Chapter 6 Late careers and labour market exit into retirement in Estonia

Transcription

1 Chapter 6 Late careers and labour market exit into retirement in Estonia Ellu Saar, Kadri Täht and Marge Unt Title 1 INTRODUCTION Economic and social reforms in Estonia have had a profound impact on population development trends, causing a decline of the population and changes in age demographics (Schlitte and Stiller, 2006). As a consequence of low fertility rates and increasing life expectancy at birth, Estonia has become an ageing society. According to the United Nations demographic projection, in Estonia the proportion of people older than 60 years will increase from 21.7% in 2005 to 26.5% in 2030 and to 31.8% in 2050 (United Nations, 2008). The growing old-age dependency ratio shows that the working-age population is expected to support an increasing number of old people (Fortuny, Nesporova and Popova, 2003). These trends call for an increase in economic participation of older people in order to prevent a future labour shortage in Estonia. Given the growing proportion of older people, late career developments and the dynamics of social inequalities in late life employment are highly relevant scientific and political issue. In this context, a key challenge becomes the type of policy the government should use in order to promote higher labour market participation of older workers. Since the 1990s there have been major changes in social policy in Europe (OECD, 2007). The aim of these reforms was the retrenchment of early retirement as well as the reduction of public pensions while strengthening the role of private pensions at the same time (for example the Lisbon strategy). The rationale behind all these reforms has been to encourage older workers to delay the transition to retirement and to foster investments in private pension schemes in order to reduce public fiscal expenses (Hofäcker, Buchholz and Blossfeld, 2006). These reforms have substantially worsened the employment situation of the older workforce as they were not accompanied by changes supporting the employability of older workers. This means that labour market risks in old age have become increasingly privatized and individualized (Ebbinghaus, 2008). The highly turbulent restructuring process in post-socialist countries, triggered by the economic exposure to global markets, as well as the system transformation, affected the working age groups in different ways. One ambiguous group has been the older workers. At the beginning of the socio-economic transition, working pensioners were the first group to be laid off and many post-socialist countries introduced early retirement schemes in order to avoid the long-term unemployment of older workers (Fultz and Ruck, 2001; Fortuny et al., 2003). This meant that the formerly secure position of older workers in the labour market weakened.

2 2 Title Cross-national studies on late career workers have focused mainly on the variation in the timing of labour market exit and the nature of the exit based on cross-sectional data (Kohli, Rein, Guillemard and van Grunsteren, 1991; Jacobs and Rein, 1994; Ebbinghaus, 2003). Longitudinal analyses of employment careers of older workers are still rare. An increasing body of literature has broadened the conceptualization of retirement behaviour as a result of interplay between individual characteristics, preferences and institutional rules. Most crossnational studies focus, however, on the effect of country-specific pension systems (Blöndal and Scarpetta, 1998; Gruber and Wise, 1998, 2004). At the same time, recent cross-national studies indicate that the labour market participation of older people is not only determined by pension systems, but also by the interaction of different aspects of the national institutional framework (see Blossfeld, Buchholz and Hofäcker, 2006). At the macro-institutional level, retirement has usually been studied from two main perspectives: the pull factors that impact on labour supply (e.g. welfare state) or the push factors that affect labour demand (e.g. economic factors) (Guillemard and Rein, 1993; Casey, 1996; Ebbinghaus, 2000). In the context of Estonia, there are comparative analyses concerning the employment prospects and work abilities of the ageing workforce (see for example Leetmaa, Võrk and Kallaste, 2004; Tiit, Leppik, Võrk and Leetmaa, 2004). These analyses focus on exit options from the labour market, such as the pension system, early retirement, social assistance and unemployment benefits, all of which affect the participation of older workers in the labour market. However, most of this research concentrates on the retirement process and little attention has been paid to the career of older worker. The aim of this chapter is to study older workers labour market careers from a longitudinal perspective, focussing not only on labour force participation rates and early retirement, but also on mobility patterns in old age. The central research questions here are whether and to what extent have recent social and economic changes affected late careers (intensity and direction of job moves), including exiting the labour market for retirement in Estonia. The chapter considers the role of both individual and structural factors in the process of older workers adjustment to the new labour market situation. We will pay particular attention to social inequality patterns, for example, which groups of older workers are more likely to face downward mobility, unemployment or early retirement? For a better understanding of the peculiarities of the country and the situation of elderly workers in the Estonian labour market, the chapter also aims at locating the Estonian institutional context in the broader frame (pension and retirement) of welfare regimes and institutional settings. We can expect that in Estonia, due to the liberalisation strategy within the labour market and the welfare state the financial risks pertaining to old age have become strongly privatised and individualised. The chapter analyzes the developments and trends in the late careers, which is career transitions both from job to job as well as from job to unemployment after age 50 and transitions into retirement through five periods: the late socialist period ( ); the early transition period ( ); the economic recovery and upturn ( ); the economic recession as an outcome of globalization ( ); the economic recovery and upturn ( ).

