Post-Communist Economies, 2015 Vol. 27, No. 4, 497–516, http://dx.doi.org/10.1080/14631377.2015.1084719

Changes in pension take-up and retirement in the context of increasing the pension age: the case of Estonia in the 2000s Allan Puur, Lauri Leppik* and Martin Klesment Estonian Institute for Population Studies, Tallinn University, Tallinn, Estonia

Downloaded by [80.235.88.165] at 01:41 24 November 2015

(Received 13 January 2015; final version received 10 July 2015) This article looks into the effects of increasing the statutory pension age on pension take-up and on employment before and after pension take-up in Estonia in 2002–11. We use individual-level administrative data from the pension register of the Estonian National Social Insurance Board and a set of analytical indicators to capture the pension take-up and employment exit processes. The results reveal a relatively successful adjustment to the increase in pension age. The age gap between take-up of normal and early retirement (old-age) pension has not widened and the increase in the statutory pension age has moved exit from employment proportionally towards a later age. At the same time, a marked increase in the take-up of incapacity pensions, accelerated during the economic recession, has significantly reduced the overall effect of pension age reform.

In the context of demographic ageing considerable hopes are placed on increasing the statutory pension age to offset the fiscal implications of ageing on public budgets. Most European Union member states have legislated pension reforms with the underlying aim of increasing the participation rate of older workers, in order to effectively channel increases in longevity into extended periods of work. Such reforms include (European Commission 2011) increases in the statutory pension age (Bulgaria, Czech Republic, Germany, Greece, Estonia, France, Lithuania, Malta, Romania, Slovenia, Slovakia, United Kingdom), convergence of women’s lower statutory pension age to that of men (Austria, Estonia, Greece), linking of statutory pension entitlements to changes in life expectancy (Denmark, Italy, Poland, Portugal, Sweden), tightening of conditions for early retirement (France, Spain, Romania), increasing the length of service requirements (Bulgaria, Cyprus, Greece, Spain, Malta) and changes in incentives affecting retirement decisions (Finland, Netherlands). While some parametric policy changes have been adopted as one-off measures, increasing the statutory pension age is as a rule phased in gradually over a period of time. Effectively there is a wide consensus among pension policy advisers and policy makers on the necessity and feasibility of increasing the pension age, in particular in the context of increasing longevity (Barr and Diamond 2010, OECD 2011, European Commission 2012a). On the other hand, many pension age reforms have been legislated only relatively recently, over the last decade, to be phased in gradually over prolonged periods of time.1 Therefore the number of empirical studies to assess the actual implications of increasing the pension age for retirement decisions – how the increase in statutory pension age is reflected in pension behaviour of *Corresponding author. Email: [email protected] © 2015 Taylor & Francis

Downloaded by [80.235.88.165] at 01:41 24 November 2015

498

A. Puur et al.

individuals, to what extent the policy change has triggered changes in pension take-up and employment exit etc. – is still limited. It is evident that increasing the statutory pension age will improve the balance of public budgets only if the policy change is accompanied by behavioural change in retirement of the affected cohorts or, in other words, if the policy change is internalised by persons to whom such policy reform is targeted and if the labour markets accommodate the increasing supply of older workers. An increase in the statutory pension age is a particular test of labour market elasticity, the ability of the labour market to absorb an increase in the supply of labour by older workers. As has been pointed out in policy debates (OECD 2011), in the negative case scenario an increase in statutory pension age may lead to an increase in unemployment of older workers, increasing pressure on early retirement provisions and increase in take-up of disability pensions. The latter is to be expected as disability prevalence increases with age (OECD 2006). Moreover, the analytical toolbox to monitor pension age reforms still needs finetuning. There is not yet any commonly agreed portfolio of indicators to monitor and evaluate the process of increasing pension age. The key indicator which has been used to characterise retirement trends is the average effective exit age from the labour market, which is one of the structural indicators in the EU portfolio of social protection indicators (European Commission 2013). However, the average effective labour market exit age is calculated from the labour force survey data and captures only the labour market aspect of the retirement process. It suffers from certain methodological problems as pointed out by Stiglbauer (2013). This article looks into the effects of increasing statutory pension age on pension take-up as well as employment before and after pension take-up in Estonia in 2002–11 using pension register data. We use ‘pension age’ to denote the statutory/legal age at which a person becomes eligible for a pension, whereas ‘retirement age’ denotes the age at which a person withdraws from employment. We speak about ‘pension take-up’ as the act of drawing a pension and ‘retirement’ as the act of exit from employment. In Estonia a gradual increase in statutory pension age commenced in 1994 from the starting level of 60 for men and 55 for women, to be equalised for both sexes at the level of 63 years by 2016 and, according to the legislation, to continue to be increased to 65 by 2026. Registration of individual contributions from 1999 has made it possible to link employment and pension data and observe the trends in retirement with greater precision. We employ a set of analytical indicators tailored to capture the processes of pension entry and employment exit. The study extends the existing research on the impact of pension age reforms in multiple ways. Previous studies on this issue focus predominantly on Western Europe and North America; by adding an Eastern European context, this study contributes to a more comprehensive account of the outcomes of pension age reforms. To illuminate the reform outcomes, the asynchronous increase of statutory pension age among men and women provides us with a quasi-experimental design which divides the older population into exposed and non-exposed parts. The use of longitudinal register data allows us to scrutinise both aggregate and individual-level changes in employment status associated with pension take-up. The alternation of periods of rapid economic growth and recession casts light on the role of varying macroeconomic conditions on retirement behaviour. Finally, we employ analytical indicators that are seldom used in the analysis of retirement. For these reasons, we expect our findings on the impact of pension age reform to be of interest beyond the contribution of a case study.

Post-Communist Economies

499

Structurally, the article consists of seven sections. Following the introduction, the second section provides a brief overview of previous studies on pension age reforms. The third and fourth sections describe the Estonian setting and present research questions. The fifth section explains data sources and methods used in the study. The sixth section presents the empirical results and the concluding section includes a summary and a discussion of the findings.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

Previous research pertaining to pension age reforms Duval (2004) has pointed out that under the standard life-cycle models of retirement, at least in theory the age of eligibility for a pension should not necessarily affect the actual age of retirement as rational forward-looking individuals could set their retirement age at the optimal level to maximise their welfare, by trading off between consumption and leisure/retirement over their life cycle. However, referring to the evidence collected by Gruber and Wise (2002), Duval also argues that actual retirement decisions do not fully conform to standard life-cycle models as these models cannot explain why many workers actually choose to retire at early and standard pension ages. Gruber and Wise (2002) demonstrated that retirement decisions were strongly influenced by social security incentives, in particular eligibility age criteria, and such effects are observed in countries with different histories, labour market institutions and other characteristics. They indicate that while country contexts differ in many respects, employees tend to react rather similarly to social security retirement incentives. Using the concept of ‘hazard rate’ to capture the probability of retiring at any given age, they demonstrated the empirical relationship between hazard rates and pension eligibility ages across countries. Data from Germany and the USA showed that hazard rates were close to zero until the age at which individuals become eligible for an early pension, followed by upsurges at early retirement and statutory pension ages. The authors also observed the effect of pension age reforms on hazard rates in Germany. However, this evidence related to reforms which introduced and widened early retirement possibilities and consequently lowered the effective retirement age. In line with the results of Gruber and Wise (2002), the EU 2012 Ageing Report (European Commission 2011) notes that although pension systems vary markedly across European countries, the age profiles of the probability of retirement reveal a close connection between pension age provisions and retirement decisions. A typical age profile of retirement tends to have spikes/modes at the minimum age for early retirement and the normal (statutory) pension age. Common features also include the skewness of retirement profiles towards the earliest possible age for drawing a pension. Nonetheless, empirical evidence from monitoring reforms that have increased the statutory pension age is still limited, partly due to the relatively short history of such reforms. Whereas both the OECD and the EU (OECD 2011, European Commission 2012b) monitor the development and project the impact of pension reforms on the average exit age from the labour force, only a limited number of studies have used comprehensive register data for a detailed ex post impact assessment of pension age reforms. Among those studies, Staubli and Zweimüller (2013) analysed the impact of raising the pension age on employment of older workers in Austria using register data from the Austrian Social Security Database. In the context of increasing the early retirement age from 60 to 62 for men and from 55 to 58.25 for women, they observed a reduction in the take-up of early retirement (by 19 percentage points among affected men and 22 percentage points among affected women). The reported outcomes also include an

