ABSTRACT
Using data from the Swiss Household Panel (1999–2012) and the German Socio-Economic Panel (1994–2010), this paper compares the impact of mothers’ education and her partners’ income on maternal employment within the second to the fourth year after childbirth in Switzerland and West Germany. The broadly similar institutional context in the two countries makes for a more controlled and narrower comparison. Around the turn of the millennium, both family-policy regimes did little to foster dual-earner families. However, they differed in their support for families’ caring role (familialistic policies), with West Germany being much more generous. It is expected that these familialistic policies widen the educational gap in maternal employment, by selectively encouraging less-educated mothers to stay at home. Moreover, they are also expected to lower the economic pressure on low-income families to have a second income, thus diminishing the impact of partners’ income. Results confirm this expectation only within the fourth year after childbirth but not within the years before. This is somehow surprising, as central country-differences with respect to familialistic policies refer to the first three years after childbirth.
1. Introduction
In recent years there has been increasing recognition of class differences in cross-country comparisons of gender equality (Esping-Andersen 2009; Mandel and Shalev 2009; Shalev 2008). Empirical findings show, for example, that not all women are evenly affected in their labour market attachment and attainment by a given institutional context, or by given policies (Korpi et al. 2013; Mandel 2012; Stadelmann-Steffen 2011). Treating women as one homogeneous group, therefore, obscures potential inequalities among women. Even though not yet the object of much research, literature dealing with the cross-country variation of individual-level effects on female employment gives some evidence as to the effect heterogeneity of women's education and partners’ income (Steiber and Haas 2012: 354–55).
Maternal employment can per se be understood as an indicator of gender equality. Continued employment is crucial in guaranteeing (at least to some degree) independence from a male breadwinner. Financial autonomy enables women to make claims within the family, and to exit from dissatisfying relationships (Hobson 1990). Moreover, family-related labour market interruptions can evoke serious career penalties (Aisenbrey et al. 2009; Giudici and Gauthier 2009), and can also lower women's long-term wage growth (Ondrich et al. 2003). Finally, the taking of parental leave also tends to increase the gender gap in domestic work (Schober 2013a).
It is, therefore, of importance to see whether and how a given family-policy context is related to group-specific differences in maternal employment. This paper investigates two potential group-specific outcomes, by asking how a woman's education and her partner's income affect her likelihood to be employed, depending on two different family-policy regimes. In particular, resource-related inequalities in maternal employment after childbirth, in both a strong and a weak familialistic context, are examined, contrasting West Germany and Switzerland. The choice of these two countries allows for a controlled and narrow comparison: both these countries have conservative gender policies, and are shifting from a male-breadwinner model towards a modernized-breadwinner model. However, they differ considerably in their support for families in their caring role (Pfau-Effinger 2005). It is argued that this country-difference affects the extent to which both the mother's education and her partner's income have an impact on mothers’ employment. Thus, the main focus of this study is on a comparison between resource-related inequalities in maternal employment, rather than a focus on how family policies affect maternal employment in general.
Maternal employment, however, is not only important with respect to gender inequality, but also with respect to social inequalities in general. Families relying on one market-income only, in general, are more vulnerable, in cases of illness or job loss of the breadwinner. Furthermore, in combination with the ubiquity of educational homogamy within couples in Western societies (Blossfeld and Timm 2003), class-selective female employment alters the income distribution between households (Esping-Andersen 2007).
In what follows, the family-policy regimes in Switzerland and West Germany are described, with similarities and differences being highlighted (Section 2). In Section 3, firstly, the general effects of women's education and partners’ income on maternal employment are discussed. Thereafter, two hypotheses on country-differences in resource-related inequalities in maternal employment are derived, drawing upon a consideration of the group-specific outcomes of the two family-policy regimes. Sections 4 and 5 describe the methods and results, respectively. The paper closes with a discussion and conclusion (Section 6).
2. The family-policy regime in Switzerland and West Germany
As the paper focuses on female employment after childbirth, regulations regarding parental leave policies are of particular relevance. However, the outcome of parental leave policies has been found to co-vary mainly with day-care provision, taxation, and the possibility of part-time work (Hegewisch and Gornick 2011: 124). Besides parental leave regulations, the institutional aspects of these co-variants are therefore also sketched, in both countries.
