Generation and the propensity of long career interruptions due to childcare under different family policy regimes: A multilevel approach

Abstract

Spanish

French

This article analyses the generation gap in the duration of long-term career interruptions due to childcare among mothers of two children, and how the differences are moderated by a country’s predominant family policy regime. The outcomes of the multilevel analysis reveal that mothers born after 1960 have significantly lower odds of interrupting their career for longer than 10 years compared with older women. A country’s predominant family policy model plays a significant role in explaining the propensity of long career breaks. Mothers from countries with post-socialist, Southern European and pro-egalitarian models exhibit lower odds of having long-term career interruptions than those in pro-traditionalist countries. Differences between generations are moderated by countries’ family policy models. Among younger generations, the propensity to take long career breaks is lower in post-socialist and non-interventionist regimes than in countries with a pro-traditionalist family policy legacy.

Keywords

Career interruptions childcare family policy multilevel analysis policy regimes

Introduction

Empirical evidence suggests that women with young children are less likely to be engaged in the labour market than women without children or women with grown-up children (Connelly, 1992). Childbirth is a major event in a mother’s progress in the labour market (Vlasblom and Schippers, 2006). Career interruptions, and in particular long career breaks, may have a negative impact on both a woman’s individual wellbeing and societal wellbeing. On an individual level, long career breaks due to childcare affect the human capital of the parents concerned. They inhibit further career progression due to missed opportunities (Math and Meilland, 2004; Ruhm, 1998), leading to less improvement in income (Dex et al., 2008; Waldfogel, 1998) and pension entitlement, and consequently impact their wellbeing during old age (Walby, 1997). At the societal level, prolonged interruptions in a woman’s career might impose pressure on welfare state expenditure, as a result of lower contributions to the redistributive system (Sainsbury, 1996; Smeaton, 2006; Walby, 1997). This implies that understanding the factors associated with career interruptions is important both for academic research and for policymakers.

It has been stipulated in relevant literature that, among many other factors, the duration of career interruptions due to childcare is a generational issue and that younger cohorts of women show higher levels of labour market attachment and reduce the time spent caring for their child on a full-time basis compared with older generations (Esping-Andersen, 2002; Gornick and Meyers, 2003; Hakim, 2002; Walby, 1997; Ziefle and Gangl, 2014). It has also been argued that the labour market attachment of women, and consequently the duration of career interruptions due to childcare, is significantly affected by a country’s family policies (Budig et al., 2012; Dex and Joshi, 1999; Ejrnaes, 2011; Joshi et al., 1996). However, to our knowledge, there is no previous study that provides empirical evidence on how the effect of generation on the duration of career interruptions is moderated by countries’ family policy characteristics in a broader cross-country comparative perspective. Analysing generational differences in career interruptions due to childcare under different family policy regimes may help us to understand under which family policy regime younger generations show a lower propensity of long career interruptions compared with their older counterparts. This can consequently also indicate which regimes have improved the efficiency of their efforts towards helping work–family reconciliation.

In the present article, I aim to bridge this gap in the empirical literature by (1) exploring the generational differences in the propensity of experiencing long career interruptions due to childcare among mothers of two children in 18 EU countries clustered in five family policy regimes, and (2) examining whether these generational differences are moderated by a country’s family policy regime,¹ i.e. whether generational differences do vary significantly between the five family policy regimes analysed.

The findings presented in this article make several contributions to the current body of research. First, due to the retrospective information available in the European Social Survey (ESS) from 2004, it is possible to analyse child-related career interruptions among different generations of mothers in a unique number of European countries. Second, I simultaneously analyse the individual and the country-level determinants of the duration of career breaks due to childcare. At the individual level, special attention is paid to the effects of generation and level of education and their interactions. At the country level, the focus is on the effect of family policy regimes on the duration of career interruptions. Third, and most importantly, the article examines the micro–macro level interaction effect of the generation and the country’s family policy regime, which provides us with innovative empirical evidence on how different generations of mothers have reconciled family and professional life in different family policy contexts.

Career interruptions due to childcare

Among women, the most frequent forms of career interruptions due to childcare are child-related leave (maternity leave and parental leave) and labour market inactivity with no guarantee of returning to the same position (Chan et al., 2012).

Maternity leave is a period of job-protected leave for mothers, taken during the last weeks of pregnancy and during the weeks following childbirth. It is a health-related provision, allowing mothers to recover physically after child delivery. Parental leave policies have been implemented at different paces in different countries. The duration of parental leave, salary replacement rates, flexibility and take-up rates vary significantly from country to country.

Parents who are not eligible for parental leave or who have no other means of arranging childcare for their children, become or remain inactive in the labour market (i.e. officially out of the labour force), and their re-entry into their previous job position is not legally guaranteed. There are two forms of inactivity: inactivity without any payment for childcare responsibility, or inactivity with cash-for-care benefits that provide income support to parents who withdraw from the labour market due to care responsibilities, but that do not guarantee resuming the previous job position.

Factors associated with career interruptions due to childcare

Generation effect

The literature studied (Esping Andersen, 2002; Hakim, 2002; Pfau-Effinger, 2004; Smeaton, 2006; Walby, 1997; Ziefle and Gangl, 2014) mentions generations or birth cohorts as one of the important determinants of female labour market participation, and consequently of the duration of career breaks due to childcare. Walby (1997) claims that in industrial societies, age and generation define the regime that one has embodied and predetermines labour market attachment. Younger women who grow up in the public gender regime² are better equipped than older women for participation in the public sphere, because their education allows access to qualified, quality jobs. Older women, whose capacities, values, motivation and preferences were formed under the domestic regime, have different resources and vulnerabilities compared with younger women. They also have greater difficulty in adapting to structural changes.

In conformity with this, Fourage et al. (2010) state that younger generations of women are more likely than their older counterparts to continue working after childbirth if they are in jobs with long hours. Similar claims can be found in the preference theory of Hakim (2002), who differentiates between women born before and after 1960. The younger generation proved better able to choose their lifestyles as a result of greater affluence, educational attainment and labour market opportunities. This is in line with Pfau-Effinger’s (2004) observation that the generational change in labour market attachment is especially notable among young mothers interviewed in the 1980s. Similar findings are presented by Ziefle and Gangl (2014), who claim that younger women have shorter duration child-related employment interruptions, and that especially among these younger generations, the duration of these interruptions depends increasingly on a woman’s own education, occupational status and earnings.

Given the above arguments, it can be hypothesized that when using pooled data from all the analysed countries, the younger generation of mothers (born after 1960) exhibits a lower propensity to take long-term career breaks due to childcare than their older counterparts (H1).

The above literature review suggests that the generational differences are strongly related to the educational attainment of women (the younger generation tends to be higher educated than their older counterparts). Thus, it is hypothesized that among all the educational groups (i.e. women with post-secondary, secondary and primary education), the younger generation of mothers shows lower odds of taking long career breaks. However, this trend is significantly more pronounced among women with a high level of education than among those with lower educational attainment (H1a).