Downloaded by [80.235.88.165] at 01:41 24 November 2015

500

A. Puur et al.

increase in unemployment rate by roughly 10 percentage points among both men and women affected by the reform, negligible effects on disability claims and a sizable reduction in net government pension expenditure. Stiglbauer (2013) calculated the average pension take-up age and its trends in Austria in 2000–11 based on register data, observing a modest overall increase from 57.7 years in 2000 to 58.3 years by 2011. In the USA Song and Manchester (2007) and Kopczuk and Song (2008) used administrative data demonstrating that the increase in the social security retirement age from 65 to 66 caused the spike in claiming at age 65 to shift nearly completely to 66 for the affected cohorts. Similar results were confirmed by Behaghel and Blau (2012) based on survey data. However, Kopczuk and Song (2008) relate this behavioural change not only to the change in pension eligibility age but indicate that Medicare eligibility also has an impact on claiming retirement benefits. There have also been other studies using survey rather than register data. Hanel and Riphahn (2012) used data from the Swiss Labour Force Survey and analysed the 1991 reform of the Swiss mandatory retirement insurance, which increased the normal retirement age for females in two steps from 62 to 64 while opening the possibility of early retirement at the expense of a benefit discount. They observed a strong retirement behaviour response to the shift of the normal retirement age and to changes in financial incentives – about half of those who would have left the labour force at the age of 62 prior to the reform now remained in the labour force. The Austrian, Swiss and US reforms, like the Estonian reform described below, were tied to the year of birth as a fixed individual characteristic. The present study seeks to complement the empirical evidence pertaining to the effects of increasing pension age, drawing on the Estonian context. The Estonian setting The state pension insurance scheme in Estonia has nearly universal coverage as all economically active persons are compulsorily covered through the payment of social tax, while a guaranteed minimum (national) pension is paid to permanent residents.2 It is a contributory scheme financed predominantly from the pension insurance component of the social tax charged on employers while the amount of tax is recorded individually for each employee.3 Beyond the general scheme, which covers all working persons (including the self-employed) there are some special schemes for certain categories of public employees (e.g. police, military). The statutory pension age in Estonia has been increased since 1994, starting from the level of 55 years for women and 60 for men. The pension age of men reached the initial target level of 63 in 2001, while the pension age of women continues to be increased to be equalised with the pension age of men at the level of 63 by 2016. Accordingly, there is by now experience of nearly 20 years of gradual increase in pension age. Effectively, the pension age is defined separately for each birth cohort of men and women. The scale of increase in the pension age over the period covered by the current study is illustrated in Table 1. Beyond 2016, when the pension age of both sexes will be equalised, a gradual increase in pension age for both men and women will continue to the level of 65 years by 2026. From 2000 an option of an early old-age pension up to three years before the statutory pension age was enacted with a reduction in pension of 0.4% for each month remaining to the normal pension age. This ‘general’ early old-age pension supplemented the categorical rights to early old-age pensions for specific occupational groups that

Post-Communist Economies Table 1.

Schedule of increasing the statutory pension age for cohorts born 1938–53. Pension age

Downloaded by [80.235.88.165] at 01:41 24 November 2015

Year of birth 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953

501

Year of reaching statutory pension age

Women

Men

55 55 y 6 m/ 56 56 y 6 m 57 57 y 6 m 58 58 y 6 m 59 59 y 6 m 60 60 y 6 m 61 61 y 6 m 62 62 y 6 m 63

63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63

Women 1993 1994 1996 1998 1999 2001 2002 2004 2005 2007 2008 2010 2011 2013 2014 2016

second half, 1995 second half, 1997 first half second half, 2000 first half second half, 2003 first half second half, 2006 first half second half, 2009 first half second half, 2012 first half second half, 2015 first half

Men 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Source: State Pension Insurance Act.

were inherited from the Soviet period.4 The qualification period for entitlement to old-age pension is 15 years of payment of contributions.5 Without going into further details of various types of state pensions in Estonia, for the purposes of the current study we have categorised all state pensions into three broad categories:6 (i) normal old-age pensions – pensions payable at or after the statutory pension age; (ii) early old-age pensions – pensions payable before the statutory pension age; (iii) incapacity pensions – pensions payable to persons aged 16 to statutory pension age on the basis of incapacity (disability) status. The second category comprises two sub-categories: early old-age pension up to three years before the statutory pension age and early old-age pension on special conditions. The first is generally available to all workers who have fulfilled the qualification period of 15 years, while the latter applies to special categories of workers with hazardous and hard working conditions and to specified social groups (e.g. one of parents with three or more children). Whereas the statutory pension age has been increased, the conditions of the early old-age pension have remained unchanged over the period covered by this study. The average net replacement rate of the old-age pension over 2000–11 has been in the range of 40–45%, which is at the low end in the European context. The net theoretical replacement rate has been slightly higher, 43–46%, but is still the lowest among the EU-27 (European Commission 2013). It is possible to supplement pension income with earnings from work without restrictions (with the exception of the general early old-age pension payable before the statutory pension age, which is paid only to non-working persons). This feature of the system makes it important to distinguish analytically between pension take-up – the event of starting to receive the pension – and retirement – exit from the labour market.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