In Switzerland, maternity insurance was only implemented in 2005. Conditional on previous employment, mothers get paid 80% of their former wage, for 14 weeks after childbirth. Before 2005, women were not allowed to work for 8 weeks after childbirth, and usually got fully paid (EKF Eidgenössische Kommission für Frauenfragen 2009). In addition, many employers granted maternity leave of up to four months to employed women (Thoenen 2010: 24). By contrast, parental leave policy in Germany is much more generous, and has expanded over recent decades. From 1993 onwards job-protected leave duration has been 36 months and child-rearing benefits have been means-tested, in terms of household income.1 The duration of means-tested child-rearing payment has been 24 months since1993. Between 2001 and 2006, parents entitled to benefits had the possibility of choosing a higher amount during one year, instead of a lower amount during two years. In 2007, a paradigm shift occurred when benefits were no longer means-tested, but rather became income-related and restricted to one year only.2 Therefore, the empirical analysis for West Germany considers the period from 1993 to 2006 only. During this period: job-protected leave lasted for three years; means-tested child-rearing benefits were independent of previous employment status, and were paid during maximally two years (see, e.g. Drasch 2013: 984, Table 1; Spiess 2011: 7, Table 1).
Even though parental leave policy in Germany is nominally gender-neutral (i.e. mothers as well as fathers are entitled), its outcomes are gendered because of the low level of benefits, the segregated labour markets, and norms (Morgan and Zippel 2003: 71–73; Steiber and Haas 2010: 254). This paper therefore assumes that parental leave policies in both countries mainly affect mothers’ labour market participation decisions.
Both countries expanded day-care facilities considerably, mainly in the first decade of the twenty-first century. An impulse programme, introduced in 2003, which gave financial support to new day-care facilities, helped to increase day-care provision in Switzerland. However, even by 2009/2010, full-time day-care provision was only available for 1 out of 10 pre-school children (Felfe et al. 2013). Germany introduced a law in 1996 guaranteeing part-time day care in kindergarten from the third birthday of a child. Further legal attempts were made to improve day care for children below the age of three (Spiess 2011: 13). In 2008, the full-time child-care coverage rate of children aged 0–3 amounted to 9% in West Germany (Pfau-Effinger and Smidt 2011: 219). Thus, in spite of some improvements, both countries lack comprehensive formal child-care coverage for small children (Pfau-Effinger 2005). Both countries, furthermore, tax marriage jointly. Joint taxation, compared to individualized taxation systems, penalizes dual-earner families (Dingeldey 2001). Finally, with respect to working-time regulations, there is no right to part-time work in Switzerland. However, maternal part-time work is very common and Switzerland has one of the highest female part-time rates in European comparison (Strub and Bauer 2002). Female part-time work is also widespread in Germany (Strub and Bauer 2002). Germany, furthermore, introduced a law in 2001 that entitles employees to reduce hours of paid work, and therefore allows parents to work part-time, even after parental leave has expired (Rüling 2004: 123).
To sum up, within the period under observation, both countries did little to support dual-earner families, or to foster an equal sharing of unpaid work. However, they differ in the degree of state support for the caring of small children at home. Leave policy regulations differ with respect to length and payment; the second year after childbirth protected leave and payment have already exceed in Switzerland but still last in Germany. The third year after childbirth protected leave still lasts in Germany but benefits are no longer paid. Finally, from the fourth year after childbirth protected leave and benefits have exceeded in both countries.
The family-policy typology developed by Leitner (2003) distinguishes between a familialistic dimension, that is, the extent to which the caring function of families are maintained and strengthened, and a de-familialistic dimension, that is, the extent to which families are relieved of their care obligations.3 According to this typology, Switzerland can be classified as implicit familialistic, given the weak indicator on both dimensions. Germany, in contrast, has stronger familialistic measures, and therefore belongs to the explicit familialistic type. Thus, both family-policy regimes rely on the family as the main caregiver. Germany actively supports the care provision within families, whilst, in Switzerland, the families’ care responsibility results mainly from a lack of alternatives.
3. Literature review and hypotheses
Before turning to potential country-differences in individual-level effects, two general mechanisms are first discussed, which might account for resource-related inequalities in maternal employment (among coupled mothers). On the one hand, from a human capital point of view, women with higher income potential are more attached to the labour market, due to higher opportunity costs of not being employed. By interrupting employment, they risk a bigger loss in human capital and income prospects, compared with women with lower human capital and income potential (Mincer 1974). Therefore, greater income potential should increase the chances of maternal employment. Indeed, research has consistently shown that women with higher educational attainment and income have higher employment rates after childbirth and re-enter the labour market faster after a family-related interruption (e.g. Drasch 2013; Grunow et al. 2011; Kenjoh 2005; Schober 2013b). Gustafsson et al. (1996), furthermore, found that women's resources, as determinants of labour market participation and re-entry timing after childbirth, are more important in a social-policy context conforming to the male-breadwinner model. Thus, if maternal employment is not explicitly supported (i.e. weak de-familialization in the sense of Leitner 2003), individual resources, in particular, become relevant for realizing a family arrangement that does not conform to the hegemonic family model.
