. It is important to emphasize, however, that the eligibility criteria for social assistance differ from those for social benefits in general. For instance, even if actual income falls below the social assistance threshold, a person’s potential income (for example, through job search) may disqualify that person.


Marginal income tax rates range from zero percent of taxable income on incomes up to SKr 8,800 a year to 51 percent on annual taxable incomes above SKr 202,000. The budgetary savings package announced in November 1994 will increase marginal tax rates by 5 percentage points from January 1995 through 1998. The average marginal tax rate will thus increase from 31 percent to 36 percent, while the top marginal rate will increase to 56 percent. There are no deductions for children and few other exclusions.


A number of other benefits could also be classified as being health related, such as pregnancy cash benefit and temporary parental leave. These are discussed below in the context of family-related benefits.


The program’s parameters have complex effects. A higher re-placement rate would tend to increase the demand by workers for the benefit. Empirical evidence tends to confirm that there is a fairly strong relation between the replacement rate and the rate of absenteeism. See, for instance, Burtless (1987) and Gustafsson and Klevmarken (1993) for good surveys of the evidence. The latter study finds that a general decrease of the compensation rate by 10 percentage points could cause a decrease in the absence rate by 1.4—1.6 percentage points. While a waiting period raises the costs of initiating a benefit period, it may also lengthen the benefit period by raising the costs of a premature return to work following illness.


The duration of benefit payment varies between 26 weeks (Italy and Spain) to 1,095 days (Portugal); in most European Union countries, sick pay is provided up to 52 weeks only.


With the changes that have been adopted since 1992, the effective replacement rate in Sweden has been reduced. However, for episodes that exceed the 14-day period during which employers are responsible, the sick pay provisions in Swedish social insurance remain attractive. This may be one reason why the number of days compensated by social insurance remains high, despite having fallen somewhat. Whereas the number of compensated days averaged around 25 during the period 1988-91, it fell to around 17.5 in 1992. The introduction of the one-day waiting period in 1993 appears to have reduced this only slightly to 17.2 days. The sustained high claim rate may also be attributable to the increased level of unemployment in Sweden since the early 1990s.


There are several possible reasons for this. First, the increase may reflect a substitution of disability pensions for the now less-generous sickness insurance benefits for long spells of illness. Second, the poor prevailing employment conditions may also ex-plain an important part of the surge.


It would seem possible for this system to be open to considerable abuse insofar as eligibility is apparently loosely controlled, parents having to provide little if any verification of a child’s illness. In view of the significant cutbacks in state-provided sick pay, one might conjecture that some substitution of temporary parental leave may be made for the now less attractive sick pay.


For instance, in Belgium, parental leave is allowed for certain reasons, but on condition that an unemployed person be hired as a replacement. In Germany, where parental leave was introduced in 1986, the benefit is payable only until the child reaches two years of age. In France, paid parental leave is restricted to families with at least three children, one of whom must be under age three. Parental leave is not provided in the Netherlands or in the United Kingdom.


Data obtained from the Swedish National Accounts and National Social Insurance Board.


In France, for instance, allowances are paid for second and subsequent children, but the amounts for each child are higher when there are three or more children in the family.


An actuarially fair pension system is one in which the present value of the benefits expected by a beneficiary equals the present value of the beneficiary’s lifetime contributions paid into the scheme.


Beginning in 1993, pensions are calculated on the basis of a base amount reduced by 2 percent. As noted earlier, for disabled persons otherwise satisfying the qualifying conditions, a temporary or permanent disability pension is available under the basic pension scheme. The scheme also provides survivor’s and children’s pensions. Supplements to the basic pension are provided to some retirees with little or no earnings-related pension, as are housing and child care allowances under certain circumstances.


Qualifying income refers to wages and social insurance benefits (sick pay, parental leave benefits, unemployment compensation, and so on) between 1 and 7.5 base amounts. Years spent caring for a child under age three are also creditable.


The full pension was originally payable only upon reaching age 67, but the retirement age was reduced to age 65 in 1976. Reduced and augmented pensions are payable, respectively, for early and delayed retirement between age 60 and 70. As in the case of the basic pension, disability and survivor’s pensions are also available. “Partial pensions” are also provided to persons between the age of 60 and 65 who reduce their time spent in gainful employment.