3 Title 3 LATE CAREERS OF OLDER WORKERS Macro-level push and pull factors On the macro level, whether older workers stay on in the labour market or exit to retirement is affected by various production-related factors, which have an impact on the demand of labour force as well as by the social production-related factors. These factors primarily act on the supply of labour (Ebbinghaus, 2003). Decline of primary and secondary sectors, agerelated skill profiles and mass unemployment are structural push factors that affect the demand for labour and late-career employees position in the labour market. The proportion of older workers in declining industries and occupations tends to be comparatively high. Older workers often work in the manufacturing and agricultural sectors, while their proportion is quite low in the developing sectors, e.g. public and social services (Blöndal and Scarpetta, 1998), and indeed many older workers are blue-collar workers. Older men, rather than women, frequently work in sectors and jobs that are shrinking and are strongly exposed to rationalization. High unemployment has applied strong pressure on the labour market to reduce the supply of workers. Attempts to withdraw older workers from the labour market have been related to hopes that this might create new employment opportunities for younger job seekers and decrease youth unemployment (Gruber, Milligan and Wise, 2009). Early retirement was often legitimated by intergenerational solidarity, in that older workers would retire and thus opening up new job opportunities for younger people (Ebbinghaus, 2000). Nevertheless, whether younger workers are actually substitutes for older workers remains doubtful (Guillemard, 1991). Since most early exit options are not conditional on replacement by a young job seeker, their impact can only be indirect. However, the impact of these structural push factors on the mobility patterns of older workers will differ according to the organization of production, on the level of protection of the labour market and on the welfare system. There are two different ideal types of reactions of employers to structural changes (Buchholz, Hofäcker and Blossfeld, 2006). First, employers may try to keep their workers by offering them retraining and requalification, which may imply re-allocation of workers within a firm and may promote intra-firm mobility. Second, employers may dismiss their older workers and replace them with younger, better-qualified workers (Blossfeld and Stockmann, 1999). The ability of firms to shed older workers depends on a variety of factors. The most important of these factors are: (i) The employment policy, specifically the ease with which firms can implement mandatory retirement or dismiss older workers without any penalty. (ii) The educational policy regarding the skill gap between older and younger workers and retraining possibilities. (iii) The production system. (iv) The nature of employment relations pertaining to the existence of seniority wages and job tenure, which raises the cost of maintaining older workers. Employment protection may have an impact on older workers employment and mobility patterns by constraining the employment and dismissal of older workers. In countries where employment protection is low, mobility between jobs as well as exits to unemployment tends to be higher (Buchholz et al., 2006). Strong legal restrictions concerning dismissal and seniority rules protect older workers. By protecting older employees, these rules may

4 4 Title have the opposite effect for older job seekers because employers will not hire older workers due to the costs of these age or service related rules. Push factors are often conceptualized according to the typology of production systems (Hall and Soskice, 2001). In uncoordinated market economies, workers tend to have low or broad skills and turnover is relatively high. On the other hand, coordinated economies rely on internal labour markets with skilled workers and low turnover. Employees in these types of economies retrain workers and often use seniority wages to keep workers (Estevez-Abe, Iversen and Soskice, 2001). After losing a job, older employees have two main alternatives: either looking for a new job or retirement. The opportunities of finding a new job depend on the employees requalification opportunities. In many cases, their continued presence in the labour market is related with downward mobility. Another possibility is to exit into early retirement. Early retirement schemes provide a pathway for older employees to exit the labour market before the statutory retirement age, and compared to regular retirement they often do not lose out in terms of money. Early retirement programs have been widely introduced in conservative and Southern European countries. The choice between these two alternatives may depend on existence of pull factors, which in turn depend on the institutional environment, primarily the welfare state regime. The existence of qualificational and occupational boundaries is also important, in that rigidity of labour market boundaries may create restrictions to move within the labour market and promote early retirement. Options for bridging the period between withdrawal from the labour market and retirement vary by country. There are a variety of possible pathways (Ebbinghaus, 2000). First is the institutionalization of a complete exit from work at the normal, in most cases statutory, retirement age 1. The second consists of partial pensions provided for a gradual transition from work to retirement. In addition, some governments have introduced special early retirement programs. Unemployment benefits or disability pensions 2 also provide a pathway to retirement as these benefits allow individuals to bridge the time from ceasing to work to statutory retirement age. The decision to withdraw from the labour market also depends on the pension replacement rate and its relation to the retirement age and further working opportunities. If the replacement rate is low, older workers may opt to remain in the labour market due to the need to earn an income. The rate at which pension rights accrue also has an important impact on the retirement decision. If the pension accrual rate is low there are practically no financial penalties in terms of reduced old-age pension income when withdrawing from the labour market (Auer & Fortuny, 2000). In the same way, the duration of pension contributions may play a decisive role, especially in systems where pensions are contribution-oriented. Some countries have also included penalties for working beyond the retirement age (Buchholz et al., 2006). The more favourable the benefits and pensions, the less worthwhile it is for an individual to continue employment and the stronger the tendency to quit working early. The various factors affecting older people s stay in the labour market or their exit are interconnected. Figure 1 depicts ideal patterns of reaction of employers and older employees to such factors. Countries can be positioned on a scale, where at one end are those with dominant support for older people to stay in the labour market while at the other end are those countries, which encourage older workers to exit the labour market early. The logic of historically developed configurations of national institutions (production regimes and labour market regulation, educational system and the national infrastructure of lifelong