On the other hand, among couples it is expected that the partner's resources have a negative effect on the mother's labour market participation. This effect occurs mainly in male-breadwinner regimes with little support for maternal employment (e.g. low provision of child-care facilities, taxation systems penalizing double-income couples, and traditional attitudes toward maternal employment) (Drobnic and Blossfeld 2004; Matysiak and Steinmetz 2008). Using, again, the labels of Leitner's (2003) typology, the negative impact of partners’ resource should arise especially in policy contexts with weak de-familialistic measures. Hoherz (2014) found that fathers’ economic resources in Germany delay mothers’ return to full-time employment after a birth-related interruption, but have no effect on her return to part-time work. Another study shows a negative association between mothers’ full- and part-time work, and partners’ education in Germany. However, part-time work in general is less affected by partners’ resource and its influence declined over time (Konietzka and Kreyenfeld 2010; Kreyenfeld et al. 2007). Similar results apply to Swiss households, with respect to husbands’ educational attainment (Liechti 2014; Stadelmann-Steffen 2011) and earnings (Bauer 2000).
In summary, literature gives evidence for a positive relationship between women's education and her labour market participation after childbirth, due to increasing opportunity costs. Conversely, there can be found a negative association between partners’ income and mothers’ labour market participation after childbirth, due to decreasing economic needs for an additional income. Since de-familialization is rather weak in both Switzerland and Germany, and maternal employment is therefore hardly supported, the two resource-related impacts on maternal employment are assumed to apply to both countries under observation. However, given that job-protected leave and benefits alter the incentives of maternal employment for different income and educational groups, a moderating effect of the country context on resource-related inequalities in maternal employment is expected. Thus, it is argued that the country-differences in the extent of familialization are reflected in group-specific outcomes.
In line with previous findings (e.g. Cantillon et al. 2001; Del Boca et al. 2009; Korpi et al. 2013; Stadelmann-Steffen 2011), it is argued that women with higher resources are less dependent on family policies to realize employment, and that those women, in general, show high labour market participation rates. Cross-national variation in class-specific maternal labour market participation rates is, therefore, not primarily due to differences among the higher educational strata, but rather due to differences among the lower strata. It is assumed that means-tested benefits mainly influence less-educated mothers’ employment, given that benefits compensate for low-income potential. Parental leave policy regulations, including low means-tested benefits, therefore, can widen class inequalities between mothers, as they mainly encourage less-educated mothers to exit the labour market (for a similar argumentation see Hegewisch and Gornick 2011: 123; Kreyenfeld et al. 2007: 437; Morgan and Zippel 2003). Thus, the incentives for less-educated mothers to stay at home are even bigger in Germany than in Switzerland. Previous empirical findings from (West) Germany indeed show that the educational gap in maternal employment has widened with the expansion of familialistic policies (Drasch 2013; Gottschall and Bird 2003; Konietzka and Kreyenfeld 2010).
The reverse country-differences are expected with respect to the effect of partners’ income. Since financial benefits in Germany are means-tested, economic pressure on low-income families after childbirth is less pronounced in Germany than in Switzerland. The benefits paid in Germany allow for a greater range of families to live in a traditional family model (for a similar argumentation see Konietzka and Kreyenfeld 2010: 264). Therefore, men's resources should be more relevant in Switzerland than in Germany, given that, after childbirth, the male-breadwinner model is less generously subsidized in Switzerland.
To sum up, in addition to the general effects of education and partner's income on mothers’ employment, the two following hypotheses on family-policy-related effects will be tested:
Hypothesis 1: Inequalities in mothers’ labour market participation due to women's education are more pronounced in Germany than in Switzerland.
Hypothesis 2: Inequalities in mothers’ labour market participation due to partners’ income are less pronounced in Germany than in Switzerland.
As benefits are paid up to two years in Germany, it is expected that country-differences occur especially within the second year after childbirth. But given that job-protected leave last up to three year in Germany and the male-breadwinner model therefore is still more generously supported in Germany compared to Switzerland, country-differences are expected to last. Within the fourth year after childbirth, when leave and benefits have expired in both countries, country-differences in resource-related inequalities should diminish.