502

A. Puur et al.

Employment rates of older workers in Estonia have been relatively high. In 2011 the employment rate in the 55–64 age group was 57.2%, putting Estonia among the EU top five along with Sweden, Germany, Denmark and Finland (Eurostat 2013). Obviously, in contrast to the Scandinavian countries, one of the factors triggering higher employment for older age groups is the relatively low replacement rate, which makes continuous employment even beyond pension age an economic necessity for many older persons. In addition, as indicated by Tiit et al. (2004), the unrestricted possibility to supplement pension income with earnings from further work stimulates employment beyond pension age. However, Tiit et al. (2004) also revealed the existence of markedly diverse groups of older workers – while many continue to work beyond pension age, there is also a significant group of older persons withdrawing from the labour market well before the statutory pension age. Consistent with the high employment rates of older workers, according to Eurostat (2013) the average effective exit age from the labour market in Estonia stands above the EU average – at 63.6 in 2010, compared to the EU27 average of 62.1 – and above the statutory pension age in Estonia in the same year. In contrast, life expectancy at 65 was below the EU27 average in 2011, standing at 17.9 years, with a sharp 5.3 year difference between men and women (14.7 years for men against 20.0 years for women) compared to the EU average of 19.2 (17.5 for men and 21 for women). The number of healthy life years in Estonia is among the lowest in the EU, standing at 54.2 for men and 57.9 for women in 2011, against the EU27 average of 61.8 for men and 62.2 for women. These indicators describe the general context to be kept in mind when assessing the impact of pension age reforms. Research questions In this study we investigate how the policy change – increase in the statutory pension age – is reflected in the pension behaviour of individuals in Estonia in 2002–11. We examine what have been the implications of the rise in the statutory pension age, in the context of macroeconomic changes, on (i) pension take-up and (ii) employment before and after pension take-up. We look at three main categories of pensions: normal old-age pensions, early old-age pensions and incapacity pensions. We suspect that the increase in statutory pension age may have increased the take-up of early old-age and incapacity pensions and thereby cancelled out part of the gains expected from the pension age reform. Our expectations concerning employment before and after pension take-up were mixed. On one hand, the relatively poor health status and low expenditure on active labour market policies may have made keeping a job increasingly difficult for older workers in the context of increasing pension age. However, on the other hand, the low replacement rate of old-age pensions may have given older persons a strong incentive to overcome these difficulties. Data and analytical approach The study is based on individual-level data from administrative registers. The main source of information is the national pension register operated by the Estonian Social Insurance Board (ENSIB). The extract of the pension register used in this study includes episodes of pension payment starting from 2000. For each episode, the extract contains start and end date, type of pension and a few additional characteristics. If arranged in time sequence, the episodes make up longitudinal pension histories of individuals. The

Downloaded by [80.235.88.165] at 01:41 24 November 2015

Post-Communist Economies

503

ENSIB also provided us with individual-level data on monthly pension insurance contributions, including the type and amount of taxable income7. From the analytical point of view, the register data used in the study have important advantages: the large scale of the dataset ensures representation of the entire population, administrative procedures used in collection and processing ensure that the data are detailed and accurate, for instance with start and end dates of episodes precisely recorded by day, while keeping any bias from reporting errors and non-response minimal. The analysis consists of two main stages. At the first stage, we examine trends in the take-up of new old-age and incapacity pensions. Pension take-up is operationalised based on the start date of the earliest pension episode of a particular type. Ideally, this approach would have required complete retrospective pension histories of the individuals but the episodes that began prior to 2000 are not included in the register. To overcome this limitation, we shifted the beginning of the analysis to 2002: we define new pensioners as persons who according to the register were not entitled to pension benefits in 2000–01 and whose entitlement started in 2002 or later.8 This approach leaves us with 217,495 new pensioners in 2002–11 (121,475 old-age pensioners and 96,023 incapacity pensioners). At the second stage we address changes in employment associated with pension take-up. We exploit the longitudinal capacity of our register data to follow, retrospectively and prospectively, the 2002–11 cohorts of new old-age and incapacity pensioners. Using personal identification numbers, individual pension histories and their monthly pension insurance contributions were linked, so that the employment status of each new pensioner can be observed at pre-defined intervals before and after pension take-up. The change in employment rates between these dates provides an indicator of the degree to which exit from employment is contingent on pension take-up. The study focuses on trends in 2002–11 and uses calendar years as the main unit of analysis. To investigate the trends in the intensity and timing of pension take-up, and the changes in employment associated with them, the study employs a set of analytical measures explained in the following sections. Given the difference in statutory pension age between men and women in Estonia, the analysis is gender-specific throughout. Compared to previous analyses based on the micro-data of the ENSIB pension register (Tiit et al. 2004, Aaviksoo et al. 2011), our study extends the analytical use of register data in multiple ways. First, it explicitly focuses on the take-up of new old-age and incapacity pensions. In the life course perspective, this represents a pivotal and in most cases irreversible event by which individuals are transformed from contributors to beneficiaries of the pension system. Second, using the personal identification numbers, the study links pension data to pension insurance contributions, allowing us to capitalise on the longitudinal capacity of the registers and examine the change in employment status associated with pension take-up at the individual level. Finally, we employ analytical measures that are only seldom used in the analysis of retirement patterns. The details pertaining to data and methods are discussed in the following sections. Results Changes in pension take-up Number of new pensioners and intensity of pension take-up Unlike the stock of old-age pensioners, which featured moderate growth throughout the 2000s, the total number of new old-age pensioners exhibits considerable variation by

Downloaded by [80.235.88.165] at 01:41 24 November 2015

504

A. Puur et al.

calendar years. In particular, this holds true for the take-up of normal old-age pensions, with the number of new pensioners ranging from less than 5800 in 2003 to almost 10,000 in 2010 (Table 2). The variation follows a regular pattern with peaks occurring in every third year and smaller numbers of new pensioners in the intermediate years. A closer look reveals that this peculiar pattern is driven by the cohort-specific increase of statutory pension age among women. The peaks occur in the years during which an entire female birth cohort reaches pension age: in 2004 women born in 1945 (statutory pension age 59), in 2007 the 1947 birth cohort (60) and in 2010 the 1949 cohort (61). In the intermediate years, the same mechanism curtails the number of women who are entitled to the normal old-age pension. For instance, the increase in statutory pension age from 59 to 59.5 years in the 1946 birth cohort implies that only about 50% of that generation (women born in January–June) became eligible in 2005. Women born in July–December 1946 reach pension age during the first half of 2006, respectively reducing the number of new old-age pensioners in the previous year. In the following year (2006), the number of new old-age pensioners was curtailed by the rise of pension age in the next cohort. Had it remained unchanged, women born in 1947 would have started to reach pension age after the middle of 2006. However, the postponement of eligibility by another six months (until age 60) in the 1947 birth cohort shifted the take-up until the beginning of 2007. The difference between peak and intermediate years suggests that without any increase in the pension age in 2002–11 the total number of new old-age pensioners would have been 28% higher than actually reported in Table 2. Among women, the estimated reduction amounts to 53%.9 The observed effect will persist until 2026 when the statutory pension age reaches the planned 65 years for both men and women (for cohorts born in 1961 and later). To stimulate deferment of the old-age pension, in 2002 a special provision was introduced that allows individuals to increase their pension amounts by postponing takeup of the old-age pension beyond the statutory pension age. Although the provision is rather generous, with the increment exceeding the actuarially neutral adjustment,10 the measure has not proved popular: in the period covered in the study, the relative number of deferred pensions increased only marginally, from 0.8% of new old-age pensions in 2002 to 1.0% in 2011. Evidently the low income replacement capacity of pensions in combination with relatively modest post-retirement life expectancy makes voluntary postponement attractive only to a small fraction of new old-age pensioners.11 In contrast to deferred pensions, the take-up of early old-age pensions is rather widespread. In the context of pension reform aiming to increase the pension age, an increase in the ratio of early to normal old-age pensions could easily cancel out much of the gain expected from the increase in statutory pension age. Table 2 suggests that concerns about the spread of early old-age pensions are not groundless — on average, in 2002– 11 42% of new old-age pensions were taken up before statutory pension age. At the same time, the number of new early old-age pensioners was relatively stable for most of the period. This stability came to an end with the onset of economic recession from 2008, which was comparatively steep in Estonia.12 The rapidly shrinking employment opportunities generated a marked upsurge in the take-up of early old-age pensions in 2009–10, when a significant number of newly dismissed older workers, who were a few years short of the statutory pension age, resorted to early retirement. The role of macroeconomic conditions is further illuminated by the distinction between the two sub-groups of early old-age pensioners. One sub-group – on average 59% of new early pensioners in 2002–11 – consists of persons who are eligible for an

