Reversing Early Retirement in Advanced Welfare Economies A Paradigm Shift to Overcome Push and Pull Factors Bernhard Ebbinghaus, Dirk Hofäcker Abstract: Recent reform efforts of advanced welfare states have attempted to re- verse trends in early retirement and increase the statutory retirement age. This paradigm shift often occurred against the protest of unions, fi rms and their employ- ees. As a consequence of expanding welfare states and as response to economic challenges since the 1970s early exit from work has become a widespread practice. Early retirement has been part of Continental Europe’s welfare without work prob- lem, while the Scandinavian welfare states, the Anglophone liberal economies and the Japanese welfare society were able to maintain higher levels of employment for older workers. Since the 1990s, an international consensus to reverse early exit from work emerged among international organisations and national policy experts. Based on a comparative historical analysis of selected OECD countries, this study analyses the cross-national variations in the institutionalisation of early exit regimes and its recent reversal using macro-indictors on early exit trends and stylised in- formation on institutional arrangements. Comparing the interaction of social policy and economic institutions, it reviews the cross-national differences in welfare state “pull” and economic “push” factors that have contributed to early exit from work and discusses the likely impact of welfare retrenchment and assesses the importance of “retention” factors such as activation policies for decreasing early exit from work. Keywords: Early retirement · Activation · Reforms · Labour market policy · Welfare state · Pension · Retirement · Social policy 1 Introduction Although living longer, workers tend to retire earlier than in the past. Today’s wel- fare states are facing the dual challenges of ageing societies and early retirement. Both trends have boosted public expenditure on old-age and disability pensions as well as other passive employment policies. The post-war expansion of pension systems in advanced economies has led to the institutionalisation of retirement as Comparative Population Studies – Zeitschrift für Bevölkerungswissenschaft Vol. 38, 4 (2013): 807-840 (Date of release: 31.12.2013) © Federal Institute for Population Research 2013 URL: www.comparativepopulationstudies.de DOI: 10.12765/CPoS-2013-24en URN: urn:nbn:de:bib-cpos-2013-24en2 • Bernhard Ebbinghaus, Dirk Hofäcker808 full exit from work at a statutory pension age, a major transition in the tripartite life course (Kohli 1987). Since the fi rst oil shock in the early 1970s, early exit from work has become a widespread practice in most advanced economies in Europe and overseas (Ebbinghaus 2006). From an economic point of view, it has been argued that the welfare state provides disincentives to continue working beyond the earli- est possible retirement age, and favourable pre-retirement arrangements have pro- vided a “pull” on early exit from work (Gruber/Wise 1999). There have been multiple “pathways” to retirement via pre-retirement, disability or unemployment benefi ts as well as company policies (Kohli 1987). Moreover, early retirement was also a reaction to economic “push” factors, in particular mass unemployment and economic restructuring. Politicians, employ- ers and unions sought early retirement as passive labour market policies and as a socially acceptable strategy to alleviate the labour market and to externalise re- structuring costs on public social protection (Naschold/de Vroom 1994). However, the alleged solution subsequently became criticised as the “welfare without work” problem (Esping-Andersen 1996) or “Continental dilemma” (Scharpf 2001): Shed- ding labour via early exit from work in response to unemployment increased pas- sive social expenditure, this in turn increased labour costs and thus pushed even more employees out of work. Following the activation paradigm shift, not least as a consequence of the OECD’s Job Study in 1994 as well as the EU’s 2010 Lisbon Strategy and the revised “Europe 2020” targets, politicians now seek measures to increase employment among older workers and delay the exit from work. In our macro-comparative study, we seek to provide a long-term analysis of the scope of early exit from work and the more recent reversal of early retirement across several welfare states in Europe, USA and Japan. Our main two dependent variables are (1) the level of early exit from work for men and women, and (2) the change in early exit over the last decades since its previous peak. Based on these two indica- tors we discuss the ranking of different early exit regimes in an “ordinal compari- son” (Ebbinghaus 2008; Mahoney 2003), comparing cross- and intra-regime differ- ences. To explain the observable cross-national variations of early exit from work and its reversal trend, we focus on macro-level institutional variations and structural differences in nation-specifi c welfare state arrangements and market regulations. Although the decision to retire or to continue working is the result of an indi- vidual decision, it is nevertheless taken under given institutional opportunities (e.g. a preretirement pension for older unemployed) and structural constraints (e.g. eco- nomic pressures on a company to downsize in a socially acceptable way). There are different research strategies to study this phenomenon. Some empirical analyses seek to combine both the institutional and the individual level in a multi-level model to determine the contingent decision process. However, such a systematic, cross- national comparison of macro- and micro-data across time remains very diffi cult due to the lack of comparable national surveys with a suffi cient number of countries and a long time horizon that includes the most recent fi gures available. Since our interest here is to explore cross-national variations and long-term trends toward early retirement and recent reversal of early exit from work, we have to rely on macro-level aggregated data in order to compare a larger number of countries from Reversing Early Retirement in Advanced Welfare Economies • 809 the “heydays” of early retirement to the most recent development under the “para- digmatic shift” towards “active ageing”. Even the frequently used SHARE (Survey of Health, Ageing and Retirement in Europe) survey does not include data on non- European “late exit” countries (such as the US and Japan), and even within the Eu- ropean context, countries following a “liberal” institutional approach are underrep- resented. Furthermore, it is diffi cult to reconstruct retirement transitions since the origins of the early retirement trend because the survey initiative started only after the turn of the millennium. In our analysis, we thus restrict ourselves to an explora- tory overview of the relevant nation-specifi c economic and institutional contextual factors that shape (but not causally determine) the opportunities and constraints for individual retirement decisions, highlighting variations across countries and over time. These macro-contextual conditions may themselves serve as background for more detailed assessments of late career transitions by future survey research (see also the other contributions to this special issue, CPoS 38,4). In the following, we will fi rst describe the differences in early exit from work, both in relative level and relative rate of decline. Following the main explanatory ap- proaches in the literature (Ebbinghaus 2006; Hofäcker 2010) we discuss the impact of welfare-related “pull” and economic “push” factors, though our macro-analysis conceives these as opportunities and constraints as shaped by institutional and structural differences across countries. However, we will not test their impact on individual retirement behaviour or company labour shedding strategies but – in line with earlier literature – assume that these factors increase incentives to exit from work earlier than otherwise. The “pull” factors comprise the social protection-relat- ed incentives and the availability of multiple pathways for an early exit from work of specifi c welfare state regimes. We also describe the cut backs (retrenchment) of these welfare programmes, aiming to control expenditure and lower incentives to early retirement. The “pull” factors include the economic factors such as the unem- ployment level, labour market rigidity and employment tenure. Finally, we discuss “retention” factors, in particular active labour market policies and continued train- ing policies which aim at reintegrating or retaining older people in work. By way of conclusion, we map the different factors contributing to the overall level of and the reversal in early exit patterns. From an institutional perspective, we consider both the pull as well as the push factors that shape the cross-national differences in early exit regimes. At the same time, we argue that we also need to investigate the contribution of welfare policy retrenchment and retention factors to the success or failure of reversing early exit from work in recent years. It is indeed the interaction between the factors that matters. 2 Early exit from work in cross-national comparison Previous comparative studies have mapped cross-national similarities and differ- ences in the development of early retirement in the late twentieth century (see for example Ebbinghaus 2006; Hofäcker/Pollnerová 2006; Kohli et al. 1991; OECD 2006). A major advancement have been micro-level studies that use event history data • Bernhard Ebbinghaus, Dirk Hofäcker810 to understand the individual level factors in explaining early or late exit from work (Blossfeld et al. 2006; Blossfeld et al. 2011), though they do not cover the most re- cent period due to survey availability issues. Until today there have been only few comparative studies that analyse the recent gradual reversal of early exit with the help of age-group specifi c employment fi gures (Ebbinghaus 2008; Eichhorst 2011; Hofäcker/Pollnerová 2006; Zaidi/Fuchs 2006). To illustrate most recent trends in early retirement and its reversal, we provide a stylised comparison of employment trends among older workers in advanced economies. The European Union set the Lisbon 2010 target to achieve employment rates above 50 percent among older workers (55-64). In a fi rst step, we map age- specifi c employment rates, i.e. the share of those gainfully working (employed or self-employed) within a given age group. In order to describe the decline in gainful employment among the older workforce and its growing reliance on transfers, we use employment rates as the appropriate measure. The alternative measure of la- bour force participation rates includes the unemployed, which makes comparisons across welfare states diffi cult (see Ebbinghaus 2006; Hofäcker/Pollnerová 2006): While in some countries, early retirement pensions or disability benefi ts were intro- duced to reduce unemployment among the elderly, other countries (e.g. Denmark, Germany, the Netherlands) have used long-term unemployment benefi ts as path- ways into pre-retirement. Since by the late-1960s, most OECD countries have insti- tutionalized the age of 65 (if not earlier) as the normal statutory pension age (Kohli et al. 1991), we count any working person leaving employment before the age of 65 as having exited from work prematurely, even if a statutory pension can be drawn ear- lier in a particular country. This approach follows the international common practice (OECD, EU) and current political discourse, such as the EU 2010 target of raising the employment level of older workers aged between 55 and 64. This benchmark allows us to compare early exit from work before age 65 across countries and irrespective of cultural or institutional practices. Figure 1 thus compares the employment rates at age 60-64 – i.e. directly prior to the benchmark retirement age of 65 – for the years 2000 and 2010. As female employment rates combine the contradictory trends of increasing labour force par- ticipation across cohorts and the early exit from work, only male employment rates are presented here to illustrate the economically most important impact of early retirement on employment patterns. Figure 1 indicates remarkable