If the ceiling of 7.5 base amounts were to continue to be indexed by inflation only, an increasing proportion of workers over time would progressively have wages above this amount. In turn, a smaller percentage of pre-retirement income would be included in the pension base, resulting therefore in a lower pension. Abstracting from other factors, this effect would ease the burden of population aging, by transforming the system from an earnings-related scheme to a flat benefit. One study (Hagemann and Nicoletti (1989)) simulated that the average replacement rate (average pensions divided by average covered wage) of the combined basic and ATP old-age pension could fall from approximately 54 percent in 1990 to 29 percent in 2050 if the ceiling were indexed by prices only.


The sensitivity of pension costs to the rate of real growth arises from the fact that the pension benefit is based on a pensioner’s highest 15 years’ earnings. When the real rate of growth falls considerably, as has been the case in recent years, a higher share of current wages is needed to pay current pensions.


Although the previous government succeeded in obtaining agreement to raise the age of retirement to 67, implementation was postponed.


The new system, as well as its phase-in during the next 20 years, is extremely complex. For a more detailed description, see Ministry of Health and Social Affairs (1994).


Disability pensions are to be reformed as well, but under a separate program that is currently being designed.


The greater (lower) the life expectancy, the lower (higher) will be initial pension level, reflecting thereby the greater number of years during which the pension is to be received.


The criteria for setting the norm are obvious.

This can be seen as follows. Let Pt refer to the pension in year t, it refer to the rate of inflation in year t, Wt refer to the nominal wage in year t, rt refer to the rate of real wage growth, and nt refer to the norm. The ratio of the average pension to the average wage is given by
When nt > 0, the numerator will grow by less than the denominator. Hence, the replacement rate will fall.

Applying the same indexation scheme to both systems means that pensions paid in the old scheme (that is, ATP pensions) will be adjusted annually by the difference between actual real wage growth and the norm. Thus, if the norm is set at 1 ½ percent and real wages in any future period grow by 1 percent, ATP pensions would be reduced by ½ percent.


Whether the norm could be left unchanged in a context of continuously higher real wage growth would depend on the same considerations as affect the choice of indexation in the old scheme. These considerations include both questions of equity—the sharing of output gains between retirees and workers—and efficiency.

Cited By

  • Abraham, K.G., and S. Houseman, Does Employment Protection Inhibit Labor Market Flexibility? Lessons from Germany, France and Belgium,” W.E. Upjohn Institute Working Paper 93-16 (Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, May 1993).

    • Search Google Scholar
    • Export Citation
  • Alexius, Annika, in Monetary Policy Indicators (Stockholm: Sveriges Rikshauk, June 1993), pp. 6576.

  • Auerbach, Alan J., Jagadeesh Gokhale, and Laurence J. Kotlikoff, Generational Accounts: A Meaningful Alternative to Deficit Accountingin Tax Policy and the Economy, Vol. 5, David Bradford (Cambridge, Massachusetts: MIT Press, 1991).

    • Search Google Scholar
    • Export Citation
  • Auerbach, Alan J., Jagadeesh Gokhale, and Laurence J. Kotlikoff, Generational Accounting: A New Approach to Understanding the Effects of Fiscal Policy on Saving,” Scandinavian Journal of Economics. Vol, 94, No. 2 (1992), pp. 30318.

    • Search Google Scholar
    • Export Citation
  • Auerbach, Alan J., Jagadeesh Gokhale, and Laurence J. Kotlikoff, Generational Accounting: A Meaningful Way to Evaluate Fiscal Policy,” Journal of Economic Perspectives, Vol. 8 (Winter 1994), pp. 7394.

    • Search Google Scholar
    • Export Citation
  • Berg, Lennart,Household Savings and Debts: The Experience of the Nordic Countries.Oxford Review of Economic Policy, Vol. 10, No. 2 (Summer 1994), pp. 4253.