0.93 0.54 0.39 0.16 0.23 0.56 0.56 0.32 0.24 0.09 0.15 0.39

0.98 0.55 0.43 0.17 0.26 0.57 0.54 0.34 0.20 0.08 0.12 0.39

4226 2262 1964 751 1213 3791

4016 2392 1624 632 992 3714

Source: Authors calculations based on the ENSIB pension register.

age

age

age

age

5912 3537 2375 955 1420 4540

2003

6572 3721 2851 1192 1659 4616

2002

0.92 0.65 0.27 0.13 0.14 0.44

1.03 0.61 0.42 0.17 0.25 0.59

6829 4490 2339 1129 1210 4229

6538 4027 2511 931 1580 4793

2004

0.58 0.32 0.26 0.09 0.17 0.43

0.97 0.56 0.41 0.16 0.25 0.57

4829 2454 2375 817 1558 4182

6062 3517 2545 839 1706 4606

2005

Number of new pensioners and total pension take-up rate, Estonia 2002–11.

Number of new pensioners, men All old-age pensioners Normal old-age pensioners Early old-age pensioners Up to three years before statutory pension On special conditions Incapacity pensioners Number of new pensioners, women All old-age pensioners Normal old-age pensioners Early old-age pensioners Up to three years before statutory pension On special conditions Incapacity pensions Total take-up rate, men All old-age pensions Normal old-age pensions Early old-age pensions Up to three years before statutory pension On special conditions Incapacity pensions Total take-up rate, women All old-age pensions Normal old-age pensions Early old-age pensions Up to three years before statutory pension On special conditions Incapacity pensions

Table 2.

0.62 0.34 0.28 0.10 0.18 0.45

0.97 0.57 0.40 0.16 0.24 0.56

5580 2984 2596 967 1629 4358

5624 3036 2588 884 1704 4568

2006

0.91 0.65 0.26 0.12 0.14 0.43

0.99 0.61 0.38 0.14 0.24 0.55

8114 5650 2464 1156 1308 4098

5484 2918 2566 852 1714 4555

2007

Downloaded by [80.235.88.165] at 01:41 24 November 2015

0.57 0.32 0.25 0.08 0.17 0.43

1.01 0.64 0.37 0.15 0.22 0.57

5266 2904 2362 733 1629 4082

5670 3034 2636 1015 1621 4683

2008

0.64 0.32 0.32 0.13 0.19 0.51

1.20 0.71 0.49 0.24 0.25 0.69

6002 3010 2992 1182 1810 4866

7195 3706 3489 1683 1806 5739

2009

0.94 0.63 0.31 0.17 0.14 0.68

1.11 0.66 0.45 0.24 0.21 0.78

8786 5909 2877 1568 1309 6328

7167 3922 3245 1662 1583 6522

2010

0.54 0.30 0.24 0.09 0.15 0.64

0.99 0.61 0.38 0.18 0.20 0.70

5143 2850 2293 889 1404 5940

6460 3794 2666 1202 1464 5813

2011

Post-Communist Economies 505

Downloaded by [80.235.88.165] at 01:41 24 November 2015

506

A. Puur et al.

early old-age pension under special provisions (length of service in certain occupations, parents who have raised three or more children etc.). The dynamics of this sub-group are determined by eligibility, with hardly any connection to the labour market situation. For the other sub-group – on average 41% of early pensioners in 2002–11 – the opposite is the case. In 2009–10 the take-up of early old-age pensions (up to) three years before the statutory pension age almost doubled (increased 1.85 times) while the number of early pensioners under special conditions even decreased slightly in the same period. Notwithstanding the large-scale fluctuations caused by the cohort-based schedule of increasing the statutory pension age, neither the number nor the proportion of new early old-age pensioners exhibits a long-term shift in favour of early retirement. Following the peak of the economic recession in 2009–10, the take-up of early old-age pensions returned to pre-recession level; in 2011 the proportion of early pensioners among all new old-age pensioners was comparable to 2002–03. Judging by the prevalence of early retirement it is important to underline that the pathways to this are not limited to old-age pensions taken up before the statutory pension age. Working-age persons with chronic health problems that reduce their capacity to work by at least 40% can apply for incapacity pension. Table 2 shows that this possibility is intensively exploited in Estonia. In 2002–11 new incapacity pensions outnumbered early old-age pensions taken up over the same period 1.8 times. The take-up of incapacity pensions closely resembles that of the early old-age pensions discussed above. The relatively stable take-up until 2008 was followed by a sharp increase: in 2009–10 the number of new incapacity pensioners increased by 47%. However, unlike for early old-age pensions, the take-up of incapacity pensions was still about a third above the pre-recession level in 2011. As the health status of the population cannot vary to such an extent over a short period of time, the upsurge indicates that during the economic crisis the disability system provided income support for working-age population pushed out of employment. Indirect evidence backing this assertion can be derived from the disaggregation of incapacity pensions; the upsurge in 2009–10 was almost wholly driven by new pensioners with moderate incapacity (OECD 2012). Finally, Table 2 presents the total take-up rates for old-age and incapacity pensions. The total take-up rates are summary measures of the intensity of pension take-up. These measures are similar to total fertility rates in population statistics, obtained by summing the single-year age-specific take-up rates (not reported in this article) for each calendar year. Unlike the numbers of new pensioners, these measures are not affected by the variation in cohort size. The latter factor should not be overlooked because the generations which reached pension age in 2002–11 were born during the Second World War and its aftermath (1939–50). Despite large fluctuations in the size of these generations, the evidence drawn from the take-up rates generally corroborates the findings reported earlier in this section. Only with regard to gender differences do the take-up rates yield a clearly more precise account of the intensity of the take-up process. For men, the take-up rate of old-age pension averages 1.02 in 2002–11. This result fits rather well the interpretation according to which the take-up rate reflects the number of specific pension events an average person in a synthetic cohort would have during his/her lifetime if he/she were subject to the age-specific take-up rates characteristic of a given calendar period. For women, the take-up rate of the old-age pension averages 0.68, about 33% lower than the men’s level observed in the same period. The scale of gender difference in the take-up rates markedly exceeds the difference observed in the number of new old-age pensioners. The small difference in the latter – men

Downloaded by [80.235.88.165] at 01:41 24 November 2015

Post-Communist Economies

507

outnumbered women by only 6% among new old-age pensioners – does not reflect the intensity of pension take-up but the differential survival rates of men and women, which biases the numbers in favour of the latter. Earlier in this section we pointed out that among women the number of new old-age pensioners has been significantly depressed by the increase in statutory pension age. Using the take-up rates, we can measure this effect with greater precision. For that purpose we applied a method developed by Bongaarts and Feeney (1998) for the adjustment of period fertility measures in the context of delayed childbearing. Following this framework, the observed take-up rates should be adjusted for the change in pension age. In Estonia the statutory pension age of women increased at the rate of 3.6 months per annum in 2002–11. Applying the Bongaarts–Feeney method, we obtained the timeadjusted take-up rate of the old-age pension for women. The difference between the time-adjusted (0.97) and the observed take-up rate (0.68) yields an estimate of the effect of increasing the statutory pension age on the take-up rate. To sum up, the take-up rates provide us with a more refined account of the intensity of the process but as the main findings pertaining to the trends in the pension take-up remain unaltered, we omit discussion of these details. Timing of pension take-up In pension systems with near universal coverage, timing becomes a central characteristic of the take-up process. In tandem with longevity, the age at which pensions are taken determines the duration of pension payment. Owing to the considerable spread of early pensions, shifts in the timing of pension take-up cannot be directly inferred from the dynamics of statutory pension age but must be empirically estimated. Table 3 shows the mean age at take-up of old-age and incapacity pensions in Estonia in 2002–11.13 The values are directly calculated from the pension register, based on the age of new pensioners at the time of take-up. Compared to the alternative method based on age-specific take-up rates, the mean take-up age method allows us to achieve greater

Table 3.