cross-national variations in employment rates of older workers aged 60-64, ranging from a mini- mum of around 10 percent (in France and Hungary) to a high of around 80 percent in Iceland, while the rest of the OECD countries almost equally spread out between these extreme points. Moreover, employment rates of older workers have risen in the majority of countries over the last decade (indicated by cases left of the dotted line in Fig. 1). For our subsequent analysis, we will investigate these cross-national differences in early exit from work and the varying success in reversing it. For practical reasons, we restrict our analysis to a theoretically selected set of cases (marked by boxes in Fig. 1), which represent typical examples of different institutional confi gurations re- garding welfare state policies and labour market regulations (see Esping-Andersen Reversing Early Retirement in Advanced Welfare Economies • 811 1990; Ebbinghaus 2006; Blossfeld et al. 2006). The United States and the United Kingdom, but also Switzerland and Japan, represent the liberal or residual welfare regime; Germany and the Netherlands are included as typical cases of the Conti- nental conservative regime; Denmark and Sweden represent the universalist social- democratic regime; France, Italy and Spain are examples of the Latin regime; and fi nally, despite limited data availability (not least due to their late entry into the EU) we also include three Central and Eastern European (CEE) countries: Hungary, the Czech Republic and Estonia. These 14 countries cover nearly the whole spectrum of variation between “early” and late “exit regimes”. Notably, the fi nal sample of Fig. 1: Employment rates of men aged 60-64 in OECD countries, 2000 (vertical axis) vs. 2010 (horizontal axis) 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 2000 2010 IS MX JP CN NZ KR US NO SE IL IE PT TR UKAU CA GR EE CH ES DK RU IT PL DE NL FI CZ BELU AT SK FR HU Increase 2000 --> 2010 Decrease 2000 --> 2010 Notes: Countries in frames are those selected for further analysis Country labels: AT = Austria, AU= Australia, BE = Belgium, CA= Canada, CH= Switzer- land, CZ= Czech Republic, DE = Germany, DK= Denmark, ES= Spain, EE = Estonia, FR = France, FI = Finland , GR= Greece, HU = Hungary, IL= Israel, IE = Ireland, IS = Iceland, JP = Japan, KR = Korea, LU = Luxembourg, MX = Mexico, NL= Netherlands, NO = Norway, NZ = New Zealand, PT = Portugal, PL = Poland, CN = China, SE= Sweden, SK= Slovakia, TR = Turkey, UK = United Kingdom, USA= United States. Source: OECD (2011a) • Bernhard Ebbinghaus, Dirk Hofäcker812 countries includes none of the rare cases where employment rates among older workers have declined throughout the previous decade. However, these countries have predominantly been economies from different institutional origin that have suffered from declines throughout the economic and fi nancial crises since 2008, but had previously been following the dominant trend towards later exit among older workers. We thus assume that the omission of these countries does not lead to a systematic bias regarding the underlying institutional characteristics. A major aim will be to investigate in how far these regime classifi cations, mostly established throughout the “heydays” of early retirement, can be upheld even under present conditions of postponed retirement or whether signifi cant departures from regime- specifi c pathways have taken place throughout recent years. Our analysis for these selected countries focuses in particular on cohort-adjusted rates of early exit from work, that is, the decline in the employment rate for the two older age groups (60-64 or 55-59) compared to their employment fi ve years earlier (i.e. age group 55-59 or 50-54 respectively).1 This cohort adjustment is necessary when employment rates are changing across cohorts, which is particularly the case for women. Although the participation of women increased across generations, co- hort-adjusted early exit rates show to what extent a given cohort’s employment rate has declined compared to its employment fi ve years earlier. By using the relative decline in employment rates we also consider the group of unemployed to have de facto exited from work since unemployment benefi ts are one of several pathways to early retirement (Kohli et al. 1991). Figure 2 con trasts the exit rates in 2010 with the respective peak values of early exit.2 This allows us not only to consider the level of exit rates but also the recent reversal since the heydays of early exit. Early exit is more frequent after age 60 than age 55 for both men and women. Furthermore, there have been declines in exit rates compared to the initial peak level, though their magnitude varies considerably across the 14 selected countries. In order to map the main exit regimes in more detail, we juxtapose the current level (represented by the 2010 exit rate) and its relative decline (i.e. percentage) as compared to the respective peak value for the main age group prone to early retire- ment since 1985 (or fi rst available year thereafter) in Table 1. We use the relative exit rates here since absolute rates do not take into account that it is far more diffi cult for countries with a low peak to achieve a decline. We compare those aged 60 to 64 to those aged 55 to 59 (the latter age group is considered to also capture very early exit practices). Following Table 1, fi ve different ideal “exit regimes” may be differentiated. 1 For an explanation and the formula of cohort-adjusted relative exit rates see Ebbinghaus (2006: 103, 278). 2 Although there is some country-specifi c variation in peak exit levels, for most countries con- sidered here, peak values were reached in the early/mid-1990s.For Eastern European countries, comparable OECD data are often not available before the mid-1990s in which exit rates can be assumed to have been the highest due to the immediate “transformation shock” and early statutory retirement rules. Peak values for these countries thus may be slightly underestimated. Reversing Early Retirement in Advanced Welfare Economies • 813 Fig. 2: Exit rates for men and women (aged 60-64 and 55-59) in OECD countries, peak level vs. 2010 100 80 60 40 20 0 France Italy Spain Czech Republic Hungary Estonia Germany Netherlands Denmark Sweden Switzerland United Kingdom United States Japan 55-59: Peak 55-59: 2010 60-64: Peak 60-64: 2010 E. Liberal Countries D. Central Europe B. Eastern Europe C. Northern Europe A. Southern Europe 0 20 40 60 80 100 a) Men b) Women Source: own calculations based on (OECD 2011a). • Bernhard Ebbinghaus, Dirk Hofäcker814 • Early exit is most pronounced in Europe’s Latin welfare states: particularly in France, traditionally also in Italy, and to a lesser degree in Spain. From a comparative perspective, these countries take an extreme position as (persistent) early exit regimes. In France and Italy, half of all employees exit work between age 55 and 65, though among Italian women, early exit rates have declined considerably. Especially France shows very early exits even before age 60, while this is less common in Italy. Given Spain’s late demo- cratic development since 1975, early retirement often occurred later and less frequently, particularly the small group of working women tend to retention longer. Both men and women show rather stable exit rates at around 35-40 percent in their late-50s and 20-25 percent in their early-50s (including those in temporary unemployment). • Following their transition to a market economy after 1989, early exit was very acute in Eastern Europe, particularly in Hungary and the Czech Repub- lic. In these transition economies, half of all working people left employment between age 55 and 65. Especially Czech men and Hungarian older people show a very high exit rate and in Hungary also a tendency to exit very early (see Fig. 2). Unlike these countries, Estonia already succeeded in reducing early retirement in the late-1990s and continues to maintain an intermediate position between “early” and “late” exit despite the recent fi nancial crisis. • For along time, Germany and the Netherlands represented traditional “early exit regimes”; yet, over the last two decades employment rates among older workers have improved considerably. Early retirement rates among people in their late-50s and early-60s have almost halved, for those in their early-50s they were reduced to relative low levels (although the situation is less favour- able in Eastern Germany). While these countries have long been seen as stuck in the “welfare without work” dilemma (Esping-Andersen 1996), they can now be considered as successful cases of early exit reversal. • The classical example of a very late exit country is provided by Sweden where only a minority of employees leave work earlier than at the age of 65. Only during the recession of the early-1990s, early retirement had briefl y turned into a mass phenomenon, but early exit rates have almost continu- ously declined ever since. Early exit patterns in Denmark appear to be more gendered: While early exit of men is similar to other late-exit countries, Dan- ish women predominantly leave work in their late-50s without signifi cant changes in recent years. • Among the liberal or residual welfare regimes, the United States, Japan and Switzerland show exit rates (almost) as low as Sweden. In contrast to Scan- dinavia, these low exit levels are less caused by an improvement in older workers’ employment rates but rather by continuously low levels of early retirement. Exit rates never signifi cantly exceeded peak values of around a third of those in their late-50s (with the exception of Swiss women) and 20 percent of those in their early-50s. These countries are thus traditional late exit regimes. The exception is the United Kingdom, where women can al- Reversing Early Retirement in Advanced Welfare Economies • 815 ready retire at the age of 60. British men had only somewhat higher exit rates during the recessions of the 1980s and 1990s. Both exit levels and their development show signifi cant variations both across and within the described clusters. While the explanation of cross-national varia- tions in exit levels has been widely discussed in previous research, the particular conditions for success of recent policy reversal have been hardly studied thus far. Adopting the ordinal comparative method (Mahoney 2003), we compare countries from high to low exit patterns, searching for institutional and structural explanations for the observable ranking in the early exit level and the degree of change. Using our indicators, we fi nd three main clusters of countries in respect to their ability to reverse early exit from work with few exceptions (see Table 1). One group of entrenched early exit countries has not been able to reverse its high level of early exit (France and Hungary) or have made only partial progress: (Italy (particularly for women) as well as the Czech Republic (particularly for men)). A second group has maintained a medium level of early exit: Denmark (particularly for men) and the United States (particularly for men), and Spain or made partial progress as the Unit- ed Kingdom (particularly for women). Finally those with relative low levels, Japan and Switzerland, have experienced some decline. The countries with a remarkable Tab. 1: Early exit from work and its reversal Early exit Decline (relative to peak) Level 2010 Low Medium High High (>50%) France Italy* ** Hungary ** ** Czech Republic ** Denmark ** Italy* ** Czech Republic ** Medium Denmark ** Spain United Kingdom United States Estonia *** United Kingdom Estonia Germany** Netherlands** Low (<30%) United States Japan Switzerland ** Sweden E A C D B Notes: Level: exit rate 60-64 in 2010; decline: decline of exit rate 60-64 (percent peak level) between peak year and 2010; ♂ men; ♀ women; * also high very early retirement (age 55-59); ** more signifi cant decline in age group 55-59; *** less signifi cant decline in age group 55-59; Clusters: A) South, B) East, C) Centre, D) Nordic, E) Liberal; countries with consistent gender pattern or those for men only in bold; cluster outliers in italics. Source: own calculations based on OECD Labour Force Statistics (raw data available on demand). • Bernhard Ebbinghaus, Dirk Hofäcker816 reversal are Germany and the Netherlands (leaving the high-medium rank), but also Estonia (with more relative decline for women than men) and Sweden (maintaining a decline from an already medium-low level). In our subsequent analysis, we will focus particularly on explaining the last group of successful reversal. 