Family policy regimes

The literature also suggests that female labour market participation, and thus also career breaks due to childcare, depends on the institutional circumstances under which women live. A country’s welfare state characteristics (Esping-Andersen, 2002; Evertsson and Grunow, 2012; Hantrais and Letablier, 1996; Pfau-Effinger, 2004; Sainsbury, 1996; Walby, 1997) and family policy arrangements (Budig et al., 2012; Gornick and Meyers, 2003; Lewis, 1992; Lewis and Smithton, 2001) are, among others, important aspects of these institutional circumstances.

In the early twentieth century, modern welfare states were built based on full-time male employment and stable families (Lewis, 2006). The traditional male breadwinner and female homemaker model – in which men and women assumed more separated roles, where men were considered primarily as earners and women as homemakers – mainly linked women to the domestic sphere (Lewis, 1992). Since the second half of the last century, the traditional male breadwinner model has evolved towards the dual-earner model (Gornick and Meyers, 2003) or what is termed the adult worker model (Lewis, 2001), in which most men work full-time and women’s labour market engagement varies across countries: from marginal part-time to full-time employment.

Lewis and Smithton (2001) and Pfau-Effinger (2004) argue that different welfare systems have endorsed this shift from the traditional breadwinner model to a more gender-equilibrated model in different ways and at different paces. Countries with a ‘traditional’ view on the division of labour between men and women are reluctant to adopt policies where the reconciliation of care and paid work is facilitated. By contrast, countries with a stronger gender equality approach tend to introduce family policies where caring and paid work is spread more equally between both parents. These different approaches in policy efforts have led to cross-country differences in female labour market activity and inactivity, and consequently, the duration of career breaks due to childcare.

Sorrentino (1983) shows that in 1961, Sweden (i.e. representative of the Nordic countries) exhibited the highest female labour market activity rate at 44%, compared with France (41%), Germany (41%), the UK (40%) and Italy (34%). In the course of the 1960s and 1970s, labour market activity rates have continuously increased in these countries, except for Germany and Italy, where an opposite trend is observed until the beginning of the 1980s. Since the 1990s, female activity rates have continued to grow in the EU countries. Nevertheless, the differences across countries remain notable, with the lowest rates in Southern European countries and the highest in the Nordic countries (Musumeci and Solera, 2013). Prior to the fall of socialist regimes, full employment was promoted in Eastern European countries, and therefore activity rates were high (Saxonberg and Sirovátka, 2006). After the fall of these regimes at the beginning of the 1990s, the Eastern European countries have exhibited medium to low employment rates (Musumeci and Solera, 2013).

Musumeci and Solera (2013)³ suggest that among younger generations of women (those who were at reproductive age in the course of the mid and second half of the 1990s), adjustment to motherhood (i.e. labour market withdrawal or reduction of working hours) is lowest in the Southern European and Nordic countries. By contrast, among all the countries analysed, the UK demonstrates the highest levels of labour market withdrawal after birth.

The most common individual indicators that have been used in the academic literature to define family policy in a country are: the extent of public day-care services for the youngest children, paid maternity leave and parental leave (Budig et al., 2012; Fux, 2002; Mandel, 2012; Sjoberg, 2004), state homecare for the elderly, the level of child allowance for children under the age of majority, family tax benefits to a non-economically active spouse and for children, childcare leave benefits (Sjoberg, 2004), a joint taxation system (Fux, 2002) and part-time employment (Ejrnaes, 2011).

The ideo-typical regime approach highlights major similarities and dissimilarities across countries with regard to selected family policy indicators (Gauthier, 1996). Countries are attributed certain general features they share with other countries, thereby reducing the complexity of comparing a larger number of indicators in a large number of countries (Muffels et al., 2002), which is the main advantage of using the ‘regime’ approach to describe the nature of a country’s family policy.

Various classifications of family policy regimes can be found in the relevant literature (Ejrnaes, 2011; Fux, 2002; Gauthier, 1996; Sjoberg, 2004). The classifications mentioned resemble each other, even though they were constructed at different times, and using different indicators and clustering methods.⁴ To summarize the underlying categories of family policy regimes proposed in these classifications, we use Gauthier’s work (1996), distinguishing the existence of four models (i.e. regimes). In the pro-natalist model (e.g. France), encouraging childbearing is seen as a priority, emphasis is placed on cash benefits, and relatively high support is also given to maternity leave and childcare facilities. The pro-traditional model (e.g. Germany) emphasizes the traditional role of the family. It supports the family, especially by encouraging a traditional male breadwinner model, primarily through the taxation system and the low or insufficient provision of childcare. Long childcare leave is granted to mothers. The pro-egalitarian model (e.g. in the Nordic countries) primarily promotes the importance of equality between men and women in the labour market and in childcare by supporting families, especially working parents. Parental leave and childcare facilities are central to this model. The non-interventionist model (e.g. in the UK) focuses mainly on families in need and limits state intervention in family life. Cash benefits and maternity leave are kept at low levels. In this model, informal and employers’ engagement in childcare are encouraged. The Southern European countries should be grouped into a distinct model and are characterized by short leave, insufficient childcare facilities and uncommon part-time working.

Ejrnaes (2011) proposes also treating post-socialist countries as a distinct category, because they exhibit long parental leave duration, low childcare use and limited part-time work opportunities.⁵

It needs to be mentioned here, however, that this typical regime approach also has its drawbacks. For example, it tends to ignore the dynamics of policies and their discontinuities over time as well as neglecting within-regime country differences (Gauthier, 1996). There is a vast body of literature elaborating on these problems. For example, Pascall and Kwak (2005) argue that after 1989, the post-socialist countries show varying levels of support for gender equality. In line with this, Szelewa and Polakowski (2008) claim that the post-socialist countries can be clustered into four distinctive models, depending on the structure and level of work–family reconciliation policies and common patterns of childcare after 1989. Javornik (2014) also points out that there are notable within-regime differences, and claims that based on the most recent data, the ‘exceptionalism’ of the post-socialist regime is not justifiable and the countries concerned share core characteristics with developed welfare state regimes.

However, given the purposes of the current article, which focuses on general tendencies in family policies over past decades – and leaving aside the situation after 2004 (which is characterized by the most dramatic development of family policies in EU states) – I make use of the aforementioned advantages of the regime approach and set out the following hypotheses based on the existing family policy regime classifications outlined above.

In the light of previous arguments, it can be stipulated that the propensity to take long career breaks due to childcare depends on the family policy regime that women are subject to. The highest propensity of long-term career interruptions is expected in pro-traditionalist countries. Conversely, post-socialist countries and countries with pro-egalitarian family policy orientation will exhibit the lowest propensity of these breaks (H2).

Given the fact that some policy regimes create better conditions for women to reconcile family and professional life than others do, it can be expected that the generation differences in the propensity to take long-term breaks are less pronounced in countries with pro-egalitarian and post-socialist family policies (H3). This is due to the fact that in these countries, work–family reconciliation policies (and consequently female labour market participation) have been traditionally more supported and therefore the policy context of older generations is not as dramatically different from that of younger generations as in other regimes. Moreover, in the majority of post-socialist countries, it has even been observed that the family policy context of younger generations (after 1989) has been more pro-familialistic than that of older generations.

