1 David Dorn / Alfonso Sousa-Poza Motives for Early Retirement: Switzerland in an International Comparison Nr. 99 der Reihe DISKUSSIONSPAPIERE des Forschungsinstituts für Arbeit und Arbeitsrecht an der Universität St. Gallen St. Gallen, April 2004
2 1 Motives for Early Retirement: Switzerland in an International Comparison David Dorn / Alfonso Sousa-Poza * Abstract: Switzerland has one of the highest employment rates of older individuals among industrialized countries. Nevertheless, in the past decade, Switzerland has experienced a marked trend toward early retirement. This study analyzes the motives of early retirement in Switzerland in the 1990s. The three main groups of motives that cover about 70% of all early retirements are "personal reasons", company reorganizations", and "poor health". The incidence of early retirement due to corporate restructuring has significantly increased during the economic recession of the mid-1990s. Contrary to other European countries, early exits from the labor force have not been promoted by public policies. In an international comparison, Switzerland thus has a relatively low frequency of early retirement. Furthermore, involuntary early retirement is relatively rare in Switzerland, but often encountered in countries in which strict employment protection legislation forces firms to use early retirement as a means to circumvent such legislation. JEL Classification: J11, J14, J26 Keywords: early retirement, motives, Switzerland * David Dorn, Research Institute for Labor Economics and Labor Law at the University of St. Gallen (FAA- HSG), Switzerland, and currently visiting scholar at the University of North Carolina at Chapel Hill, USA. Alfonso Sousa-Poza, Economics Department, University of St. Gallen, and Research Institute for Labor Economics and Labor Law (FAA-HSG), Switzerland, This paper was presented at the 2004 annual conference of the Eastern Economic Association in Washington, D.C. The authors would like to thank the participants for valuable discussions. The usual disclaimer applies.
3 2 Motives for Early Retirement: Switzerland in an International Comparison 1. Introduction As in virtually all industrialized countries, Switzerland will experience large demographic shifts in the next few decades. Both a falling birth rate and an increasing life expectancy will have a significant impact on the public pension scheme as well as on the structure and functioning of the labor market. Inevitably, the future financing of the unfunded public pension system will be challenged. While currently four working-age individuals support a person older than 65, this support ratio will fall to 2.5 by In the labor market, the laborsupply growth will slow down, and, according to predictions by the Swiss Federal Statistical Office (see SFSO, 2003a), it will eventually become negative. The employment of older workers in Switzerland is thus gaining importance in public debate. These demographic trends have so far not been accompanied by an increase in labormarket participation of older individuals. Instead, during the 1990s, early retirement became increasingly popular: in 1991, 21.8% of all individuals who were within three years of reaching the legal retirement age had retired from the active workforce. Eight years later, in 1999, this proportion had increased to 29.8%. A notable rise in early retirements took place in 1997/1998, when an early-retirement option was introduced in the public old-age pension system. This increase also coincided with Switzerland's most severe recession since the 1930s. Such a trend to early retirement despite an aging population can also be observed in most other industrialized countries. Switzerland is, however, an interesting case: first, and despite the recent trend to early retirement in Switzerland, almost 70% of its population aged is still employed. This labor-force participation rate of older people is one of the highest among industrialized countries, as shown in figure 1. Apart from Switzerland, only some Nordic and East Asian countries, as well as the USA, have participation rates of about 60 to 70%. In the countries bordering Switzerland, however, most individuals aged 55 to 64 no longer form part of the labor force: in Germany, France, and Italy, the participation rate is at about 40%, and in Austria, this figure is as low as 30%. A second interesting characteristic of the Swiss case is that the legal retirement age corresponds very well to the actual retirement age. In other countries with similar legal retirement ages, the discrepancy between the two is considerably larger (see figure 2). Clearly, both the high employment rate of older workers as well as the small difference between the legal and actual retirement ages are strongly related to a relatively low incidence of early retirement. 1 The support ratio being defined as the number of people aged divided by number of people aged 65 and older; see Swiss Federal Statistical Office (2003a).
4 3 Figure 1: Participation rates of individuals aged in selected OECD countries 70 Participation rate (in %) Sweden Norway USA Denmark Portugal Finland Ireland Italy Greece Czech Rep. Poland Hungary Source: OECD (2002). Despite the obvious importance of the role of early retirement in an aging society, very little empirical research has been done on the motives for early retirement. Although some recent empirical evidence on the determinants of early retirement in Switzerland exists (Baldenweg-Bölle, 1997; Carnazzi, 2000; Gaillard et al., 2003; Balthasar et al., 2003; Wanner et al., 2003; Dorn and Sousa-Poza, 2004), there are still few insights into the reasons for early retirement. Furthermore, we are also not aware of studies - whether for Switzerland or for other countries - that analyze the motives for early retirement in a dynamic setting and, specifically, how motives change across a business cycle. The claim has also often been made that the incident of involuntary early retirement depends strongly on the state of the economy. An important issue is the extent to which involuntary early retirement is related to a country's employment protection legislation. Does strict employment protection legislation lead firms to circumvent such legislation by forcing early retirement? Undoubtedly, the answer to this question, as well as a better understanding of the motives of early retirement in general (and how these motives relate to changes in the economy), is crucial for policymakers.