3 Protection-related pull-factors In previous comparative studies, early exit from work prior to the recent turnaround has been frequently explained by protection-related “pull” factors, that is, through welfare state policies that provide opportunities for individual workers to leave work early (Ebbinghaus 2006: chap. 5; Hofäcker 2010: chap. 2). Economists have argued that pension schemes or other public benefi ts often provide an implicit tax on con- tinuing to work (Gruber/Wise 1999). In particular, old-age pension systems that of- fer a pre-retirement option without or with only small actuarial reduction provide a major incentive to rational individuals to stop working since the expected income stream from benefi ts and utility from leisure outweighs the gains from continued working and receiving a pension later. From a life course perspective, sociologi- cal approaches focus on how welfare regimes shape the retirement age and how they structure the transition from work to retirement through multiple “pathways” through old-age pensions, special schemes, disability income support and unem- ployment benefi ts (Kohli et al. 1991). Comparative studies using micro-level data show that welfare regimes provide very different opportunity structures for indi- vidual retirement decisions (Blossfeld et al. 2006; Blossfeld et al. 2011). Sociological studies also stress social norms on retirement timing (Radl 2012) and even “exit cultures” (Maltby et al. 2004) in societies with a high level of early retirement. These societal expectations make a reversal very diffi cult since early retirement is seen as an acquired social right, not least because pay-as-you-go pensions are underwrit- ten by an implicit social contract between generations (Myles/Pierson 2001). Among the main pathways to early exit from work are old-age pensions and special pre-retirement schemes (Ebbinghaus 2006; Kohli et al. 1991). Many pension schemes allow retirement before age 65 (and sometimes an even earlier age for women) for those with long contribution records (“seniority pensions”), for work- ers in hazardous industries, and for the long-term unemployed. Following the last decades of pension reforms as response to the challenges of ageing societies (Im- mergut et al. 2007), a reversal of early retirement has increasingly attempted to be achieved by closing down special schemes and phasing-out early exit options in the pension systems (Ebbinghaus 2006: Chap. 7). In addition, welfare retrench- ment makes pension benefi ts less generous and strengthens the actuarial principle, thereby decreasing the incentives to retire early. International organisations, in par- ticular the OECD and EU Commission, recommended increasing the retirement age (or contribution years), not least in order to harmonise women’s retirement age with that of men as required by EU law. Moreover, the shift towards a multi-pillar pension system and thus an increased share of privately funded pensions has increased the actuarial incentives to retire Reversing Early Retirement in Advanced Welfare Economies • 817 later, particularly in the case of defi ned contribution schemes (Ebbinghaus 2011). In addition, disability pensions have been criticised for providing pathways into early exit from work. Disability pensions vary considerably across countries (Prinz 2003), they have often been the last resort for early exit for those that were unable to fi nd a job due to their age and thus gain particular importance during economic crisis (Ebbinghaus 2006). Unemployment benefi ts, particularly in countries with long- term benefi ts, provided a “bridge” (Guillemard/van Gunsteren 1991) to retirement for older workers, and often a pre-retirement option is granted for unemployed old- er people. However, considerable reform efforts in recent years shifted emphasis from passive to active labour market policies as advocated by OECD and the EU (Weishaupt 2011). Focusing fi rst on protection-related “pull”, we will seek to provide an explanation of the cross-national variations in early retirement by welfare state differences in pathways to early retirement. Does the existence of a high early exit pattern cor- respond to the availability of generous multiple pathways, while lower exit levels can be related to fewer available opportunities? The very early exit in Southern and Eastern Europe can indeed be related to a traditionally low pension age which has been granted since the onset of early retirement (Ebbinghaus 2006: Chap. 5). In Italy’s “pensioner state”, men could retire at age 60 with relatively generous pen- sions and those with 35 contribution years (20 years in the public sector) even from age 55 onwards, while women could draw pensions even 5 years earlier. In France in 1983, retirement age was lowered from 65 to 60 years for those with suffi cient contributions in order to reduce old-age unemployment, while those with insuf- fi cient pensions, particularly women, tend to postpone exit until 65. An exception is Spain’s medium-level early retirement pattern and the late exit age of the – rather small – group of older women still working; this is due to a statutory pension age at 65 for both men and women with few exceptions.3 The CEE countries traditionally maintained an early pension age during the tran- sition to market democracies in the 1990s (Blossfeld et al. 2006): Czech and Hun- garian men could retire at age 60, Czech women at age 57 (if not earlier, depending on the number of children) and Hungarian women at age 55. The disability pen- sion pathway has also been very widespread in both countries. Moreover, special schemes granted even earlier exit options to (mainly male) senior workers with long contribution records (Czech Republic: 3, Hungary: 5 years). Whereas Estonia grant- ed an early pension age of 63 years to men and 58 years to women in the 1990s, the gross pension replacement rate was so low (35 percent) that many continued work- ing. Thus in all three CEE countries, we fi nd a high early exit rate during the transi- tion years since older people were strongly affected by the market-liberalisation shock. 3 In Spain, earlier retirement was available for contributors under the pre-1967 system, older un- employed (age 61, previously 60) could receive an anticipated but reduced pension, and those without work could receive means-tested assistance from age 52 onwards. • Bernhard Ebbinghaus, Dirk Hofäcker818 The Continental European countries are well known for using early retirement as a peaceful labour shedding strategy (Esping-Andersen 1996). Although normal retirement age was set at 65 years in the post-war Federal Republic of Germany, since the 1970s early retirement increased due to a new seniority pension at age 63 (with 35 contribution years), a special pre-retirement scheme during the 1980s, and special pensions at age 60 for the long-term unemployed, severely disabled and ex-housewives (with few contribution years). In the Netherlands, early exit from work was not possible through public basic pension, only granted at age 65, but was achieved through long-term unemployment benefi ts, negotiated pre-retirement pensions and particularly generous disability pensions. These passive labour mar- ket policies became known as the “Dutch disease” (Aarts et al. 1996). While in these early exit countries, early non-working benefi ts granted senior workers and partly women early pensions with little or no actuarial reduction, the Scandinavian welfare states, the Anglophone liberal market economies, Switzer- land and Japan experienced much lower early exit, partly due to more strict public welfare systems (Ebbinghaus 2006: Chap. 5). Scandinavian countries had initially high retirement ages: Old-age pension was only granted at age 67 in Denmark until 2004 (but Danish single women could retire at 62 until 1984) and Sweden’s pension age was lowered from 67 to 65 as early as 1976. Special temporary benefi ts for older unemployed existed in Denmark (phased out after 1996) and an early pension age was available for the unemployed in Sweden (until 1991), while disability pensions also provided early exit pathways (particularly for Danish women), often based on the labour market situation and social criteria. In the United Kingdom, the Beveridge-type public pension was granted at age 65 for men and 60 for women. The US social security system included a statutory pen- sion at age 65 but since the late-1960s fl exible pensions with actuarial reductions have been possible from age 62 onwards. The Japanese national pension is only granted at age 65 but the earnings-related supplementary pension at age 60 (for women 5 years earlier). Disability pensions also played some role in the UK and US (much less in Japan), while unemployment benefi ts were too short to build a bridge into retirement (though a special job release scheme had been applied in the UK until 1989). In these low exit countries, public pension benefi ts were either not avail- able or only at actuarially reduced rates, while disability pensions and sometimes unemployment schemes were partly used as alternative pathways, particularly dur- ing recessions. In Switzerland, the post-war public pension was provided for men at age 65 but for women already at age 62, while occupational pensions (mandatory since 1985) allowed early withdrawal, though with actuarial reduction. Hence, the liberal/residual multi-pillar pension systems of the United States, Switzerland and Japan provided the least opportunities to early exit; this was also the case for Swe- den before the 1990s, but to a lesser degree for Denmark and the United Kingdom as early exit was more frequent there (particularly among women). Adopting an ordinal comparison of the early retirement pathways prior to recent reforms (see Table 2), we can detect that the early exit level strongly relates to our index that measures the availability of multiple pathways provided by pension sys- tems, unemployment schemes and disability benefi ts (based on country informa- Reversing Early Retirement in Advanced Welfare Economies • 819 tion in Ebbinghaus 2006 and Blossfeld et al. 2011). The early-cum-high-exit cluster of Italy, France, Hungary and the Czech Republic has the highest score (8 or more), it is particularly characterised by a low statutory retirement age and early seniority pensions. At a high-medium level, Germany and the Netherlands also show notable levels of early exit from work with a medium index score, followed by the “outliers” Estonia, Spain and Denmark. The least generous protection systems, providing only few early exit options, are the remaining countries of liberal/residual welfare states (UK, US, Switzerland, Japan) and Scandinavian universalism (Sweden). The ordinal comparison thus indicates that the availability of pathways and the overall exit pat- tern conform in the expected way as assumed by the pull thesis. 