4. Method
4.1. Sample selection
The analyses are based on the Swiss Household Panel (SHP) and the German Socio-Economic Panel (GSOEP). Both panels collect data annually from representative households. The GSOEP has operated since 1984,4 the SHP since 1999. These data provide required and comparable variables for both countries. The sample is restricted to heterosexual married and non-married couples living together and providing valid information for three consecutive waves, when one and the same youngest child is between 12 and 48 months (i.e. the second, third and fourth year after birth).5 The focus of the analysis is therefore on resource-related inequalities in maternal employment at a particular stage of the family cycle, namely the second to the fourth year after childbirth. As described above, the second year after childbirth in Switzerland parental leave has already finished, whereas in Germany means-tested benefits and job-protected leave still continue. The third year after childbirth job-protected leave still continues in Germany but benefits are no longer paid. The fourth year after childbirth, leave and benefits have expired in both countries. Information about childbirth is taken from women's birth history. The Swiss sample makes use of all available waves (1999–2012) and therefore contains children born between 1997 and 2009. The German sample considers only youngest children born between 1993 and 2006 (i.e. waves 1994 to 2010), in order to avoid major changes in parental leave policy. During the selected period: job-protected leave lasted for three years; means-tested benefits were independent of foregone employment status, and paid up to two years; and part-time work, up to certain weekly hours during leave, is possible.6 In the German sample, only households from former West Germany are included. Even after reunification mothers’ labour market participation (Drasch 2012), public day-care provision (Spiess 2011), and normative expectations about motherhood and maternal employment (Pfau-Effinger and Smidt 2011) differ between former East and West Germany.7 If a couple satisfies the criteria for more than one youngest child, the most recent childbirth is chosen. This leaves us with 1108 couples in the German sample and 412 couples in the Swiss sample.
4.2. Plan of analysis
For each sample, three logistic regressions are applied, in order to predict the probability of women's labour market participation when the youngest child is between 12 and 24, 24 and 36, 36 and 48 months, respectively. Average marginal effects are reported, as opposed to odds ratios, in order to simplify interpretation, and to allow a comparison of the coefficients between different groups and models (e.g. Mood 2010). So as to provide insight into the absolute employment rate levels on which the resource-related differences occur, predicted probabilities are also reported.8 Using the significance test proposed by Auspurg and Hinz (2011: 67), the existence or not of significant country-differences in the effect size for the main explanatory variables (women's education and partners’ income) is determined.
4.3. Variables
The binary dependent variable indicates whether the woman is currently working in the labour market at least one hour a week or not. The first independent variable is women's highest attained educational degree at the time of interview. This variable differentiates between a degree lower (no education, compulsory school), equal (apprenticeship, full-time vocational school, maturity) or higher than high school (university, university of applied science). The second independent variable is man's (imputed) yearly gross labour income. Incomes are translated to 2010 (GSOEP) and 2011 (SHP) prices, respectively. For analysis, income is divided into three categories, indicating an income in the lower, second or third, and upper quartile of the sample-specific male income distribution. In order to improve comparability between both samples, the educational degree and income, provided by the Cross National Equivalent File (CNEF), are used. The SHP and GSOEP are both part of the CNEF, which contains harmonized data from several panel data surveys (see Frick et al. 2007).
It is assumed that educational degree directly affects values, norms, and preferences, especially when controlling for income (Blossfeld and Drobnic 2004: 37; Bonke and Esping-Andersen 2011: 46), and therefore partner's educational attainment is also included. Traditional gender ideology is usually less prevalent in higher educational strata (Lachance-Grzela and Bouchard 2010: 771). Holding income constant, a higher education of the partner should therefore positively affect maternal employment. In addition, mothers’ age, the age of the youngest child in months, and whether one, two, three, or more children live in the household are also included. Finally, the birth year of the youngest child is included, in order to control for period effects owing to changes over time on the demand and supply side of maternal employment. Table 1A in the appendix shows the descriptive statistics of the variables included in the models.
5. Results
Before turning to resource-related inequalities in maternal employment, the impact of the control variables are discussed. Table 1 shows that partner's education does not have a significant impact on maternal employment in any of the years or countries. Older women tend to be more likely to be employed in Germany but less likely in Switzerland. Effects are, however, hardly significant. Results furthermore show that there is a negative relationship between the number of children and maternal employment. In Germany, this is especially true when a woman has more than two children. In Switzerland, by contrast, this relationship is more pronounced in early motherhood. Maternal employment within a particular year after childbirth is more likely, when children get older. However, effects of the age of the youngest child are only partly significant. Finally, maternal employment has become more common over time as mothers having a child more recently are more likely to be employed.
