## Abstract

As a consequence of the reversal of the gender gap in education, the female partner in a couple now typically has as much as or more education compared with the male partner in most Western countries. This study addresses the implications for the earnings of women relative to their male partners in 16 European countries. Using the 2007 and 2011 rounds of the European Union Statistics on Income and Living Conditions (N = 58,292), we investigate the extent to which international differences in women’s relative earnings can be explained by educational pairings and their interaction with the motherhood penalty on women’s earnings, by international differences in male unemployment, or by cultural gender norms. We find that the newly emerged pattern of hypogamy is associated with higher relative earnings for women in all countries and that the motherhood penalty on relative earnings is considerably lower in hypogamous couples, but neither of these findings can explain away international country differences. Similarly, male unemployment is associated with higher relative earnings for women but cannot explain away the country differences. Against expectations, we find that the hypogamy bonus on women’s relative earnings, if anything, tends to be stronger rather than weaker in countries that exhibit more conservative gender norms.

## Introduction

Universities largely remained a male domain until well into the second half of the twentieth century. Male enrollment and completion rates in advanced education were higher than female rates virtually everywhere. Since the 1990s, however, more women than men are enrolled in college-level education, and women also graduate more successfully (Buchmann and DiPrete 2006; Esteve et al. 2016).

The reversal of the gender gap in education implies that for the first time in history, more highly educated women than highly educated men reach the ages of partnering and parenthood. Recent studies have shown that the gender-specific trends in educational enrollment have indeed undermined the traditional pattern of educational hypergamy (women marrying up). Hypogamy (women marrying down) has become more prevalent than hypergamy in most countries where the reversal of the gender gap in education has occurred (De Hauw et al. 2017; Esteve et al. 2012, 2016; Grow and Van Bavel 2015).

Here we investigate how the new pattern of educational assortative mating shapes the relative contribution of partnered women to joint couple income across European countries. Drawing on the European Union Statistics on Income and Living Conditions (EU-SILC), we attempt to explain country differences in women’s relative earnings, compared with their male partners, by accounting for micro-level compositional differences in terms of educational pairings and parenthood as well as the interaction between these two. After accounting for these micro-level factors, we investigate the extent to which the remaining country differences can be explained by male unemployment and country-level gender norms.

The interplay between the relative education and relative earnings of female and male partners is an important topic for family demography because of its potential implications for fertility and divorce (Esteve et al. 2016; Schwartz and Han 2014; Van Bavel 2012). A switch from hypergamy to hypogamy is expected to affect women’s contribution to the joint couple income. If the female partner has more education than her male partner, she will tend to have a higher earning potential in the paid labor market (Wang et al. 2013; Winkler 1998; Winkler et al. 2005). Theoretically, this shift should affect the decision-making processes concerning both partners’ labor market participation and thus the timing and quantum of fertility (Van Bavel 2012).

Studies have consistently reported that mothers tend to scale down their paid labor market activity and earn less income after childbirth (Budig and England 2001; Budig et al. 2012). Yet, a recent study by Klesment and Van Bavel (2017) showed that this motherhood penalty may be modified by the relative education of the partners. The authors found that college-educated mothers of school-aged children with less-educated partners are about as likely to be the main breadwinners as college-educated childless women who are with a similarly educated partner. The Klesment and Van Bavel (2017) study had a number of limitations, however. First, it focused on a rather crude measure of relative earnings: namely, whether the wife is earning more than one-half of the joint couple income. We offer a more nuanced approach here, addressing the full continuum of the female partner’s relative income rather than grouping it into broad categories. Second, the earlier study is based on gross income measures, which may yield misleading results because couples in Europe may be taxed differently from country to country depending on the relative earnings of each partner. The current article therefore uses net income data, after taxation (which varies widely across Europe). Third, the earlier study did not include unemployment and sick leave benefits. Because such benefits are linked to employment in European welfare states, leaving them out may yield a distorted picture of the importance of his and her employment career for the family budget. We therefore include these benefits here. Finally, the earlier study looked only at the overall picture in the pooled sample of all countries. The current study aims to explore country differences. The primary focus here is the extent to which international differences in women’s relative earnings can be explained by compositional differences in women’s relative education.

## Background

In the male breadwinnerfemale homemaker family model he earns (almost) all the income, and she allocates most of her time to childcare and household work. In contrast, the dual-earner model refers to families in which both partners earn a significant if not equal share of the family income (Nock 2001; Raley et al. 2006). In this section, we formulate a set of hypotheses about how the female contribution to the couple’s joint earnings depends on educational assortative mating and its interaction with the motherhood status, and about macro-level factors that may account for cross-country variation in women’s relative earnings.

### Educational Assortative Mating

Differences between the earnings of spouses are partly driven by their relative human capital. The traditional male breadwinner model was linked with educational hypergamy. Yet, with the expansion of female participation in college-level education, educational homogamy became the modal marriage pattern in most Western countries during the twentieth century (Blossfeld 2009; Kalmijn 1991; Mare 1991; Schwartz 2013; Schwartz and Mare 2005). Correspondingly, dual-income households became the norm, but with the share of the male partner in the joint income typically largely exceeding the share of the female partner (Blossfeld 2009; Blossfeld and Timm 2003; Buss et al. 2001; Raley et al. 2006; Schwartz 2013; Winkler et al. 2005).