5 4 Figure 2: Differences between legal retirement age and average age of transition to inactivity for older men Retirement age Belgium Austria Netherlands Finland France Germany Italy Spain Australia Canada U. K. Denmark Sweden Ireland Portugal USA Norway Switzerland Note: The right end of the bars shows the legal retirement age, while the left end refers to the average actual age of retirement. Source: based on data from Blöndal and Scarpetta (1998). The aim of this study is two-fold: first, it analyzes a new Swiss data set that captures the motives for early retirement. This will allow us to understand the motives for the increased trend towards early retirement in Switzerland in the past decade. Second, it will analyze a unique international microdata set in order to investigate the incidence of involuntary early retirement in an international setting and thereby assess the extent to which institutional aspects (employment protection legislation) influence early retirement. The paper is structured as follows: in section 2, an overview of the Swiss institutional system is presented. Section 3 discusses the data sets and methodology used in this study. Section 4 presents the results of the analysis on the motives for early retirement in Switzerland and section 5 is devoted to an international analysis on involuntary early retirement. Section 6 concludes. 2 Institutional framework The old-age provision system of a country can (by providing incentives and by defining the legal retirement age) play an important role in determining the actual retirement age of individuals. It thus influences the incidence of - and motives for - early retirement. In
6 5 Switzerland, as in most other countries, the old-age provision system consists of a public oldage insurance and private occupational pension funds. 2 a) Public old-age insurance The compulsory old-age insurance (the first pillar) aims at covering the basic living expenses of all retired individuals in Switzerland. It is financed by a pay-as-you-go system, with employees and employers each paying equal contributions of 4.2% of the employee s wage. The old-age insurance has a strongly redistributive character because contributions are proportional to income, whereas annual pensions are limited to a range between 12,660 and 25,320 CHF. 3 The legal retirement age, i.e., the age at which old-age pensions can be drawn, is currently set at 65 years for men, and 63 years for women. While men s legal retirement age has remained constant since the introduction of the public old-age insurance in 1948, women s retirement age fell from an initial 65 years to 62 years by However, in order to reduce the expenses of the insurance and to treat men and women more equally, the women's retirement age was raised to 63 years in A further increase to 64 years is scheduled for 2005, and a return to the initial level of 65 years is planned. Only since 1997 can old-age insurance pensions be claimed prior to the legal retirement age. At present, men can draw an old-age insurance pension as of age 63, and women as of the age of 62. Annual pension payments will in this case be reduced in an actuarially neutral manner. In the framework of the life-cycle models (see, for example, Mitchell and Fields, 1982; Burtless, 1986; Gustman and Steinmeier, 1986; Gustman and Steinmeier, 2000), actuarial neutrality implies that the net present value of the expected future stream of old-age pension income will be identical no matter whether an individual retires at the legal retirement age or up to two years in advance. The design of the old-pension scheme hence allows for a flexible choice of the retirement age without promoting early retirement through financial incentives. Workers may also choose to postpone their retirement by up to five years and receive accordingly increased pensions thereafter. Thus, the Swiss public pension system is characterized by a large degree of flexibility. It does allow for, but does not encourage early (or late) retirement. The fact that even the maximum public pension of 25,320 CHF is, for Swiss standards, relatively low suggests that other sources of old-age income (such as occupational pension funds) must have a greater influence on the decision about early retirement. 2 For a more comprehensive analysis of the effects of Swiss social insurances on the employment of older people see Dorn and Sousa-Poza (2003). Note that contrary to the United States, health insurance coverage in Switzerland is not associated with an individual's employment status, and thus plays no role in early retirement decisions (see, for the United States, Johnson et al., 2003; Karoly and Rogowski, 1994). 3 Based on the exchange rates in March 2004, this roughly translates to an annual pension value of 10,000 to 20,000 USD.