| Switzerland | Western Germany | |||||
|---|---|---|---|---|---|---|
| Second year | Third year | Fourth year | Second year | Third year | Fourth year | |
| Education mother (ref. low) | ||||||
| Middle | 0.217** (0.0756) | 0.0420 (0.0770) | −0.0478 (0.0761) | 0.161*** (0.0313) | 0.159*** (0.0373) | 0.165*** (0.0403) |
| High | 0.329*** (0.0834) | 0.211** (0.0812) | 0.165* (0.0785) | 0.240*** (0.0454) | 0.226*** (0.0499) | 0.285*** (0.0505) |
| Income partner (ref. low) | ||||||
| Middle | −0.123* (0.0508) | −0.137** (0.0486) | −0.139** (0.0477) | −0.0568 (0.0351) | −0.0978** (0.0363) | −0.0153 (0.0359) |
| High | −0.130* (0.0642) | −0.189** (0.0638) | −0.202** (0.0637) | −0.0602 (0.0432) | −0.148*** (0.0450) | −0.0545 (0.0458) |
| Education partner (ref. low) | ||||||
| Middle | 0.0996 (0.110) | 0.180 (0.108) | 0.121 (0.108) | 0.00507 (0.0409) | −0.0272 (0.0438) | −0.00757 (0.0422) |
| High | 0.196 (0.114) | 0.219 (0.113) | 0.193 (0.111) | 0.00728 (0.0508) | −0.0419 (0.0548) | −0.0816 (0.0535) |
| Age women | −0.000719 (0.00570) | −0.00362 (0.00556) | −0.0116* (0.00561) | 0.00436 (0.00310) | 0.00442 (0.00333) | 0.00454 (0.00335) |
| Number of children (ref. one) | ||||||
| 2 children | −0.139* (0.0549) | −0.0788 (0.0561) | −0.0162 (0.0594) | −0.0525 (0.0328) | −0.0341 (0.0351) | −0.0427 (0.0347) |
| 3+ children | −0.224*** (0.0626) | −0.145* (0.0632) | −0.0592 (0.0659) | −0.123** (0.0383) | −0.149*** (0.0418) | −0.156*** (0.0429) |
| Age of youngest child | 0.0171* (0.00695) | 0.00492 (0.00669) | 0.0105 (0.00676) | 0.00810 (0.00422) | 0.0156*** (0.00450) | 0.0132** (0.00447) |
| Year of birth of youngest child | 0.0211*** (0.00596) | 0.0161** (0.00601) | 0.0154* (0.00606) | 0.00970** (0.00367) | 0.0125** (0.00391) | 0.0153*** (0.00388) |
| N | 412 | 1108 | ||||
| Switzerland | Western Germany | |||||
|---|---|---|---|---|---|---|
| Second year | Third year | Fourth year | Second year | Third year | Fourth year | |
| Education mother (ref. low) | ||||||
| Middle | 0.217** (0.0756) | 0.0420 (0.0770) | −0.0478 (0.0761) | 0.161*** (0.0313) | 0.159*** (0.0373) | 0.165*** (0.0403) |
| High | 0.329*** (0.0834) | 0.211** (0.0812) | 0.165* (0.0785) | 0.240*** (0.0454) | 0.226*** (0.0499) | 0.285*** (0.0505) |
| Income partner (ref. low) | ||||||
| Middle | −0.123* (0.0508) | −0.137** (0.0486) | −0.139** (0.0477) | −0.0568 (0.0351) | −0.0978** (0.0363) | −0.0153 (0.0359) |
| High | −0.130* (0.0642) | −0.189** (0.0638) | −0.202** (0.0637) | −0.0602 (0.0432) | −0.148*** (0.0450) | −0.0545 (0.0458) |
| Education partner (ref. low) | ||||||
| Middle | 0.0996 (0.110) | 0.180 (0.108) | 0.121 (0.108) | 0.00507 (0.0409) | −0.0272 (0.0438) | −0.00757 (0.0422) |
| High | 0.196 (0.114) | 0.219 (0.113) | 0.193 (0.111) | 0.00728 (0.0508) | −0.0419 (0.0548) | −0.0816 (0.0535) |
| Age women | −0.000719 (0.00570) | −0.00362 (0.00556) | −0.0116* (0.00561) | 0.00436 (0.00310) | 0.00442 (0.00333) | 0.00454 (0.00335) |
| Number of children (ref. one) | ||||||
| 2 children | −0.139* (0.0549) | −0.0788 (0.0561) | −0.0162 (0.0594) | −0.0525 (0.0328) | −0.0341 (0.0351) | −0.0427 (0.0347) |
| 3+ children | −0.224*** (0.0626) | −0.145* (0.0632) | −0.0592 (0.0659) | −0.123** (0.0383) | −0.149*** (0.0418) | −0.156*** (0.0429) |
| Age of youngest child | 0.0171* (0.00695) | 0.00492 (0.00669) | 0.0105 (0.00676) | 0.00810 (0.00422) | 0.0156*** (0.00450) | 0.0132** (0.00447) |
| Year of birth of youngest child | 0.0211*** (0.00596) | 0.0161** (0.00601) | 0.0154* (0.00606) | 0.00970** (0.00367) | 0.0125** (0.00391) | 0.0153*** (0.00388) |
| N | 412 | 1108 | ||||
Notes: *p < .05, **p < .01, ***p < .001.