5 Title 5 learning and welfare regimes) suggests the adoption of one of these two strategies (Buchholz et al., 2006). Maintenance Exclusion Job mobility Employment mobility Occupational mobility Retraining Early retirement Moves to unemplyment Transitions to welfare state subsystems Need for labour market income Integrative labour market policies Generous pension system Strong insideroutsider logic Figure 1. Ideal patterns of adaptation on structural changes Pension regime typologies The peculiarities of a country s institutional environment (labour market adjustment, social security system, etc.) affect the maintenance or exclusion of older generations in the labour market. Various institutions interact with each other, leading to country-specific outcomes. Most research concerned with pension systems differentiate between four main groups of national welfare regimes: liberal, social democratic, conservative and fragmented (see Ebbinghaus, 2000; Mayer, 2004; Buchholz et al., 2006) (see Table 1). Liberal welfare regimes demonstrate market-induced maintenance regimes (Hofäcker, Buchholz and Blossfeld, 2006). They are characterized by a minimum level of public support for non-employment (Esping-Andersen, 1990) and in this regime private pensions are the main source of income for retirees (Blöndal and Scarpetta, 1998). Consequently older workers are forced to complement their private pensions with additional income from paid labour. Thus, late career workers may stay longer in the labour market and continue working beyond the normal retirement age. Older workers may, therefore, accept low paid jobs, which are unacceptable to younger workers. Pathways to retirement in a liberal regime are relatively limited. The pertinence of having an occupational pension leads to high levels of inequality in old age. A low level of employment protection and weak occupational boundaries facilitate and promote both employment and a high degree of mobility among older workers. Their employment careers tend to be generally longer, but less stable. Social-democratic welfare regime, like liberal regimes emphasise on keeping people of all age groups in employment. In contrast to the liberal, social-democratic regimes maintain older workers with support of public policy (publicly sponsored retraining as well as refresher programs) (Buchholz, Jabsen, Kurz, Marold, Schmelzer and Blossfeld, 2009). Social-democratic regimes provide multiple pathways to retirement maintaining at the same time an active employment policy, which integrates older people into the labour market through either or both part-time work options and partial pension schemes. Trends of flexibilization of work of older workers existing in liberal regimes have also been found in social-democratic regimes. However, this is less due to market forces as in liberal regimes, but rather as a result of active labour market policies (a public-induced maintenance