7 6 b) Occupational benefit plans The occupational benefit plan law (the second pillar) aims at ensuring that retired employees receive an old-age income beyond the one guaranteed by the old-age insurance in order to guarantee the accustomed standard of living. 4 One of the fundamental features of the law is the requirement that employers must insure their employees aged 25 to 65 in an occupational benefit plan if their annual income exceeds the so-called co-ordination deduction of 25,320 CHF. Occupational benefit plans are based on a fully funded system where every employee has an individual pension account to which annual contributions and interest payments are made. These contributions come from both employers and employees. Contribution rates are defined by the pension funds themselves and thus a large degree of flexibility exists. Federal law only requires that the employers contribute at least the same amount as their employees. According to the Swiss Federal Statistical Office (2002), the total contributions by employers exceed those of the employees by 60%. For the calculation of annual pensions, a relatively complicated formula is applied. 5 The first and second pillars of Swiss old-age provision should guarantee a pension that equals approximately 75% of the salary earned before retirement. As in the old-age insurance, the age at which pensions can be drawn is legally set at 65 years for men, and 63 years for women. The plans to raise women s retirement age also apply to pension funds. Once again, occupational benefit plans enjoy a considerable degree of freedom. According to the Swiss Federal Statistical Office (2002), occupational benefit plans that represented 16% of all insured men and 22% of all insured women have defined regulatory retirement ages that differ from the legal ones. Usually, these pension funds foresee a common retirement age for both men and woman. Table 1 shows that most insured employees have access to flexibility options with regard to retirement age. Nearly 90% of men and 80% of women can receive pension benefits five years prior to legal retirement age, whereas few occupational benefit plans offer the option of a deferred retirement. Judging from the design of these institutional regulations of pension funds, it seems that many companies tend to favor early retirement. This is possible because strong organizational links between a pension fund and the company whose employees it insures are common (although both entities have to be legally independent). As contributions in the second pillar vary with the age of an employee, the occupational benefit plan law makes early retirement more attractive for companies. The contributions start at 7% of the employee s salary at age 25 and reach 18% as of ten years prior to the legal 4 When the legal framework for this second pillar of Swiss old-age provision was introduced in 1985, a variety of schemes were already in place. These schemes had been set up on a voluntary basis by private and public employers. Taking this situation into account, the law did not enforce a unification of the existing pensions funds, but only defined minimum standards. Thus, occupational pension funds are still managed in a very decentralized fashion and they enjoy a large autonomy with regard to their institutional design. 5 See Dorn and Sousa-Poza (2003) for a detailed description.
8 7 retirement age. Because at least half of the contributions have to be paid by the employers, this regulation results in considerably higher wage costs for older employees, thus possibly discouraging their employment. Table 1: Proportion of pension-fund-insured employees who have access to flexible retirement options Women Men Flexible retirement age 92.5% 90.6% Early retirement five years prior to age 62/ % 89.5% Retirement later than age 62/ % 10.9% Source: Dorn and Sousa-Poza (2003), using data from Swiss Federal Statistical Office (2002). The data refers to the year 2000, when women s legal retirement age was still 62 years. Pension funds that have specified a retirement age other than the legal retirement age 62/65 are not included. In summary, the impact of occupational benefit plans on early retirement is higher than the one of the public old-age insurance. A majority of pension funds allow for early retirement several years prior to the legal retirement age. As occupational benefit plans have a substantial autonomy with regard to the institutional design of such crucial issues as the definition of a regulatory retirement age or the calculation of reduced pensions in case of early retirement, they have many possibilities to influence the early retirement behavior of their members. 3. Data and methods a) The 2002 Swiss Labor Force Survey (SLFS) The first data set analyzed in this paper is the 2002 Swiss Labor Force Survey (SLFS). 6 The SLFS is a representative survey conducted annually by the Swiss Federal Statistical Office since The choice of households to be contacted is done by a random draw from the Swiss phone register and classified according to social groups. The selected households are then contacted by phone, and an inventory of all people living in the household is made. One of the household members who is at least 15 years old is then randomly chosen to be the target respondent of the household. Target respondents are interviewed in detail. In comparison to former surveys, the 2002 SLFS contains two novel features that make it especially suitable for a study on early retirement: first, the questionnaire was supplemented with a special social security module. This module contains various questions that directly address the issue of early retirement. All respondents aged between 55 and 70 years were asked whether they have retired early or not. Moreover, a question on the reasons for early 6 An elaborate discussion of the SLFS methodology is given in Swiss Federal Statistical Office (2003b) and Swiss Federal Statistical Office (1996).
9 8 retirement allows us to analyze why people retired early. The second improvement of the 2002 SLFS is an increase in the sample size from about 15,000 to over 40,000 individuals. The analysis of the motives for early retirement is based on the SLFS question "Why did you go, or have you been sent, into early retirement?" Only individuals that reported having retired early were posed this question. In order to have a reasonably large sample size, the analysis will cover individuals who went into early retirement after The year 1996 is chosen because those men who retired at the age of 64 (i.e., the oldest possible age for an early retirement) in 1996 were 70 years old at the time of the survey and that is the age up to which the questions regarding early retirement were asked in the 2002 SLFS. Twelve possible answers to the above question were possible (including the answer "other reasons"). For the econometric analysis, these reasons were regrouped into four categories, as shown in table 2. 7 Table 2: Reasons for early retirement Reasons as defined in SLFS Company restructuring Company closing Accident, illness Invalidity No need or desire to work Could financially afford it Attractive offer by the employer Family duties Too old for the job Wish to become self-employed Fiscal reasons Other reasons Reasons as defined in explained variable Company reasons Health reasons Personal reasons Other reasons The motives for early retirement are analyzed with a multinomial logit model. The dependent variable depicts the four categories defined in table 2. Note that the reference category is "personal reasons". The dependent variable is defined for 1,005 individuals that went into early retirement between 1996 and 2002, who were aged 55-61/62 (women) or (men) at the time of early retirement, and who worked at some point in time during the 7 years preceding the survey. A variety of explanatory variables are used. They represent socio-demographic, professional, socio-economic, and employment-related characteristics of the respondents and are described in table A1 in the appendix. Note that the variables refer to a respondent s situation prior to early retirement. 7 Other ways of regrouping the motives are conceivable. For instance, "attractive offer by the employer" could have been included in the group of "company reasons", or maybe also among the reasons labeled as "personal reasons". However, our classification in table 2 intends to create three narrow and one wide group of reasons rather than several groups that are quite heterogeneous. This allows for a more meaningful interpretation of the results referring to the first three groups.
