- Split View
-
Views
-
Cite
Cite
Martin Klesment, Jan Van Bavel, Women’s Relative Resources and Couples’ Gender Balance in Financial Decision-Making, European Sociological Review, Volume 38, Issue 5, October 2022, Pages 739–753, https://doi.org/10.1093/esr/jcac019
- Share Icon Share
Abstract
We investigate how the relative education and earnings of husbands and wives are associated with self-reported decision-making within the family. Using European Union Statistics on Income and Living Conditions 2010 data for 27 European countries (n = 72,638), we find that women who earn more than their partner are more likely to report that they alone make the major financial and other important decisions. Men are more likely than women to be reported as financial decision makers if women contribute less than a quarter to joint earnings. However, in line with predictions based on traditional gender display, the association with relative earnings is not linear: among couples in which wives earn almost all of the income, we find that husbands are reported to have more say in financial decision-making than among couples in which both contribute a substantial part of the joint income. This non-linear pattern does not hold similarly for general decision-making. The discrepancy suggests that major financial issues, which were traditionally within the male realm, may be more susceptible to gender display than other family decisions.
Introduction
Over the past decades, gender role differentiation has been weakening in most Western societies. The educational attainment, labour market participation, and earnings of men and women have become more similar over time (Buchmann and DiPrete, 2006; Goldin, 2006; Lutz, Cuaresma and Sanderson, 2008). An important trend in this context has been the reversal of the gender gap in education: among most young-adult cohorts in Western countries, women have become more educated than men (Buchmann and DiPrete, 2006; Van Bavel, Schwartz and Esteve, 2018). In these countries, couples in which the wife has a higher education than her husband now outnumber couples in which the husband is more highly educated (Esteve et al., 2016; De Hauw, Grow and Van Bavel, 2017). More education often implies higher earning potential and, in effect, women’s higher relative education has clearly increased the proportion of families in which she is the main income provider (Wang, Parker and Taylor, 2013; Vitali and Arpino, 2016; Klesment and Van Bavel, 2017).
Despite these trends, gender inequality remains strong, particularly in the private family sphere (England, 2010; Pedulla and Thébaud, 2015). While ideological support for gender equality has increased over the past decades (Brewster and Padavic, 2000; Thornton and Young-DeMarco, 2001; Davis and Greenstein, 2009; Pampel, 2011), there is also evidence of a recent stall in the move towards egalitarianism (Cotter, Hermsen and Vanneman, 2011; Pedulla and Thébaud, 2015), or even a revival of traditional gender ideology within the sphere of family life (Pepin and Cotter, 2018).
Consequently, it remains to be seen how rising gender equality in education and the labour market is affecting power relations between men and women in the realm of intimate relationships. Both structural and ideological factors are likely to play a role. From a structural point of view, both the micro-economic perspective (Becker, 1993) and the marital power perspective (Blood and Wolfe, 1960) highlight the role of relative resources, including differences in human capital between partners. Many studies have found that more equal resources are associated with more equal relationships in the home (see reviews in Coltrane, 2000; Graf and Schwartz, 2011; Sullivan, 2011). However, critics argue that the distribution of resources is unable to explain persistent gender inequalities. Gender ideology matters as well: women’s higher earnings do not seem to buy them an equal position in the domestic sphere because of the persisting norms and attitudes that maintain traditional gender roles in the family (Tichenor, 2005a; England, 2010). Gender display may overrule relative resources (Bittman et al., 2003), i.e. when men or women, or both, continue ‘doing gender’ (West and Zimmerman, 1987)—acting out traditional gender roles—even when the resources have become more equal or even reversed in favour of women.