4 Welfare retrenchment Given the entrenched pathways to early exit from work due to its high popularity, reversing this course has been a major societal challenge in all welfare states (Ebb- inghaus 2006). Several governments have attempted to reverse the early exit trend by scaling down existing “pull” incentives, particularly through raising the statutory age of retirement and restricting if not abolishing previously attractive early path- ways. Surprisingly, such efforts were discussed much earlier in the Anglophone as well as Japan’s residual welfare state with internationally rather low early exit levels, while the generous welfare states of Central and Southern Europe that suffered from high levels of early exit took action much later (Ebbinghaus 2006). The Euro- pean and other OECD-countries differ in their statutory old-age pension schemes, thus in some pension systems normal retirement was granted before age 65 for everybody or for women only, while several pension reforms have been gradually increasing these limits over the last two decades (OECD 2011b, see Table 2). The fi rst plan to increase retirement age dates back to the US social security reform of 1983, though it was a long-term phasing-in plan. Retirement age for full benefi ts is slowly being raised from 65 to 67 between 2002 and 2027 (starting with those born 1938), while early retirement at age 62 remains possible, but with greater actuarial reduction. The United Kingdom decided early on to phase out the Job Release Scheme by 1989, and paid tribute to EU equality rules by phasing in retire- ment age for women from 60 to 65 only since 2010 until 2020. A major reform of the disability pension also occurred during New Labour, though incapacity benefi ts remain a pathway to early exit. Given its rapid ageing, Japan raised the basic pension to 65 years in the mid- 1980s (though earlier drawing is possible with actuarial reductions), while for the second (earnings-related) pension the age is raised for men from 60 to 65 by 2025 and for women from 55 to 60 by 2000 and to 65 years by 2030. Note that post-career jobs have become less subsidised by the gradual in-work benefi ts, while employ- ers fi nd it increasingly diffi cult to fulfi l their promise to fi nd such re-employment for their career workers. In Switzerland, the national pension age for women was raised from 62 to 63 in 2001 and to 64 in 2005, whereas it remained at 65 years for men • Bernhard Ebbinghaus, Dirk Hofäcker820 Tab. 2: Index of pathways ranked by exit opportunities Pension age 1993 / 2010 Old-age pension Flexi- pension Special scheme Unem- ployment Disability Index of pathways Exit pattern A France 60 60,5 *** *** ** * 9 early, high Italy 55 59 *** ** ** ** * {**} 10 {12} early, high Spain 65 = 65 * ** ** 5 moderate B Czech 60 61 ** * {**} * ** 8 early, Republic 57 58.7 *** 9 high Hungary 60 = 60 *** * ** * **{*} 9 {10} early, 55 59 high Estonia 60 63 ** * * ** 6 moderate 55 60.5 *** 7 C Germany 63 65 * ** {**} ** ** 7 {9} high 60 65 ** 8 Netherlands 65 = 65 **{*} ** **{*} 8 {6} high D Denmark 67 65 * **{***} ** 5 {6} : mod. { *} *** 6 {8} : high Sweden 65 = 65 * * * {**} 3 {5} moderate E Switzerland 65 = 65 * * 2 low, 62 63 ** 4 moderate UK 65 = 65 ** { **} * 3{5} : mod. 60 = 60 *** 6 : high USA 65 66 ** * 3 moderate 65 66 Japan 65 = 65 ** * 3 low 61 65 Notes: opportunities for early exit (see Ebbinghaus 2006: Table 5.5 for details): Disability: with labour market consideration; Index of pathways: *** major pathway (3 points); ** conditional pathway (dismissal, unemployment, disability); (2 points); * limited pathway (means-tested, medical-test only, partial pension, actuarial reduction) (1 point); exit trend for men (♂) and women (♀) (see Ebbinghaus 2006: Table 4.8). For additional information for the Czech Republic see Bednárik/Škorpík (2007), for Estonia, Hungary and Switzerland see Blossfeld et al. (2011), and for statutory old-age pension scheme in 1993 and 2010 see OECD (2011b). Source: Ebbinghaus 2006 and own updates Reversing Early Retirement in Advanced Welfare Economies • 821 (earlier drawing will lead to reductions), despite increased unemployment since the 1990s. Following the unemployment crisis of the early 1990s, Sweden abolished the pension at age 60 for long-term unemployed in 1991 and the gradual partial pen- sions have been phased out. A major reform of the two-tier pension system was decided in 1994, replacing the universal basic pension by means-tested supple- ments and the earnings-related pension into an NDC system plus a partially funded individual pension, while fl exible pensions are only available with actuarial reduc- tions. Since its peak of high unemployment in the mid-1990s, Danish labour market policy shifted from passive to active measures. Although the statutory pension age was lowered from 67 to 65 by the 1999 reform (but it will increase again as of 2024), there were increased incentives for those who deferred retirement, not least due to increased importance of funded occupational pensions. Likewise the Continental European countries turned away from passive labour market policies. When the Netherlands suffered from the “Dutch disease” that was epitomised by a million people on disability pensions, a sequence of reforms sought to limit the disability pathway in the late-1980s and early-1990s (Aarts/de Jong 1996). The social partners were also turning away from negotiating generous early exit options, particularly after a tripartite agreement in 1997. The fi rst major reform in Germany was passed when the Berlin wall came down, since 1992 it phased out most of the early retirement pathways (early pension for women, senior workers, and unemployed). Further steps were taken in 1996 by the Conservative-Liberal government but these were repealed again by the subsequent Red-Green govern- ment, which however introduced voluntary pensions and public pension cuts in 2001 and 2003 (Ebbinghaus 2011). In 2007, the grand coalition decided to increase the pension age from 65 to 67 by 2029. The Dutch government decided to increase public pensions from 65 to 67 years by 2023. Thus, both the Netherlands and Ger- many experienced major turnarounds from passive to active policies; indeed both countries met the EU 2010 target of a 50 percent employment rate for older workers (aged 55-64) on time. Starting in the 1990s, a policy shift also occurred in the Southern and Eastern European countries, which was, however, only more or less successful. This was not only due to the long-term fi nancial sustainability of pay-as-you-go public pen- sions caused by rapidly ageing societies and widespread early retirement, but the Maastricht criteria of the EU monetary union for France and Italy as well as the ac- cession perspective and IMF conditions for the CEE countries added urgency. After major political confl icts, the Italian government and trade unions agreed in 1995 to phase-in an increase of retirement ages for men and women, long-term gender equalisation, as well as similar conditions for both the public and private sector. Further – partly contradictory – steps were taken by subsequent governments in order to speed up or slow down the retirement age increase, and most recently an increase beyond 65 years has been discussed. In France, raising the pension age has remained politically impossible, but the necessary contribution period for full pensions has been extended for the private sector (1993), while the Conservative government raised the age by two years for both seniority and normal pensions, but • Bernhard Ebbinghaus, Dirk Hofäcker822 the reforms will be partly reversed by the Socialist government that came to power in 2012. All three CEE countries increased the retirement age in the late-1990s. The Czech Republic raises male and female pension ages to 63 by 2016 and 2019 respectively, though there remain child credits for women. Hungary increased the pension age to 62 for men by 2000 and for women by 2009, thus equalising the fi ve year difference, while an increase to 65 years is also planned for 2022. Estonia has also increased the pension age to 63 for both men (previously 60) and women (previously 55). All three countries will eventually reach a retirement age of 65, though this will be achieved faster in Estonia and Hungary than in the Czech Republic. Compared to the peak of the early retirement period, there has been some re- markable path departure but also much path dependent development in reforming early retirement. The reversal of early exit policies has been most pronounced in the formerly traditional early exit countries Germany and the Netherlands, through phasing out unemployment, preretirement and disability pensions and focusing on the activation of older workers. The Swedish reforms of the 1990s also reversed the short wave of early exit, while Denmark was later and less severe in its reforms. The liberal/residual welfare states further increased the retirement age while providing few opportunities for early exit. Although Italy has been more reform-prone in in- creasing its early statutory retirement age and Spain had less of a problem, France was more hesitant in its reversal policies which remained highly politically conten- tious. The reforms to increase retirement ages in CEE countries precede also rather slowly, particularly the Czech Republic remains behind Estonia and Hungary. 5 Economic Push factors We will now consider the role of different push factors for explaining cross-national variations in early retirement and its reversal. While pull factors primarily affect the decision of older workers to retire or to continue working, push factors assume that structural labour market conditions are crowding out older workers from em- ployment. Although pull and push factors – are typically conceptualised as oppos- ing forces, they have often represented two sides of the same coin in establishing early retirement within modern societies: In Denmark, for example, generous pen- sion incentives and early exit pathways, which are “pull” factors, were specifi cally introduced in response to persistent labour market diffi culties for older workers throughout skyrocketing unemployment in the mid-1990s, that is, “push” factors (Blossfeld et al. 2006). In a similar way, existing transfer schemes such as the Ger- man unemployment insurance and Dutch or Swedish disability benefi ts were inad- vertently used for other purposes to facilitate early exit. The cutback of such social policies may have lowered the incentives to exit, but if push factors remain strong, unemployment or poverty in old age would be high. Thus we also need to look at the economic push factors and the policies that foster reintegration of unemployed and retain older workers in employment. Reversing Early Retirement in Advanced Welfare Economies • 823 Business cycle shifts and economic restructuring Most obviously, push effects can be attributed to changes in the business cycle, es- pecially economic downturns that put pressure on national labour markets. Under these unfavourable conditions, older workers are more likely to accept the alterna- tive roles as pensioners than younger workers would accept being unemployed. Early retirement as a socially accepted form of downsizing has been preferred by unions and works councils. During economic downturns older workers were fre- quently disproportionately pushed into (voluntary) unemployment or offered early exit opportunities, in both cases leading to declining employment rates. The “ma- jor surge” of early exit indeed occurred throughout recessions (Ebbinghaus 2006: 178).4 In liberal economies (such as the UK and the US) or small and open econo- mies (such as Denmark), early retirement mirrored the economic cycle: immediate increases in times of economic recessions and declines in times of economic boom (Hofäcker 2010). Furthermore, the recent reversal of early retirement has been facili- tated by the overall economic recovery during the last decade.5 However, a closer look reveals that the relationship between business cycle and early retirement is less straightforward. Historically, it was often not large-scale economic crises that initiated early exit waves. For instance, Germany’s fi rst pre- retirement programmes were already introduced before the 1970s, they were in- tended as a targeted measure for specifi c groups and not as a general labour market strategy (Jacobs et al. 1991b: 182; Naschold et al. 1994: 450). Instead of causing it, economic crises often rather amplifi ed existing early exit trends by using available instruments in order to relieve national labour markets. Furthermore, early retire- ment trends in Continental as well as Southern European countries often outlived economic crises and persisted throughout subsequent periods of economic recov- ery, thus “decoupling” early exits from the business cycle and turning them into “institutionalised early exit” regimes (Hofäcker 2010; Hofäcker/Unt 2013). Another frequent argument has been that the processes of deindustrialisation and economic restructuring – i.e. the gradual move from an industrial to a service economy and the downsizing of the public sector following welfare state retrench- ment – may have specifi cally affected older workers who are disproportionately represented in declining sectors. Scattered evidence indeed suggests a relationship between downsizing in the above mentioned sectors and early retirement. Howev- er, these explanations merely hold for the initial phase of early retirement, and less for its long-term development. In many countries, early retirement was extended to larger parts of national economies (Jacobs et al. 1991a) and effectively developed into a general pathway far beyond individual sectors (Ebbinghaus 2006). By these 4 Examples are the declines of older workers’ employment in CEE countries following the oil crisis (Hofäcker/Pollnerová 2006) or the increase of old-age unemployment from nearly zero to more than 10 percent during the Swedish recession of the early 1990s (Sjögren Lindquist 2006). 