Data and methods

For this study, I use ESS data from 2004 (mainly the module on family, work, wellbeing and work–life balance) covering the 18 countries listed in Table 1. This dataset is the only source of retrospective comparative data on the duration of career interruptions due to childcare and which covers a large number of EU countries.⁶ The module of the ESS data regarding career breaks only includes women who had at least one child. This subsample (not restricted by age) contains 11,562 observations, out of which 3110 (26.9%) were mothers of one child, 5153 (44.6%) of two children, 2095 (18.1%) of three children and 1204 (10.4%) of four or more children.

Table 1.

Total time spent at home caring for children on a full-time basis, percentages.

	No time	<= 6 months	> 6 months, <= 12 months	>1 year, <= 2 years	> 2 years, <= 4 years	> 4 years, <= 10 years	> 10 years	Total	N	% of total N
Estonia	11.8	10.1	14.2	23.7	28.4	11.2	0.6	100.0	341	7.5
East Germany	4.2	13.9	18.1	27.8	26.4	8.3	1.4	100.0	75	1.6
Slovenia	8.1	9.0	19.2	42.3	13.2	6.0	2.1	100.0	245	5.4
Czech Republic	2.2	4.1	6.7	17.7	35.5	30.9	2.9	100.0	461	10.2
Poland	17.7	16.8	12.3	10.9	19.1	19.5	3.6	100.0	232	5.1
Portugal	21.9	49.8	7.8	6.4	7.8	1.8	4.6	100.0	235	5.2
Finland	9.2	7.3	7.7	19.6	24.6	23.8	7.7	100.0	263	5.8
Denmark	17.9	15.8	20.1	18.4	11.1	8.1	8.5	100.0	236	5.2
Sweden	2.2	2.2	8.6	33.0	32.6	11.8	9.7	100.0	281	6.2
Spain	36.2	11.8	17.3	7.9	9.4	7.1	10.2	100.0	171	3.8
Norway	6.0	4.7	9.1	29.3	18.1	19.8	12.9	100.0	232	5.1
Belgium	26.6	19.6	14.7	9.8	4.9	7.6	16.8	100.0	191	4.2
Greece	41.4	15.3	8.7	3.7	5.6	8.1	17.1	100.0	346	7.7
Luxembourg	26.4	20.1	11.8	4.9	6.9	11.8	18.1	100.0	160	3.5
Austria	9.8	3.5	8.4	16.1	20.6	22.7	18.9	100.0	301	6.7
West Germany	11.4	7.4	8.1	11.4	15.4	25.7	20.6	100.0	282	6.2
UK	12.4	4.1	2.8	10.3	12.4	28.3	29.7	100.0	220	4.9
Switzerland	16.3	9.3	5.7	5.7	9.3	17.5	36.2	100.0	251	5.5

Source: ESS, 2004. N = 4522.

In this article, I only work with a subsample of mothers with two children. The first reason is that this subsample represents a large enough group of analysed mothers. Mothers of two children are more numerous than single-child mothers or mothers of three and more children, which allows for more reliable cross-country comparisons. The second reason is based on the fact that the number of children affects labour market strategies and consequently the career interruptions of a mother (Lalive and Zweimüller, 2009; Vlasblom and Schippers, 2006). This is because having two children represents a greater risk to a woman’s further career development (OECD, 2002), which appears to be more relevant to the choice than for mothers of a single child. The third reason is that particular characteristics (sociodemographic characteristics and life preferences) change with the number of children. Therefore, to compare relatively homogeneous groups, the analysis was conducted on mothers with the same number of children.

As the focus here is only on analyses for the postwar situation, I restricted the sample to mothers born after 1926. Working with data collected in 2004 obviously means that the presented analysis cannot cover the situation after this year. Given this, the most recent dramatic developments in family policy arrangements, family–work reconciliation patterns and the labour market engagement of mothers cannot be captured or commented on. The findings in the present article should therefore be interpreted in light of this fact.

As a consequence of the restrictions, the final sample analysed contains data for 4522 mothers. The last two columns in Table 1 provide more information regarding the number of analysed observations by country. Summary statistics of selected sociodemographic characteristics split by country (i.e. sample profiles) are available in the Appendix (Table A1). The differences in the main sociodemographic characteristics among different family policy regimes are discussed in detail in the section concerning control variables.

Method

The analysis was conducted using a multilevel analysis approach, which allows for the simultaneous estimation of individual-level and macro-level (in this case, country) effects. The multilevel approach is ideal for the calculation of cross-level effects, while controlling for factors at both levels (Mandel, 2012). Due to the binary nature of the dependent variable, the software STATA xtmelogit command was used.

When working with a subsample of mothers of two children it is necessary to control for the self-selection (Heckman, 1979) of mothers (Fourage et al., 2010; Schober, 2013) and selection bias in multilevel models (Bunel and Guironnet, 2011). In the first step, a set of probit models was estimated. This measures the propensity of having two children (having two children = 1, otherwise = 0). The covariates in the model are: religiosity, measured by church attendance⁷ (Zhang, 2008); age; age squared; educational level; and gender-role attitudes (Schober, 2013). For further details concerning the covariates, please refer to the section on control variables. The outcomes of this estimation served to calculate an Inverse Mills Ratio (IMR). The IMR was then used in multilevel models as an individual-level control variable.

In multilevel models with selection, the precise formulation of variance-covariance is problematic and therefore the bootstrap method allows for the substitution of the calculation of the asymptotic for such distribution (Bunel and Guironnet, 2011). In total, 100 bootstrap samples were drawn to estimate standard errors in the multilevel models.

A set of multilevel models were estimated. First, I estimated a null (i.e. unconditional) model (M0). This model has no predictors and includes solely the dependent variable. It was used to calculate the intra-cluster correlation coefficient (ICC), which estimates how much unexplained variation in the outcome exists between country-level units. This informed me about the usability of the multilevel modelling framework. I then estimated separately the following models. First, an individual-level model (M1), including the dependent variable, the main explanatory variable at the individual level (i.e. generation variable) and individual-level controls. Second, an individual-level model with interactions (M1a) adding to the previous model the interaction terms between the generation variable and educational attainment. Third, a country-level model (M2), augmenting model M1 by family policy regime and country-level covariates. Last, a model (M3) adding to M2 the interaction terms of generation and family policy regimes.