10 9 b) The 1997 International Social Survey Program (ISSP) In this paper, an international comparison of forced early retirement is conducted using data from the 1997 International Social Survey Program (ISSP). The ISSP is a continuing annual program of cross-national collaboration, which started in The data are collected by independent institutions in the participating countries using the same questionnaire. The topics of the annual surveys change from year to year. In 1997, 34,835 individuals were interviewed on "Work Orientations", covering issues on work content and organization as well as general attitudes towards work and leisure. Our analysis includes the following 19 countries: Canada, Czech Republic, Denmark, France, Germany, Great Britain, Hungary, Israel, Italy, Japan, Norway, Poland, Portugal, Russia, Slovenia, Spain, Sweden, Switzerland, and the USA. Non-employed workers were asked the following question: "What was the main reason that your job ended?" Two of the possible answers were "I retired early - by choice and I retired early - not by choice". These questions are used to define a dichotomous dependent variable having a value equal to one if the respondent retired early in an involuntary manner, and equal to zero if early retirement was voluntary. The analysis will cover the early retirement of individuals aged 45 to 64 who retired between 1983 and This broad definition allows for a reasonable sample size, which includes 722 early retirees. This variable is analyzed with the aid of a logit model. Table A2 in the appendix summarizes the corresponding explanatory variables. In addition to the ISSP variables, our analysis will include three variables referring to the labor-market characteristics of the different countries. The data for the variables "Average unemployment rate ", "Change in unemployment rate ", and "Employment protection legislation index" are taken from OECD (2002, 1999). 4. Reasons for early retirement in Switzerland Table 3 shows the reasons for going into early retirement between 1996 and 2002 in Switzerland. Together, personal motivation, company reasons, and health reasons were responsible for two thirds of all early retirements. The remaining third of retirements was caused by a variety of other reasons. Table 3: Incidence of reasons for early retirement Number of observations Percentage Personal reasons % No desire/need to work % Could financially afford it % Company reasons % Company restructuring % Company closing %
11 10 Health reasons % Accident, illness % Invalidity % Other reasons % Attractive offer by the employer % Too old for the job % Family duties % Fiscal reasons % Wish to become self-employed % Other reasons % No answer / does not know % The incidence of the three dominating reasons has, however, changed considerably over the last decade. Figure 3 shows the evolution of their relative importance since Most striking is the development of early retirement caused by company restructurings or company closings. At the beginning of the 1990s, they only accounted for some 16% of all early retirements. Thereafter, they steadily gained importance until 1997, when about 35% of all early retirements were due to "company reasons". Since then, this percentage has again returned to its initial level. Figure 3: Evolution of major reasons for early retirement since Percentage of total observations /91 92/93 94/ /02 Personal reasons Company reasons Health reasons Figure 4 shows that the incidence of early retirement for company reasons closely follows the unemployment rate. During the economic recession that started in the early 1990s, many companies were forced to reduce staff, and applied various measures to do this (see also Sousa-Poza, 2004). With regard to older employees, supporting or even enforcing early retirement was a frequent alternative to layoffs. Hammer et al. (2003) portray several pension
12 11 funds that installed early retirement programs that offered employees favorable conditions for a sometimes compulsory early retirement. A further economic development that contributed to the increasing incidence of company-driven early retirements was a series of major company mergers in the second half of the 1990s. In 1996, two pharmaceutical companies teamed up to form Novartis, and only a year later, UBS was created by the merger of two large banks, while Credit Suisse, another major bank, took over Winterthur, a large insurance company. Again, these mergers were followed by restructuring programs that included early retirements. The percentage of early retirements that can be attributed to personal reasons fluctuated around 25% throughout the 1990s. Since its low in 1997, when the relative importance of company reasons was much higher, it has consistently risen to 33% in 2001/02. The third major group of retirement reasons, health reasons, represented a relatively constant or even a slightly falling portion of all early retirements. These results clearly show that the reasons for going into early retirement are strongly influenced by the state of the external economy. Figure 4: Evolution of early retirement for company reasons and unemployment rate /91 92/93 94/ /02 Percentage early retirement for company reasons Unemployment rate Early retirement for company reasons Unemployment rate 0 Source for unemployment data: OECD (2002). The last data point for unemployment refers to 2001 only. The results of a multinomial logit model are presented in table 4. The category "personal reasons" is used as the reference category. The results displayed for company and health reasons hence denote the impact of the explanatory variables on the probability of retiring for company or health reasons relative to the probability of retiring for personal reasons. 8 Table 4 shows that workers with different socio-demographic characteristics indeed tend to retire early for different reasons. For many determinants, both company and health reasons have, compared to personal reasons, equally directed significant coefficients. These effects suggest 8 Note that, for the sake of clarity of presentation, the (mostly insignificant) results for the heterogeneous group "other reasons" are not displayed in the following tables.