5 While on average, total unemployment (25-64 years) in the EU-21 declined from around 9 per- cent in 2000 to 6 percent in 2008, the employment rate of senior workers increased from 36.5 percent to 45.6 percent in the same countries and time span (OECD 2011a). • Bernhard Ebbinghaus, Dirk Hofäcker824 means, early retirement became institutionalised as an acquired social right across larger parts of society. Although variations in the business cycle and deindustriali- sation may play a role in explaining the emergence of early exit from work, these accounts of economic processes largely fail to adequately explain cross-national variations in the long-term early retirement development (Ebbinghaus 2008). Similar claims as for deindustrialisation are frequently made with regard to ac- companying changes in occupational structures, arguing that shifts in industrial structures together with rapid technological advances in a globalising economy have transformed skill requirements (Blossfeld et al. 2006). Older workers’ quali- fi cations have usually been acquired early in their careers and they often have not adequately been updated. They thus run a higher risk of becoming redundant. The renewal of their qualifi cations comes at a high cost, particularly if they can easily be replaced by younger, better qualifi ed employees. For the initial early exit wave, this argument about lower qualifi cation replacement is a useful explanation. However, it can be assumed that as early retirement continued, workers with redundant quali- fi cations were selectively shed (Ebbinghaus 2006: 175-176). At the same time, the process of educational expansion gradually replaces older by younger cohorts that possess better qualifi cations, allowing them to remain in employment longer (Ber- nardi/Garrido 2006). Some recent fi ndings indicate that it is no longer an individual’s occupational status that is decisive for exit (Henkens/Kalmijn 2006; Hofäcker 2010). Thus one could argue that the explanatory power of occupational changes as a main driver of early exit has been weakened. Production regimes and labour market institutions Some recent approaches shifted the attention to the role of nation-specifi c labour market institutions in mediating the employment prospects of older workers. Most prominently, the Varieties of Capitalism approach (Hall/Soskice 2001) claims two ideal representations of market economies with different types of production re- gimes, skill formation and employment relations, which also have consequences for older workers (see Ebbinghaus 2006 for an overview). On the one hand, there are liberal market economies (LMEs) with largely Fordist mass production, an unregulated labour market, high staff turnover and little em- ployment protection. This fl exible labour market is supplemented by an unstand- ardised educational and occupational training system with a strong emphasis on general skills and the acquisition of company-specifi c qualifi cations via on-the-job- training (Blossfeld/Stockmann 1999). Due to short-termism in fi nancial and corpo- rate governance, such companies are forced to react to downturns by downsizing, that is, via numerical staff fl exibility. Employees in LMEs thus are at a high risk of being dismissed, while the fl exible labour market provides more ample opportuni- ties for them to re-enter employment. Older workers make frequent use of this fl ex- ibility, not least because low public pensions and actuarial private pensions neces- sitate postponing retirement. Only for higher qualifi ed staff with company-specifi c knowledge, employers are willing to invest into long-term maintenance of their staff by granting specifi c fringe benefi ts, such as occupational pension plans. In times of Reversing Early Retirement in Advanced Welfare Economies • 825 economic downturns, these schemes could be (and in fact were frequently) used to promote early exit via “golden handshakes”. On the other hand, in coordinated market economies (CMEs), production is based on quality products, which require more specifi c skills that are acquired in standard- ised (vocational) training and education. Due to the higher importance of long-term trust between employers and employees, labour relations in CMEs are more insti- tutionalised (Hall/Soskice 2001). Moreover, more regulated forms of employment protection are common. Maintenance of a qualifi ed workforce in such companies is supported by higher wages’ and dismissal protection depending on tenure. How- ever, earlier studies have pointed to the ambivalent nature of such regulations (see, for example, Blossfeld et al. 2006). Although such rules seem to promote employ- ment security for older workers, they simultaneously create comparative labour market disadvantages compared to younger workers: In recent decades, following an increasing availability of high-skilled younger workers with more fl exible working contracts and lower starting wages (Blossfeld et al. 2005), older workers in CMEs are more and more regarded as a costly workforce with insuffi cient qualifi cations and are thus frequently shed into early retirement, occasionally with the “help” of occupational benefi ts. This allowed for a socially peaceful solution to labour market imbalances, while at the same time companies were able to maintain their tenure employment relationships. Table 3 illustrates these institutional characteristics using established stylised indicators and contrasts key labour market characteristics of the 14 countries under study. The index of employment protection legislation (EPL), a summary indicator provided by the OECD, allows for a general assessment of the overall strength of labour market regulation on a yearly basis since the late 1980s.6 However, since this index does not capture age-specifi c regulations, we supplement it with two further measures to illustrate labour market consequences of existing regulations for older workers. First, for employees aged 55 years and older, we report average tenure within the company as well as the percentage of those with tenure of more than 10 years. High tenure rates indicate a strong regulation of employment as a result of well-developed internal labour markets and seniority rights. Second, we report the share of long-term unemployment (as a percentage of total unemployment) to refl ect labour market rigidities and re-entry barriers for those who lost their jobs. In fl exible labour markets, their re-entry chances (indicated by low rates of long-term unemployment) will probably be more favourable, whereas in coordinated market economies, redundant workers will fi nd it much harder to re-enter these internal labour markets. The synthesis of indicators in Table 3 indeed confi rms the clusters outlined above (similar patterns for women are not shown here). In the countries of Continental (Germany, the Netherlands) and Latin Europe (France, Italy and Spain) older workers 6 The OECD EPL Index combines measures of (i) job protection for workers with regular con- tracts, (ii) for those with temporary work and (iii) regulations for collective dismissal. For further details see OECD (2004). • Bernhard Ebbinghaus, Dirk Hofäcker826 occupy a safe “insider” position in internal labour markets with high employment protections and long tenure. But the high share of long-term unemployment indi- cates that those older workers out of work are at a high risk of becoming permanent “outsiders”. This simultaneity may have fostered the consensual implementation of rather generous early exit solutions. In the long run, this mutual understanding pro- moted the emergence of complementary “early retirement expectations” among both employers (Schröder et al. 2009) and employees (Frommert et al. 2009), thus creating a self-reinforcing “early retirement deadlock” that may have promoted the broad spread of early retirement respectively and long-term persistence and decou- pling from the business cycle. Quite in contrast, the United States combines a highly fl exible labour market with only minimal employment protection and low tenure (Warner/Hofmeister 2006). Lower and more similar shares of long-term unemployment among older and mid- career workers indicate that redundant workers of virtually any age group can return to work. Patterns in Denmark and – to a lesser extent – in Switzerland approximate this classic fl exible labour market. The absence of employment protection and inter- nal labour markets combined with little re-entry barriers for those out of work has Tab. 3: Overview of selected institutional push factors, 2000-2010 Average tenure (in years) Tenure 10+ years (% overall) EPL-Index (Version 1) Long-term unemployed (1 year and longer) as % of total unemployment, by age group Age group: 55-64 55-64 All 55+ 25-54 55+ 25-54 Year 2000 2009 2000 2009 2000 2008 2000 2000 2010 2010 A France 21.8 22.7 77.2 75.1 2.98 3.05 67.7 42.8 56.3 41.8 Italy 23.6 23.9 82.4 78.3 2.51 1.89 63.1 62.7 59.2 49.0 Spain 22.4 23.0 78.1 76.2 2.93 2.98 61.8 50.9 62.2 45.2 B Czech Republic 16.9 16.5 54.6 63.6 1.90 1.96 45.6 53.3 40.8 46.4 Hungary 16.6 15.6 56.0 57.8 1.27 1.65 57.9 52.6 55.7 52.3 Estonia — — — — — *2.39 50.9 51.9 43.0 48.5 C Germany 21.7 21.9 74.9 74.5 2.34 2.12 69.1 51.0 62.3 48.4 Netherlands 22.2 23.0 75.6 75.3 2.12 1.95 63.0 **30.0 52.0 30.5 D Denmark 18.5 17.7 65.3 65.4 1.50 1.50 47.5 21.6 35.1 22.4 Sweden 21.0 20.2 75.7 — 2.24 1.87 49.3 26.6 29.7 21.2 E Switzerland 20.6 20.2 74.6 70.6 1.14 1.14 *** *** *** *** United Kingdom 15.3 16.1 56.5 55.5 0.68 0.75 42.1 33.2 42.9 37.0 United States — — — — 0.21 0.21 11.9 6.6 38.1 31.5 Japan — — — — 1.43 1.43 36.0 22.5 40.5 38.7 Notes: *: EPL Index, Version 2; **: Data from 2002; ***: no data available Source: OECD (2011a) Reversing Early Retirement in Advanced Welfare Economies • 827 contributed signifi cantly to the persistently low exit rates among older workers and their higher sensitivity to changes in the business cycle. There, however, remain some ambiguous cases. According to Table 3, the rather weak push factors in Eastern European countries are apparently at odds with the simultaneous prevalence of high early exit rates. One possible explanation is that the standardised indicators may not be fully capable of capturing the highly dy- namic transition processes of post-communist labour markets. Furthermore, evi- dence from case studies suggests that especially throughout the early transition years, many governments used early retirement schemes to cope with the dramatic “push” exerted through initially skyrocketing unemployment and the inability of then still rigid labour markets with high rates of over-employment to accommodate the crisis (e.g. Hamplová/Pollnerová 2006). At fi rst glance, the favourable employment performance of older Japanese work- ers also appears surprising since Japan is seen as a CME with regulated employ- ment relations, high tenure as well as well-developed seniority regulations (OECD 2004), thus suggesting a high prevalence of push factors. However, especially in larger companies, these employment relations have been partly fl exibilised through the opportunity of mandatory retirement from career jobs at age 60 and a subse- quent re-employment in supplier fi rms or secondary labour markets (Ebbinghaus 2006). Thus large Japanese companies dismiss career workers early but they fi nd them secondary jobs elsewhere, while occupational pensions are used to partially subsidise earning losses. Finally, Sweden stands out as a clear exception with a low early exit rate but simultaneously rather regulated labour markets thanks to strong unions. Sweden complements the public guarantee of employment security with employment- supporting measures that bring about more favourable employment outcomes for older workers. While there thus is some explanatory power of institutional push factors regard- ing the emergence and development of early exit until the 1990s, their value in accounting for the recent reversal is more limited. In some cases, such as Italy or Sweden, recent declines in older workers’ exit rates were accompanied by a decline in the strength of push factors (i.e. a move towards lower employment protection and higher labour market fl exibility). However, the general picture concerning insti- tutional push factors is one of relative stability. As rather long-lasting institutions, they apparently only contribute little to the explanation of the recent, sometimes remarkable increase in older workers’ employment. We will therefore need to turn to further institutional factors that explain the ability to reintegrate and retain older workers in employment better. 