    • Search Google Scholar
    • Export Citation
  • Burtless, Gary,Taxes, Transfers, and Swedish Labor Supply,” in The Swedish Economy, Barry P. Bosworth Alice M. Rivlin (Washington: Brookings Institution, 1987).

    • Search Google Scholar
    • Export Citation
  • Calmfors, Lars (1993a), Centralization of Wage Bargaining and Macroeconomic Performance: A Survey, Seminar Paper No. 536 (Stockholm: Institute for International Economic Studies, 1993).

    • Search Google Scholar
    • Export Citation
  • Calmfors, Lars (1993b), “Lessons from the Macroeconomic Experience of Sweden.European Journal of Political Economy, Vol. 9 (March 1993), pp. 2572.

    • Search Google Scholar
    • Export Citation
  • Catāo, Luis, and Ramana Ramaswamy, Recession and Recovery in the United Kingdom in the 1990s: A Vector Autoregression Approach”, IMF Working Paper No. 95/40 (Washington: International Monetary Fund, April 1995).

    • Search Google Scholar
    • Export Citation
  • Eliasson, Gunnar, and others, The Knowledge Based Information Economy (Stockholm: Industrial Institute for Economic and Social Research, 1990).

    • Search Google Scholar
    • Export Citation
  • EUROSTAT (Statistical Office of the European Communities), Social Protection Expenditure and Receipts, 1980-1992 (Luxembourg, 1994).

  • Expert Group on Public Finance (ESO), Social Security in Sweden—How to Reform the System (Stockholm: Ministry of Finance, 1994).

  • Franco, D., J. Gokhale, L. Guiso, L.J. Kotlikoff, and N. Sailor, General Accounting: The Case of Italy.Temi di Discussione 171 (Rome: Bank of Italy, 1992).

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, and Mohsin S. Khan, Income and Price Effects in Foreign Trade,” Handbook of international Economics, Vol. 2, ed. by Ronald W. Jones and Peter B. Kenen (Amsterdam: North-Holland, 1985).

    • Search Google Scholar
    • Export Citation
  • Gustafsson, Björn, and N. Anders Klevmarken, Taxes and Transfers in Sweden: Incentive Effects on Labor Supply”, in Welfare and Work Incentives. A North European Perspective, ed. by A.B. Atkinson and Gunnar Viby Mogeuses (Oxford: Clarendon Press, 1993).

    • Search Google Scholar
    • Export Citation
  • Hagemann, Robert P., and Giuseppe Nicoletti,Ageing Populations: Economic Effects and Implications for Public Finance”, Organization for Economic Cooperation and Development. Department of Economics and Statistics. Working Paper No. 61 (Paris, 1989).

    • Search Google Scholar
    • Export Citation
  • Hagemann, Robert P., and Christopher John, Fiscal Policy in Sweden: A Generational Perspective.IMF Working Paper (Washington: International Monetary Fund, forthcoming, 1995).

    • Search Google Scholar
    • Export Citation
  • Hansen, HansSocial Security in Sweden and Four Other European Countries,” in Social Security in Sweden and other European Countries—Three Essays (Stockholm: Finansdeparttnentet, 1994).

    • Search Google Scholar
    • Export Citation
  • Hansen, Hans Henrekson, MagnusA New Framework for Testing the Effect of Government Spending on Growth and Productivity.Public Choice, Vol. 81 (December 1994), pp. 381401.

    • Search Google Scholar
    • Export Citation
  • Hansen, Hans, “Catching up in Industrialized Countries: A Desegregated Study.Journal of International Trade and Economic Development (forthcoming).

    • Search Google Scholar
    • Export Citation
  • Haveman, Robert,Should Generational Accounts Replace Public Budgets and Deficits?Journal of Economic Perspectives. Vol. 8 (Winter 1994), pp. 95111.

    • Search Google Scholar
    • Export Citation
  • Hedström. P.,Disability Pension: Welfare or Misfortune?in The Scandinavian Model: Welfare States and Welfare Research, ed. by R. Erikson and others (New York: M.E. Sharp, 1987).

    • Search Google Scholar
    • Export Citation
  • Heller, Peter S., Richard D. Haas, and Ahsan S. Mansur, A Review of the Fiscal Impulse Measure. IMF Occasional Paper No. 44 (Washington: International Monetary Fund, 1986).