The reversal of the gender gap in education implies that more highly educated women than men enter the marriage market. Some of the highly educated women will then either remain single or select a partner with less education. Recent evidence indeed showed that hypogamy has become more prevalent than hypergamy in many countries, including the United States and most European countries (De Hauw et al. 2017; Esteve et al. 2012; Grow and Van Bavel 2015). Our baseline hypothesis is that a women’s relative education is positively associated with her relative earnings in the couple (Hypothesis 1). That is, women who are in hypogamous partnerships will have higher relative earnings than those who are in other pairing types. If so, compositional differences in educational pairings might at least partially explain country differences in women’s relative earnings.

### Motherhood and Relative Income

Women with children earn less relative to both their male partners and women without children. The presence of dependent children tends to lead to lower involvement in the paid labor market and, hence, to lower relative earnings (Budig and England 2001; Budig et al. 2012; Dotti Sani 2015; Gangl and Ziefle 2009). Several explanations for this motherhood penalty have been advanced, including selection effects as well as causal mechanisms (Anderson et al. 2002; Budig and England 2001; Petersen et al. 2010; Waldfogel 1998). Selection effects involve cases in which women who are more family-oriented opt out of better-paying but competitive and time-consuming careers in order to spend more time with their children (Chevalier 2007; Hakim 2003; Lück and Hofäcker 2008). However, the lower earnings of mothers may also be caused directly by the fact that most women at least temporarily retreat from paid work following childbirth (Budig et al. 2012; OECD 2011; Stier et al. 2001). Thus, motherhood implies an opportunity cost via reduced income simultaneous with the household need for increased income to account for the significant costs that children entail.

The extent of retreat from the labor market after entry into parenthood is expected to depend on his and her income potential, which is determined partly by the partners’ education. Both the level and the field of education affect income potential (Van Bavel 2010), but because our data lack information on field of study, we consider only educational attainment. For both men and women with lower levels of attainment, the gains from labor market activity and the opportunity costs of staying at home are relatively low. Although the need for additional income might produce the incentive to work more hours in paid labor, the opportunity costs of staying at home to provide childcare are low. For those with higher levels of education, the gains from paid labor and the opportunity costs of staying at home are higher. Parents with greater earning potential would need less work hours to achieve a given income level, but they will also lose more financially by staying at home (Becker 1993; Mincer 1963). Research in wealthy countries has typically found the opportunity costs predominating for women and the income effect for men (England 2010).

Women more often stay at home when they have children, given that the costs of outsourcing childcare undermine the incentives for paid labor, particularly among women with a low level of education. But when a woman has more education than her partner, the gender balance of income effects and opportunity costs may shift because she is then more likely to have a higher earning potential than her partner. In this case, her attachment to the labor market will be stronger because the income she generates is relatively more important to the family’s standard of living. In micro-economic terms, if she has higher potential income than he, the opportunity costs of staying at home to take care of children will tend to be relatively higher for her than for him. Among childless couples, the influence of opportunity costs is less critical given the lack of imperative to divide partners’ time between paid work and childcare. We therefore expect a smaller motherhood penalty on relative earnings in case of educational hypogamy and a larger motherhood penalty in case of hypergamy compared with homogamy (Hypothesis 2). If that is the case, compositional differences between countries in terms of the combined distribution of educational pairings and motherhood may explain at least part of the country gradient.

### Male Unemployment

Both economic and cultural factors may affect women’s relative earnings. Male unemployment is a major economic factor because the share of the female partner in the joint couple income will obviously be higher when her male partner is unemployed (Bettio et al. 2013; Chesley 2011; Vitali and Arpino 2016). Therefore, the economic crisis in the wake of the 2008 financial crisis may have affected women’s relative earnings because of high male unemployment rates independent of their relative education or any prevailing motherhood penalty. The hardest-hit economic sectors tended to employ more men than women (such as heavy industry), while sectors that were less affected tended to employ more women (such as education). Recent evidence indeed suggests that women’s relative income share increased in the wake of the 2008 financial crisis (Klesment and Van Bavel 2015; Vitali and Mendola 2014). In Europe, Bettio et al. (2013) found that the proportion of dual-earner couples declined during the downturn, while the share of female breadwinner couples increased to almost 10 %. Similarly, in the United States, Chesley (2011) saw the decline of men’s employment as a major factor explaining the rise of breadwinner women. Vitali and Arpino (2016) found that regional male unemployment rates were positively associated with women’s self-reported share in the household income in Europe. Based on these prior studies, we expect that part of the country gradient in women’s relative income is due to male unemployment. Hypothesis 3 therefore states that the country gradient in women’s relative earnings will be reduced after we account for male unemployment.