13 12 that the characteristics of the people who retire early for personal reasons strongly differ from those of people who retire due to company or health reasons. Women more often retire for personal reasons, while company and health reasons are relatively more important for men. This result reflects women's weaker attachment to the labor market, i.e., it primarily underlines the traditional gender roles in Switzerland, especially among older individuals (Sousa-Poza et al., 2001). In the case of health reasons, it can also be argued that men more often do physically demanding work with corresponding health implications. Compared to retirement for personal motivation, both company and health reasons are more frequent for individuals who are still relatively far away from the legal retirement age. This result implies that workers only retire for personal motives when they are able to financially abridge a shorter period until legal retirement age. Table 4: Socio-demographic determinants of reasons for early retirement - multinomial logit model Company reasons Health reasons Coeff. / S.E. Risk Ratio Coeff. / S.E. Risk Ratio Intercept ** (0.346) Years until legal retirement age (in years) 0.114** (0.044) Male 0.770** (0.211) Lives with active partner (0.277) Lives with non-active partner ** (0.213) German-speaking region 0.462* (0.210) Foreigner (0.295) Low education 0.661* (0.259) High education * (0.228) ** (0.370) * (0.049) * (0.233) (0.297) ** (0.245) (0.230) (0.344) ** (0.274) * (0.284) Early retired in 2001/02 Reference Early retired in * (0.295) Early retired in ** (0.303) Early retired in ** (0.328) Early retired in ** (0.346) Early retired in ** (0.379) (0.312) (0.343) (0.379) ** (0.383) (0.427) Number of observations 1023 Log likelihood McFadden Pseudo-R All variables are dummies, if not indicated otherwise in brackets. Reference category is "personal reasons". */** denotes significance on the 5%/1% level.
14 13 Workers who have a non-active partner retire more often for personal motives. This reason for early retirement is also more frequent among people with a high education, while its importance is relatively small for workers with a low education. As educational levels are positively correlated with income, this result suggests that early retirement due to personal reasons is more frequent among individuals with higher incomes. It is also worth noting that the strongest effect with regard to educational levels is found for the relative probability that people with a low education retire for health reasons. Low-educated individuals more often do physical work, which increases the probability of a retirement for health reasons. With regard to the region of residence of the respondents, a regional effect can be found in the case of company reasons. In the German-speaking region of Switzerland, company reasons were more important than in the French- and Italian-speaking areas. This effect may be explained by differences in economic structures. Most large companies are based in the Germanic region. In general, they seem to favor early retirement. The dummy variables for the retirement years closely reflect the trend of the early retirement reasons that was displayed in figure 4. During the second half of the 1990s, retirement for company reasons was much more frequent than in 2001/2002. The relative probability of these reasons compared to retirement for personal reasons was between twice and almost seven times higher than in the reference period. Only in 1997 was the relative probability of an early retirement for health reasons significantly higher than in In table 5, the results of a multinomial logit model are presented that include dummy variables denoting the professional groups. Clerks are chosen as the reference group. The results in table 5 show that professions usually do not have similar impacts on the relative probabilities to retire for company and for health reasons. In the case of company reasons, differences between the professional groups are modest. Most professionals have a somewhat smaller probability of retiring for company reasons than clerks. However, this effect is only significant for the sales and service industry. A very different situation is found for health reasons. The assumption that physical labor increases the incidence of an early retirement due to poor health is confirmed. For craftsmen, plant workers, and elementary workers, it is significantly higher than for the reference group (clerks). At the same time, people in scientific or technical professions rarely retire for health reasons.
