PARENTHOOD AND RETIREMENT: Gender, cohort, and welfare regime differences

The association between women's fertility and employment has received considerable attention in the social science literature, which documents significant variation in the observed correlations across time and between welfare states (e.g., Engelhardt and Prskawetz 2004; Matysiak and Vignoli 2008; Hilgeman and Butts 2009). Moreover, the employment-fertility-nexus has been shown to vary by gender: mothers might suffer from limited opportunities for paid employment under the same institutional regime that allows for positive income effects of fatherhood (e.g., Blossfeld et al.2001; Pollmann-Schult and Diewald 2007; also see Kmec 2011).

Parenthood and its immediate consequences for labour market outcomes are also likely to exhibit sustained influences on employment patterns over the family life course, particularly for women (e.g., Drobnič et al.1999; Brugiavini et al.2011). Even though children need not necessarily reduce one's total years in the labour force, they may create discontinuities in employment careers eventually affecting individuals' pension receipt. Thus ‘the retirement process is appropriately viewed as temporarily embedded in current incentive-disincentive structures that mediate retirement decision-making and in long-term family relations that constitute the joint role pathways of couples through work and family domains' (O'Rand et al.1992: 82; italics in the original).

Only few studies have yet investigated the long-term relationship between individuals' reproductive history and retirement, though. The limited evidence collected so far tends to suggest that having children might actually delay women's exit from the work force (e.g., Pienta 1999; Hank 2004). These and related findings have been interpreted as reflections of gendered role patterns developed across the family life course (e.g., Henretta et al.1993; Pienta et al.1994), but they are also likely to be affected by welfare state policies and institutions – which are not gender neutral either (cf. Daly and Rake 2003).

Investigations of the interplay between childbearing and later life labour force exits should thus ideally take a cross-national and gendered perspective. The third wave of the Survey of Health, Ageing and Retirement in Europe collected a broad set of retrospective life history information from respondents in 13 Continental European countries, offering unique opportunities to conduct such kind of research. Our analysis of these so-called SHARELIFE data allows us to add to the existing literature by providing initial answers to three questions in particular: first, are there systematic gender differences in the association between reproductive history and retirement (e.g., O'Rand and Landerman 1984)? Second, are longstanding differences in fertility, employment, pensions, etc., reflected in welfare regime differences in the association between individuals' reproductive history and retirement? One might assume, for example, that institutional contexts fostering younger mothers’ employment (e.g., Sweden) and such prohibiting women's labour force participation after childbirth (e.g., Italy) also shape older individuals' retirement decisions. Third, and finally, we ask whether there are cohort differences in the association between reproductive history and retirement (e.g., Ginn 2003).

The remainder of this article is structured as follows: the next section provides a brief overview of previous research on childbearing patterns and later life labour market behaviours. The two succeeding sections describe our data and measures, as well as our results. The final section concludes.

Research investigating men's and women's retirement behaviour needs to consider both the role of institutional (i.e., welfare state) contexts and the role of the individual's family life course in shaping employment decisions:

It is by now widely acknowledged that there is a structural gender bias in all welfare states which shapes women's behaviour differently from that of men (e.g., Daly and Rake 2003; Frericks et al.2009: 713). Macro-level gender inequalities on the labour market may be promoted or ameliorated by direct as well as by indirect welfare state interventions. For example, public services are a large employer of women in many countries (e.g., Gornick and Jacobs 1998; Okun et al.2007), whereas – at the same time – the tax system or the poor provision of public day care for children discourages female employment (e.g., Kreyenfeld and Hank 2000; Dingeldey 2001). Within social security, men are therefore generally over-represented in programmes granting rather generous entitlements based on contribution payments, whereas women are over-represented within means-tested welfare programmes providing less generous benefits and granting entitlements based on needs (cf. Leitner 2001). The earnings-related structure of social security systems and the privileged treatment of traditional family arrangements (including the traditional division of labour between the sexes) have been identified as main causes of women's discrimination within these systems.

Depending on a country's pension system (as well as on the individual's work-and-family-biography) this has important consequences particularly for women's old age security (for overviews see Döring et al.1994; Jefferson 2009). Ginn and Arber (1992) distinguish three western European models of pension systems: the ‘residual model’ (in the United Kingdom, for example), with low minimum pensions that are conditional on the contribution record, the ‘income security model’ (in Germany, France, and Italy, for example), where earnings-related schemes are the main pillar of the old age security system, and ‘basic security model’ (in The Netherlands and – to some extent – in the Scandinavian countries), providing a basic pension regardless of the individual work record. The authors conclude that ‘those who are for any reason disadvantaged in employment, especially women with domestic responsibilities, are more likely to receive an adequate pension in their own right where the pension system approximates to the “basic security” model (such as Denmark) rather than the “income security” (Germany) or “residual” (UK) model’ (Ginn and Arber 1992: 268). Which country would be preferable for women's old age security, though, very much depends on their work behaviour and care arrangements (e.g., Leitner 2001; also see Frericks et al.2009).