Mean age at pension take-up, Estonia 2002–11. 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Men (years) All old-age pensions Normal old-age pensions Early old-age pensions Up to three years before statutory pension age On special conditions Incapacity pensions Women (years) All old-age pensions Normal old-age pensions Early old-age pensions Up to three years before statutory pension age On special conditions Incapacity pensions

61.1 63.0 58.5 60.6

61.2 63.0 58.4 60.6

61.2 63.0 58.2 60.5

61.0 63.1 58.1 60.5

60.8 63.1 58.0 60.5

60.7 63.1 58.0 60.4

61.0 63.1 58.5 60.6

61.1 63.1 59.1 60.6

61.3 63.0 59.2 60.7

61.3 63.1 58.9 60.5

57.1 46.1

57.0 45.8

56.8 45.9

57.0 45.9

56.8 45.8

56.8 46.1

57.2 45.8

57.7 45.4

57.6 45.1

57.6 45.0

57.6 58.5 56.3 56.9

57.7 58.5 56.7 57.0

58.3 59.0 56.8 57.3

58.3 59.5 57.0 57.8

58.6 59.5 57.6 58.0

59.3 60.0 57.7 58.3

59.4 60.5 57.9 58.8

59.7 60.6 58.8 59.2

60.3 61.0 58.9 59.4

60.4 61.5 59.0 59.9

55.9 45.2

56.5 45.6

56.4 45.8

56.6 46.5

57.4 46.9

57.3 46.5

57.5 46.8

58.6 46.3

58.3 46.4

58.5 47.0

Source: Authors calculations based on the ENSIB pension register.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

508

A. Puur et al.

accuracy as the age at take-up was recorded with precision to a day in the register data.14 Sensitivity tests showed that our substantive results are not significantly affected by the choice of method. As regards normal old-age pensions, the mean age at take-up closely follows the statutory pension age. Among men, the mean age at take-up of normal old-age pensions is stable (63.0–63.1 years) throughout the period of study. Among women, the mean age at take-up increased by three years, from 58.5 in 2002 to 61.5 in 2011. The results show that, in addition to fluctuations in the number of new old-age pensioners, the cohort-specific increase of statutory pension age produces a stair-like trendline for women. The mean age at take-up normally increased 0.5 years per annum but in every third year (2003, 2006 and 2009) the upward trend temporarily ceased. The take-up of early old-age pension occurs at noticeably younger ages. In 2002–11 men took an early old-age pension on average at the age of 59.1 years, i.e. 4.5 years before the take-up of the normal old-age pension. Among women, the gap between the take-up of normal and early old-age pension is smaller; in the period covered in the study, the difference averaged 3.4 years. Closer scrutiny of the evidence suggests that the observed gender difference does not boil down to a compositional effect caused by varying proportions between the sub-groups of early pensions among men and women. Figures 1 and 2 illustrate the age-specific take-up rates for men and women respectively in 2001, 2004, 2007 and 2010.15 For men the take-up of general early old-age pensions (up to three years before pension age) declined by 9 percentage points by 2007 as compared to 2001 due to a decline in unemployment during the years of high economic growth. In contrast, the take-up of pensions at the standard pension age of 63 increased by 9 percentage points by 2010 compared to 2001. At the same time the increase in pension age for women has drawn the take-up of pensions towards higher ages on a par with shifts in legal entitlements while the age-specific take-up rates of different sub-types of pensions for successive birth cohorts have been rather stable. If the labour market is not sufficiently flexible to absorb an increased supply of older workers, the increase in statutory pension age may result in a widening gap between the statutory pension age and the mean age of drawing an early pension. The evidence presented in Table 3 indicates that this negative scenario has not materialised in Estonia. The same phenomenon is illustrated in Figure 3. Among women, the increase in statutory pension age has drawn the take-up of early old-age pensions

700 normal old-age pension 600 500 2001 400

general early old-age pensions

300

2004

special early old-age pensions

200

2007 2010

100 0 50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Figure 1. Age-specific take-up rates of old-age pensions of men in 2001–10, per 1000 persons. Source: Authors’ calculations based on ENSIB pension register.

Post-Communist Economies

509

700 normal old-age pension 600 500 2001 400

general early old-age pensions

300

2004 2007

200

2010

special early old-age pensions

100 0 50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

Downloaded by [80.235.88.165] at 01:41 24 November 2015

Figure 2. Age-specific take-up rates of old-age pensions of women in 2001–10, per 1000 persons. Source: Authors’ calculations based on ENSIB pension register.

proportionally towards a later age. Among men, the economic recession has curtailed rather than widened the age gap between the take-up of early and normal old-age pensions. The distinction between the two sub-groups of early pensioners suggests that this unexpected shift is driven by somewhat later take-up of early old-age pensions on special conditions after 2008. It seems that men who were eligible for certain types of early old-age pensions preferred to stay longer in their previous jobs, probably due to shrinking opportunities of alternative employment caused by the recession.16 The overall timing of old-age pensions (early and normal old-age pensions aggregated) almost perfectly follows the rise in statutory pension age. Among women, the mean age at take-up of the old-age pension increased from 57.6 years in 2002 to 60.4 years in 2011, i.e. only 0.15 years (5%) less than the increase in statutory pension age in the same period. Among men, the mean age at take-up of old-age pension did not undergo any major change; in 2011 men took their old-age pension on average at the age of 61.3 years (61.1 in 2002). This implies that by the end of the observation period the gender difference in the age at which old-age pensions are taken was reduced to 0.9

64 63 62

2 years Normal old-age pensions, men

61 Normal old-age pensions, women

60

All old-age pensions, men 59

0.9 years All old-age pensions, women

58 57 56 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Figure 3. Timing of pension take-up: normal versus early old-age pensions, men and women 2002–11, mean values. Source: Authors’ calculations based on ENSIB pension register.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