### Gender Norms

International differences in women’s relative earnings may also be due to cultural factors related to gender roles. Female labor market participation and earnings of mothers vary substantially across countries in accordance with prevailing gender role expectations and cultural family ideals (Budig et al. 2012; Craig and Mullan 2010; Harkness and Waldfogel 2003; Stier et al. 2001). The effect of gender role attitudes on relative earnings may be indirect as well as direct. In some countries, widespread reluctance to enroll young children in formal childcare may prolong career interruptions after childbearing (Davies and Pierre 2005; Lundberg and Rose 2000; Molina and Montuenga 2009; Sigle-Rushton and Waldfogel 2007), especially lowering the earnings of mothers. Beliefs, values, and attitudes about gender roles may also sustain gender inequality more directly (Lewis 1992; Pascall and Lewis 2004). Institutional arrangements hindering gender equality in the labor market (Lewis 1992; Neyer and Andersson 2008; Thévenon 2011) typically reflect cultural beliefs and attitudes that may underpin the male breadwinner–female homemaker family model (Creighton 1999; Esping-Andersen 2009; Janssens 1997; Lewis 2001; Pfau-Effinger 1998). Gender roles embedded in this family model are explicitly at odds with women earning more than their partners, especially if they have children. To the extent that such traditional views prevail, women may modify their labor supply in order to avoid a gender role reversal in earnings. Hence, traditional gender attitudes may undermine the hypogamy bonus on relative earnings by preventing women to harvest the returns to education in the labor market. We therefore expect the hypogamy bonus on relative earnings to be smaller in countries that have more conservative gender role attitudes (Hypothesis 4).

## Data and Methods

### Sample Selection

We use the EU-SILC, an annual household survey collecting both cross-sectional and longitudinal data. In most countries, the survey applies a rotating panel design with a length of four waves. Each subsequent wave replaces part of the sample, and the entire sample is renewed across a four-year period (Atkinson and Marlier 2010). We analyze cross-sectional data from the 2007 and 2011 waves, ensuring that samples do not overlap. The income reference period is the year before the year of interview, so our income data cover earnings in 2006 and 2010, respectively. Our focus on woman’s contribution to the joint net earnings of the couple requires limiting the study sample to the 16 countries that provide this measure (see Table 1).

We select women who are living with a partner at the time of the survey, either married or unmarried, and who are aged 25 to 45. To calculate the contribution of the female partner to the couple’s combined income, only couples in which at least one partner had earned income are included. We include income earned through both employment and self-employment. Our measure incorporates unemployment benefits and sick leave benefits, which compensate for the temporary absence from the labor market (Dotti Sani 2015). Excluding individuals with missing information, couples without any earnings, and observations with extreme values of the relative earnings measure (as explained in the next section) yields a study sample of 58,292 couples (30,702 in 2007 and 27,590 in 2011, respectively). Table 1 displays sample sizes per country included, along with basic descriptive statistics of the main variables featuring in the analysis.

### Measures

Our main dependent variable is the women’s relative income, defined as follows:
$wicf=yicf−yicmycm.$
1

In this definition, $yicf$ represents the earnings of the female partner of couple i in country c, $yicm$ is the earnings of the male partner in the same couple and country, and the denominator $ycm$ represents the average earnings of men in that country. The relative income is standardized by the country’s average male earnings to prevent scores from reflecting country differences in income level rather than gender difference because the potential raw differences between his and her income are greater in high-earning countries than in low-income countries. Women’s relative earnings, as defined in Eq. (1), therefore represent her relative earnings proportional to the average male earnings level in a given country. If he and she earn exactly the same amount, this variable equals 0. Positive values signify that she makes more money than he, and negative values signify that she earns less than he. For example, a value of 0.10 means that she outearns him by 10 % of the national male earnings; a value of –0.50 means that she earns 50 % of the national male average earnings less than her partner. We exclude 118 observations with very extreme values in which her relative earnings were less than –5.00 (i.e., earnings difference between the spouses in favor of the man is five times the average male earnings); we exclude 24 more cases in which relative earnings are above +4.00 (i.e., earnings difference in favor of the woman is four times the average male earnings). Figure S1 in Online Resource 1 plots the country-specific distributions of the dependent variable.

Our baseline Hypothesis 1 states that educational hypogamy is associated with higher relative earnings for women. To measure the educational attainment of the male and female partners, EU-SILC used a measure to facilitate international comparison, the ISCED-97 scale (UNESCO 2003). We collapse this scale into three categories: low (ISCED levels 0–2, up to the second stage of basic education, equivalent to 7th to 9th grades in the United States); medium (ISCED 3–4, secondary education or postsecondary but not tertiary; up to 12th grades, vocational education, or junior and community colleges); and higher education (ISCED 5–6, university level bachelors, masters, and PhD). Using these broad categories implies a loss of information and obviously amplifies the level of educational homogamy, but it ensures that the categories represent substantive differences in educational attainment in Europe.

Hypothesis 2 holds that the motherhood penalty on relative earnings will be smaller in case of educational hypogamy. To measure motherhood status, we discern whether any children are living in the couple’s household. EU-SILC does not collect direct fertility information on the number of children ever born. Hence, what we will capture in this article is really the effect of having a child living at home rather than the effect of parenthood per se.

To address Hypothesis 3 concerning male unemployment, we use individual-level data from the EU-SILC on the number of months men spent unemployed during the income reference period. We enter the months of unemployment as a continuous variable in regression models. Average values of the unemployment variable are shown in Table 1.

The EU-SILC does not contain data that can be used to test Hypothesis 4, which relates to attitudes about gender roles. To obtain relevant country-level indicators, we incorporate data from the European Social Survey (ESS), which samples people aged 15 and older for almost all European countries (Jowell et al. 2007). Two questions about conservative male breadwinner–female homemaker attitudes were asked in Rounds 2 and 5 (fieldwork around 2004 and 2010, respectively). Respondents were asked to indicate on a five-point scale whether they strongly agreed, agreed, neither agreed nor disagreed, disagreed, or strongly disagreed with the following statements: “Women should be prepared to cut down on paid work for sake of family,” and “Men should have more right to job than women when jobs are scarce.” We calculated the percentages agreeing or strongly agreeing with these items and took the average of the two ESS rounds. Bulgaria and Lithuania participated in Round 5 but not in Round 2, whereas Italy and Luxembourg participated in Round 2 but not Round 5; thus, we can use data for only one survey year for these four countries. Table 1 gives country-specific means of both ESS variables.