Closely related to this latter issue is research indicating that the retirement decision is a family decision (cf. O'Rand et al.1992). There is ample evidence suggesting a joint retirement decision of currently married husbands and wives (e.g., Henretta et al.1993; O'Rand and Farkas 2002), but individuals' marital and reproductive history has also been shown to be associated with older men's and women's labour force transitions and pension receipt (e.g., Yabiku 2000; Hank 2004). While recent evidence from Germany, for example, suggests that marriage and fatherhood tend to bear a positive association with men's career development (e.g., Pollmann-Schult and Diewald 2007; Tölke 2007), women with young children are almost universally less likely to be engaged in market work at all, and working mothers are more likely to be part-time employed than their childless counterparts (e.g., Brewster and Rindfuss 2000; Konietzka and Kreyenfeld 2010); also see Kmec (2011). Although this is by now quite well investigated, we know only relatively little about the work and retirement patterns of parents in their later years, i.e., in the period after intensive childrearing (e.g., Pienta et al.1994; Ruhm 1996): will parents exit the workforce earlier than the childless, or is parenthood associated with longer working lives (e.g., Pienta 1999; Hank 2004)?¹

Mothers should be more likely to have a male breadwinner in the household, who is expected to provide economic security throughout their life course. Along these lines, Hank and Erlinghagen (2011) found that – different from their female counterparts – older male workers in Europe are more likely to perceive their job security as poor if they are co-residing with children, independent of the household's actual economic situation. This might indicate that even close to retirement age, fathers of co-resident children who lose their job would also lose their traditional provider role, making them more sensitive to potential job loss. This would speak in favour of the hypothesis that paternity goes along with a later retreat from the labour market, whereas the reverse should be true for mothers.

However, independent of a potential partner's employment status, mothers might as well decide to work longer than childless women in order to make-up for losses in their own personal income and retirement benefits due to reduced work hours or employment interruptions during their reproductive phase. While such discontinuities need not necessarily reduce women's total years in the labour force, they may create discontinuities in employment careers, including temporary part-time work. The effect on (public and/or private) pension receipt might still be the same: various studies show that women are disadvantaged by childbearing and family care duties (e.g., Yabiku 2000; Evandrou and Glaser 2003; also see Sefton et al.2011a). This is of considerable relevance for women's economic well-being in old-age, particularly in a situation where the greater overall access or entitlement to pensions for younger cohorts of women is likely to be paralleled by future cuts in public retirement income sources. This shifts even more saving and investment responsibility to the individual worker, particularly if she needs to buffer the effects of (temporary) labour exits resulting from conflicts between family and market contingencies (e.g., Ginn and Arber 1999; Frericks et al.2007).

This study uses data from SHARELIFE, the third wave of the Survey of Health, Ageing and Retirement in Europe (SHARE; a detailed description is given in Schröder 2011; also see Börsch-Supan et al.2010). Between October 2008 and August 2009 individual life histories of respondents aged 50 or older who had already participated in at least one of the previous SHARE waves were collected in 13 countries: Austria, Belgium, the Czech Republic, Denmark, France, Germany, Greece, Italy, The Netherlands, Poland, Spain, Sweden, and Switzerland.

From this database we derived annual employment histories of 15,920 ever married men and women aged 50+ who reported at least one episode of gainful employment. These individuals contribute a total of 150,825 person-years of observation to the analysis. Running separate regressions for men and women, we estimate discrete-time logit models to analyse the individual's entry into retirement. These models estimate the effect of the covariates on the log-odds of the event. Different from conventional logistic regression analysis, discrete-time models use multiple observations for each individual in the sample, i.e., each time unit during which an individual is observed contributes a separate and independent observation to the input data (e.g., Jenkins 1995). Since the transition to retirement is defined here as a non-repeatable event, the individual is excluded from further observation once the event has occurred. Regression results are presented as odds ratios.

For our dependent variable ‘retirement’ there is no exclusive definition (e.g., Ekerdt and DeViney 1990). It is operationalised here as the last observed exit from paid employment after age 50, assuming inactivity as an absorbing state (note that observations are censored at age 80 or at the time of the SHARELIFE interview); see Blanchet et al. (2005) for an overview of different pathways to retirement in SHARE countries.