6 6 Title pattern). Tools for maintaining individuals employability include publicly sponsored retraining programs as well as refresher programs. The age at full retirement in socialdemocratic regimes is relatively high with low variability in timing. Conservative welfare regimes provide multiple pathways with often generous preretirement benefits, which rely on shedding labour from the workforce. Rigid employment protection and an expensive seniority wage system motivate employers to induce early retirement, which has been institutionalized as a social right. Without a strong system of active labour market programs, the generous welfare state provides a strong pull effect and fosters older worker s early withdrawal from the labour market. Rigid labour market boundaries do not promote occupational mobility. Additionally, the infrastructure for lifelong learning is rather weak (Buchholz et al., 2009). Fragmented welfare regimes are characteristic to Southern European countries. In many respect, these regimes, in their institutional settings, strongly resemble conservative countries. A distinctive feature of the fragmented regime is that low public welfare expenditure is distributed unequally and favours labour market insiders (Ferrera, 1996). Insiders are shielded against growing uncertainty and the requirements of flexibility. As a result, older workers enjoy high welfare provisions as the pension system is extremely generous and offers security even in case of an early labour market withdrawal (Hofäcker et al., 2006). However, there is a strong segmentation of welfare support due to a rather high level of self-employment and informal work (Hofäcker, 2006). The types of regimes described so far refer to traditional market economies. In contrast, former socialist states have experienced profound changes in their welfare regimes since the 1990s. Hofäcker et al. (2006) indicate that post-socialist countries make up a very distinct case within the dichotomy of employment exit versus maintenance regimes. In the socialist, pre-transition period, old age employment was high and stable and pensions were a major responsibility of government. Pensions were financed on a pay-as-you-go (PAYG or PAYGO) basis (Fultz and Ruck, 2000). Labour market exit patterns were homogeneous (see Bukodi and Róbert, 2006; Täht and Saar, 2006). The transition from a planned to a market economy set restructuring and rationalization demands on enterprises and on the economy as a whole. As a response to these changes, most of the former socialist countries initially decided to follow a strategy of early retirement and introduced profound reforms to adjust their pension systems to the new conditions. The main feature of these reforms was the shift from redistributive policies towards a system of contribution-based benefits. These reforms aimed at increasing the average retirement age, combined with low levels of pensions, which might have forced pensioners to earn additional income (Fortuny et al., 2003). In the development of the new system, the influence of several international organizations (World Bank, OECD) was decisive (Fox, 1997). However, compared to the other regimes the post-socialist countries now form a relatively heterogeneous cluster. In order to cope with the reorganization of a planned economy into a market economy these countries followed very different paths and developed various strategies (Buchholz et al., 2009).

7 Title 7 Table 1 Institutions and their impact on late careers Liberal Social democratic Conservative Fragmented Estonia Countries England, Ireland, USA Sweden, Denmark, Finland, Norway Germany, the Netherlands Italy, Spain Labor market Highly flexible with low compensation of risks Coordinated with high level of state intervention Rigid with strong insider-outsider segmentation Rigid with strong insider-outsider segmentation Strong insideroutsider segmentation Occupational boundaries Weak due to low importance of occupational certificates and strong commitment to lifelong learning Weak due to strong commitment to publicly supported lifelong learning Very strong due to high importance of occupational certificates and age discrimination in continued education Strong due to low commitment to lifelong learning Strong due to low commitment to lifelong learning Employment sustaining policies Low involvement of the state; activating labour market policies Active, employment supportive programs Few active employment programs Few employment sustaining policies Few employment sustaining policies Employment exit policies Strongly privatized pension systems; low generosity of public pension; low replacement Highly compensating universal pension system with strong public pillar; high replacement Contribution-based pension systems; moderate replacement; strong early retirement incentives Contributionbased pension systems; high replacement; strong early retirement incentives Privatized pension systems; low replacement Exit pathways Limited pathways Gradual pathways Multiple pathways Bridging unemployment Disability pensions as bridging pension; limited pathways Impact on late career workers Moderate to high labour force participation; high rates of job-to-job/ occupational mobility; high variance Moderate to high labour force participation; relatively continuous careers; few early exits; low variance Low labour force participation; strong tendency to early exit; few cross-occupational mobility; low variance Low labour force participation; strong tendency to early exit; few crossoccupational mobility; high variance Moderate labour force participation; high rates of jobto-job mobility; high variance Overwhelming strategy Market-induced maintenance Public-induced maintenance Employment exit Source: Based on Ebbinghaus, 2000; Mayer, 2004; Saar, and Täht, 2005; Buchholz et al., 2006; Buchholz et al., Combination of market-induced maintenance and employment exit strategies

8 8 Title ESTONIAN INSTITUTIONAL CONTEXT Industrial restructuring The transition process brought fundamental changes to the composition of employment, both by economic sector and branches. The changes in the proportion of employment in the industrial sector have been rather modest, while the changes in the agricultural and service sectors are impressive. In agriculture, total employment dropped from (1989) to in 2004 (Eamets, Paas and Masso, 2006). Blue-collar workers were particularly affected, their numbers declined by 36% between years 1989 and 2008, while the number of white-collar workers declined by only 6%. The occupational structure of different age groups was also affected. In 1989, years old workers were overrepresented in basic unskilled occupations as well as in agricultural occupations. Both of these occupational groups were strongly affected by the structural changes. Despite the fact that the proportion of late career workers in basic unskilled occupations did not change between , they were quite successful in relocating to higher level occupational groups as the percentage of older employees in white collar positions has increased from 40% in 1989 to 52% in 2006 (for the total working age population, the increase was from 47% to 57%) (Statistics Estonia). Lifelong learning An important feature of the Estonian labour market has been the lack of opportunity for older workers to enter the educational system or to get (re)training over the life course (Saar, 2002). Despite the introduction of the life-long learning concept, the actual participation of older age groups in re-training and further education programs has remained very low. In the Soviet period, any return to the educational system was difficult. This seems to have been carried over into the new labour market environment because in 2007, the majority of the 50+ age group considered themselves as too old for studies. An Adult Education Survey carried out in 2007 indicated that for those older than 55 the participation rate in training was half that for year olds. According to the same survey, retraining and further education were, in terms of supply and means of participation, accessible largely by white-collar workers. Thus the probability for a quick and successful return to the labour market for those losing a job in the industrial or agricultural sectors was lower. In addition, further training is restricted to the working population - the participation rate of the unemployed and pensioners is very low. Welfare regime and pension system The welfare regime in Estonia does not exactly follow any of the models used in other EU countries and has been quite dynamic over the last decade. In the development of Estonia s new system, the influence of several international organizations has also been influential. According to Masso and Paas (2006), at the outset of the transition period, Estonia mainly used the Central European (corporative, labour-market-based) model. In the middle of the 1990s, elements of a more liberal model were introduced (Bernatas and Guogis, 2004). This model contributes more to the economic efficiency than the social cohesion of society. The pre-reform statutory retirement age was fairly low in Estonia, being set since 1956 at 60 years for men and 55 years for women. At the beginning of 1990s, the statutory