As control variables, we include the woman’s age and age squared in years, a dummy variable for the year of income reference period (2006 vs. 2010), and the absolute level of the couple’s joint earnings (because women in poorer families tend to be more often the main earners; see Winslow-Bowe 2006) in country-specific quartiles.

### Analytic Strategy

To test our hypotheses, we fit linear models with country fixed effects and with standard errors adjusted to account for the clustering at the country level. We also fit binary logistic versions of our fixed-effects models with the dependent variable redefined as the probability that the female earns more than her male partner. Details about the binary approach are reported in Online Resource 1, but the picture that emerges from it is the same as that reported here.

We first test whether the woman’s relative earnings are associated with being in a hypergamous, hypogamous, or homogamous union, controlling for her own level of educational attainment (Hypothesis 1). Next, we test whether the motherhood penalty on relative earnings is smaller in hypogamous unions than in other union types (Hypothesis 2) by including an interaction term between motherhood and the educational pairing variable. For Hypothesis 3, we test whether the variance of the country fixed effects becomes smaller after we add a measure of male unemployment to the model. Finally, we evaluate the correlation between (1) the country-level indicators for conservative gender role attitudes and (2) the country-specific effects of hypogamy on women’s relative earnings (Hypothesis 4). Given the limited number of countries that can be included from the EU-SILC data, fitting a multilevel model with both individual- and country-level variables is not feasible. Instead, we interact country effects with the effect of educational pairings to obtain country-specific estimates of the effect of educational hypogamy and hypergamy compared with homogamy.

## Results

Table 1 shows basic descriptive statistics about the distribution of couples by type of educational pairing in each country. Equally educated couples are the majority of the sample in every country. Among heterogamous unions, hypogamy is more common than hypergamy in all countries except Austria and Romania. The observation that hypogamy has become more prevalent than hypergamy confirms findings from other recent studies (De Hauw et al. 2017; Esteve et al. 2012; Grow and Van Bavel 2015).

Figure 1 shows women’s relative earnings by educational pairing and country. Overall, across pairing types, women’s relative earnings are lowest in Austria, Italy, Greece, and Estonia (where women’s earnings are lower than their partners’ by approximately 50 % of the national male average) and highest in Portugal, Sweden, and Slovenia (where the earnings gaps are less than 35 %). In all countries, women in hypogamous couples have higher relative earnings than women in homogamous or hypergamous couples, although the differences between pairing types are very small in Sweden. Figure 2 shows that in all countries, mothers consistently have lower relative earnings than childless women.

Table 2 reports the estimates from the regression models. The baseline Model 1 contains only control variables: her age (centered on age 35) and age squared, her educational level, the joint couple earnings, the income reference year, and country fixed effects. Unsurprisingly, higher educational attainment is associated with women’s higher relative earnings. The couple’s combined earnings are negatively correlated with relative income: the higher the joint earnings, the lower the woman’s share tends to be. As expected from the economic climate of the period, women’s relative earnings were somewhat higher in 2010 than in 2006.

The country fixed effects for Model 1, plotted in Fig. 3, are very much in line with the country differences described with the black solid points in Figs. 1 and 2. The major difference is that after controls, Belgium is found in the middle of the country distribution, whereas it was near the top for women’s relative earnings in the unadjusted distribution. This shift reflects the fact that the female population of Belgium is very highly educated. After we control for education, we find that Belgium lands in the middle of the distribution.

In Model 2, we focus on the key substantive variables of the study: educational pairing and motherhood status. Mothers have lower relative earnings than childless women. On average across countries, they earn approximately 23 % less than their male partners (–0.232, standardized to the country-specific male average). The parameters for the educational pairings are consistent with Hypothesis 1: net of the control variables and motherhood status, women in hypogamous unions have almost 14 % higher relative earnings than those in homogamous ones and approximately 28 % higher than the women in hypergamous couples.

The results shown in Model 3 for the interaction between motherhood and educational pairings are consistent with Hypothesis 2. The motherhood effect on relative earnings is significantly smaller for women in hypogamous unions than for their homogamously paired peers; the difference is approximately 7 percentage points (as implied by parameter estimate 0.074 in Model 3) and is statistically significant at the p < .01 level. The same interaction can also be considered the other way around: the hypogamy bonus on relative earnings is approximately 7 percentage points larger for mothers than for childless women. Adding the effects of educational pairings and the interactions with motherhood, the differences in predicted values are quite substantial. For mothers in hypogamous unions, the predicted relative earnings are approximately –10 % (–0.254 + 0.080 + 0.740 = –0.100; all other variables at their reference levels), compared with approximately –25 % for mothers in homogamous unions and –39 % for mothers in hypergamous unions (–0.254 – 1.154 + 0.015 = –0.393).

We have no clear evidence in support of the second part of Hypothesis 2, regarding the difference between hypergamous and homogamous women. The difference in motherhood effect between these two groups is not statistically significant. In additional tests not reported here, we also distinguished between couples with children younger than 5 and those with children older than 5, but this hardly made a difference for the estimates and does not affect our conclusions. The negative motherhood effect is stronger when the youngest child is younger than 5, but the interaction with hypogamy stays the same.