Our main explanatory variables are three indicators of individuals' fertility: first, a dummy variable indicating whether the individual ever had a child (note that 8–9% of the respondents in our sample remained childless); second, the number of children ever born (with an average of 2.1 in our sample);and third, binary indicators of an early (late, respectively) first birth, where the threshold was set at age 24 for women and age 27 for men (i.e., the median of the distribution for the pooled sample).

Further control variables include the individual's age, level of education (‘low’ = lower secondary level of education or less; ‘medium’ = upper secondary or post-secondary, non-tertiary level of education; ‘high’ = first stage of tertiary education or higher), and work experience (years in the labour force and number of jobs till the age of 50), as well as time-varying indicators of marital status (married vs. all other states), self-employment, and homeownership (cf. Hank 2004). To account for possible cohort differences in the relationship between fertility and retirement, we distinguish respondents born up until 1940 (who account for roughly one third of females and somewhat less than half of the men in our sample) from those born later.

For our regionalized analyses, we start from Esping-Andersen's (1990) initial welfare state typology to group the countries represented in SHARELIFE into four clusters which we label as ‘Social-democratic’ (DK, NL, SE), ‘Conservative’ (AT, BE, CH, DE, FR), ‘Mediterranean’ (ES, GR, IT), and ‘Post-communist’ (CZ, PL); see Table 1 for descriptive sample statistics.

Before turning to the discussion of the main explanatory variables, which also considers cohort and welfare regime interactions (see Table 3), we first describe the results for all other control variables in an initial model for the pooled SHARELIFE sample (Table 2). Most of these findings are consistent with previous research (e.g., Hank 2004) and very similar for men and women.

To begin with, the odds of entering retirement increases significantly with individuals' age, sky-rocketing after age 59 and yet again at ages 64 and over. Moreover, women belonging to the post-1940 cohorts retire later than those born before World War II. With regard to the role of marital status, there is evidence for married individuals entering retirement earlier than their unmarried counterparts, which might be explained by couples’ income pooling.

Women with higher educational degrees tend to stay in the labour force longer than their less qualified counterparts, whereas we find some indication for a reverse association between (medium) education and retirement among men. The number of years in employment till the age of 50 significantly influences the individual's decision to retire. A longer employment record results in earlier eligibility for old-age pensions and tends to speed up the transition to retirement. Moreover, those who were self-employed in their last reported job exhibit markedly lower risks of leaving the labour force than all others. Career interruptions (indicated by the number of employment spells) as well as homeownership, however, do not bear any association with the timing of retirement.

Table 3 summarises the results for different indicators of men's and women's fertility. In the initial model (also exhibited in Table 2) we operationalised individuals' reproductive history by an indicator of whether they ever had a child. Looking at the full samples of men and women, we do not detect any differences in the retirement behaviour of parents and the childless. If, however, women born pre-1940 are distinguished from those born later (Table 3a), we find that mothers belonging to the older cohorts tend to exit the labour force later (especially in conservative countries), whereas those belonging to the post-1940 cohorts are more likely to retire earlier than their childless counterparts. An alternative specification, based on the respondents’ number of children, confirms these cohort differences in the relationship between motherhood and women's retreat from the labour force (Table 3b). It seems worth noting that the positive correlation between a greater number of children and retiring earlier, which is statistically significant in the post-1940 cohorts, is particularly strong in social-democratic and post-communist countries, that is, in social contexts characterized by the highest levels of labour force participation among mothers with young children.

The association we detect between the number of children and retirement in the pooled sample of men is a weak negative one (Table 3f). If fatherhood as such (that is, independent of family size) is considered (Table 3e), we find a significantly lower odds of entering retirement in social-democratic countries among fathers born 1940 or later. The timing of fertility might matter here, because having had a late first birth seems to be a main driver of the observed relationship (see Table 3h). However, overall, the timing of parenthood does not seem to matter a great deal. Early fatherhood does not bear any correlation with men's retirement (Table 3g) and although there is some indication that mothers who experienced an early first birth exhibit higher odds of leaving the labour force than their childless counterparts (Table 3c), one clearly needs to acknowledge that – by and large – the coefficients of the variables representing women's early and late childbearing (Table 3d) are unlikely to be significantly different from each other. Moreover, if looking at the timing of fertility and women's entry into retirement particularly in conservative countries, we once more observe clear cohort differences along the lines already suggested above: both early and late mothers born pre-1940 exhibit a greater propensity to retire later than childless women, whereas (early) mothers belonging to the post-1940 cohorts tend to exit the labour force earlier.

Different from the employment-fertility-nexus earlier in life, the relationship between men's and women's reproductive history and later life labour force participation has received relatively little attention yet. Exploiting retrospective life-history data from the SHARELIFE project, this study set out to assess the long-term relationship between individuals' fertility and retirement in 13 Continental European countries. Our analysis aimed at providing initial answers to three issues in particular, namely whether there are gender, regime, and cohort differences in the proposed association between reproductive history and retirement.