Source: SHP 1999–2012, GSOEP 1994–2010, own calculations.
Determinants of mothers’ labour market participation during the second, third and fourth year after childbirth, (a) average marginal effects of and (b) predicted probabilities for women's education. Source: SHP 1999–2012, GSOEP 1994–2010, own calculations.
Notes: *p < .05, **p < .01, ***p < .001. The reference groups in part (a) are lower educated women. Models control for partner's income and education, women's age, number of children, age, and year of birth of the youngest child.
Determinants of mothers’ labour market participation during the second, third and fourth year after childbirth, (a) average marginal effects of and (b) predicted probabilities for women's education. Source: SHP 1999–2012, GSOEP 1994–2010, own calculations.
Notes: *p < .05, **p < .01, ***p < .001. The reference groups in part (a) are lower educated women. Models control for partner's income and education, women's age, number of children, age, and year of birth of the youngest child.
The results, however, undermine Hypothesis 1, contradicting the expected country-difference, given that the effects of mothers’ education during the second year after childbirth are larger in Switzerland than in Germany. Mothers in Switzerland holding an educational degree beyond high school have a 33 percentage points higher chance of being employed compared to lower educated mothers. The corresponding educational gap accounts for only 24 percentage points in Germany. Only from the third year onward educational gaps are larger in Germany compared to Switzerland. However, differences between the average marginal effects of mothers’ education on their chance of employment in Switzerland and Germany are not significant within the second and third year after childbirth. The only significant country difference occurs during the fourth year after childbirth, among middle educated mothers (see Table A2 in the appendix).
Predicted probabilities show that the educational gap in maternal employment occurs at a higher absolute employment level in Switzerland (Figure 1(b)). Compared to mothers in Germany, all educational groups in Switzerland have higher employment probabilities. Even higher educated mothers in Germany are less likely to be employed than middle and lower educated mothers in Switzerland. Thus, the country-difference seems to trump even education in this respect. Only in the fourth year after childbirth are higher educated mothers in Germany slightly more likely to be employed than middle and lower educated mothers in Switzerland. Rising maternal employment probability is found in both countries, from the second to the fourth year after childbirth. In Switzerland, however, the difference between the third and fourth year is only marginal, whereas in Germany, the rise is more pronounced and roughly constant over time. The declining educational gap from the second to the fourth year after childbirth in Switzerland is mainly due to a large increase in maternal employment among lower educated women. The rising educational gap in Germany, by contrast, is mainly due to a large increase in maternal employment among higher educated women.
Determinants of mothers’ labour market participation during the second, third and fourth year after childbirth, (a) average marginal effects of and (b) predicted probabilities for partners’ income. Source: SHP 1999–2012, GSOEP 1994–2010, own calculations.
Notes: *p < 0.05, **p < 0.01, ***p < 0.001. The reference groups in part (a) are low-income men. Models control for women's and partners’ education, women's age, number of children, age, and year of birth of the youngest child.
Determinants of mothers’ labour market participation during the second, third and fourth year after childbirth, (a) average marginal effects of and (b) predicted probabilities for partners’ income. Source: SHP 1999–2012, GSOEP 1994–2010, own calculations.
Notes: *p < 0.05, **p < 0.01, ***p < 0.001. The reference groups in part (a) are low-income men. Models control for women's and partners’ education, women's age, number of children, age, and year of birth of the youngest child.
Here again, predicted probabilities in maternal employment for different income strata do not change a lot over time in Switzerland (Figure 2(b)). The biggest change occurs between the second and third year for mothers of low-income partners. In Germany, predicted probabilities rise over time for all income quartiles. The rise and fall of income-related inequalities are mainly driven by a large increase in maternal employment among mothers partnered to low-income men between the second and third year, in contrast to a smaller increase in the following year.