So far, the findings suggest that women’s relative earnings are a function of their absolute and relative education, motherhood status, and the interaction between the latter two variables. International differences in relative earnings may therefore be due to the fact that these variables are differentially distributed in different countries. The country effects plotted in Fig. 3 address this issue. These reflect the residual country differences in women’s relative earnings after accounting for the other variables in the model. The country effects from Model 3 are considerably closer to the zero line than the ones from baseline Model 1, confirming that motherhood and (relative) education are important reasons why women’s relative earnings are generally lower than men’s in all countries. However, accounting for these factors does not reduce the country gradient. The variability of unexplained country differences is not smaller for Model 3 compared with Model 1; the standard deviation of the country fixed effects in Model 3 is 0.0843, which is actually a bit larger than the one for Model 1 (0.0831).

Model 4 includes male unemployment with the expectation that this factor would account for some of the remaining cross-country differences (Hypothesis 3). Male unemployment is clearly associated with higher relative earnings on the micro level: a month of unemployment for the male partner is associated with an increase of 5 % of the female partner’s yearly relative earnings. However, this factor cannot explain the country differences given that the standard deviation for the residual country effects in Model 4 is 0.0854, which is not smaller than the one for the country effects in Model 3 (0.0843). Interestingly, all the residual country effects are more strongly negative in Model 4 than in Model 3, indicating that the share of women’s earnings in the joint couple income is elevated by male unemployment in all countries and that women’s expected relative earnings are lower again after this factor is accounted for. We illustrate this point by comparing Spain and Sweden. These two countries are similar in the baseline rank ordering of women’s relative income (Model 1). However, after we account for male unemployment, the expected relative earnings of women are much lower in Spain than in Sweden. The rank order based on Model 4 places Spain in a much lower position—somewhere between Italy and Poland—because the male unemployment rate in Spain is three times as high as in Sweden (see Table 1), which results in the comparably high relative earnings in Spain in the baseline ranking. In sum, male unemployment explains an important part of the gender earnings gap in all countries, and it explains part of the relative position between specific countries, but it cannot explain the overall country gradient in women’s relative earnings.

Figure 4 plots our two country-level measures of conservative gender role attitudes against country-specific estimates of the effect of hypogamy compared with homogamy (upper panels) and hypergamy (lower panels). Hypothesis 4 is that the positive effect of hypogamy on women’s relative earnings is lower in countries that exhibit more traditional gender norms. Yet, if anything, we find the opposite to be true, as indicated by the positive slopes of the lines in Fig. 4: the more people tend to agree with these conservative statements about gender roles, the higher the positive effect of hypogamy tends to be. The correlations are weak, however, and most of them are not statistically significant at the 5 % or even 10 % level. The only exception is the contrast between hypogamy and hypergamy in the lower-left panel, which shows that the positive effect of hypogamy on women’s relative earnings compared with women in hypergamous unions is statistically significantly correlated with the percentage strongly agreeing/agreeing that women should be prepared to reduce paid work for the sake of family (r = .56, p < .04). Again, against Hypothesis 4, the correlation is positive, not negative.

Additional descriptive data explorations, depicted in Fig. 5, suggest that the reason for the significant correlation in the case of the contrast between hypogamy and hypergamy is the fact that hypergamous women tend to have lower relative earnings in countries that exhibit more conservative gender norms, while the country-level correlation between gender norms and relative earnings is absent or much weaker for hypogamous women.

## Conclusion

The reversal of the gender gap in education represents a major social development, not just in the West but also in many other regions of the world (Esteve et al. 2016). It has potentially wide-ranging implications for gender roles, including female and male economic roles in the family. Our aim was to investigate the extent to which international differences in women’s relative earnings could be explained by couple-level gender differences in educational attainment and their interaction with parenthood status. We also investigated male unemployment and attitudes about gender roles as additional candidates to explain country differences. Based on EU-SILC data on net earnings of partnered men and women in 16 European countries, we found considerable heterogeneity in women’s relative income, even though women, on average, earned less than their male partners in all countries.

As a baseline, we examined educational pairings and found that in all countries included, women who are more educated than their male partners have higher relative earnings. Next, considering the reversed gender gap in education led us to hypothesize that the negative association between motherhood and relative earnings would be weaker for women who are more educated than their male partners, compared with women who have the same or a lower attainment level than their male partners. Indeed, our results indicate that the motherhood penalty on relative earnings is weaker for hypogamous women than for women in other pairing types and further implies that the hypogamy bonus on women’s relative earnings is stronger for mothers than for childless women.

At the macro level, we found that neither the country distributions of women’s absolute and relative education nor differences in motherhood status and their interaction with relative education can explain the country heterogeneity in women’s relative earnings. Additional analyses showed that male unemployment has a positive effect on women’s relative earnings in all countries, but it cannot explain the remaining country heterogeneity of those earnings. Male unemployment helps to account for why women’s relative earnings tend to be higher in some particular countries than in others (much in line with recent findings by Vitali and Arpino 2016), but the unexplained variance in country differences remains just as high as before unemployment is accounted for.