With regard to gender differences, our findings provide some weak evidence that, on the one hand, having (more) children is associated with later retirement among men. This is likely to reflect child-related financial obligations (including potential support for grandchildren; e.g., Wang and Marcotte 2007) and thus a greater responsibility of fathers – especially if their first child was born late – as breadwinners lasting well into their late career phase. In the pooled sample of women, on the other hand, we observed no statistically significant association between motherhood (the number of children, respectively) and the individual's exit from the labour force.

Distinguishing between older (born pre-1940) and younger (born post-1940) cohorts, however, revealed strikingly different retirement behaviours of mothers and childless women. Mothers belonging to the older cohorts tend to exit the labour force later, whereas those born post-1940 exhibit a greater propensity to retire earlier than their childless counterparts. Our findings for women of the pre-1940 cohorts are consistent with results of previous research arguing that mothers’ delay retirement to compensate for reduced work hours or employment interruptions experienced during their reproductive phase (e.g., Pienta 1999; Hank 2004). The pattern observed for the post-1940 cohorts is new and requires further discussion.

Overall, our results show that women belonging to the post-1940 cohorts retire later than those born before World War II (see Table 2), suggesting a greater labour market attachment among the former, resulting from changes in both individual (e.g., human capital endowment) and institutional (e.g., termination of early retirement programmes) characteristics. Against this background in particular, it might be considered a surprise to find a reverse association among mothers in the younger cohort. A possible explanation for this is that the economic and institutional opportunities for an earlier exit of mothers in the post-1940 cohorts in the SHARELIFE sample – the youngest of which were born in 1956 – might have been greater than those experienced by older cohorts. In most Continental European countries, mothers in younger cohorts are likely to have enjoyed a greater overall generosity of public pension systems as well as relatively high individual pension benefits acquired through their own rather than their husbands’ employment and contributions.

The latter argument is supported by our observation that the association between childbearing and earlier retirement appears to be particularly strong in countries characterized by social-democratic and post-communist welfare regimes, which exhibit the highest levels of labour force participation among mothers with young children. However, against the background of a generally decreasing generosity of public pension systems and increased individual responsibility for old age security (e.g., Whiteford and Whitehouse 2006), one might expect to observe a return to the pre-1940 pattern of entering retirement in more recent and future cohorts of women, with mothers’ working longer (or not significantly shorter) than the childless.

Our study has several limitations. Unfortunately, SHARELIFE does not provide us with information from countries representing a ‘liberal’ welfare regime (but see, for example, Pienta 1999). Moreover, we lack of information on respondents’ pension schemes or their pension wealth at exit from the labour force. Another important issue that we could not address in our study but that clearly deserves attention in future research is how women's (mothers’, respectively) employment over the life course translates into later life economic well-being. To what extent might longer years of labour force participation actually improve individuals' retirement incomes (e.g., Sefton et al.2011a,b)? And what consequences will a further shift – parallel to increasing pressures of population ageing on pension systems – in income composition, from primarily public pension benefits to diverse forms of private pensions (e.g., Farkas and O'Rand 1998; Frericks et al.2007), bring about? So far, these issues have almost exclusively been addressed from a female perspective. However, as greater numbers of fathers get involved in family work, such as childcare (e.g., Geisler and Kreyenfeld 2011; Duvander and Johansson 2012), social scientists and policy makers will need to acknowledge more the role of fatherhood in later life labour force participation and pension receipt.

This paper uses data from SHARELIFE release 1, as of November 24th, 2010. The SHARE data collection has been primarily funded by the European Commission through the 5th framework programme (project QLK6-CT-2001-00360 in the thematic programme Quality of Life), through the 6th framework programme (projects SHARE-I3, RII-CT-2006-062193, COMPARE, CIT5-CT-2005-028857, and SHARELIFE, CIT4-CT-2006-028812) and through the 7th framework programme (SHARE-PREP, 211909 and SHARE-LEAP, 227822). Additional funding from the U.S. National Institute on Aging (U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, Y1-AG-4553-01 and OGHA 04-064, IAG BSR06-11, R21 AG025169) as well as from various national sources is gratefully acknowledged (see www.share-project.org for a full list of funding institutions).

Karsten Hank is Professor of Sociology at the University of Cologne and Research Professor at the German Institute for Economic Research (DIW), Berlin. His main research interests are in the fields of aging and families.

Julie M. Korbmacher is with the Max Planck Institute for Social Law and Social Policy, Munich. Her main field of research is survey methodology.