9 Title 9 retirement age was revised and after a couple of reforms was eventually equalized at age 63 for both genders, with a gradual rise of 0.5 years annually. Some authors claimed that in post-socialist countries, an early retirement policy was used as a substitute for welfare and unemployment benefits (Müller, 2002). As Fortunyi et al. (2003: 42) stated, labour offices were sceptical about the employment chances of many older jobseekers and therefore often gave up intensive job search assistance and offered older workers either an early retirement or disability pension. In Estonia, special early retirement schemes existed both during the Soviet era and after the pension reforms of the 1990s, but such options were limited to specific professional groups, such as pilots, seamen, miners and to the parents of families with numerous or disabled children. The 1998 State Pension Insurance Act was the first to provide the opportunity to take early retirement up to 3 years before the statutory retirement age. However, the choice is penalized by reducing the pension by 0.4% for each month before the statutory retirement age. Those who choose to postpone retirement can defer their pension, the benefit of which is a pension level that is 0.9% higher per month of continuing to work without drawing a pension. In 2007, 4.2% of all pensioners received early-retirement pensions (Sotsiaalvaldkonna, 2008: 41). More than 80% of those drawing early retirement pensions were previously unemployed (Leppik and Kruuda, 2003). Also, disability pensions have been quite frequently used as an early retirement pathway. In 2007, the proportion of persons receiving a disability pension was 17% of all pension recipients (Sotsiaalvaldkonna, 2008: 40). Thus, the disability pension is one of the reasons for the quite remarkable outflow of mainly older workers from the labour market. While Estonia was part of the Soviet Union, the pension system worked as a PAYG (PAYGO) system. Pensions were calculated from the reference wage of monthly earnings received in the last twelve months, which often led workers to attempt to increase the reference wage during the last twelve months of employment by for example, multiple appointments, changing to a blue-collar or other better paid job, etc. These attempts to increase their pensions increased the job mobility of older workers. In 1992, the pension system changed all old age pensioners received pensions of a similar amount, depending on the minimum wage. In 1993, a differentiation according to the length of employment was introduced. In 1998, Parliament adopted a reform program, which aimed at producing the three-pillar system used by most other post-socialist countries (Müller, 2002; Ferge and Juhász, 2004). The three pillars were: 1) a mandatory PAYG (PAYGO) public pension system to insure the minimum standard of living; 2) a mandatory funded and privately managed pension system, based on personal finances (the Latin American approach); 3) a voluntary pension plan system. The pension replacement ratio has remained rather low and stable in Estonia since the introduction of the pension scheme in 1956, fluctuating between 30-40% (Figure 2). In the early 1990s, the replacement capacity dropped to half the previous level, and has only gradually increased reaching the level of the late 1960s by the late 1990s. As pension replacement ratio is low and there is still very little differentiation between pension receivers, becoming a pensioner significantly increases an individual s poverty risk. A quarter of year olds and more than a third of those aged over 65 belonged to the lowest income quintile in 2000 (Sotsiaalvaldkonna, 2008: 103). Although in the relative poverty risk of children and working age population has decreased, it has increased for the elderly (65 years and older). During , the relative poverty rate increased among the elderly by a factor of 1.56 and in 2005 a quarter of the elderly belonged to the poor (Tervis, töö ja sotsiaalelu, 2008: 18).