Dependent variable: Long career interruptions due to childcare

The dependent variable ‘long career interruptions due to childcare’ was constructed from the ESS question: ‘Including any time spent on maternity and parental leave, approximately how long, in total, have you spent at home caring for your child(ren) on a full-time basis?’ There were seven response categories considered for analysis (see Table 1). The ‘do not know’ category was treated as missing cases. For the purposes of this article, the variable was dichotomized: category 1 includes women who had taken a career break of 10 years or more and the remaining categories were categorized as 0. This step can be justified, because according to Moss and Deven (1999), it is mainly those interruptions which extend beyond the period of parental leave that tend to promote labour market exit and therefore the accumulation of all the negative effects of career interruption. By distinguishing between mothers who spent 10 or more years on childcare and those who spent less time, we made certain that the duration of career breaks in all the analysed countries exceeded the summed duration of two consecutive statutory parental and maternity leaves. Choosing a duration of career breaks that would be (for a mother of two children) shorter than the sum of the duration of two maternity leaves and two parental leaves, could lead to the risk of mixing career breaks due to leave that guarantees mothers a return to the labour market, and career breaks where women exited the labour market. Another reason for dichotomization of the dependent variable is that there are insufficient observations per country. A more detailed categorization would leave only a few observations in some categories, which would negatively affect the statistical power of our analysis. The limitations of this dependent variable are discussed in the Conclusions section.

Table 1 shows the descriptive analysis of the original variable indicating total time spent at home caring for children on a full-time basis by country. The data reveal that taking long-term career breaks (10 years or more) varies significantly across countries, from 0.6% in Estonia to 36% in Switzerland.

Key explanatory variable at the individual level: Generations

The term ‘generation’ used in this article refers to birth cohorts, i.e. people who were born in the same period of time and socialized in a similar way (Edmunds and Turner, 2002; Walby, 1997). In line with Kertzer (1983), generations are defined as a wider range of birth cohorts that are characterized by living in particular historical periods. Even though it is a specific historical event or set of events that define(s) such generations, they are often linked to cohorts of people who are thought to be particularly affected by this event or these events. Members of a generation are therefore exposed to similar period effects (i.e. particular institutional settings, norms, workplace cultures, etc.).

Based on the arguments found in literature (Fourage et al., 2010; Hakim, 2002; Pfau-Effinger, 2004; Smeaton, 2006), mothers were divided into two generations. Generation 1 includes mothers born between 1927 and 1960 (i.e. women who reached the age of 18 after 1945 and who reached their reproductive age after the Second World War). Generation 2 comprises women born in or after 1961 (i.e. women whose socialization process took place during or after the 1960s and who were at the peak of their fertile age during the 1970s or 1980s).

To make sure that all respondents had an equal chance to report career interruptions of more than 10 years, only women whose youngest child was more than 10 years old (i.e. born in 1994 or before) and who had been active in the labour market at the time of the interview were included in the analysed sample. This indicates that these women had already terminated their caring responsibilities and thus could have reported the correct and accomplished duration. In the analysed sample, 66.1% of women belong to Generation 1 and 33.9% to Generation 2.

Using the generation variable, it is not possible to clearly distinguish whether the effect of this variable on the duration of career breaks can be attributed to the effect of the period during which the women had their children (period effect) or to the generation effect (the fact that younger generations of women in general tend to be more work and career oriented compared with older generations) (Esping-Andersen, 2002; Hakim, 1992).

Key explanatory variable at country level: Family policy regimes

The creation of the instrument measuring family policy regimes is based on the above-described classifications of family policy models proposed by Gauthier (1996) and Ejrnaes (2011). Hence, to account for the family policy regime, a set of five dummy variables was used, representing each regime: post-socialist countries (the Czech Republic, Poland, Estonia and Slovenia), pro-traditionalist (Luxembourg, Switzerland, Belgium, Germany and Austria), non-interventionist (UK), Southern European (Spain, Portugal and Greece) and pro-egalitarian (Sweden, Finland, Norway and Denmark). The pro-traditional regime was taken as the reference category.

To provide empirical cross-validation of this regime typology, hierarchical cluster analysis (Gough, 2001) was applied to the selected attributes of family policy efforts such as: childcare usage under three years of age, the duration of parental leave, parental leave income replacement rate, expenditure on family services (childcare), duration of existence of parental leave arrangements in the country (in years) and the tradition of joint taxation (one of the policy measures that stimulate the labour market withdrawal of women after getting married or having a child). These indicators cover the situation in 2004 as proxies for long-term policy efforts based on the pass-dependency paradigm (Pierson, 2000), because comparable longitudinal data concerning the required indicators are not available. A detailed summary of the aforementioned attributes of family policies separated by countries is presented in the Appendix (Table A2). A description of the cluster analysis is available on request. The empirically obtained clusters of family regimes in general corroborate the classifications of Gauthier (1996) and Ejrnaes (2011).

Control variables

At the individual level, the study controls for human capital measured by the highest level of educational attainment: post-secondary, secondary or primary. As a proxy for vulnerability in the labour market, we used a retrospective variable measuring whether the respondents had ever been unemployed and seeking work for a period of more than three months. To account for the possible impact of family tradition in the labour market participation of women, a set of dummies measuring whether the mother of the interviewed woman was working or not when the interviewee was 14 years old was included in the analysis. The model also controls for individual gender-role attitudes, as existing literature suggests that women’s work–family preferences (Hakim, 2002) and gender-role attitudes (Crompton et al., 2005; Kangas and Rostgaard, 2007) are important factors associated with women’s labour market attachment and behaviour. Crompton et al. (2005) and Kangas and Rostgaard (2007) suggest that there is a positive relationship between gender egalitarian attitudes and the labour market supply of women. It has also been proven that egalitarian gender-role attitudes are associated with a more gender-egalitarian division of household chores (Shelton and John, 1996). Gender-role attitudes are measured by a summed index that is composed of two highly correlated items in the ESS questionnaire: ‘A woman should be prepared to cut down on her paid work for the sake of her family’ and ‘When jobs are scarce, men should have more right to a job than women’.

It is important to be aware of the differences in basic observed individual sociodemographic characteristics of analysed mothers among different regimes. A descriptive inspection of the data reveals that with regard to education, the highest proportion of mothers with post-secondary education is found in pro-egalitarian countries (39.6%), followed by post-socialist (23.2%), non-interventionist (21.8%), pro-traditionalist (19.1%) and Southern European (13.7%) countries. The highest percentage of mothers who reported having been unemployed in the past are found in post-socialist and Southern European countries (about 28% for each) and the lowest in non-interventionist countries (about 17%). Approximately 72% of sampled mothers in post-socialist, 63% in pro-egalitarian, 48% in non-interventionist, 43% in Southern European and 39% in pro-traditionalist countries reported that their mother was working when they were 14 years old, which indicates the historical level of female labour market engagement in these respective regimes. Looking at the proportion of mothers who occupied a supervisory position in their last job, the highest proportion is identified in non-interventionist regimes (37%), followed by pro-egalitarian (24.4%), post-socialist (21.5%), pro-traditionalist (20.1%) and Southern European (10.7%) countries. In this article, the differences found in basic individual characteristics are accounted for by using multilevel regression analysis.⁸

Marital status, working part-time versus full-time, being employed either in the public or the private sector, household income, occupation, social class, partner’s occupation and current labour market position could not be included in the analysis, because they refer to 2004 and therefore do not reflect the situation of the mothers when they were considering taking up childcare responsibilities.