510

A. Puur et al.

years in Estonia. Hence, with regard to timing, early old-age pensions play a more salient role among men, in part reflecting their wider access to old-age pensions on special conditions. Against the backdrop of early old-age pensions, the take-up of incapacity pensions occurs at an even younger age. In 2002–11 men and women took incapacity pensions on average at the ages of 45.7 and 46.3 respectively. Assuming that entitlement is not withdrawn after take-up (on account of recovery from incapacity), this would translate into about 17 years of benefit payment. Although incapacity benefits are usually smaller than old-age pensions, the duration of payment lucidly illuminates the burden that disability support may impose on the pension system over the longer run. 17 With regard to trends, Table 3 shows that the change in the timing of incapacity pensions varies between genders. Among women, the mean age at take-up increased from 45.2 years in 2002 to nearly 47 years in 2006–08. At the present stage, it is difficult to establish whether the observed increase reflects an improvement in the health status of women or is driven by the rise in statutory pension age. However, the relative stability of the mean age at take-up of incapacity pensions among men until 2008 lends some support to the latter possibility. Unlike for early old-age pensions, the recession shifted the timing of incapacity pensions to an earlier age. Among men, the mean age at take-up decreased from 46.1 years in 2007 to 45.0 years in 2011; among women, the recession temporarily halted the previous upward trend. In 2011 men took incapacity pension on average two years earlier than was the case for women. Finally, the aggregation of old-age and incapacity pensions (not shown in Table 3) reduces the increase in the timing of pension take-up to nil. The increase in the mean age at take-up of new old-age and incapacity pensions among women (from 51.3 years in 2002 to 51.9 years in 2011) falls short of outweighing the concurrent decrease for men (from 53.7 to 51.8 years). This means that, notwithstanding the increase in statutory pension age, in 2011 the take-up of old-age and incapacity pensions occurred on average at an earlier age in Estonia than in 2002. Employment before and after pension take-up Table 4 presents employment rates calculated from the linked register data, with time reference 12 months prior to and 12 months after pension take-up.18 The data reveal two distinct periods in the dynamics of employment rates associated with pension takeup. Among men, in 2002–07 the employment rate 12 months prior to take-up of the normal old-age pension increased by 5.1 percentage points, from 59.7% to 64.8%. For women, the increase exceeded 2 percentage points, from 71.2% to 73.2%. Likewise, a shift toward higher levels was characteristic of employment following take-up of the old-age pension. In 2002–07 the post-pension employment rate increased by 10.6 percentage points for men and 5.0 percentage points for women. In 2008 the global economic recession reversed the preceding trend. By 2010 employment rates prior to pension take-up had decreased by 3.8 percentage points among men and 5.8 percentage points among women. The reduction in post-pension employment rates proved more extensive, amounting to 11.5 and 10.9 percentage points respectively. In 2011 the postpension employment rates exhibited the first signs of recovery but its extent is no match for the reduction that took place at the height of the recession.19 The difference between pre- and post-pension employment rates casts light on the degree to which exit from employment is contingent on old-age pension take-up, and how the association between the two dimensions of retirement transition has altered in

Post-Communist Economies Table 4.

511

Employment rate before and after take-up of pension, Estonia 2002–11.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 12 months before take-up, men (%) All old-age pensions 57.3 59.8 Normal old-age pensions 60.3 59.7 Early old-age pensions 53.3 59.9 Up to three years before 32.0 41.2 statutory pension age On special conditions 68.7 72.5 Incapacity pensions 44.0 48.4 12 months before take-up, women (%) All old-age pensions 63.5 66.6 Normal old-age pensions 69.0 71.2 Early old-age pensions 55.5 61.3 Up to three years before 37.2 47.0 statutory pension age On special conditions 67.2 70.2 Incapacity pensions 51.6 57.7 12 months after take-up, men (%) All old-age pensions 37.4 38.4 Normal old-age pensions 43.3 42.7 Early old-age pensions 29.7 32.0 Up to three years before 1.6 1.2 statutory pension age On special conditions 50.0 52.7 Incapacity pensions 22.1 25.4 12 months after take-up, women (%) All old-age pensions 44.6 44.4 Normal old-age pensions 53.8 54.6 Early old-age pensions 31.0 32.7 Up to three years before 0.6 1.2 statutory pension age On special conditions 50.4 52.2 Incapacity pensions 31.6 35.0

60.9 62.4 58.5 34.4

61.9 61.5 62.6 36.7

62.2 62.6 61.7 34.7

62.5 63.0 61.8 33.2

65.3 63.4 67.6 48.2

64.8 64.9 64.7 49.6

63.7 64.8 62.3 52.8

58.8 61.1 55.4 37.4

72.7 45.8

75.3 46.5

75.8 49.1

76.1 51.9

79.7 59.2

78.7 55.9

72.3 51.1

70.1 47.1

65.3 70.1 56.1 41.0

66.3 70.5 62.1 48.8

68.2 72.1 63.8 52.1

67.3 70.7 59.4 46.6

71.8 73.2 70.0 59.2

71.1 72.3 69.9 62.6

69.4 72.1 63.9 58.5

64.9 67.5 61.7 50.4

70.2 57.7

69.1 57.3

70.7 61.1

70.7 62.7

74.9 69.4

74.6 67.0

70.4 64.5

68.9 63.4

42.3 45.8 36.6 2.9

43.7 46.7 39.5 1.3

45.8 49.4 41.6 1.7

45.1 48.2 41.5 1.8

37.6 40.6 34.2 1.1

31.8 37.9 25.4 1.5

33.5 41.1 24.3 1.7

NA NA NA NA

56.5 27.9

58.3 30.6

62.3 34.5

61.3 33.8

54.9 32.6

47.6 29.5

48.1 32.2

NA NA

46.4 55.4 29.1 3.0

47.5 57.7 37.0 2.1

47.8 58.0 36.2 1.0

48.7 56.7 30.2 1.4

46.6 51.9 40.0 2.0

39.4 47.1 31.7 2.4

40.9 49.5 23.4 1.2

NA NA NA NA

53.4 41.6

55.3 41.1

57.0 48.5

55.7 49.4

57.1 47.1

50.8 43.0

50.0 44.7

NA NA

Source: Authors’ calculations based on ENSIB pension register.

the period covered by the study. In general, the relatively small difference between preand post-pension employment rates indicates that the association between pension take-up and exit from employment is relatively weak in Estonia. The evidence presented in Table 4 also suggests that the association between these dimensions is strongly influenced by economic circumstances. In the period of rapid economic growth, exit from employment became increasingly detached from pension take-up. The difference between pre- and post-pension employment rate decreased to 13.2 percentage points among men and 14 percentage points among women. In contrast, during the recession, the difference doubled, driven by excessive decrease in employment rates in the post-pension stage. In the context of pension age reform, the central finding in this section pertains to the dynamics of female employment. Although the increase in women’s employment rates prior to the recession was less extensive than that of men, the rise in statutory pension age did not prevent an upward shift in female pre- and post-pension employment rates. The rise in pension age precipitated neither an expansion in pre-term withdrawal from employment nor a reduction in take-up of employment in the post-pension stage.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