6 Retention Factors In addition to reducing early exit incentives and existing barriers to re-employment, societies can also actively foster older workers’ labour force participation through targeted, state-funded programmes. By means of active labour market policies, wel- • Bernhard Ebbinghaus, Dirk Hofäcker828 fare states may support job search through a wide range of activities, ranging from direct job creation to targeted counselling. In addition, through institutionalised op- portunities for lifelong learning, public policies (together with companies and social partners) may enhance individual qualifi cations and thus improve older workers’ employability. In contrast to pull and push factors, such “retention” policies, that have been implemented for the explicit purpose of reversing early exit are a rather new phenomenon whose importance has only risen under the more recent paradig- matic turn towards “active ageing” (Jepsen et al. 2002). In other words, they do not just represent a “scaling back” of previous pull or push factors, but an own set of unique programmes. Adding to previous literature, we thus treat them as a separate analytical category of our explanation of the more recent early exit reversal. Tab. 4: Overview of selected institutional retention factors, 1985-2010 ALMP expenditure, in % GDP PLMP expenditure, in % GDP Training in last four weeks, 55-64 years 1985-99 2000-09 1985-99 2000-09 1992-99 2000-10 A France 0.95 1.00 0.52 1.46 0.2 1.7 Italy — 0.51 0.93 0.77 0.7 1.6 Spain 0.58 0.77 2.50 1.63 0.4 2.9 B Czech Republic 0.16 0.22 0.18 0.27 — 2.2 Hungary 0.48 0.32 1.10 0.39 0.3 0.4 Estonia — 0.09 — 0.30 2.3 4.5 C Germany 1.04 1.03 1.82 1.82 0.8 2.3 Netherlands 1.32 1.32 2.77 1.24 4.7 7.5 D Denmark 1.41 1.65 4.09 2.10 6.0 17.8 Sweden 2.25 1.32 1.61 0.97 14.6 12.9 E Switzerland 0.39 0.63 0.73 0.76 17.5 19.8 United Kingdom 0.39 0.34 0.88 0.23 5.7 15.3 USA 0.21 0.15 0.40 0.45 — — Japan 0.31 0.29 0.35 0.45 — — Source: for expenditure: OECD (2011a), training: Eurostat (2011) Reversing Early Retirement in Advanced Welfare Economies • 829 In order to compare the relative importance of passive (PLMP) and active (ALMP) labour market policies, we analyse OECD indicators (available since 1985) for ex- penditure on both types of policies (as percentage of the GDP) in Table 4.7 Notably, such expenditure data refers to expenditure for the labour market as a whole, and thus inhibits the identifi cation of expenditure specifi cally targeted at older work- ers.8 Cross-national differences thus can only be regarded as approximations of a country’s general political commitment towards either a labour-maintaining “active strategy” or a labour-shedding “passive” strategy. Table 4 contrasts average expenditure for two periods: the heydays of early exit from 1985 to 1999 and the decade of the 2000s in which its reversal has been on the political agenda. In addition, we analyse participation rates of older workers (aged 55-64) in continued training within a time span of four weeks as indicator of lifelong learning. We assume that the more often older workers are given continued educa- tion opportunities, the less likely they will suffer from qualifi cation disadvantages compared to younger workers and will therefore be more likely to remain in employ- ment. Again, average values are compared for the early exit period (1992-99) and its more recent reversal (2000-10). Scandinavian countries display high levels of engagement in both areas (see Table 4). Especially Sweden stands out as the only country in which ALMP exceeds PLMP expenditure. Only throughout Sweden’s economic crisis in the 1990s, passive expenditure signifi cantly increased, but this was largely reversed during economic recovery. For a long time, Denmark deviated with PLMP exceeding ALMP expendi- ture more than three times in the 1990s as a consequence of introducing various pathways into early retirement during the 1980s recessions. However, a turnaround happened with Danish activation policies in the mid-1990s, when ALMP and PLMP converged, thus moving Denmark closer to the Swedish model. With a strong pub- lic engagement in promoting lifelong learning, both countries display high participa- tion rates in continued education and training. In both countries, the actively-orient- ed labour market policy as well as the high commitment to lifelong learning played an important role in the prevention of massive early exits and helped to reduce exit rates during recent years. Quite in sharp contrast, both Eastern and Latin European countries have contin- ued to follow residual labour market policies with low expenditure for both ALMP and PLMP. Moreover, participation in lifelong learning is low and has only marginally 7 ALMP are primarily aimed at helping the unemployed back into work and include, for example, employment services, training, hiring subsidies, direct job creation in the public sector, sup- ported employment and rehabilitation and start-up incentives (OECD 2011a). PLMP, in contrast, aim more at maintaining the living standard of those who have become unemployed, and en- compass unemployment benefi ts and preretirement programmes (but not early old-age pen- sions or disability benefi ts). 8 Even when older workers are considered separately – as for example in the share of expendi- tures on early retirement for labour market reasons as reported by the OECD – data are incom- plete and only focus on explicit early retirement programmes through the national pension system and do not take alternative pathways from employment to retirement into account, e.g. through unemployment or disability insurance. • Bernhard Ebbinghaus, Dirk Hofäcker830 improved in recent years (only Estonia shows a greater increase in recent years). These countries thus continue to rely on “front-loaded” education systems that train workers in the early parts of their career but provide little skill renewal thereafter (Dieckhoff et al. 2007). The mere absence of active employment and educational policies for older workers may be one reason why it has remained diffi cult for these countries to lower their high exit rates during the last decade. Although labour market policy expenditure is similarly low in the late exit regimes of liberal welfare states, active state policies in these countries are not considered necessary to enable (older) workers to continue their employment since fl exible labour markets pose little re-entry barriers. Furthermore, training rates within these fl exible labour markets are still high, probably because of high job turnover and reli- ance on on-the-job training.9 A similar picture can be observed for Japan, where the welfare state takes a rather residual stance in labour market policies, as employer- employee relations are strong and training is extensively provided by companies (OECD 2004). Despite little active state engagement, both liberal market economies and Japan thus are able to maintain a similarly favourable labour market perfor- mance as Scandinavia. The central European countries, Germany and the Netherlands stand out given their recent reversal in early exit. Since the 1990s, the Netherlands has promoted active policies for older workers, refl ected in a more balanced ratio between ALMP and PLMP (see Table 4). In addition to cutting back on previously attractive path- ways into retirement, more emphasis had been placed on extending lifelong learn- ing also to older workers. In Germany, however, trends in active policies appear to be more cautious, though national evidence suggests that training for older workers has been enhanced (Kuwan et al. 2006). Thus other factors (such as reforms in pen- sion incentives) may have been more decisive in promoting the recent reversal than retention factors. Overall, there appears to be some (though no perfect) correspondence between the extent and development in retention policies and the level of early exit and its reversal. Countries in which older workers’ employment is being supported through either active labour market policies or lifelong learning are frequently among those that are doing relatively well in terms of reversing early exit. 7 Comparing Macro-Confi gurations In a meta-analysis we now compare our results on early retirement and its institu- tional and structural determinants by mapping the main confi gurations by regime and country (see Table 5). In our analysis we fi rst described the evolution of early exit and its development over time in the 14 selected advanced economies. We 9 Since the training indicator in Table 4 is based on the European Labour Force Surveys, there is no data available for the United States and Japan, but both countries come close to the other lifelong learning regime. Reversing Early Retirement in Advanced Welfare Economies • 831 were generally able to identify a ranking of early exit patterns. Both Liberal-residual welfare states and Scandinavian countries can be regarded as “late exit” regimes with traditionally low levels of early retirement, though this was interrupted by a brief early exit boom in Sweden during the 1990s recession, whereas Denmark has generally shown a higher level of early retirement among women and until the mid- 1990s also among men. In contrast, Latin and CEE countries are marked by strong and persistent early exit trends that are facilitated by strong pull and push factors, although Spain and Estonia provide fewer preretirement incentives for older work- ers. While the above clusters are largely stable over time and broadly refl ect previ- ous classifi cations of “early” and “late exit regimes” (Ebbinghaus 2006; Blossfeld et al. 2006), Germany and the Netherlands are a notable exception, moving from an early exit regime towards a late retirement model over the last two decades. In our meta-analysis we use comparative indicators to provide an institutional and structural explanation for these long-term exit patterns and their reversals (see Table 5): pull factors (differentiated into characteristics of mandatory pension sys- tems, additional early exit pathways), the reform efforts leading to a retrenchment of such welfare arrangements; push factors (differentiated into unemployment and regulation of national labour markets) and retention factors (differentiated into ALMP and lifelong learning). The combination of these factors refl ects the “cornerstones” of the Swedish “late exit” model and its further development. During the 1970s and 1980s, Sweden only exhibited a low level of early exit due to both pull and push factors: A rather stringent pension system with high retirement ages and little early exit opportuni- ties, while unemployment was historically low and high employment across all age- groups on formally regulated labour markets was fostered by active labour market policies and lifelong learning. It was only in the economic crisis of the early 1990s when long-term unemployment and early exit from work became widespread, though these push and pull factors were soon reversed as the economy recovered and early exit declined almost to its pre-crisis level. Denmark initially followed the Swedish model, but due to persistent unemploy- ment, early retirement pathways were increasingly used since the 1980s, thus mov- ing Denmark into an intermediate position between early and late exit countries. Only since the mid-1990s, activation policies were strengthened and the most gen- erous early exit pathways were closed. Nonetheless, the only modest declines in exit rates suggest that further cutbacks in pull incentives (such as phasing out pre- retirement pathways) will be required to further improve older workers’ employ- ment. As our analysis indicates, similarly high employment levels among older work- ers in liberal welfare states rest on an alternative institutional combination. These welfare states only invest little in social protection and labour market policies. In- stead, high employment rates of older workers are maintained by a fl exible, weakly regulated labour market (with little statutory employment protection) – i.e. weak “push” forces – combined with a residual and largely privatised pension system that simultaneously provides little “pull” incentives for early exit. An unregulated educa- tional system strongly relying on on-the-job training fosters life-long learning and • Bernhard Ebbinghaus, Dirk Hofäcker832 S o u rc e : O w n c o m p ila ti o n b as e d o n T ab le s 1 -4 a n d p as si m . W h e n c o n fi g u ra ti o n f o r w o m e n d if fe rs t h e n m ar ke d i n i ta lic s. C o n fi g u ra - ti o n al c o n d it io n s fu lly i n l in e w it h t h e si s (p u ll an d p u sh f o r le v e l, re tr e n ch m e n t an d r e te n ti o n f ac to rs f o r re v e rs al ) ar e u n d e r- lin e d . T a b . 