At the country level, I controlled for predominant gender-role norms. Pfau-Effinger (2012) and Crompton (2006) state that dominant beliefs in society about appropriate gender relations (which Pfau-Effinger labels as ‘gender culture’ and Crompton ‘cultural norms’) and social policies interlink in their relationship with female employment. In countries demonstrating more egalitarian norms, support for women’s employment is greater than in less egalitarian ones (Fuwa, 2004; Musumeci and Solera, 2013). Thus, controlling for gender-role norms at the country level allows for separating the effect of family policy regimes (measured mainly by institutional characteristics) from that of norms. In this article, the variable of gender-role norms is defined as an aggregated country mean of two items measuring gender attitudes.⁹ Because generations in the analysed countries might differ with regard to the level of educational attainment, another control – the mean difference in the obtained educational level between generations – was calculated for each country.

All the continuous control variables, both at the individual and the country level, were grand mean centred (Hox, 2010). Descriptive statistics of the control variables are available in the Appendix (Table A1).

Results

A descriptive analysis of the distribution of the dependent variable by family policy regime (see Table 2) shows that mothers from pro-traditionalist and non-interventionist countries are more likely than expected to have long-term career interruptions (26.6% and 29.5% respectively). This contrasts with women from pro-egalitarian (9.7%) and post-socialist countries, where the incidence of career breaks exceeding 10 years is notably small (2.1%).

Table 2.

Variable on long career breaks by family policy regimes.

Regimes		Career breaks
		< 10 years	=>10 years
Pro-traditionalist	%	77.4%	22.6%
	ADR	−13.1	13.1
Pro-egalitarian	%	90.3%	9.7%
	ADR	2.5	−2.5
Non-interventionist	%	70.5%	29.5%
	ADR	−6.7	6.7
Southern European	%	88.3%	11.7%
	ADR	0.1	−0.1
Post-socialist	%	97.9%	2.1%
	ADR	12.9	−12.9

Source: ESS, 2004. N = 4522.

Note: ADR – adjusted residual.

The data also reveal that in the majority of the countries analysed, the younger generation is less likely to report long career breaks. Inter-generation differences are particularly notable in pro-traditionalist and non-interventionist countries, where a higher percentage of women from the older generation indicate long career breaks (approximately 30–40%) than the younger generation (less than 15%). An opposite tendency can be found in the post-socialist countries (and to a lesser extent in the pro-egalitarian countries), where only a few women report career interruptions of 10 years and longer. Thus, the inter-generation difference is very small (approximately 2–5%) and in the Czech Republic and Estonia, the younger generation actually reports slightly longer career breaks than their older counterparts (approximately 1%).

The following text presents the outcomes of the multilevel models (see Table 3). In model M0, the intra-cluster correlation (ICC) is equal to 0.28, which indicates that 28% of the total unexplained variance can be attributed to country-level characteristics. Therefore, the use of country-level variables is justified. Model M1 shows that when controlling for key individual characteristics, the generation variable has a significant effect on the propensity to take long career interruptions. Women born after 1961 are four times less likely than the older generation to have career interruptions of 10 years or more (the odds ratio is 0.25 and significant at the .001 level). This corroborates Hypothesis 1.

Table 3.

Multilevel analysis: individual- and country-level factors affecting the propensity of experiencing long career breaks due to childcare, odds ratios.

	M0	M1	M1a	M2	M3
Personal-level factors
Education
Post-secondary (ref. cat.)
Primary		2.971***	2.624	2.887***	2.897***
Secondary		2.525***	2.234	2.564***	2.561***
Gender-role attitudes – centred		0.722***	0.722***	0.718***	0.714***
Supervisory position: No (ref. cat.)
Yes		0.612**	0.611**	0.619*	0.614*
Missing		1.862	1.848	1.784	1.800
Experience with unemployment: No (ref. cat.)
Yes		1.135	1.131	1.121	1.105
Mother worked when 14: Yes (ref. cat.)
No		1.301*	1.301*	1.286*	1.277
Other		0.855	0.848	0.840	0.812
IMR (Inverse Mills Ratio)		1.299	1.316	1.322	1.262
Generation (born before 1961 ref. cat.)		0.250***	0.107	0.247***	0.188***
Interactions of education and generation
Generation*primary			2.721*
Generation*secondary			2.510*
Country-level factors
Family policy regime
Pro-traditionalist (ref. cat.)
Pro-egalitarian				0.632	0.635
Non-interventionist				1.565***	1.284
Southern European				0.221***	0.202***
Post-socialist				0.066***	0.052***
Gender-role norms – centred				0.687	0.688
Difference in educational level				1.628	1.596
Interactions of family policy regimes and generation
Pro-egalitarian*generation					0.785
Non-interventionist*generation					3.334*
Southern European*generation					1.826
Post-socialist*generation					4.352**
N	4196	4196	4196	4196	4196
Log likelihood (LL)	−1352	−1223	−1219	−1204	−1197
BIC	2704	2317	2312	2281	2268
Intra-cluster correlation (ICC)	0.28	0.29	0.29	0.035	0.035

Source: ESS, 2004; *p < .05, **p < .01, ***p < .001; ref. cat. = reference category.

The results of model M1 are in line with Hypothesis 1a, which states that among all the education groups, the generation variable affects the odds of taking long career breaks in a similar way (in the same direction). Specifically, regardless of educational attainment, younger mothers are less likely to take long career breaks. However, it needs to be noted here that despite this similar trend, there are notable differences in the magnitude of the intra-generation gap across different educational groups. The largest generational difference in the odds of taking long labour market interruptions due to childcare is found among women with post-secondary education, whereas the smallest difference between the younger and older generations is observed among women with primary education.

After introducing macro-level variables (M2), it also becomes apparent that the family policy regime is associated with the propensity of long-term career breaks. Mothers in post-socialist and Southern European countries exhibit significantly lower odds of reporting long career interruptions compared with their counterparts in pro-traditionalist countries (the odds ratios are 0.06 and 0.22 respectively, and are both significant at the .001 level). The opposite is found among non-interventionist counties, where analysed mothers have higher odds of taking long interruptions compared with mothers from pro-traditionalist countries (the odds ratio is 1.57 and is significant at the .001 level). No significant differences are found between mothers (generations combined) in pro-egalitarian regimes when controlling for selected individual characteristics, and a country’s gender norms and differences in average educational attainment.

To examine how the effect of the generation variable is moderated by the family policy regime in the countries analysed, the results presented in the interaction model (M3) should be examined. It appears that in post-socialist and non-interventionist regimes, the differences in the propensity of long career breaks between generations are significantly smaller than in countries with a pro-traditional family policy regime. Southern European and pro-egalitarian countries do not differ significantly from the reference group.

Conclusions

The main aims of the article were, first, to examine the propensity of having a long career interruption due to childcare among mothers of two children, depending on different generations of mothers from different European countries, and second, to analyse how these generational differences are related to the dominant family policy regimes in those countries. Multilevel analysis was applied. This accounted for the selection of mothers with two children and controlled for a set of key individual and contextual factors.