15 14 Table 5: Socio-demographic and professional determinants of reasons for early retirement - multinomial logit model Company reasons Health reasons Coeff. / S.E. Risk Ratio Coeff. / S.E. Risk Ratio Intercept ** (0.396) ** (0.443) Years until legal retirement age (in years) 0.116** (0.044) Male 0.736** (0.226) Lives with active partner (0.280) Lives with non-active partner ** (0.217) Germanic region 0.513* (0.213) Foreigner (0.299) Low education 0.629* (0.274) High education (0.261) Early retired in 2001/02 Reference Early retired in ** (0.300) Early retired in ** (0.307) Early retired in ** (0.333) Early retired in ** (0.353) Early retired in ** (0.385) Managers and legislators (0.376) Academic professionals (0.357) Technicians and associated profess (0.288) Clerks Reference Service and sales people * (0.391) Skilled agricultural workers (0.789) Craftsmen (0.389) Machine operators and assemblers (0.600) Workers in elementary occupations (0.508) Number of observations 1023 Log likelihood McFadden Pseudo-R ** (0.051) (0.258) (0.307) ** (0.252) (0.237) (0.350) ** (0.293) * (0.328) (0.322) (0.350) (0.393) ** (0.396) (0.438) (0.425) (0.462) (0.372) (0.437) (0.745) * (0.431) * (0.613) ** (0.483) All variables are dummies, if not indicated otherwise in brackets. Reference category is "personal reasons". */** denotes significance on the 5%/1% level.
16 15 In table 6, employment-related characteristics, including industry variables, are included. Moreover, a dummy variable for asset income is added. Table 6: Socio-demographic and employment-related determinants of reasons for early retirement - multinomial logit model Company reasons Health reasons Coeff. / S.E. Risk Ratio Coeff. / S.E. Risk Ratio Intercept (0.787) (0.828) Years until legal retirement age (in years) (0.061) Male 0.909** (0.337) Lives with active partner (0.406) Lives with non-active partner (0.279) Germanic region (0.277) Foreigner (0.381) Low education (0.340) High education (0.333) Early retired in 2001/02 Reference Early retired in * (0.449) Early retired in ** (0.444) Early retired in ** (0.479) Early retired in ** (0.478) Early retired in ** (0.528) Self-employed * (0.827) Employee in supervisory position (0.266) Unemployed within last 10 years 1.333** (0.496) Work experience (in years) (0.014) Agriculture (1.106) Manufacturing 1.587** (0.480) Energy and water supply (0.990) Construction (0.720) (0.068) * (0.367) (0.437) * (0.316) (0.300) (0.422) * (0.360) (0.420) (0.467) (0.468) (0.505) ** (0.494) (0.556) (0.679) (0.302) (0.580) ** (1.370) (1.086) * (0.580) (1.370) ** (0.711)
17 16 Wholesale and retail trade (0.602) Transport and communication 1.091* (0.547) Financial sector 1.121* (0.511) Real estate, renting, business act (0.578) Public administration Reference Education (0.636) Health and social work (0.589) Hotels and restaurants and other activities (0.564) Asset income > 1000 CHF ** (0.261) Number of observations 640 Log likelihood McFadden Pseudo-R (0.661) (0.641) (0.681) (0.814) (0.699) (0.672) * (0.611) ** (0.313) All variables are dummies, if not indicated otherwise in brackets. Reference category is "personal reasons". */** denotes significance on the 5%/1% level. Table 6 shows that, not surprisingly, self-employed workers differ from other employees in that they hardly retire for company reasons. For individuals that have been unemployed, company reasons are an important cause for early retirement. The more years of work experience a person has, the higher is the probability of retiring for personal motivation rather than for company or health reasons. The probability of early retirement due to company or health reasons is very low in the public sector. Stated differently, a large part of all early retirements in the public sector is due to personal reasons. In most other sectors, the probability of a company- or health-caused early retirement is higher. In the financial sector, and in the energy and water supply industries, the risk ratio for the probability of an early retirement due to company reasons is relatively large. These industries are dominated by large firms that tend to favor early retirement. In the construction industry and in agriculture, early retirement for health reasons is most prominent. At the same time, early retirement for personal reasons is rare, as there is no generous financial support of retirement by employers. As expected, having an asset income greatly increases the probability of retiring for personal motives. 5. An international, comparative analysis Our results for Switzerland show that the increase in the number of early retirements in the 1990s can be largely attributed to the severe recession in that decade. Despite this recent trend to early retirement, almost 70% of Switzerland s population aged is still
18 17 employed. This participation rate of older individuals is one of the highest among industrialized countries. Only some Nordic and East Asian countries as well as the USA have similar participation rates. A major reason for the international differences in participation rates of older people is the varying incidence of early retirement. Figure 5 displays the ratio of early retirees per worker for individuals aged 45 to 69. The countries with the highest participation rates, i.e. Sweden, Switzerland, Norway, Japan, and the USA, all have low early retirement ratios. A high frequency of early retirement is found in many continental European countries. 9 Figure 5: Early retirees per worker in age group Slovenia Hungary Spain Poland Early retirees per worker U. K. Germany France Portugal Italy Russia Denmark Israel Canada Switzerland Czech Rep. Sweden USA Norway Japan Source: based on ISSP 1997 data. Both the participation rates and early retirement ratios reveal that an early exit from the labor force is much more common in some industrialized countries than in others. This observation can hardly be explained by international differences in worker's preferences for leisure. Instead, workers seem to adapt their retirement behavior to the different institutional frameworks set by social security programs and pension funds (see hereto Weaver, 2003). An obvious impact of social security policies on retirement behavior comes from the definition of the legal retirement age, i.e., the standard age of entitlement to a public old-age pension. It is generally assumed that countries with a low legal retirement age also have low participation rates of older individuals. However, in a majority of OECD countries, the legal retirement age is 65 years, the same as for men in Switzerland. Among the exceptions are France, Japan, the Czech Republic, and Hungary, where the regular entitlement to public old- 9 Although participation and early-retirement rankings display a similar pattern, they do not fully coincide. One reason is that some countries, including Norway and the Czech Republic, have a very high proportion of older individuals that are out of the labor force due to disability. These high disability ratios among older workers may reflect the presence of a practice of "hidden" early retirement via disability insurance schemes.