5: C o m p a ra ti v e M e ta -A n a ly si s o f M a cr o -C o n fi g u ra ti o n s R e g im e E ar ly e x it P u ll R e tr e n ch m e n t P u sh R e te n ti o n P e a k le ve l R e ve rs a l R e ti re m e n t a g e E a rl y e xi t in ce n ti ve s R e fo rm in te n si ty U n e m p lo y- m e n t La b o u r m a rk e t re g u la ti o n La b o u r m a rk e t p o lic y Li fe lo n g le a rn in g N o rd ic S E L o w H ig h H ig h M e d iu m H ig h M e d iu m M e d iu m H ig h H ig h D K M e d iu m / H ig h L o w H ig h M e d iu m H ig h H ig h M e d iu m M e d iu m H ig h H ig h Li b e ra l U K M e d iu m M e d iu m H ig h Lo w M e d iu m Lo w Lo w Lo w H ig h H ig h Lo w M e d iu m M e d iu m U S L o w L o w H ig h Lo w M e d iu m Lo w Lo w Lo w H ig h C H J P L o w L o w M e d iu m , b u t re -e m p lo y m e n t Lo w M e d iu m Lo w M e d iu m Lo w H ig h C e n tr al D E H ig h H ig h M e d iu m H ig h H ig h M e d iu m , N L: M e d iu m M e d iu m N L: M e d iu m N L st ru ct u ra l D : H ig h D : Lo w La ti n F R (v e ry ) h ig h L o w Lo w H ig h Lo w M e d iu m , H ig h Lo w Lo w IT st ru ct u ra l E S M e d iu m M e d iu m M e d iu m M e d iu m Lo w H ig h , M e d iu m Lo w Lo w st ru ct u ra l E as t H U H ig h L o w Lo w H ig h Lo w H ig h , M e d iu m Lo w Lo w C Z st ru ct u ra l E E M e d iu m M e d iu m M e d iu m M e d iu m M e d iu m H ig h , M e d iu m Lo w Lo w , H ig h st ru ct u ra l in cr e as in g Reversing Early Retirement in Advanced Welfare Economies • 833 prevents the obsolescence of older workers’ skills. In contrast to the Scandinavian focus on explicit retention policies, it is the mere absence of institutional pull and push factors that explains the high employment among older workers. Apart from some recent increases in statutory retirement ages, these institutional patterns have largely remained stable since the millennium switch which is a possible expla- nation of the equally stable employment rates of older workers in the last decade. Japan shows some similarities with the liberal strategy, with a largely residual role of the welfare state, but the regulation of employment relations is bound by strong employer-employee ties in larger Japanese companies. This is refl ected in the promise of lifelong employment tenure until mandatory retirement when having a career job, and the company’s commitment to secondary re-employment thereaf- ter. As a consequence, employment careers have been stable and long-lasting, and a high level of employment could be achieved among older workers. Only recently, this model has been modifi ed as companies found it increasingly diffi cult to main- tain their social responsibilities and as the Japanese government increased statu- tory retirement ages to prepare for demographic ageing (Ebbinghaus 2008: 284). Nevertheless, Japan has been able to maintain its status as a late retirement country among the OECD countries. In contrast to these low exit regimes, central European countries have virtu- ally displayed all institutional characteristics of an “early exit regime” for a long time. High “push” forces, refl ected in rigid labour markets combined with persistent structural unemployment since the late 1970s, have promoted the introduction of strong “pull” incentives for an early exit, either through public seniority pensions (Germany), occupational pensions (Netherlands) or additional welfare state pro- grammes (unemployment insurance in Germany, disability insurance in the Neth- erlands). Labour market policy largely focussed on maintaining the living standard of those unemployed, while only moderate employment support was provided to older workers. Given the signifi cance of both pull and push factors and the mere absence of effective retention policies, it is not surprising that exit rates of older workers remained high until the mid-1990s. However, various institutional reforms paved the way for an effective “recalibration” (Pierson 2001: 425) and thus laid the foundation for a subsequent “early exit” reversal. Most importantly, pull incentives have been reduced in both countries, while active retention policies for older work- ers have been strengthened, particularly in the Netherlands. Up until the mid-1990s, the early exit regimes of CEE and Latin European coun- tries shared some relevant conditions with the continental European countries. Strong early exit incentives were introduced as a means to overcome high unem- ployment and structural economic changes (such as the transition into market econ- omies in Spain and Eastern Europe). While these polices became institutionalised, labour market policies remained marginal and – if at all – were oriented at fi nancially supporting employment exits. Yet, in contrast to Germany and the Netherlands, recent institutional reforms have been far less pervasive. Though some countries have implemented higher pension ages, they have done so with rather long periods of phasing-in. At the same time, generous early retirement pathways have largely been maintained (for Italy see Barbieri/Scherer 2011). Strong internal labour mar- • Bernhard Ebbinghaus, Dirk Hofäcker834 kets and the high signifi cance of seniority regulations led to disadvantages for re- employment chances of older jobless people, while little active measures promoted older workers’ employment. Only cautious reforms of pull and push factors as well as largely residual labour market policies are reasons for the, so far, only modest reversal of early exit. Spain and Estonia, however, appear to deviate from the re- gional regime pattern by either showing historically lower early exit rates (Spain) or exhibiting a recent early exit reversal (Estonia). Unlike the early exit regimes, both countries maintained (Spain) or promoted (Estonia) a higher formal retirement age and reduced early exit pathways. At the same time, the increased engagement in active policies may have contributed to the countries’ more favourable performance compared to their counterparts. 8 Outlook Our comparative analysis shows that early exit trends and their most recent reversal can be understood better when refl ecting it as a combined effect of pull, push, re- trenchment and retention factors. These four different explanatory approaches are no suffi cient accounts by themselves, but represent complementary explanations that need to be considered simultaneously in order to achieve a comprehensive un- derstanding of early exit and its reversal. Push and pull factors have played a more central role in establishing and maintaining the early retirement trend. As the exam- ples of early exit countries show, generous pull-incentives were often implemented or utilised by employers, unions and governments to mediate pressures through different types of economic push, either due to structural economic changes or rigid labour markets. Over time, these measures often became institutionalised in both national regulations as well as expectations of employers and employees. Yet, reversing early retirement does not only require a gradual scaling back or abolish- ment of these institutional forces (retrenchment of pull factors), but it also depends on active ageing policies that help maintain older workers’ employment. Together with welfare reforms, these retention policies have developed into a vital element in overcoming entrenched early exit routines during the last two decades. The institutional patterns studied here are themselves embedded in a respective “cultural” setting, which could further enhance the understanding of exit transitions over time (see Hofäcker/Unt 2013). A vivid example for the mutual interplay between the nation-specifi c policies and such “exit cultures” (Guillemard/Argoud 2004) or “retirement age norms” (Radl 2012) is France, where the Socialist government intro- duced a formal retirement age of 60 years for those with long contribution records in 1983 to curb the already widespread societal practice to retire early via long-term unemployment insurances. Thereafter, changes to this “early exit culture” remained diffi cult for French politicians. Due to limited survey data, the infl uence of cultural contexts could not be controlled systematically for all selected countries. Especially for the recent reversal of the early retirement trend it can, however, be assumed that cultural norms did not change as fast as public policy efforts, which means that they may even have “slowed down” the pace of reversal. Either a case study design for Reversing Early Retirement in Advanced Welfare Economies • 835 cultural specifi cities or longitudinal survey analysis would be required to ascertain these complex relationships over time. The post-2008 fi nancial and economic crisis has made the future of successful reversal of early exit a more contingent but also a more pressing policy issue. It may be too early to judge whether the recent crisis will again lead to a new wave of early retirement among older workers due to higher unemployment or whether the fi scal need for further welfare state recalibration will be intensifi ed due to the current sov- ereign debt crisis. In any case, future policies to further reduce early exit from work would be well-advised not to rely solely on singular strategies, such as a scaling back of pension incentives. As long as structural labour market rigidities or inade- quate qualifi cations do not allow older workers to remain in employment, these wel- fare retrenchment measures may be largely ineffective and only shift income and labour market risks to older individuals. Instead, public policies by governments in cooperation with social partners will have to develop integrated strategies – aimed at both reducing early exit incentives and simultaneously enhancing older workers employability – to arrive at a sustainable and socially acceptable solution to the still persistent problem of early retirement in ageing societies. References Aarts, Leo J.M.; de Jong, Philip R. 1996: Evaluating the 1987 and 1993 Social Welfare Reforms: From Disappointment to Potential Success. In: Aarts, Leo J.M.; Burkhauser, Richard V.; de Jong, Philip R. (Eds.): Curing the Dutch Disease: An International Per- spective on Disability Policy Reform. Aldershot: Avebury: 47-69. Aarts, Leo J.M.; Burkhauser, Richard V.; de Jong, Philip R. (Eds.) 1996: Curing the Dutch Disease: An International Perspective on Disability Policy Reform. Aldershot: Ave- bury. Barbieri, Paolo; Scherer, Stefani 2011: Retirement in Italy: Rising inequalities across gen- erations. In: Blossfeld, Hans-Peter; Buchholz, Sandra; Kurz, Karin (Eds.): Aging Popula- tions, Globalization and the Labor Market. Cheltenham, UK: Edward Elgar: 91-120. Bednárik, Rastislav; Škorpík, Jan 2007: Transition from Work to Retirement in the Czech Republic and Slovakia. In: Vienna Centre Policy Brief, 2007. Bernardi, Fabrizio; Garrido, Luis 2006: Men‘s late career and career exits in Spain. In: Blossfeld, Hans-Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.): Globalization, Uncer- tainty and Late Careers in Society. London: Routledge: 119-39. Blossfeld, Hans-Peter; Stockmann, Reinhard 1999: The German Dual System in Com- parative Perspective. In: International Journal of Sociology 28,1: 3-28. Blossfeld, Hans-Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.) 