The presented analysis has certain limitations that have an effect on the results. One such impact relates to the way in which the duration of career breaks due to childcare was measured in the ESS questionnaire. The variable is of a categorical nature, with categories having relatively broad and unequal durations. It would be preferable to have exact information on the duration of career interruptions in months. However, the data were used because it is the only available variable that provides retrospective information on the duration of career interruptions for mothers of different generations in many European countries. Another limitation might relate to the retrospective nature of the dependent variable. The further in the past the event that is being retrospectively recalled happened, the less precise the recollection tends to be (Jacobs, 2002). Therefore, older women may have more difficulty remembering the exact duration of their withdrawal from the labour market. However, this recall bias might have been minimized by the abovementioned crude categorization of the duration variable.

It should also be kept in mind that the dependent variable used encompasses all variations of career breaks due to childcare and does not allow us to discern between women who were inactive in the labour market and those who took maternity or parental leave (i.e. officially employed). Nevertheless, Chan et al. (2012) argue that this measurement of career interruptions may yield better results than would be obtained from separating leave breaks and career interruptions without adequately specifying in the questionnaire what these terms exactly mean and whether they are exclusive.

Another possible drawback is the use of family policy regimes as the key country-level factor, which might mask some inter-cluster country differences. However, in the context of this study – which covers a long period of time – it was necessary to find a categorization of countries that would describe the essence of the family policy efforts over time. Moreover, the regime variable used is validated by existing categorizations realized at different points in time (Ejrnaes, 2011; Fux, 2002; Gauthier, 1996). It would have also been preferable and more informative to work with a more detailed generation variable, rather than a somewhat broadly defined one. However, this was not possible due to the relatively small number of cases available in the ESS dataset.

Despite these possible limitations, the article presents clear and original findings. For example, the outcome of the analysis reveals that there is a significant difference between generations in the odds of taking long-term career breaks due to childcare. The younger generation – i.e. women born after 1961 and therefore bringing up their children in the course of the late 1970s and 1980s – have lower odds of interrupting their career for 10 years or more compared with the older generation. This finding is not surprising and is in conformity with general expectations derived from the literature studied (Esping-Andersen, 2002; Fourage et al., 2010; Hakim, 2002; Pfau-Effinger, 2004; Smeaton, 2006; Walby, 1997).

The outcomes of empirical analysis also reveal that this general trend can be found among all the educational groups of analysed mothers. This implies that younger women, regardless of their educational level, take shorter career breaks. However, the magnitude of the generation gap varies across different educational groups. The greatest generation gap is found among women with post-secondary education and the smallest is for women with primary education. This finding indicates that young women with a higher educational level (and consequently greater human capital) demonstrate the most significant generational change with regard to their family–work life reconciliation, which is in line with Hypothesis 1a. This finding may be explained by the fact that women with a higher level of education have greater opportunities to benefit from the prevailing labour market settings (in concrete terms, they can obtain jobs with higher pay and better stability, which motivates them to re-enter the labour market earlier due to relatively high income loss). Another possible explanation may be that women with post-secondary education exhibit less traditional gender-role attitudes compared with their lower-educated counterparts. Hence, they prefer the dual-earner family model to the traditional breadwinner model (in line with Knudsen and Waerness, 2001) and select partners who comply with these views.

Country-level characteristics relate significantly to the propensity to take long career breaks. The outcomes of the analysis suggest that women in countries with post-socialist and Southern European family policy regimes are significantly less likely to take long career breaks due to childcare than women from countries with a pro-traditionalist regime. This only partially corroborates Hypothesis 2 and suggests that in non-interventionist regimes, the labour market participation of mothers with two children is even lower than in pro-traditionalist countries.

One possible explanation for the above findings regarding post-socialist regimes is that full employment was promoted under socialist rule. Career interruptions due to childcare were limited only to maternity or parental leave, and labour market inactivity due to childcare was a rare phenomenon. A high level of female labour market participation was maintained, even after the fall of the socialist regimes (Saxonberg and Sirovátka, 2006).

The findings for the Southern European regime (mainly the fact that mothers tend to report shorter career breaks compared with those in pro-traditionalist regimes) may have several explanations. The first stems from the empirical findings (see Table 1) that a relatively high proportion of mothers in these countries (more than one-third) report no full-time career breaks due to childcare. This is approximately twice as many mothers as in pro-traditionalist countries. This phenomenon may be due to the fact that older generations of mothers from these countries worked in the agricultural sector to a greater extent than in the other countries analysed (Roser, 2015). Roser indicates that in the 1980s, agricultural employment as a share of total employment accounted for approximately 17% in Spain, 24% in Portugal and 28% in Greece, whereas in pro-traditionalist countries it was approximately four times lower (e.g. 3.1% in Belgium, 8.8% in Austria and 6% in Switzerland). In the agricultural sector, mainly consisting of family businesses, formal full-time career interruptions were likely to be rare. In the subsequent decades, employment in the agricultural sector decreased significantly in all countries, including Southern European ones. The ESS data used in this article support this line of argumentation, and show that the proportion of mothers analysed whose last job was in the agricultural sector was substantially higher in Southern European regimes (14.5%: 19% among the older generation and 5.1% among the younger generation) compared with post-socialist (6.1%), pro-egalitarian (2.1%), pro-traditionalist (1.7%) and non-interventionist (0.4%) countries.

Another explanation may be related to the labour market specificities of the Southern European counties. Namely, that in these countries irregular and often ‘black’ employment is widespread (Ferrera, 1996), during which a formal career break due to childcare cannot be reported. A related explanation may stem from the fact that the incidence of temporary contracts is most prevalent in Southern European countries (Adsera, 2005). Another explanation may be short and unpaid parental leave combined with insufficient childcare and part-time work being uncommon (Ejrnaes, 2011), which makes mothers likely to use informal childcare arrangements and not to leave the labour market for longer periods.

The results regarding non-interventionist regimes, which demonstrate the highest overall propensity for long career interruptions due to childcare among all the regimes analysed, could be explained as follows. Among older generations, the proportion of mothers from non-interventionist regimes who reported long career interruptions is similar to that in pro-traditionalist countries (about 30%). However, when looking at the younger generations, mothers from non-interventionist countries exhibit the highest percentage of long career interruption due to childcare of all the regimes analysed (see Table 4).

Table 4.

Distribution of the variable on long career breaks by generation and family policy regimes, percentages.

Family policy regimes			Dependent variable – long term breaks => 10 years		Total
Family policy regimes			No	Yes	Total
Pro-traditionalist	Generation 1961+	0	68.6%	31.4%	100.0%
	Generation 1961+	1	92.0%	8.0%	100.0%
	Total		77.4%	22.6%	100.0%
Pro-egalitarian	Generation 1961+	0	86.6%	13.4%	100.0%
	Generation 1961+	1	98.2%	1.8%	100.0%
	Total		90.3%	9.7%	100.0%
Non-interventionist	Generation 1961+	0	67.8%	32.2%	100.0%
	Generation 1961+	1	80.0%	20.0%	100.0%
	Total		70.3%	29.7%	100.0%
Southern European	Generation 1961+	0	84.6%	15.4%	100.0%
	Generation 1961+	1	95.2%	4.8%	100.0%
	Total		88.3%	11.7%	100.0%
Post-socialist	Generation 1961+	0	97.6%	2.4%	100.0%
	Generation 1961+	1	98.3%	1.7%	100.0%
	Total		97.8%	2.2%	100.0%

Source: ESS, 2004. N = 4522.