10 10 Title Figure 2 Pension replacement ratio in Estonia, (percent average gross wage) Source: Põldma, 2000: 260; Ministry of Social Affairs, The mandatory state pension insurance system in Estonia includes many incentives for working. First, the amount of pension insurance paid for an employee as part of their Social Tax increases the insurance component and thus also the future pension amount. Second, according to the State Pension Insurance Act (2004) it is possible to receive remuneration for work and a pension simultaneously without being either restricted or penalised. Third, the extremely low replacement rate of pensions, compared to other European countries, fosters the need to remain in the labour market. The Estonian pension system also contains feature aimed at stimulating career exits before statutory retirement age: (i) the statutory right of certain professions to retire before the retirement age; (ii) the opportunity to choose the age of retirement; (iii) liberalized disability provisions; and (iv) a combination of the existing unemployment insurance and unemployment benefit scheme may reduce late employment. As unemployment insurance and unemployment benefit only cover short-term unemployment risk in Estonia, their stimulating impact on late employment is low and motivates unemployed late career workers to retire as soon as this option becomes available. Hence, the institutional regulation of Estonia (see also Table 1) gives a rather ambiguous message as to whether the overall aim is to keep the older generations in the labour market or make them to leave early. On the one hand, employees are encouraged, even forced, to stay in the labour market as long as possible, for which the increasing retirement age, low replacement rate of pensions, low unemployment benefits and opportunities to combine pensions with salaries are clear evidence. On the other hand, extremely modest investments are made into keeping older workers in the labour market through training and retraining.

11 Title 11 HYPOTHESES Social and periodic change hypotheses Restructuring of economy and institutional settings in the early 1990s changed the former stable status of late career workers in the labour market into insecure and instable. The emerging insider/outsider labour market did no longer protect older workers by means of seniority rules, but excluded them from regular employment. Outflows from employment to inactivity exceeded significantly outflows to unemployment as labour market pressures were primarily solved at the expense of most vulnerable groups (primarily older workers) (Helemäe and Saar, 2008). We therefore expect to find an increase in the labour market mobility (especially downward mobility) of late career workers in the first half of 1990s. After the period of accelerated economic restructuring, the labour market gradually stabilized during the period of recovery and stabilization between 1995 and 1998: both the rate and the number of people who experienced job-to-job moves decreased significantly (Haltiwanger and Vodopivec, 1999). In the second half of 1998, the Russian economic crisis took place. The loss of competitiveness in the Russian market due to the devaluation of the rouble forced Estonian manufacturing industry to carry out extensive restructuring which resulted in a significant reduction in employment. We hypothesize that these economic changes increased late career workers transitions from employment to unemployment. The labour market situation improved in the context of the economic upturn experienced by Estonia Unemployment declined and employment rates increased. We expect that it has also affected the labour market chances of the elderly and improved their late career possibilities. Due to the pension reforms in the 1990s (i.e. postponement of statutory retirement age, introduction of deferred pension, etc.), older workers were expected or even encouraged to stay in the labour market as long as possible. At the same time, the combination of the restructuring of industries with lay-offs of often older workers, increasing competition, the introduction of new technologies and increasing labour market competition has oriented the labour market towards younger people. As a result, older workers had more difficulty in either remaining in the labour market or returning to employment after unemployment. The status of unemployment creates a high risk of exclusion, which may make early retirement especially attractive for older workers. We therefore expect that the increase in labour market insecurity also affects the retirement process, making remaining in the labour market during economic upturns more attractive and making early retirement more likely during economic downturns. At the same time, we expect a general increase in retirement age over the observed period, which is however mostly due to the increased statutory retirement age. Still, the statutory and actual retirement ages are expected to converge over time. Social selectivity hypothesis In the context of the labour market and the welfare regime, Estonia chose a broad liberalization strategy. As a result, financial risks were strongly individualized. Here our main hypothesis for a late career is that there are increasing differences for a variety of social groups both in terms of staying in the labour market as well as leaving it for