2006: Globalization, Un- certainty and Late Careers in Society. London: Routledge. Blossfeld, Hans-Peter; Buchholz, Sandra; Kurz, Karin (Eds.) 2011: Aging Populations, Globalization and the Labor Market: Comparing Late Working Life and Retriement in Modern Societies. Cheltenham, UK: Edward Elgar. Blossfeld, Hans-Peter; Klijzing, Erik; Mills, Melinda; Kurz, Karin (Eds.) 2005: Globaliza- tion, Uncertainty and Youth in Society. London: Routledge. • Bernhard Ebbinghaus, Dirk Hofäcker836 Dieckhoff, Martina; Jungblut, Jean-Marie; O‘Connell, Philip J. 2007: Job-Related Train- ing in Europe: Do Institutions Matter? In: Gallie, Duncan (Eds.): Employment Regimes and the Quality of Work. Oxford: Oxford U.P.: 77-104. Ebbinghaus, Bernhard 2006: Reforming Early Retirement in Europe, Japan and the USA. Oxford: Oxford University Press. Ebbinghaus, Bernhard 2008: Comparative Regime Analysis: Early Exit from Work in Europe, Japan, and the United States. In: Kenworthy, Lane; Hicks, Alexander (Eds.): Method and Substance in Macrocomparative Analysis. London: Palgrave: 260-289. Ebbinghaus, Bernhard (Eds.) 2011: The Varieties of Pension Governance: Pension Pri- vatization in Europe. Oxford: Oxford University Press. Eichhorst, Werner 2011: The Transition from Work to Retirement. In: German Policy Studies 7,1: 107-33 [http://www.spaef.com/article/1271/The-Transition-from-Work-to- Retirement]. Esping-Andersen, Gøsta 1990: The Three Worlds of Welfare Capitalism. Cambridge: Pol- ity Press. Esping-Andersen, Gøsta 1996: Welfare States without Work: The Impasse of Labour Shedding and Familialism in Continental European Social Policy. In: Esping-Andersen, Gøsta (Eds.): Welfare States in Transition: National Adaptations in Global Economies. London: Sage: 66-87. Eurostat 2011: NEW CRONOS Database. Luxembourg: Eurostat [http://epp.eurostat. ec.europa.eu, 24.08.2011]. Frommert, Dina; Hofäcker, Dirk; Heien, Thorsten; Andreß, Hans-Jürgen 2009: Pension Systems and the Challenge of Population Ageing: What does the Public Think? In: Edmondson, Ricca; Kondratowitz, Hans-Joachim von (Eds.): Valuing Older People: A Humanist Approach to Ageing. Bristol: Policy Press: 139-160. Gruber, Jonathan; Wise, David A. (Eds.) 1999: Social Security and Retirement around the World. Chicago, IL: University of Chicago Press. Guillemard, Anne-Marie; Argoud, Dominique 2004: France: A country with a deep early exit culture. In: De Vroom, Bert; Maltby, Tony; Mirabile, Maria Luisa ML.; Øverbye, Einar (Eds.): Ageing and Transition to Retirement. Ashgate: 147-165. Guillemard, Anne-Marie; van Gunsteren, Herman 1991: Pathways and their Prospects: A Comparative Interpretation of the Meaning of Early Exit. In: Kohli, Martin; Rein, Martin; Guillemard, Anne-Marie; van Gunsteren, Herman (Eds.): Time for Retirement: Comparative Studies on Early Exit from the Labor Force. New York, NY: Cambridge University Press: 362-387. Hall, Peter A.; Soskice, David 2001: An Introduction to Varieties of Capitalism. In: Hall, Peter A.; Soskice, David (Eds.): Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. New York, NY: Oxford University Press: 1-68. Hamplová, Dana; Pollnerová, Štepánka 2006: Employment exits of older workers in the Czech Republic. In: Blossfeld, Hans-Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.): Globalization, Uncertainty and Late Careers in Society. London: Routledge: 281-387. Henkens, Kène; Kalmijn, Matthijs 2006: Labor market exits of older men in the Neth- erlands: An analysis of survey data 1979-1999. In: Blossfeld, Hans-Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.): Globalization, Uncertainty and Late Careers in Society. London: Routledge: 79-100. Reversing Early Retirement in Advanced Welfare Economies • 837 Hofäcker, Dirk 2010: Older Workers in a Globalizing World. An International Comparison of Retirement and Late-Career Patterns in Western Industrialized Societies. Chelten- ham, UK: Edward Elgar. Hofäcker, Dirk; Pollnerová, Štepánka 2006: Late careers and career exits. An interna- tional comparison of trends and institutional background patterns. In: Blossfeld, Hans- Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.): Globalization, Uncertainty and Late Careers in Society. London: Routledge: 25-53. Hofäcker, Dirk; Unt, Marge 2013: Exploring the ‘new worlds’ of (late?) retirement in Eu- rope. In: Journal of International and Comparative Social Policy 29,2: 163-183 [doi: 10.1080/21699763.2013.836979]. Immergut, Ellen; Anderson, Karen; Schulze, Isabelle (Eds.) 2007: The Handbook of West European Pension Politics. Oxford: Oxford University Press. Jacobs, Klaus; Kohli, Martin; Rein, Martin 1991a: Testing the Industry-Mix Hypothesis of Early Exit. In: Kohli, Martin; Rein, Martin; Guillemard, Anne-Marie; van Gunsteren, Herman (Eds.): Time for Retirement: Comparative Studies on Early Exit from the Labor Force. New York, NY: Cambridge University Press: 67-96. Jacobs, Klaus; Kohli, Martin; Rein, Martin 1991b: Germany: The Diversity of Pathways. In: Kohli, Martin; Rein, Martin; Guillemard, Anne-Marie; van Gunsteren, Herman (Eds.): Time for Retirement: Comparative Studies on Early Exit from the Labor Force. New York, NY: Cambridge University Press: 181-221. Jepsen, Maria; Foden, David; Hutsebaut, Martin (Eds.) 2002: Active Strategies for Older Workers. Brussels: ETUI. Kohli, Martin 1987: Retirement and the Moral Economy: A Historical Interpretation of the German Case. In: Journal of Ageing Studies 1,2: 125-44 [doi: http://dx.doi. org/10.1016/0890-4065(87)90003-X]. Kohli, Martin; Rein, Martin; Guillemard, Anne-Marie; van Gunsteren, Herman (Eds.) 1991: Time for Retirement: Comparative Studies on Early Exit from the Labor Force. New York, NY: Cambridge University Press. Kuwan, Helmut; Bilger, Frauke; Gnahs, Dieter; Seidel, Sabine 2006: Berichtssystem Wei- terbildung IX: Integrierter Gesamtbericht zur Weiterbildungssituation in Deutschland. Berlin: Bundesministerium für Bildung und Forschung. Mahoney, James 2003: Strategies of Causal Assessment in Comparative Historical Analysis. In: Mahoney, James; Rueschemeyer, Dietrich (Eds.): Comparative Historical Analysis in the Social Sciences. New York: Cambridge University Press: 337-72. Maltby, Tony; De Vroom, Bert; Mirabile, Maria Luisa; Øverbye, Einar (Eds.) 2004: Ageing and the Transition to Retirement. A Comparative Analysis of European Welfare States. Aldershot: Ashgate. Myles, John; Pierson, Paul 2001: The Comparative Political Economy of Pension Reform. In: Pierson, Paul (Eds.): The New Politics of the Welfare State. New York: Oxford Uni- versity Press: 305-333. Naschold, Frieder; de Vroom, Bert (Eds.) 1994: Regulating Employment and Welfare: Company and National Policies of Labour Force Participation at the End of Worklife in Industrial Countries. Berlin: W. de Gruyter. Naschold, Frieder; de Vroom, Bert; Casey, Bernard 1994: Regulating Employment and Retirement: An International Comparison between Firms and Countries. In: Naschold, Frieder; de Vroom, Bert (Eds.): Regulating Employment and Welfare: Company and National Policies of Labour Force Participation at the End of Worklife in Industrial Countries. Berlin: W. de Gruyter: 433-489. • Bernhard Ebbinghaus, Dirk Hofäcker838 OECD 2004: Aeging and Employment Policies in Japan. Paris: OECD. OECD 2006: Live longer, work longer: A synthesis report. Paris: OECD. OECD 2011a: OECD Labour Forces Statistics. Paris OECD [http://stats.oecd.org, 24.08.2011]. OECD 2011b: Pensions at a Glance 2011: Retirement-income systems in OECD and G20 countries. Paris: OECD. Pierson, Paul 2001: Coping with Permanent Austerity: Welfare State Restructuring in Affl uent Democracies. In: Pierson, Paul (Eds.): The New Politics of the Welfare State. New York, NY: Oxford University Press: 410-456. Prinz, Christopher (Eds.) 2003: European Disability Pension Policies: 11 Country Trends 1970-2002. Aldershot: Ashgate. Radl, Jonas 2012: Too old to work, or too young to retire? The pervasiveness of age norms in Western Europe. In: Work Employment & Society 26,5: 755-771 [doi: 10.1177/0950017012451644]. Scharpf, Fritz W. 2001: Employment and the Welfare State: A Continental Dilemma. In: Ebbinghaus, Bernhard; Manow, Philip (Eds.): Comparing Welfare Capitalism: Social Policy and Political Economy in Europe, Japan and the USA. London: Routledge: 270- 283. Schröder, Heike; Hofäcker, Dirk; Müller-Camen, Michael 2009: HRM and the em- ployment of older workers: Germany and Britain compared. In: International Jour- nal of Human Resource Management and Development 9,2: 162-179 [doi: 10.1504/ IJHRDM.2009.023451]. Sjögren Lindquist, Gabriella 2006: Late careers and career exits in Sweden. In: Bloss- feld, Hans-Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.): Globalization, Uncertainty and Late Careers in Society. London: Routledge: 211-233. Warner, David; Hofmeister, Heather 2006: Late career transitions among men and wom- en in the United States. In: Blossfeld, Hans-Peter; Buchholz, Sandra; Hofäcker, Dirk (Eds.): Globalization, Uncertainty, and Late Careers in Society. London: Routledge: 141-181. Weishaupt, J. Timo 2011: From the Manpower Revolution to the Activation Paradigm: Explaining Institutional Continuity and Change in an Integrated Europe. Amsterdam: Amsterdam University Press. Zaidi, Asghar; Fuchs, Michael 2006: Transition from Work to Retirement in EU25. In: Vienna Centre Policy Brief [http:www.euro.centre.org/data/1164874912_40611.pdf]. Reversing Early Retirement in Advanced Welfare Economies • 839 Prof. Dr. Bernhard Ebbinghaus (). School of Social Sciences and Mannheim Centre for European Social Research (MZES), University of Mannheim, Germany. E-Mail: ebbinghaus@uni-mannheim.de, URL: http://www.mzes.uni-mannheim.de/d7/en/profi les/bernhard-ebbinghaus Dr. Dirk Hofäcker. University of Duisburg-Essen and Mannheim Centre for European Social Research (MZES), University of Mannheim, Germany. E-Mail: dirk.hofaecker@mzes.uni-mannheim.de, URL: http://www.mzes.uni-mannheim.de/d7/en/profi les/dirk-hofacker A German translation of this reviewed and authors’ authorised original article by the Federal In- stitute for Population Research is available under the title “Trendwende bei der Frühverrentung in modernen Wohlfahrtsstaaten. Paradigmenwechsel zur Überwindung von Push- und Pull-Faktor- en”, DOI 10.12765/CPoS-2013-24de or URN urn:nbn:de:bib-cpos-2013-24de3, at http://www.com- parativepopulationstudies.de. Date of submission: 18.01.2012 Date of Acceptance: 05.03.2013 © Federal Institute for Population Research 2013 – All rights reserved Published by / Herausgegeben von Prof. Dr. Norbert F. Schneider Federal Institute for Population Research D-65180 Wiesbaden / Germany Managing Editor / Verantwortlicher Redakteur Frank Swiaczny Assistant Managing Editor / Stellvertretende Redakteurin Katrin Schiefer Language & Copy Editor (English) / Lektorat & Übersetzungen (englisch) Amelie Franke Copy Editor (German) / Lektorat (deutsch) Dr. Evelyn Grünheid Layout / Satz Beatriz Feiler-Fuchs E-mail: cpos@bib.bund.de Scientifi c Advisory Board / Wissenschaftlicher Beirat Jürgen Dorbritz (Wiesbaden) Paul Gans (Mannheim) Johannes Huinink (Bremen) Marc Luy (Wien) Clara H. Mulder (Groningen) Notburga Ott (Bochum) Peter Preisendörfer (Mainz) Board of Reviewers / Gutachterbeirat Martin Abraham (Erlangen) Laura Bernardi (Lausanne) Hansjörg Bucher (Bonn) Claudia Diehl (Konstanz) Andreas Diekmann (Zürich) Gabriele Doblhammer-Reiter (Rostock) Henriette Engelhardt-Wölfl er (Bamberg) E.-Jürgen Flöthmann (Bielefeld) Alexia Fürnkranz-Prskawetz (Wien) Beat Fux (Zürich) Joshua Goldstein (Rostock) Karsten Hank (Köln) Sonja Haug (Regensburg) Franz-Josef Kemper (Berlin) † Michaela Kreyenfeld (Rostock) Aart C. Liefbroer (Den Haag) Kurt Lüscher (Konstanz) Dimiter Philipov (Wien) Tomáš Sobotka (Wien) Heike Trappe (Rostock) Comparative Population Studies – Zeitschrift für Bevölkerungswissenschaft www.comparativepopulationstudies.de ISSN: 1869-8980 (Print) – 1869-8999 (Internet)