6. Discussion and conclusion
The aim of the paper was to explore how resource-related inequalities in maternal employment differ along two family-policy regimes. Switzerland and Germany are good examples of weak and strong familialistic policy intervention, while otherwise sharing many policies affecting maternal employment. Two mechanisms were discussed that state a positive relation between mother's education and her labour market participation, but a negative association between maternal employment and partner's income level. It was assumed that the degree of these resource-related inequalities differ between the two countries. The main argument was that means-tested child-rearing benefits, paid in Germany up to two years after childbirth, reinforce lower educated women's weaker labour market incentives, and might also reduce the economic need for an additional market income for low-income families. The two hypotheses therefore claimed that (1) inequalities in maternal employment due to women's education are larger, whereas (2) inequalities due to partners’ income are smaller, in Germany, compared with Switzerland.
It was found that, in both Switzerland and Germany, labour market participation is higher among higher educated mothers. There is, however, one limitation: middle and lower educated mothers’ labour market participation in Switzerland hardly differs during the third and fourth year after childbirth. Mothers’ individual resources in a weak familialistic policy context, therefore, seem to be more relevant at the very beginning of parenthood.
With respect to country-differences in educational inequalities, familialistic policies in Germany do not seem to widen the educational gap between mothers, compared to the laissez-faire family policy in Switzerland. The lack of support for Hypothesis 1 might be due to cross-country-differences in job protection after childbirth. In Switzerland, higher educated mothers might be reluctant to take longer (unpaid) leave that could signal lower career orientation towards the employer. In Germany, by contrast, parental leave regulation supports and protects employment interruptions to a much broader extent, given the guarantee of being able to return to the previous job (see Grunow et al. 2011: 405). Therefore, extended and job-protected leave also encourages higher educated mothers to withdraw from paid work. Gottschall and Bird (2003) conclude that normative pressure to make use of parental leave has increased with the expansion of parental leave duration in Germany, and labour market interruptions after childbirth have become institutionalized in the female life course. This might explain the smaller educational gap in Germany, as well as the non-significant country-differences in effect size during the first years after childbirth.
Moreover, the results show that mothers with higher income partners tend to be less frequently in gainful employment. However, income-related inequalities in maternal employment are rather small in Germany, and only become significant in the third year after childbirth. This suggests that the need-for-income effect in Germany occurs when familialistic policies reduce mothers’ time constraints in the absence of economic relief for low-income families (i.e. job-protected leave still continues but means-tested benefits are no longer paid). Another reason why the need-for-income effect only partly occurs in Germany might be that the analysis does not distinguish between part-time and full-time work. Other studies for German couples found only weak (Konietzka and Kreyenfeld 2010) or no (Hoherz 2014) effects of partners’ resources on mothers’ part-time work. Combining different employment levels might mask significant effects in the German sample.
Even though income-related inequalities in maternal employment are more pronounced in Switzerland during all three years studied, significant country-differences only arise during the fourth year after childbirth. Similar to the educational gap, significant country-differences, in terms of the hypotheses, occur when the intervening effects of family policy are no longer expected (i.e. when job-protected leave and benefits have expired in both countries). Means-tested child-rearing benefits paid in Germany, therefore, do not seem to be a relevant enough component of a family's income to account for significant country-differences in effect size of partners’ income.
Compared to Switzerland, maternal employment rates in Germany are considerably lower, but become higher each year after childbirth. This is in line with previous findings, that is, countries with shorter parental leave entitlements show higher maternal employment rates (Hegewisch and Gornick 2011: 124). Moreover, longer leave duration, in both between-country and within-country comparisons, results in longer labour market interruptions after childbirth (Drasch 2013; Gottschall and Bird 2003; Gustafsson et al. 1996; Haas and Hartel 2010; Rønsen and Sundström 2002). Given that female labour market participation rates can be understood as indicators of gender (in)equality, Germany fares worse on this equality dimension. It should, however, be kept in mind that mothers’ work in both countries is mainly organized as part-time work. Independence from a breadwinner due to employment is therefore moderate, at most (cf. Evertsson et al. 2009: 218), and gender equality within the labour market is, at best, partly satisfied. Furthermore, we must take into consideration that increased maternal employment is not always desirable from a gender equality point of view. Firstly, following Fraser's Caregiver Parity model (1994), it could be argued that, not only state support for maternal employment, but also state support for the caring of children at home mitigates within family dependencies. From this perspective, mothers in Germany do not necessarily fare worse, compared with mothers in Switzerland. Secondly, statistical data used in this paper do not reveal anything about the circumstances under which the employment of mothers with small children is realized. It is possible that higher maternal employment rates in Switzerland entail more work–family conflicts, as well as more precarious care arrangements, given the weak de-familialization.