Turning to cultural factors, we examined the association between the country-specific effects of hypogamy on women relative earnings and two country-level measures of prevailing attitudes about gender roles regarding women’s position in the labor market. We expected that the positive effect of hypogamy on relative earnings would be more limited in countries where conservative attitudes toward gender roles prevail. Yet, if anything, we found the opposite to be the case. The rationale for our hypothesis was that conservative gender norms may hinder well-educated women to cash the returns to their degrees in the labor market, but this is apparently not the case. Instead, we found that conservative norms hinder the earnings of less-educated, hypergamous women. The more-educated, hypogamous women are not really bothered by traditional gender norms that may prevail in their country: for them, we found at best a weak association between their relative earnings and country-level gender role attitudes, whereas less-educated hypergamous women clearly earn less (compared with their partners) in countries with traditional attitudes. Yet, our macro-level correlations should not be considered as conclusive evidence because our tests were based on a limited number of countries and because our sample-based measures of gender role attitudes obviously are subject to error.

Apart from that, our study has other important limitations. First, the number of countries in the study sample was limited to 16 because the data for the other European countries did not include information on net income. Given the important differences between European countries in fiscal arrangements, it was important to use a net income measure to evaluate international differences in women’s relative contribution to the joint disposable income. This requirement reduced the number of countries to a level that precluded robust estimation of multilevel models. For the analysis of gender role attitudes, we were further limited to 14 countries, thus undermining the statistical power of our macro-level analysis.

A second limitation is that we examined only the educational attainment level of male and female partners. Information on their respective fields of education was not available in our data set. However, the field of study clearly has strong implications for the earning potential in the labor market for women and men and is strongly related to the transition to motherhood as well as to attitudes about gender roles (Van Bavel 2010). Choice of study disciplines along gender-stereotypical lines might explain why women often earn less than their partners even in hypogamous couples. Third, the reported associations between educational pairings and women’s share in the family income cannot be interpreted as pure causal effects. The selection into union, union type, and motherhood status are likely to be correlated with women’s earnings and educational level (Dribe and Nystedt 2013). Our results therefore reflect both selection and causal effects.

Nevertheless, the micro-level findings from this study are very robust. They have important implications for family demography, regardless of whether they reflect causal or selection effects. The rising prevalence of hypogamy, the decline of hypergamy, and high male unemployment mean that women’s earnings represent a growing share of joint couple income. This is bound to affect the couple’s negotiations about the division of paid and unpaid work and, hence, their decision-making about the timing and quantum of fertility (Esping-Andersen and Billari 2015; Goldscheider et al. 2015).

## Acknowledgments

The research leading to these results has received funding from the European Research Council under the European Union Seventh Framework Programme (FP/2007-2013)/ERC Grant Agreement no. 312290 for the GENDERBALL project. The authors thank Gray Swicegood as well as the anonymous reviewers for many useful comments and suggestions.