    • Search Google Scholar
    • Export Citation
  • Henrekson, M., L. Jonung, and J. Stymne, Economic Growth and the Swedish Model” (unpublished, 1994).

  • Henrekson, Magnus, and others, The Crisis of the Swedish Welfare State. SNS Economic Policy Group. Occasional Paper No. 55 (Stockholm, May 1994).

    • Search Google Scholar
    • Export Citation
  • Kotlikoff, Laurence J., Generational Accounting: Knowing Who Pays, and When, for What We Spend (New York: Free Press, 1992).

  • Lindbeck, Assar, and others, The Swedish Experience, Institute for International Economic Studies, Seminar Paper No. 482 (Stockholm, 1990).

    • Search Google Scholar
    • Export Citation
  • Lindbeck, Assar, and others, Turning Sweden Around (Cambridge, Massachusetts: MIT Press, 1994).

  • Milgrom, Paul, and John Roberts, The Economics of Modern Manufacturing: Technology, Strategy, and Organisation,” American Economic Review. Vol. 80 (June 1990), pp. 51128.

    • Search Google Scholar
    • Export Citation
  • Ministry of Finance, “Complement to the Supplementary Budget Bill,” in Sweden’s Economy (Stockholm, April 25, 1994).

  • Ministry of Finance, Sweden’s Economy (Stockholm, April 1995).

  • Ministry of Health and Social Affairs, Pension Reform in Sweden, A Short Summary (Stockholm, 1994).

  • Muellbauer, John,The Assessment: Consumer Expenditure,” Oxford Review of Economic Policy, Vol. 10, No. 2, Summer 1994, pp. 141.

    • Search Google Scholar
    • Export Citation
  • Muellbauer, John,, “Comment on Auerbach, A.J. J. Gokhale, and L.J. Kotlikoff, ‘Generational Accounting: A New Approach to Understanding the Effects of Fiscal Policy on Saving,’Scandinavian Journal of Economics. Vol. 94, No. 2 (1992), pp. 31922.

    • Search Google Scholar
    • Export Citation
  • National Social Insurance Board, Sweden, Annual Report, 1992/93 (Stockholm, 1993).

  • Nilson, C., and J. Nilson, in “Monetary Policy Indicators” (Stockholm: Sveriges Riksbank, June 1993), pp. 4554.

  • Organization for Economic Cooperation and Development (1993a), OECD Economic Outlook, No. 54 (Paris, December 1993).

  • Organization for Economic Cooperation and Development (1993b), OECD Employment Outlook (Paris, 1993).

  • Organization for Economic Cooperation and Development (1994a). OECD Economic Surveys, 1993-1994; Sweden (Paris, 1994).

  • Organization for Economic Cooperation and Development (1994b), OECD Employment Outlook (Paris, 1994).

  • Olsson, Sven E., Social Policy and Welfare in Sweden (Lund: Arkiv, 1990).

  • Oxley, Howard, and John P. Martin, Controlling Government Spending and Deficits: Trends in the 1980s and Prospects for the 1990s.OECD Economic Studies. No. 17, Autumn 1991.

    • Search Google Scholar
    • Export Citation
  • Piore, Michael J., and Charles F. Sabel, The Second Industrial Divide: Possibilities for Prosperity (New York: Basic Books, 1984).

  • Ramaswamy, Ramana,The Structural Crisis in the Swedish Economy: Role of Labor Markets,” Staff Papers, International Monetary Fund, Vol. 41 (June 1994), pp. 36779.

    • Search Google Scholar
    • Export Citation
  • Sandmo, Agnar,Economists and the Welfare State,” European Economic Review, Vol. 35, 1991.

  • Sveriges Rikshank, Inflation and Inflation Expectations in Sweden (Stockholm, October 1994).

  • United Stales, Office of Management and Budget, Budget of the United States Government: Analytical Perspectives. Fiscal Year 1995 (Washington: U.S. Government Printing Office, 1994).

    • Search Google Scholar
    • Export Citation