12 12 Title retirement in the 1990s. The balancing between the tendency to stay in the labour market or to leave is expected to result from various individual as well as institutional characteristics. Restructuring combined with the privatization of the economy exerted a dual pressure on old workers labour market behaviour in the 1990s. Previous analysis indicates that two groups of older male workers, the highly educated and the low-skilled agricultural workers, had better chances to stay in the labour market and also had lower risks of being excluded from employment (Täht and Saar, 2006). We therefore predict that industrial workers, especially those with low qualifications will show a higher probability of moving into inactivity as well as into unemployment than those in other sectors and those with higher qualifications. We also expect interplay between labour market opportunities and education to influence the labour market behaviour of older workers. For older workers whose labour market location was most endangered by the economic restructuring, education was a highly important resource to help avoiding the risks of unemployment and dropping out of employment. Previous analysis has also indicates that workers in the social service sector were more shielded against withdrawal from the labour market in the 1990s (Täht and Saar, 2006). As both women and Estonians (as opposed to non-estonians) are overrepresented in this sector we expect that these groups have a slightly lower risk dropping out from labour market. Also previous findings show that women s unemployment rate was lower than of men, especially at the beginning and in the middle of 1990s when unemployment appeared (Täht and Unt, 2002). We, therefore, hypothesize that men and non-estonians will suffer more from economic downturns. DATA AND VARIABLES For our analysis, we used data from the Estonian Social Survey (ESS), which was collected in 2004 by the Estonian Statistical Office. The target population of ESS was all private households in Estonia, with the original sample of approximately 4,500 households and 10,000 individuals. All household members aged 15 or older were interviewed. The final number of respondents in the survey was approximately 8,900 individuals. For studying the late careers and labour market exit, our sub-sample consisted of those who reached 50 years of age after 1980 and before the time of the interview in An additional requirement for the respondents of this sub-sample was to have had a job at the age of 50 years, which meant we studied the late careers and labour market exits of those who were in the labour market at the age of 50 years. This requirement reduced the population of the sub-sample by about 11 percent leaving a total of 2,532 individuals. Depending on the specific research question and respective analysis, additional requirements were applied. The data on employment careers is retrospective; as the younger cohorts have experienced fewer career events, they may be slightly over-represented in unemployment transitions and definitely underrepresented in retirement transitions. However, the analysis of the empirical part is carried out and presented as a comparison of historical economic periods and is this way less sensitive to any birth cohort issue. Besides, the analysis controls the effect of age in assessing the transition probabilities. Although due to potential sample bias we have also excluded from our analysis those born before the 1930s, it still

13 Title 13 has to be kept in mind that due to the higher life expectancy of women, women may be slightly overrepresented in our sample. As both men s and women s labour market participation has been high throughout the studied periods, our analysis includes both genders and controls for gender effect. In order to study the instabilities of late careers, we analyze the risk of entering long-term unemployment (lasting longer than 6 months) during the first labour market transition after 50 years of age. As for the Estonian case, the issue of unemployment becomes relevant only since 1990, so we restricted our analysis of unemployment to respondents who started their late career only after year This leaves us with an effective sample of 1,592 individuals. As a method of analysis we use binary logistic regression models. Late careers are studied by analyzing upward and downward mobility during all the employment episodes after the age of 50 years. One respondent can have several transition episodes in the analysis. In order to measure the upward and downward mobility, we studied the difference between two consecutive work episodes measured on the International Socio-Economic Index (ISEI) scale (Ganzeboom, de Graaf and Treiman, 1992). If there is an unemployment or inactivity episode between two jobs, we compare the ISEI of the occupations before and after the unemployment or inactivity episode (with control for that effect). The change in the ISEI of more than 10 percentage-points upwards counts as upward mobility, whereas a change more than 10 percentage-points downwards counts as downward mobility. In this context, we use multinomial logistic regression, where the probability of upward and downward moves is compared to the absence of movement or a lateral move. We analyze also the transition to retirement (as the main status) from the last employment episode. At the same time, in the analysis we control for all unemployment and inactivity spells before and after the age of 50 years and before starting the last work episode that leads to retirement. Here we use transition rate models. As independent variables, we included gender, age, ethnicity, education and occupation. The time periods when the employment transitions took place were divided as follows: ; ; ; and Industry is measured using the collapsed classification introduced by Singelmann (1978). For measuring occupational status we use the abbreviated ISCO-88. Depending on the analysis, we also control for the effect of the unemployment rate. In addition to the ESS, we also used in a descriptive manner data from the Estonian Labour Force Survey (ELFS) This data provides us with time series on labour market transitions from the middle of 1990s until RESULTS Labour force participation of the elderly in a comparative perspective For a better understanding of the Estonian labour market situation regarding the older workers, Figure 3 presents the employment rates of people over the age of 50 in other EU countries and in the USA. Grouping the countries shows systematic differences existing across the countries. The employment rate of older workers is very low in conservative welfare regimes (Austria, Germany, France) and fragmented regimes (Spain, Greece, Italy). These countries rely most heavily on the labour shedding strategy. In liberal welfare

14 14 Title regimes the employment rate is at the middle level. The Scandinavian social democratic welfare regime shows high employment performance during old age: the employment rates of older workers in Sweden and Denmark are respectively just above 70% and just below 60%. In Estonia the employment rate is above 60%, which is substantially higher than most of the other post-socialist countries and comparable with figures for liberal or social democratic welfare regimes. Figure 3 Employment rate of older people (over age 50) in 2008 across welfare regimes, a country comparison, % Source: Eurostat. The labour market participation rate of the elderly in Estonia has, however, not been static over recent decades, as the various age groups of the elderly have experienced different trends (Figure 4). The labour market tensions at the beginning of 1990s were initially solved by pushing older workers (and especially working pensioners) out of the labour market, resulting in a drop of the employment rate of elderly. After Estonia had achieved economic recovery, labour force participation rate started to increase again, including the older participants in the labour market. The latter trend is partly also due to the deferred pension age (especially for women for whom the pension age was increased from 55 to 63 years of age), low pensions and pension laws favouring working pensioners. As pensioners receive a pension whether they work or not, they were more willing to accept less stable and low paid jobs (see also Fortunyi et al., 2003).