19 18 age pensions starts already at age 60, and Norway, Denmark, and Iceland with a legal retirement age of 67 years. It is also not surprising that an increase in the legal retirement age has often been proposed as a measure to increase the participation rates of older workers. The OECD (2000) for instance recommends that the legal retirement age in Switzerland be raised to 67 years. But while a raise of the legal retirement age might ceteris paribus lead to a higher participation rate in a given country, the international comparison reveals that a higher legal retirement age need not go hand in hand with a higher average actual retirement age. According to Blöndal and Scarpetta (1998), the actual retirement age has consistently fallen in virtually all industrialized countries since 1950, although the legal retirement age has remained constant in most countries. Thus, the average age of transition to inactivity has, with the exception of Japan, fallen below the legal retirement age. The size of the differences between legal retirement age and average age of transition varies considerably between countries, as shown in figure 2 above. 10 In Switzerland, the deviation between the actual and the legal retirement age is still quite small. In other countries, however, a wide gap has opened. Despite a legal retirement age of 65 years, workers in Finland, Austria, the Netherlands, and Belgium on average retire before the age of 60. A main reason for the gap between legal and actual retirement age are social security systems that set incentives to retire prior to the legal retirement age. Gruber and Wise (1999) show for various countries that retirement tends to take place around the age when early retirement benefits can be claimed. In the USA, for instance, more people retire at the age of 62, when an early draw of old-age benefits is possible, than at 65 - the legal retirement age. In many countries, social insurances do not only allow for, but also support early retirement by various measures: In most OECD countries, a deferment of retirement from the age of 55 to 64 reduces the lifetime old-age pension income. People who decide to keep working until age 64 are hence confronted with an implicit tax on their labor income: while they receive a wage for every additional year of labor, they at the same time loose part of their potential pension income. According to Blöndal and Scarpetta (1998), the resulting implicit tax rate on labor income is substantial in many countries. For men, it amounts to more than 20% in Belgium and Finland, and even to 34% in Austria. It is thus not surprising that a wide gap between legal and actual retirement age has opened in these countries. France, Germany, and the USA have implicit tax rates of more than 10%. 11 Several countries foresee a facilitated access to old-age pensions for older unemployed individuals. Börsch-Supan and Schnabel (1999) describe such a scheme for Germany: an 10 Figure 2 does not include Japan, which is the only notable exception from the rule that the average retirement age is below the legal retirement age. In Japan, pension entitlement starts at age 60 for men (58 for women), while the average retirement age is at 66.5 years (63.7 for women). According to OECD (1995b), the Japanese government actively supports the continued employment of employees past the retirement age by granting subsidies to their employers. Moreover, Yashiro and Oshio (1999) argue that the trend to earlier retirement has stopped in the mid-1990s because the workforce started to shrink and the demand for older employees began to raise. 11 For the case of Italy, Brugiavini (1999) shows that the largest probabilities of retirement are found at those ages when the implicit tax rates on continued work are highest.