In the following paragraphs, I discuss the main findings regarding generational differences in career interruptions among mothers in different family policy regimes. To see these differences, I examined interactions between the generation variable at the individual level and the family policy regime at the country level. It is found that the generational difference in the odds of taking long career breaks is moderated by the predominant family policy welfare model of the country. In concrete terms, in post-socialist and non-interventionist regimes, the differences in the propensity to take long career breaks between generations are significantly smaller than in countries with a pro-traditionalist family policy regime. In other words, in countries with post-socialist and non-interventionist regimes, the inter-generation gap has closed to a significantly lesser extent than in the pro-traditionalist, pro-egalitarian and Southern European countries.

There may be several explanations for the fact that in post-socialist and non-interventionist regimes the generation differences are smaller than in other regimes. The reasons behind this finding for post-socialist countries can be found in the developments in family policy over recent decades (i.e. the decades following the fall of socialist regimes at the beginning of the 1990s), which has shifted in a more pro-traditionalist direction (Rostgaard, 2004; Saxonberg and Sirovátka, 2006). This shift involved a dramatic reduction in childcare provisions, which evidently hinders the labour market participation of women. In the case of the non-interventionist regimes, this situation might be related especially to the relative lack of reconciliation policies such as paid parental leave or the availability of childcare services, and also to the high cost of formal childcare (Schober, 2013). Another explanation regarding post-socialist countries may be related to the fact that the incidence of long career breaks is also extremely low among women from the older generation (i.e. 2.4% compared with 1.7% among the younger generation). Given this, a significant inter-generation change was not possible.

The results regarding non-interventionist regimes are in line with the findings of Musumeci and Solera (2013), who demonstrate that mothers from the younger generation in the UK are notably more likely to withdraw from the labour market after birth compared with mothers in other European countries. One possible explanation for this phenomenon can be as follows. According to OECD data from 2004, the UK (i.e. representative of non-interventionist regimes) scores the highest of all countries analysed with regard to the childcare cost as a percentage of family net income compared with other countries. It is also the second highest for childcare costs and benefits as a percentage of average wages, immediately behind Ireland. In combination with unpaid and relatively short parental leave, this has an impact on mothers’ labour market attachment.

This article should stimulate further research into the under-investigated domain of empirical cross-country comparisons of career interruption due to childcare. A more detailed public register of data should be used in specific countries representing family policy models to verify the findings presented here and to be able to examine the differences between less crudely defined generations.

Footnotes

Appendix

Table A2.

Description of indicators in identifying family policy clusters by country.

Country	Duration of parental leave (weeks)	PL income replacement rate	Expenditure on fam. services (childcare)	Childcare usage under 3 years	Duration of existence of parental leave (years)	Joint taxation
Austria	104.0	21.0	0.6	6.6	43	1
Belgium	12.0	12.0	0.9	33.6	6	1
Switzerland	0.0	0.0	0.4	3.0	0	1
Czech Republic	156.0	20.0	0.6	3.0	40	1
East Germany	156.0	11.0	0.8	9.0	40	1
Denmark	32.0	90.0	2.3	61.7	20	0
West Germany	156.0	11.0	0.8	9.0	18	1
Estonia	104.0	15.0	0.6	7.0	38	0
Spain	52.0	0.0	0.7	20.7	33	1
Finland	26.0	60.0	1.4	22.4	26	0
UK	13.0	0.0	0.8	25.8	4	0
Greece	28.0	0.0	0.4	7.0	19	1
Luxembourg	26.0	62.0	0.6	14.0	5	1
Norway	42.0	100.0	1.5	43.7	26	1
Poland	156.0	14.6	0.5	2.0	36	1
Portugal	26.0	0.0	0.9	23.5	20	1
Sweden	51.0	80.0	1.9	39.5	30	0
Slovenia	52.0	100.0	0.6	20.0	38	0

Source: OECD, 2012.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

Author biography

Marie Valentova, PhD in sociology, has worked as researcher at LISER in Luxembourg since 2002. She is also a visiting fellow at the Catholic University of Leuven (Belgium). Her main research interests concern: reconciliation of family and paid work, female labour market participation, gender issues, social policy analysis and evaluation, integration of immigrants and social cohesion. Her research papers have been published in international peer-reviewed journals such as Journal of Social Policy, International Journal of Intercultural Relations, Work, Employment and Society, Social Indicators Research, Survey Research Methods etc. She is a member of the Expert’s Forum of the European Institute for Gender Equality (EIGE) and also a member of the Network of International Review of Leave Policies and Related Research.

References

Adsera

(2005) Where are the babies? Labor market conditions and fertility in Europe. IZA Working Paper, IZA DP No. 1576.

Budig

Misra

Boeckmann

(2012) The motherhood penalty in cross-national perspective: The importance of work–family policies and cultural attitudes. Social Politics 19(2): 163–193.

Bunel

Guironnet

J-P

(2011) Earning inequalities between and within nests. Working Paper WP-2011–18, CREM, University of Caen and University of Rennes 1.

Chan

Hamplova

Le Bourdais

(2012) Are parental leaves considered as work interruptions by survey respondents? Canadian Studies in Population 39(1–2): 31–44.

Connelly

(1992) The effect of child care costs on married women’s labor force participation. The Review of Economics and Statistics LXXIV(1): 83–90.

Crompton

(2006) Employment and the Family: The Reconfiguration of Work and Family Life in Contemporary Societies. Cambridge: Cambridge University Press.

Crompton

Brockmann

Lyonette

(2005) Attitudes, women’s employment and the domestic division of labour. Work, Employment and Society 19(2): 213–233.

Dex

Joshi

(1999) Careers and motherhood: Policies for compatibility. Cambridge Journal of Economics 23(5): 641–659.

Dex

Ward

Joshi

(2008) Gender differences in occupational wage mobility in the 1958 cohort. Work, Employment and Society 22(2): 263–280.

10.

Edmunds

Turner

(2002) Generations, Culture and Society. Maidenhead: Open University Press.

11.

Ejrnaes

(2011) The impact of family policy and career interruptions on women’s perceptions of negative occupational consequences of full-time home care. European Societies 13(2): 239–256.

12.

Esping-Andersen

(ed.) (2002) Why We Need a New Welfare State. Oxford: Oxford University Press.

13.

Evertsson

Grunow

(2012) Women’s work interruptions and career prospects in Germany and Sweden. International Journal of Sociology and Social Policy 32(9): 561–575.

14.

Ferrera

(1996) The ‘Southern Model’ of welfare in social Europe. Journal of European Social Policy 6(1): 17–37.

15.

Fourage

Manzoni

Muffels

Luijkx

(2010) Childbirth and cohort effects on mothers’ labour supply. Work, Employment and Society 24(3): 487–507.

16.