512

A. Puur et al.

As a result of these shifts, the gender difference in employment and labour force participation rates reversed. In Estonia women feature systematically higher pre- and post-pension employment rates than men. The evidence based on the labour force survey corroborates the reversal.20 With regard to early old-age pensions, employment patterns differ markedly across sub-groups of early pensioners. Older persons who resort to early retirement because of difficulties in keeping or finding a job feature systematically lower employment rates than retired people who opt for a normal old-age pension. In that sub-group the postpension employment rate does not exceed 1–2%. In contrast, persons who took an early old-age pension on special conditions exhibit a strong attachment to the labour market, with employment rates before and after take-up systematically exceeding those characteristic of normal old-age pensioners. Among new incapacity pensioners, employment rates appear lower than those for old-age pensioners. The trends in employment rates before and after take-up of early old-age and incapacity pensions generally conform to the pattern reported for normal old-age pensioners earlier in this section. Summary and discussion In this study we investigated the change in retirement behaviour in Estonia in the 2000s and, particularly, how the increase in statutory pension age has affected pension take-up and the employment dynamics surrounding the latter, based on data from administrative registers. In comparative perspective, Estonia offers a particularly interesting context for analysing these issues. The reform package adopted early in the 1990s stipulated the increase of statutory pension age from 60 (men) and 55 (women) to 63 years. The rise, aimed at reducing the fiscal burden on the pension system, was scheduled for the period 1994–2016, so that in each birth cohort the age would increase by 0.5 years. The asynchronous schedule of the rise – among men the target was already reached in 2001 while women will attain it only in 2016 – provided us with a quasi-experiment that divides the population into exposed and non-exposed parts. In addition, the alternation of rapid economic growth and recession allows us to cast light on the impact of varying macroeconomic conditions on retirement outcomes. The results show that the increase in statutory pension age significantly reduced the take-up of new old-age pensions. Among women, who were affected by the increase in our study period, the estimated reduction amounts to slightly over 50% of the new oldage pensioners in 2002–11. We suspected that the increase in entitlement age to normal old-age pensions might have increased the take-up of early retirement and thereby cancelled out part of the gains expected from the reform. However, the results do not confirm these concerns. Neither the number of early old-age pensioners nor their proportion among the pool of new old-age pensioners reveals a persistent shift towards early retirement. Similarly, our results with regard to the timing of take-up show no widening of the age gap between normal and early old-age pensions. The increase in statutory pension age has shifted the take-up of early old-age pensions almost proportionally towards a later age. Our expectations concerning employment before and after old-age pension take-up were mixed. On one hand, the relatively poor health status of older people, low expenditure on active labour market policies and considerable skill mismatch (Jagger et al. 2008, OECD 2012) could have made keeping a job increasingly difficult for older workers in the context of increasing pension age. On the other hand, the relatively high

Downloaded by [80.235.88.165] at 01:41 24 November 2015

Post-Communist Economies

513

employment rates among older age groups suggest that other factors may have effectively outweighed these difficulties. Our results lend greater support to the latter assertion – although the increase in women’s employment was less extensive than that for men, a significant rise in statutory pension age among women precipitated neither an expansion in pre-term exit from employment nor a decrease in propensity to remain employed in post-pension age. Moreover, it seems likely that the postponement of entitlement itself made a contribution to the rise in employment among older age groups and is partly responsible for the country’s high ranking in this respect in international comparisons. Retirement patterns closely corroborate these findings – the analysis disclosed no explicit negative effects that could be associated with the increase in statutory pension age. Several factors may have contributed to the absence of adverse effects of the pension age reform. First, favourable macroeconomic conditions until 2008 should be included among the plausible correlates of the observed outcome. The salient role of economic context is demonstrated in multiple changes after 2008, including the upsurge in the take-up of early old-age and incapacity pensions, fall in employment rates prior to and after pension take-up and concurrent shifts in retirement patterns. The second factor relates to women’s strong attachment to the labour force, which is instrumental in sustaining high employment rates before and after pension age. In terms of female employment, Estonia has ranked high in international comparisons ever since the 1960s (Puur 2000, ESA 2006). Unlike in many former state socialist countries, the scale of employment decline during the transition to a market economy was similar among men and women. On the eve of the 2008 economic recession the full-time equivalent employment rate of Estonian women was the highest of all EU member states (European Commission 2009). Third, modest income support for non-employed persons of working age and the relatively low income replacement capacity of old-age pensions, coupled with the possibility to claim a full old-age pension while employed, can be seen as strong incentives not to leave employment before entitlement to old-age pension and if possible, to continue working for some years beyond that. Reflecting a social policy stance that is focused on self-responsibility and work incentives, in comparative perspective Estonia stands out for relatively low social spending directed at the working-age population (OECD 2012). The modest income-replacement capacity of old-age pensions becomes particularly evident in comparison with the old EU member states. At the beginning of the period covered in this study, the theoretical net replacement rates of old-age pensions in the EU-15 countries started from 66% while in Estonia it was only 43% (Tiit et al. 2004).21 By 2010 the net theoretical replacement rate in Estonia had increased to 46% but remained by far the lowest among the enlarged EU-27 (European Commission 2013). Finally, the evidence from the Population Policy Acceptance survey suggests that in Estonia the relatively late pension age has been internalised in the expectations of the population (Jozwiak et al. 2008). Against the background of successful adjustment to the rise in statutory pension age, the findings draw attention to the concurrent expansion in the take-up of incapacity pensions, which offers an alternative pathway to early retirement. In the period covered in the study, new incapacity pensions outnumbered new early old-age pensions 1.8 times. Moreover, the relatively stable take-up of incapacity pensions until 2008 was followed by a sharp increase, nearly 50%, in 2009–10. As the health status of the population cannot undergo such a large fluctuation in a matter of a few years, our results lend support to the view that, with limited availability of income support for the working-age

Downloaded by [80.235.88.165] at 01:41 24 November 2015

514

A. Puur et al.

population, the disability system seems to have increasingly served this purpose. Not surprisingly, the intensive take-up of incapacity pensions in Estonia catches the eye in international comparisons. Among the OECD countries, Estonia ranks first for increase in disability-related entitlements (OECD 2012). Several considerations make the observed large-scale increase in the take-up of incapacity pensions highly undesirable. As shown in this article, the proliferation of incapacity pensions has become so extensive that it threatens to effectively cancel out the expected effects of the pension age reform. No less importantly, income support through incapacity pensions implies high long-term costs and, moreover, lacks the activation role that is essential for enhancing the earnings potential of benefit recipients. An important policy implication of our findings therefore underlines the need for a holistic approach to pension system and social protection at large. Optimisation in specific policy sectors should not lose a broader view and needs to consider side-effects that entail costs in related sectors. This lesson has been learnt in Estonia and reforms of the work capacity benefit and rehabilitation systems are under way. Finally, although the results demonstrate the usefulness of the Estonian pension register for the analysis of retirement patterns, a few limitations of this study should be mentioned. First, the lack of a broader range of socio-demographic variables in the dataset prohibited investigation of how pension take-up and exit from employment vary among different sub-groups of the population, aside from men and women. In future research, bringing in information from other sources could provide interesting insights into the factors that shape retirement decisions in population groups with different resources as well as potentially different preferences with regard to retirement. Acknowledgement This article has benefited from support for research theme SF0130018s11 by the Estonian Ministry of Education and Science and the Estonian Science Foundation grant no. 8904. The authors would like to thank Andres Roostalu from the IT-department of the Estonian National Social Insurance Board for preparing the register extract used in this study and for consultation.

Notes 1.

2. 3. 4.

5. 6. 7.

For example, in the Czech Republic following a reform legislated in 2011 the pension age of men and women will be gradually equalised by 2044 and thereafter will be further increased by two months every year for both men and women without any predetermined limit (Loužek 2014). For a more detailed account of the Estonian pension system and its evolution, see for instance Leppik (2006). The social tax, being an ear-marked tax to finance pension and health insurance, while the payment of this tax builds entitlements to pension and health care, may be called a social insurance contribution. A crucial difference between the conditions of the two types of early retirement pension is that the ‘general’ early old-age pension cannot be supplemented with earnings from work before the statutory pension age, whereas the ‘special’ early old-age pension for listed occupational groups can be supplemented with earnings even before the statutory pension age provided the person switched employment to a non-listed occupation. For entitlement periods prior to 1999 there were some non-contributory equalised periods (e.g. periods of university studies), whereas from 1999 onwards entitlement periods are directly linked to the payment of contributions. Survivors’ pensions payable to persons below pension age are not included in this analysis. A register-based system for collecting individualised data on pension insurance contributions was introduced in 1999 in Estonia. This makes 2000 the earliest year for which the

Post-Communist Economies

8.