## References

Anderson, D. J., Binder, M., & Krause, K. (
2002
).
The motherhood wage penalty: Which mothers pay it and why?
.
American Economic Review
,
92
,
354
358
. 10.1257/000282802320191606
Atkinson, A. B., & Marlier, E. (
2010
).
Income and living conditions in Europe
.
Luxembourg City, Luxembourg
:
European Union
.
Becker, G. S. (
1993
).
A treatise on the family
(Enlarged ed.).
Cambridge, MA
:
Harvard University Press
.
Bettio, F., Corsi, M., D’Ippoliti, C., Lyberaki, A., Lodovici, M. S., & Verashchagina, A. (
2013
).
The impact of the economic crisis on the situation of women and men and on gender equality policies
.
Luxembourg City, Luxembourg
:
European Commission
.
Blossfeld, H-P (
2009
).
Educational assortative marriage in comparative perspective
.
Annual Review of Sociology
,
35
,
513
530
. 10.1146/annurev-soc-070308-115913
Blossfeld, H.-P., & Timm, A. (
2003
).
Educational systems as marriage markets in modern societies: A conceptual framework
. In H.-P. Blossfeld & A. Timm (Eds.),
Who marries whom? Educational systems as marriage markets in modern societies
(pp.
1
18
).
Dordrecht, the Netherlands
:
.
Buchmann, C., & DiPrete, T. A. (
2006
).
The growing female advantage in college completion: The role of family background and academic achievement
.
American Sociological Review
,
71
,
515
541
. 10.1177/000312240607100401
Budig, M. J., & England, P. (
2001
).
The wage penalty for motherhood
.
American Sociological Review
,
66
,
204
225
. 10.2307/2657415
Budig, M. J., Misra, J., & Boeckmann, I. (
2012
).
The motherhood penalty in cross-national perspective: The importance of work-family policies and cultural attitudes
.
Social Politics
,
19
,
163
193
. 10.1093/sp/jxs006
Buss, D. M., Shackelford, T. K., Kirkpatrick, L. A., & Larsen, R. J. (
2001
).
A half century of mate preferences: The cultural evolution of values
.
Journal of Marriage and Family
,
63
,
491
503
. 10.1111/j.1741-3737.2001.00491.x
Chesley, N. (
2011
).
Stay-at-home fathers and breadwinning mothers: Gender, couple dynamics, and social change
.
Gender & Society
,
25
,
642
664
. 10.1177/0891243211417433
Chevalier, A. (
2007
).
Education, occupation and career expectations: Determinants of the gender pay gap for UK graduates
.
Oxford Bulletin of Economics and Statistics
,
69
,
819
842
. 10.1111/j.1468-0084.2007.00483.x
Craig, L., & Mullan, K. (
2010
).
Parenthood, gender and work family time in the United States, Australia, Italy, France, and Denmark
.
Journal of Marriage and Family
,
72
,
1344
1361
. 10.1111/j.1741-3737.2010.00769.x
Creighton, C. (
1999
).
The rise and decline of the male breadwinner family in Britain
.
Cambridge Journal of Economics
,
23
,
519
541
. 10.1093/cje/23.5.519
Davies, R., & Pierre, G. (
2005
).
The family gap in pay in Europe: A cross-country study
.
Labour Economics
,
12
,
469
486
. 10.1016/j.labeco.2005.05.003
De Hauw, Y., Grow, A., & Van Bavel, J. (
2017
).
The reversed gender gap in education and assortative mating in Europe
.
European Journal of Population
Dotti Sani, G. M. (
2015
).
Within-couple inequality in earnings and the relative motherhood penalty. A cross-national study of European countries
.
European Sociological Review
,
31
,
667
682
. 10.1093/esr/jcv066
Dribe, M., & Nystedt, P. (
2013
).
Educational homogamy and gender-specific earnings: Sweden 1990–2009
.
Demography
,
50
,
1197
1216
. 10.1007/s13524-012-0188-7
England, P. (
2010
).
The gender revolution: Uneven and stalled
.
Gender & Society
,
24
,
149
166
. 10.1177/0891243210361475
Esping-Andersen, G. (
2009
).
The incomplete revolution. Adapting to women’s new roles
.
Cambridge, UK
:
Polity Press
.
Esping-Andersen, G., & Billari, F. C. (
2015
).
Re-theorizing family demographics
.
Population and Development Review
,
41
,
1
31
. 10.1111/j.1728-4457.2015.00024.x
Esteve, A., García-Román, J., & Permanyer, I. (
2012
).
The gender-gap reversal in education and its effect on union formation: The end of hypergamy?
.
Population and Development Review
,
38
,
535
546
. 10.1111/j.1728-4457.2012.00515.x
Esteve, A., Schwartz, C., Van Bavel, J., Permanyer, I., Klesment, M., & Garcia, J. (
2016
).
The end of hypergamy: Global trends and implications
.
Population and Development Review
,
42
,
615
625
Gangl, M., & Ziefle, A. (
2009
).
Motherhood, labor force behavior, and women’s careers: An empirical assessment of the wage penalty for motherhood in Britain, Germany, and the United States
.
Demography
,
46
,
341
369
. 10.1353/dem.0.0056
Goldscheider, F., Bernhardt, E., & Lappegård, T. (
2015
).
The gender revolution: A framework for understanding changing family and demographic behavior
.
Population and Development Review
,
41
,
207
239
. 10.1111/j.1728-4457.2015.00045.x
Grow, A., & Van Bavel, J. (
2015
).
Assortative mating and the reversal of gender inequality in education in Europe: An agent-based model
.
PLoS ONE
,
10
(
6
),
1
24
. 10.1371/journal.pone.0127806
Hakim, C. (
2003
).
A new approach to explaining fertility patterns: Preference theory
.
Population and Development Review
,
29
,
349
374
. 10.1111/j.1728-4457.2003.00349.x
Harkness, S., & Waldfogel, J. (
2003
).
The family gap in pay: Evidence from seven industrialized countries
.
Research in Labor Economics
,
22
,
369
413
. 10.1016/S0147-9121(03)22012-4
Janssens, A. (
1997
).
The rise and decline of the male breadwinner family? An overview of the debate
.
International Review of Social History
,
42
(
Suppl. 5
),
1
24
. 10.1017/S0020859000114774
Jowell, R., Roberts, C., Fitzgerald, R., & Gillian, E. (
2007
).
Measuring attitudes cross-nationally: Lessons from the European Social Survey
.
London, UK
:
Sage
.
Kalmijn, M. (
1991
).
Shifting boundaries: Trends in religious and educational homogamy
.
American Sociological Review
,
56
,
786
800
. 10.2307/2096256
Klesment, M., & Van Bavel, J. (
2015
).