Fuwa

(2004) Macro-level gender inequality and the division of household labour in 22 countries. American Sociological Review 69(6): 751–767.

17.

Fux

(2002) Which models of the family are encouraged or discouraged by different family policies? In: Kaufmann

F-X

Kuijsten

Schulze

H-J

Strohmeier

(eds) Family Life and Family Policies in Europe. Oxford: Oxford University Press, pp. 363–409.

18.

Gauthier

(1996) The State and the Family: A Comparative Analysis of Family Policies in Industrialized Countries. Oxford: Clarendon Press.

19.

Gornick

Meyers

(2003) Families that Work: Policies for Reconciling Parenthood and Employment. New York: Russell Sage Foundation.

20.

Gough

(2001) Social assistance regimes: A cluster analysis. Journal of European Social Policy 11(2): 165–170.

21.

Hakim

(1992) Explaining trends in occupational segregation: The measurement, causes, and consequences of the sexual division of labour. European Sociological Review 8(2): 127–152.

22.

Hakim

(2002) Lifestyle preferences as determinants of women’s differentiated labour market careers. Work and Occupations 29(4): 428–459.

23.

Hantrais

Letablier

(1996) Families and Family Policies in Europe. London: Longman.

24.

Heckman

(1979) Sample selection bias as a specification error. Econometrica 47(1): 153–161.

25.

Hox

(2010) Multilevel Analysis: Techniques and Applications. Abingdon: Routledge.

26.

Jacobs

(2002) Reliability and recall of unemployment events using retrospective data. Work, Employment and Society 16(3): 537–548.

27.

Javornik

(2014) Measuring state de-familialism: Contesting post-socialist exceptionalism. European Journal of Social Policy 24(3): 240–257.

28.

Joshi

Macran

Dex

(1996) Employment after childbearing and women’s subsequent labour force participation. Journal of Population Economics 9(3): 325–348.

29.

Kangas

Rostgaard

(2007) Preferences or care context? Predictors for working life opportunities in seven European countries. Journal of European Social Policy 17(8): 240–256.

30.

Kertzer

(1983) Generation as a sociological problem. Annual Review of Sociology 9: 125–149.

31.

Knudsen

Waerness

(2001) National context, individual characteristics and attitudes on mothers’ employment. Acta Sociologica 44(1): 67–79.

32.

Lalive

Zweimüller

(2009) How does parental leave affect fertility and return to work? Evidence from two natural experiments. Quarterly Journal of Economics 124(3): 1363–1402.

33.

Lewis

(1992) Gender and development of the welfare regimes. Journal of European Social Policy 2(3): 159–173.

34.

Lewis

(2001) The decline of the male breadwinner model: Implications for work and care. Social Politics 8(2): 152–169.

35.

Lewis

(2006) Employment and care: The policy problem, gender equality and the issue of choice. Journal of Comparative Policy Analysis 8(2): 103–114.

36.

Lewis

Smithton

(2001) Sense of entitlement to support for the reconciliation of employment and family life. Human Relations 54(11): 1455–1481.

37.

Mandel

(2012) Winners and losers: The consequences of welfare state policies for gender wage inequality. European Sociological Review 28(2): 241–262.

38.

Math

Meilland

(2004) Family related leave and industrial relations. European Foundation for the Improvement of Living and Working Conditions, European Industrial Relations Observatory online. Available at: www.eiro.eurofound.eu.int/about/2004/study/tn0403101s.html.

39.

Moss

Deven

(eds) (1999) Parental Leave: Progress or Pitfall? Research and Policy Issues. Brussels: Vlaamse Gemeenschap.

40.

Muffels

Wilthagen

Van den Heuvel

(2002) Labour Market Trajectories and Employment Regimes. Berlin: Wissenschaftszentrum Berlin fur Sozialforschung.

41.

Musumeci

Solera

(2013) Women’s and men’s career interruptions in Europe: The role of social policies. Observatoire de la Société Britanique. No. 14. Available at: osb.revues.org/1511.

42.

OECD (2002) Employment Outlook. Paris: OECD.

43.

OECD (2012) Family and children. Available at: www.oecd.org/social/family/

44.

Pascall

Kwak

(2005) Gender Regimes in Transition in Central and Eastern Europe. Bristol: Policy Press.

45.

Pfau-Effinger

(2004) Development of Culture, Welfare States and Women’s Employment in Europe. Farnham: Ashgate.

46.

Pfau-Effinger

(2012) Women’s employment in institutional and cultural context. International Journal of Sociology and Social Policy 32(9): 530–543.

47.

Pierson

(2000) Increasing returns, path dependence, and the study of politics. American Political Science Review 94(2): 251–267.

48.

Roser

(2015) World maps of employment in agriculture as share of total employment. Available at: ourworldindata.org/data/food-agriculture/agricultural-employment/

49.

Rostgaard

(2004) Early childhood policy in Central and Eastern Europe – a decade and half of transition. Early Childhood and Family Policies Series, Issue 8. Available at: unesdoc.unesco.org/images/0013/001337/133733e.pdf.

50.

Ruhm

(1998) The economic consequences of parental leave mandates: Lesson from Europe. The Quarterly Journal of Economics 113(1): 285–317.

51.

Sainsbury

(ed.) (1996) Gender and Welfare State Regimes. Oxford: Oxford University Press.

52.

Saxonberg

Sirovátka

(2006) Failing family policy in post-Communist Central Europe. Journal of Comparative Policy Analysis: Research and Practice 8(2): 189–207.

53.

Shelton

John

(1996) The division of household labour. Annual Review of Sociology 22: 299–322.

54.

Schober

(2013) The parenthood effect on gender inequality: Explaining the change in paid and domestic work when British couples become parents. European Sociological Review 29(1): 74–85.

55.

Sjoberg

(2004) The role of family policy institutions in explaining gender-role attitudes. Journal of European Social Policy 14(2): 107–125.

56.

Smeaton

(2006) Work return rates after childbirth in the UK – trends, determinants and implications: A comparison of cohorts born in 1958 and 1970. Work, Employment and Society 20(1): 5–25.

57.

Sorrentino

(1983) International comparisons of labor force participation, 1960–81. Monthly Labour Market Review. International Comparisons of Labor Force Participation. Available at: www.bls.gov/opub/mlr/1983/02/art3full.pdf

58.

Szelewa

Polakowski

(2008) Who cares? Changing patterns of childcare in Central and Eastern Europe. Journal of European Social Policy 18(2): 115–131.

59.

Vlasblom

Schippers

(2006) Changing dynamics in female employment around childbirth: Evidence from Germany, the Netherlands and the UK. Work, Employment and Society 20(2): 329–347.

60.

Walby

(1997) Gender Transformations. London: Routledge.

61.

Waldfogel

(1998) The family gap for young women in the United States and Britain: Can maternity leave make a difference? Journal of Labor Economics 16(3): 505–545.

62.

Zhang

(2008) Religious affiliation, religiosity, and male and female fertility. Demographic Research 18(8): 233–262.

63.

Ziefle

Gangl

(2014) Do women respond to changes in family policy? European Sociological Review 30(5): 562–581.