Taken together, comparing resource-related inequalities in maternal employment in a weak and strong familialistic context, policies encouraging mothers to care for small children at home do not – in the short run – either increase the educational gap among mothers, or significantly narrow the impact of partners’ resources. However, the stronger familialistic policy context in Germany seems to come at the cost of lower overall maternal employment rates. The country-differences in individual-level effects stated in the hypotheses only occur over the longer term. This suggests that familialistic measures that continue after parental leave (e.g. tax deductions and child allowance) are, in terms of a component of the household income, more relevant in Germany than in Switzerland. They, therefore, disproportionally widen the educational gap in Germany, by selectively encouraging lower educated mothers to withdraw from paid work, on the one hand, whilst lowering the economic pressure for a second income, on the other hand. Moreover, in Switzerland, there are negative incentives for a second income, especially in the upper income classes (Bütler 2007). This might contribute to the significantly stronger impact of partners’ income in Switzerland. It is, of course, also possible that other policy measures, or country specificity, account(s) for these findings. There is, for example, evidence that labour market opportunities in Germany have worsened, especially for low-skilled individuals (Konietzka and Kreyenfeld 2010; Kreyenfeld et al. 2007), and that unemployment rates are, when averaged over the long term, considerably higher in Germany than in Switzerland (ILO 2014). Thus, structural constraints might contribute to the discrepancy between theoretically expected and empirically observed country-differences in resource-related inequalities in maternal employment. Another blind spot of this paper is the cultural dimension of maternal employment. Even though cultural and political dimensions are not always strictly different but mutually reinforcing (Charles and Cech 2010; Pfau-Effinger 2004), research found that both gender culture and policies affect maternal employment (e.g. Epple et al. 2015; Hummelsheim and Hirschle 2010) and that the impact of a given work–family policy on maternal employment is dependent on cultural factors regarding mothers’ paid work (Hegewisch and Gornick 2011). The non-consideration of country-specific gender culture might therefore partly explain results. Further research should therefore also include structural and cultural aspects in order to understand more thoroughly the potential of family policies to deepen or narrow class division among women.
Even though educational gaps in maternal employment hardly differ (significantly) between Germany and Switzerland, educational gaps are considerable in both countries. This supports previous literature (e.g. Baumgartner 2008: 303; Gustafsson et al. 1996) which argues that, without sufficient support for maternal employment (i.e. weak de-familialization), individual resources become more salient, and potentially perpetuate the class division among women, after transition to motherhood.
Differences in maternal employment, explained by partner's income, have consequences for the economic inequality between family households. In Switzerland, mothers compensate, with their own employment, for a husband's lower income. Dual-earner families are more common when the husband has a lower income, and mothers’ employment thus potentially mitigates economic inequalities between family households, in terms of total labour income. In Germany, by contrast, family households may differ more in terms of total labour income, because maternal employment is less likely to compensate for a partner's lower income. However, the German welfare state redistributes, and thus reduces, inequalities in market incomes to a greater extent than the Swiss welfare state (Grabka and Kuhn 2012; Kenworthy and Pontusson 2005).
Disclosure statement
No potential conflict of interest was reported by the author.
Notes on contributor
Lena Liechti is a scientific collaborator at the Swiss Federal Statistical Office. She did her Ph.D. at the Division of Sociology, Social Policy, and Social Work at the University of Fribourg. Her research focuses on gender inequality across different social classes and the development of gender arrangements within families.
Footnotes
In order to get means-tested benefits, the household income must not exceed a certain level.
There are two extra months paid, if both parents take leave.
The GSOEP started with West German households only and extended to East Germany in 1989.
Considering three consecutive waves precludes pregnant women having an additional child soon, which is likely to affect women's labour market decision.
Drasch (2013: 985) argues that the possibility of receiving more means-tested benefits for a shorter period after 2000 could narrow the educational gap in the timing of mothers’ labour market return after a family-related interruption. However, the author could not find empirical evidence for depolarizing trends in the reformed parental leave cohort. Moreover, Swiss parental leave policy regulation has also changed, with the introduction of maternity insurance in July 2005. However, theoretically, the mechanism remains the same. One could interpret the compulsory (and paid) eight-week mother protection after childbirth as a shorter version of maternity insurance.
Average marginal effects, and predicted probabilities, are obtained in STATA using the post-estimation command margins (see Williams 2012).
Models excluding partner's educational attainment were also tested, in order to check for multicollinearity with partner's income. However, in the German sample, significances do not and effect sizes hardly change. In the models for Switzerland, excluding partner's education weakens the income effects, as the two effect signs point in the inverse direction.
This might be counter-intuitive, as the effect of partner's income during the second year after childbirth is significant in Switzerland but not in Germany. However, as Gelman and Stern (2006) show, the difference between being significant and non-significant is not itself statistically significant.