15 Title 15 Figure 4 Labor force participation rate by age, Source: Estonian Labour Force Survey, Labour market status in late career The dynamics of labour market status and the late careers of older workers across time (from 1980 to the first years of 2000) are presented in Table 2. The proportion of men and women employed in the 50 year age group has been rather high, for example for men being 95.7 % in the 1980s and 80.1% at the beginning of 2000s. At the same time, the data reveals a decreasing trend of employment among the 50 year age group, the rates dropping about 15% (men) and 10% (women). Unemployment can account for part of this drop. Although unemployment had not existed before the 1990s, due to the soviet policy of full employment, by the end of the decade the gendered unemployment rates for the 50 year age group were men at 8.1% and women at 5.9%.. The proportion of inactivity (including both disability and old age pension) for this age group has stayed quite stable over time and fairly comparable between the genders. Nevertheless there has been a slight increase in the inactivity status among men and women from 1980s until 2000 when it starts again showing a slight decrease. For women, an increasingly important category for characterizing the social status at this age is care leave or staying at home. The latter may be, however, an indicator of hidden unemployment of women, especially as the official unemployment rate for women compared to men is lower. The picture looks slightly different when analyzing the labour market status for the 60 year age group. While during the 1980s more than 80% of the men in this group (60 years at the time was the official retirement age) and 46.9% of the women were still employed at

16 16 Title the beginning of 1990s, after which a significant drop in the labour market participation rate for both genders occurred. For men, the trend has remained downward throughout the analysed years, whereas for women a slight increase occurred at the end of the 1990s. The latter may be due to the increase of statutory retirement age for women. The status of being unemployed becomes marginal as most of this age group are either disability or old age pensioners. The status of being unemployed becomes more important at the beginning of 2000s, which may be also partly due to increased retirement age and therefore a lack of chances to leave labour market earlier (especially for women). Another reason is that the economy was experiencing an upturn and many of this age group decided to return to employment. Although women s unemployment rate at that period was lower than this of men, still an increasing amount of women were out of employment and staying at home. At 65, beyond the statutory retirement age, there are elderly who continue to work. The employment rates were higher in the 1980s and have dropped both for men and women throughout the 1990s, showing a slight increase again at the beginning of As we can see, 38.9% of the men and 28.2% of the women () at 65 were still employed. By the time of the Russian financial crisis, , the employment rates drop to 25.3% for men and 11.3% for women. When the economic upturn started at the beginning of 2000s, the opportunities for this age group to work improved.

17 Table 2 Descriptive indicators for late career transitions and retirement Title 17 Men Women Employment status at age 50 (%) Employed (in %) Long-term unemployed (in %) Disability/old age pension Care leave, staying home Employment status at age 60 (%) Employed (in %) Long-term unemployed (in %) Disability/old age pension Care leave, staying home Employment status at age 65 (%) Employed (in %) Long-term unemployed (in %) Disability/old age pension Care leave, staying home Late career characteristics a (%) Unemployment experience b Upward mobility c Downward mobility c Lateral mobility c Retirement c Transition to retirement b Mean retirement age d Standard deviation Source: Own calculations based on Estonian Social Survey 2004 Note: a - based on those employed at the age of 50 years; b from all the transitions that took place in respective period, how many were transitions into long-term unemployment; c from all the transitions that took place in respective period, how many were upward, downward or lateral career moves or a move into retirement; career mobility transitions include both transition directly from one job to other, as well as transitions with an inactivity spell between two consecutive job episodes; d - based on those employed or unemployed at the age of 50 years; e mean age calculated from the retirement age of those respondents who have retired by the time of interview, i.e. before Unemployment risk in late career While in the 1980s unemployment did not really exist, it has become an important feature in the late careers since then. The unemployment rates in Estonia differ also for age groups, and tend to decline with age. Until the 2000s, the difference between the prime-age group (29-45 years) and the old-age group (50 years and older) accounted for 3 percentage points. Lower unemployment rate of late career workers was mainly the result of early withdrawal