20 19 unemployed worker is entitled to an old-age pension at the age of 60, provided that he has been unemployed for at least one year and that he has paid social security contributions for 15 years. 12 According to OECD (1998), in more than half of all OECD countries, regular unemployment benefits can be claimed for several years, thereby allowing workers to receive unemployment-related payments until the legal retirement age of 65. Because older unemployed workers are usually either officially or tacitly exempted from actively searching for a new job, they are de facto often already in early retirement before having access to an old-age pension. 13 In several countries, a form of "hidden" early retirement takes place via disability insurance schemes. According to OECD (1998), disability benefits are used to aid early withdrawal from the labor force "to some extent at least". Some countries have explicitly or implicitly eased the entitlement conditions to disability benefits for unemployed workers. These practices contrast sharply with regulations of the public social insurances in Switzerland, which barely set any incentives for early retirement: old-age pensions are calculated in an actuarially neutral manner, and there is thus no implicit tax on labor income between the age of 55 and 64. The Swiss unemployment insurance does not offer an early retirement option to a worker aged 55, since unemployment benefits can only be claimed for a maximum period of two years. Furthermore, the Swiss disability scheme does not appear to facilitate early retirement (see Dorn and Sousa-Poza, 2003). In conclusion, there is no doubt that the design of the Swiss social security system has strongly contributed to the high participation rate of older people and the relatively low incidence of early retirement in Switzerland. In other countries, social security systems are a major reason of an early exit from the labor force. However, as discussed above, the institutional framework for early retirement is also shaped by the policies of private occupational benefits plans. In Switzerland, the often company-controlled pension funds tend to promote early retirement. The OECD (1998) suggests that pension funds tend to favor early retirement in many countries. Disney (2001) describes the example of Great Britain, where most pension funds offer early retirement options on favorable terms. Early retirement has been especially promoted during economic recessions and in times of structural change, such as in the privatization process of the late 1980s. 14 As shown above, this is also the case in Switzerland. Such practices obviously raise the important question of the extent to which early retirements are voluntary or not. The analysis of the reasons for early retirement in Switzerland has already indicated that some early retirements are taken voluntarily, while others are not. However, distinguishing voluntary from involuntary early retirements with data from the SLFS is not possible as some 12 According to Blöndal and Scarpetta (1998) and OECD (1996), similar programs exist in Australia, Austria, Denmark, Finland, Ireland, Italy, and Portugal. In other countries, including Great Britain and the Netherlands, the unemployment insurance allows for an early exit from work since there is no time limit for unemployment benefits. 13 The French experience with such policies is discussed in Gray (2002). 14 According to Guillemard (2003), employers also promote early retirement in France. Some companies have established retirement schemes that specify an early retirement age below 50 years.
21 20 reasons (such as health reasons) could be voluntary or involuntary. The ISSP data allows us to distinguish between voluntary and involuntary early retirement. In figure 6, a country ranking of involuntary early retirements is depicted. Figure 6: Ratio of involuntary early retirement in 19 countries 70 Percentage of involuntary retirement Russia Hungary Czech Rep. Portugal Slovenia Germany Poland Italy Israel France Spain Sweden U. K. Canada Switzerland Japan USA Norway Denmark Source: based on ISSP 1997 data. The Swiss ratio of forced retirement is relatively low in an international comparison (about 20% of all early retirements). While involuntary retirement is even less frequent in some countries, including the USA and Japan, more than one third of the early retirees in France, Italy, or Germany have been forced into early retirement. In some Eastern European countries, a majority of early retirements have been involuntary. The analysis for Switzerland has shown that people with distinctive socio-demographic characteristics tend to retire for different reasons. Accordingly, the first model analyzing forced retirement tests the impact of some basic socio-demographic variables on the probability that an early retirement has been involuntary. The analyzed sample contains observations referring to early retired persons from 19 countries. 15 The analysis includes individuals who went into early retirement between 1983 and 1997 and at ages between 45 and 64. The results in table 7 show that the effects of the investigated socio-demographic characteristics on the probability of an involuntary retirement largely coincide with previously discussed findings for retirement for personal reasons in Switzerland: the relative probability of an involuntary early retirement is smaller for women, married individuals, older workers, 15 The models below were also estimated for the G7 and Western European countries only. The results are similar to those of the entire sample.
22 21 and people with a high education. For men, singles, younger persons, and individuals with a low education, a higher probability of a forced early retirement can be observed. 16 Table 7: Socio-demographic determinants of involuntary early retirement in 19 countries - logit models Coefficient S.E. Odds Ratio Constant ** (0.283) Male 0.676** (0.189) Married * (0.214) Low Education 0.464* (0.279) High Education (0.322) Retirement age Reference Retirement age ** (0.209) Retirement age ** (0.245) Retirement age ** (0.362) Early retired in Reference Early retired in (0.224) Early retired in (0.233) Early retired in * (0.307) Early retired in (0.309) Number of observations 718 Log likelihood McFadden Pseudo-R All variables are dummies. */** denotes significance on the 5%/1% level. As involuntary early retirement can be linked to certain socio-demographic characteristics, the high incidence of forced early retirement in a given country might be mainly explained by the fact that people with these characteristics dominate among early retirees in that country. However, it is also possible that country-specific influences on the probability of an involuntary early retirement remain large even after controlling for sociodemographic differences. To test for country-specific effects, country variables were included (Switzerland being the reference country). Table 8 shows that significant country-specific differences in the probability of involuntary early retirement remain even after controlling for socio-demographic differences. 16 Note that no dynamic trend can be observed. In the reference time period of 1995 to 1997, involuntary retirement has been somewhat more frequent than in previous years. During these years, several European countries experienced an economic recession. However, the same years fell into an economically favorable phase in other countries, including the USA.