The reversal of the gender gap in education and female breadwinners in Europe
(FamiliesAndSocieties Working Paper Series 26).
Stockholm, Sweden
:
Stockholm University Department of Sociology
.
Klesment, M., & Van Bavel, J. (
2017
).
The reversal of the gender gap in education, motherhood, and women as main earners in Europe
.
European Sociological Review
,
33
,
465
481
.
Lewis, J. (
1992
).
Gender and the development of welfare regimes
.
Journal of European Social Policy
,
2
,
159
173
. 10.1177/095892879200200301
Lewis, J. (
2001
).
The decline of the male breadwinner model: Implications for work and care
.
Social Politics: International Studies in Gender, State & Society
,
8
,
152
169
. 10.1093/sp/8.2.152
Lück, D., & Hofäcker, D. (
2008
).
The values of work and care among women in modern societies
. In van Oorschot, W., Opielka, M., & Pfau-Effinger, B. (Eds.),
Culture and welfare state: Values and social policy in comparative perspective
(pp.
289
313
).
Cheltenham, UK
:
Edward Elgar
.
Lundberg, S., & Rose, E. (
2000
).
Parenthood and the earnings of married men and women
.
Labour Economics
,
7
,
689
710
. 10.1016/S0927-5371(00)00020-8
Mare, R. D. (
1991
).
Five decades of educational assortative mating
.
American Sociological Review
,
56
,
15
32
. 10.2307/2095670
Mincer, J. (
1963
).
Market prices, opportunity costs, and income effects
. In Christ, C. (Ed.),
Measurement in economics
(pp.
67
82
).
Stanford, CA
:
Stanford University Press
.
Molina, J. A., & Montuenga, V. M. (
2009
).
The motherhood wage penalty in Spain
.
Journal of Family and Economic Issues
,
30
,
237
251
. 10.1007/s10834-009-9153-z
Neyer, G., & Andersson, G. (
2008
).
Consequences of family policies on childbearing behavior: Effects or artifacts?
.
Population and Development Review
,
34
,
699
724
. 10.1111/j.1728-4457.2008.00246.x
Nock, S. L. (
2001
).
The marriages of equally dependent spouses
.
Journal of Family Issues
,
22
,
755
775
. 10.1177/019251301022006005
Organisation for Economic Co-operation and Development (OECD)
. (
2011
).
Doing better for families
.
Paris, France
:
OECD Publishing
.
Pascall, G., & Lewis, J. (
2004
).
Emerging gender regimes and policies for gender equality in a wider Europe
.
Journal of Social Policy
,
33
,
373
394
. 10.1017/S004727940400772X
Petersen, T., Penner, A. M., & Høgsnes, G. (
2010
).
The within-job motherhood wage penalty in Norway, 1979–1996
.
Journal of Marriage and Family
,
72
,
1274
1288
. 10.1111/j.1741-3737.2010.00764.x
Pfau-Effinger, B. (
1998
).
Gender cultures and the gender arrangement—A theoretical framework for cross-national gender research
.
Innovation: The European Journal of Social Science Research
,
11
,
147
166
.
Raley, S. B., Mattingly, M. J., & Bianchi, S. M. (
2006
).
How dual are dual-income couples? Documenting change from 1970 to 2001
.
Journal of Marriage and Family
,
68
,
11
28
. 10.1111/j.1741-3737.2006.00230.x
Schwartz, C. R. (
2013
).
Trends and variation in assortative mating: Causes and consequences
.
Annual Review of Sociology
,
39
,
1
41
. 10.1146/annurev-soc-071312-145544
Schwartz, C. R., & Han, H. (
2014
).
The reversal of the gender gap in education and trends in marital dissolution
.
American Sociological Review
,
79
,
605
629
. 10.1177/0003122414539682
Schwartz, C. R., & Mare, R. D. (
2005
).
Trends in educational assortative marriage from 1940 to 2003
.
Demography
,
42
,
621
646
. 10.1353/dem.2005.0036
Sigle-Rushton, W., & Waldfogel, J. (
2007
).
Motherhood and women’s earnings in Anglo-American, continental European, and Nordic countries
.
Feminist Economics
,
13
(
2
),
55
91
. 10.1080/13545700601184849
Stier, H., Lewin-Epstein, N., & Braun, M. (
2001
).
Welfare regimes, family-supportive policies, and women’s employment along the life-course
.
American Journal of Sociology
,
106
,
1731
1760
. 10.1086/321302
Thévenon, O. (
2011
).
Family policies in OECD countries: A comparative analysis
.
Population and Development Review
,
37
,
57
87
. 10.1111/j.1728-4457.2011.00390.x
United Nations Educational, Scientific and Cultural Organization (UNESCO)
. (
2003
).
International standard classification of education, ISCED 1997
. In J. H. P. Hoffmeyer-Slotnik & C. Wolf (Eds.),
Advances in cross-national comparison. A European working book for demographic and socio-economic variables
(pp.
195
220
).
New York, NY
:
.
Van Bavel, J. (
2010
).
Choice of study discipline and the postponement of motherhood in Europe: The impact of expected earnings, gender composition and family attitudes
.
Demography
,
47
,
439
458
. 10.1353/dem.0.0108
Van Bavel, J. (
2012
).
The reversal of gender inequality in education, union formation and fertility in Europe
.
Vienna Yearbook of Population Research
,
10
,
127
154
. 10.1553/populationyearbook2012s127
Vitali, A., & Arpino, B. (
2016
).
Who brings home the bacon? The influence of context on partners’ contributions to the household income
.
Demographic Research
,
35
(
41
),
1213
1244
. 10.4054/DemRes.2016.35.41
Vitali, A., & Mendola, D. (
2014
).
Women as main earners in Europe
(Centre for Population Change Working Paper No. 56).
Southampton, UK
:
Economic & Social Research Council Centre for Population Change
.
Waldfogel, J. (
1998
).
Understanding the “family gap” in pay for women with children
.
Journal of Economic Perspectives
,
12
(
1
),
137
156
. 10.1257/jep.12.1.137
Wang, W., Parker, K., & Taylor, P. (
2013
).
.
Washington, DC
:
Pew Research Center
.
Winkler, A. E. (
1998
).
Earnings of husbands and wives in dual-earner families
.
Monthly Labor Review
,
121
(
4
),
42
48
.
Winkler, A. E., McBride, T. D., & Andrews, C. (
2005
).
Wives who outearn their husbands: A transitory or persistent phenomenon for couples?
.
Demography
,
42
,
523
535
. 10.1353/dem.2005.0028
Winslow-Bowe, S. (
2006
).
The persistence of wives’ income advantage
.
Journal of Marriage and Family
,
68
,
824
842
. 10.1111/j.1741-3737.2006.00298.x