9. 10. 11.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

12.

13. 14. 15. 16. 17. 18.

19. 20.

21.

515

micro-data on pension insurance contributions are available. Data availability considerations were central in defining the timeframe of the study. The near-continuous receipt of pension after take-up and the exceptionality of pension episodes with a duration exceeding one year imply a very small bias from left-censoring. In our data, only 1.3% of pensioners had gaps between successive pension episodes. The gaps were typically not characteristic of old-age pensioners. The reduction is not precisely 50% owing to variation in the size of birth cohorts reaching retirement age. Deferment is rewarded by a 0.9% increase in pension benefit per month deferred beyond statutory pension age. Because of their low incidence, the deferred old-age pensions are not shown separately but merged with normal old-age pensions in Table 2. In 2002–07 GDP growth varied between 6.3% and 10.1% in Estonia. In 2008 GDP began to decrease (-4.2%) which accelerated in 2009 (-14.1%). In 2010–11 economic growth resumed (growth rates of 3.3% and 8.3% respectively) but the GDP level had not fully recovered to the pre-recession level by the end of our observation period (ESA 2013). Busemeyer (2006) and Stiglbauer (2013) use the term ‘pension entry age’ to denote the same measure. The alternative calculation method based on age-specific rates relies on specific assumptions about the distribution of events within single-year age intervals which may not always correspond to reality. These were years when a full cohort of women reached pension age. Intermediate years have been omitted as only half of the cohort reached pension age then. For instance, this applies to employees in the law enforcement and military sectors. They can claim an early old-age pension on special conditions only if they leave their work in these sectors. In 2011 the average incapacity pension amounted to 59% of the average old-age pension. To validate the register-based employment data, we compared the number of employed estimated from pension insurance contributions to the estimates based on the labour force survey. In 2002–11 the average net discrepancy between the two series of estimates did not exceed 1.3%. Age-specific employment rates calculated from the two sources appear very similar in pre- as well as post-pension age groups. As the register data used in this study were extracted in 2012, employment rates 12 months after take-up were not available for new pensions that were taken up in the latter year. In the age group 55–59 women reached parity with men in terms of employment rates in 2004. In 2011 the employment rate of 55–59-year old women exceeded that of men by 6.4 percentage points. The female excess in labour force participation rates first appeared in 2009; it amounted to 2.4 percentage points in 2011. In the age group 60–64 women closed much of the gender gap and it seems quite plausible that by 2016, when women catch up with men for statutory pension age, the parity, if not reversal, will extend to 60–64-year olds (ESA 2013). Theoretical replacement rate is calculated for a hypothetical person who has worked 40 years and earned average income, participates in the most common pension scheme and retires at age 65.

References Aaviksoo, A. et al., 2011. Eesti sotsiaalkindlustussüsteemi jätkusuutliku rahastamise võimalused [Options for sustainable financing of the Estonian social security system]. Tallinn: PRAXIS Center for Policy Studies. Barr, N. and Diamond, P., 2010. Pension reform: a short guide. New York and Oxford: Oxford University Press. Behaghel, L. and Blau, D.M., 2012. Framing social security reform: behavioral responses to changes in the full retirement age. American economic journal-economic policy, 4 (4). Bongaarts, J. and Feeney, G., 1998. On the quantum and tempo of fertility. Population and development review, 24 (2), 271–291. Busemeyer, R.E., 2006. Moving the unmovable: political strategies of pension reform in Germany. German policy studies, 3 (3), 400–445.

Downloaded by [80.235.88.165] at 01:41 24 November 2015

516

A. Puur et al.

Duval, R., 2004. Retirement behavior in OECD countries: impact of old-age pension schemes and other social transfer programmes. OECD economic studies, no. 37, 2003/2. ESA, 2006. Women and men. Tallinn: Statistics Estonia. ESA, 2013. Statistics Estonia. Statistical database [online]. Available from: http://pub.stat.ee/pxweb.2001/dialog/statfile1.asp [Accessed 20 November 2013]. European Commission, 2009. Employment in Europe 2009. Luxembourg: Office for Official Publications of the European Communities. European Commission, 2011. The 2012 ageing report: underlying assumptions and projection methodologies. European Economy 4/2011. European Commission, 2012a. White paper: an agenda for adequate, safe and sustainable pensions [online]. Available from: http://ec.europa.eu/social/BlobServlet?docId=7341&langId=en. European Commission, 2012b. The 2012 ageing report: economic and budgetary projections for the EU27 member states (2010–2060). European economy 2/2012. European Commission, 2013. Adequacy and sustainability of pensions [online]. Available from: http://ec.europa.eu/europe2020/pdf/themes/04_pensions.pdf. Eurostat, 2013. Statistics database [online]. Available from: http://epp.eurostat.ec.europa.eu [Accessed 20 November 2013]. Gruber, J. and Wise, D.A., 2002. Social security programs and retirement around the world: micro estimation. NBER working paper no. 9407. Hanel, B. and Riphahn, R. T., 2012. The timing of retirement – new evidence from Swiss female workers. Labour economics 19, 718–728. Jagger, C. et al., 2008. Inequalities in healthy life years in the 25 countries of the European Union in 2005: a cross-sectional meta-regression analysis. Lancet 372, 2124–2131. Jozwiak, J., Kotowska, I. and Abramowska, A., 2008. Activating older workers: policies versus opinions and expectations. In: C. Höhn, D. Avramov and I.E. Kotowska, eds. People, population change and policies, 2. Berlin: Springer, 221–243. Kopczuk, W. and Song, J., 2008. Stylized facts and incentive effects related to claiming of retirement benefits based on social security administration data. Michigan Retirement Research Center working paper 2008–200. Leppik, L., 2006. Transformation of the Estonian pension system: policy choices and policy outcomes. Tallinn: Tallinn University Press. Loužek, M., 2014. Pension reform in the Czech Republic after 2010. Post-communist economies, 26 (1), 89–102. OECD, 2006. Sickness, disability and work: breaking the barriers (vol. 1). Norway, Poland and Switzerland. Paris: OECD Publishing. OECD, 2011. Trends in retirement and in working at older ages. In: Pensions at a glance 2011: retirement-income systems in OECD and G20 countries. Paris: OECD Publishing. OECD, 2012. OECD economic surveys. Estonia. Paris: OECD Publishing. Puur, A., 2000. Female labour force participation during the economic transition: the case of Estonia. RU Series B No 44. Tallinn: Estonian Interuniversity Population Research Centre. Song, J. and Manchester, J., 2007. Have people delayed claiming social security benefits? Responses to changes in social security rules. Social security bulletin, 67 (2). Staubli, S. and Zweimüller, J., 2013. Does raising the early retirement age increase employment of older workers? Journal of public economics, 108, 17–32. Stiglbauer, A., 2013. Effective retirement age in Austria – a review of changes since 2000. Monetary policy and the economy, Q1/13 [online]. Available from: http://www.oenb.at/de/ img/mop_2013_q1_analyses3_tcm14-254584.pdf. Tiit, E.-M. et al., 2004. Euroopa Liidu ühiste pensionieesmärkide mõju Eesti pensionisüsteemile [The impact of EU pension objectives on the Estonian pension system]. Praxise Toimetised Nr. 14.