Earlier studies of gender inequality in the family sphere have mostly looked at the sharing of housework and childcare. This paper examines another dimension of gender equality within couples, namely, the balance in major financial decision-making. Equality in decision-making is argued to be one of the most important pillars supporting fairness in partnerships (Rosenbluth, Steil and Whitcomb, 1998). In this article, we investigate how women’s relative education and earnings, compared with their male partner’s, determine who makes important decisions for the family. We focus on decisions to spend significant sums of money, to borrow money, or spend savings. We call these three items ‘major outlay decisions’. Such decisions are made less frequently than decisions about everyday spending or management of the household. They are rarely part of the daily routine, used to be a more male-dominated domain in the past (Blood and Wolfe, 1960; Davis and Rigaux, 1974), and are therefore an aspect that differentiates our scope of analysis from the more common housework and childcare research. In order to put our findings about financial decision-making in a broader perspective, we compare them with findings about who in the couple is more likely to have ‘the final say’ when taking all types of important decisions.
Our analysis covers 27 European countries and is based on the 2010 ad-hoc module of European Union Statistics on Income and Living Conditions (EU-SILC). In this survey, respondents were asked to give their subjective opinion of how certain types of decisions are made in the household. We use this information to calculate a summary score for financial decision-making and then to model the probability that either the woman or the man would be more likely to be reported as the decision maker, compared with a more balanced decision-making process. The final say in general decision-making in the couple is captured by a separate question in the same module.
Our results support the notion that women in non-traditional unions, i.e. who are more educated than their partner and/or who earn a larger share of the couple’s income, tend to have more influence over major outlay decisions in the household, compared with women in traditional types of unions. But, in line with predictions based on traditional gender display, the association with relative earnings is not linear: when wives earn almost all of the income, we find that husbands are reported to have more say in financial decision-making than among couples in which both contribute a substantial part of the joint income. This non-linear pattern does not hold similarly for general decision-making. The discrepancy suggests that major financial issues, which were traditionally within the male realm, may be more susceptible to gender display than other family decisions.
Theoretical Perspectives
Relative Earnings and Women’s Decision-Making
From the perspective of relative resources, differences in individual resources such as income can explain the gender differentials in power and decision-making in the family (Blood and Wolfe, 1960; Blumberg and Coleman, 1989; Steil and Weltman, 1991). In the past, married women generally had less education than their husbands. Human Capital Theory holds that this supported a gendered division of labour within the couple (Becker, 1993). Since the husband focused on work in the labour market and was usually the main monetary contributor to the family budget, he was typically regarded as the ‘head of household’, i.e. the person who ‘had more say’ about critical household decisions. The wife was more involved in day-to-day spending in the domestic sphere while the breadwinning husband made the decisions about larger expenditures (Edgell, 1980; Nyman, 1999). Although typical of traditional male-breadwinner families, this pattern of decision-making has been observed among contemporary dual-earner couples as well (Bartley, Blanton and Gilliard, 2005). The higher earning spouse is more likely to maintain overall control and have ‘the final say’ while the less earning partner is occupied with the daily routine decision-making (Burgoyne et al., 2007). Some studies have suggested that the same principle applies to the division of housework, with the less earning spouse expected to do more housework in exchange for economic resources from the breadwinning partner (Brines, 1994; Greenstein, 2000), but more recent work has cast doubt on this (see Van Bavel, Schwartz and Esteve, 2018: pp. 347–351 for a review).
Previous research reports that women who earn more than their husband are more likely to be in control of decisions in the household and decide alone on making major purchases (Treas and Tai, 2012). Experimental studies demonstrate an effect of increased cash resources on female decision-making in the household, female autonomy, and family decisions such as the migration of one spouse (Nobles and McKelvey, 2015; Urbina, 2020). From the perspective of relative resources, our basic expectation is that the partner with a higher income would be more likely to make the major outlay decisions and have the final say in important decisions, irrespective of gender. We formulate our first hypothesis about the role of relative earnings accordingly:
Hypothesis 1: Higher relative earnings of a spouse are positively associated with the probability of him or her making decisions, and the higher the relative earnings, the greater the likelihood of being the principal decision maker.
Female Breadwinners and Gender Display
Resource Theory is gender-neutral in the sense that it assumes that the effect of relative earnings applies equally to women and men. However, earlier research has found that in households in which the woman earned a larger part of the couple’s income, the spouses often did not switch roles as the theory would predict (Atkinson and Boles, 1984). For example, some studies of the distribution of housework indicate that the relationship between the partners’ relative income and their share of the housework is not linear; men are not doing more housework as their female partners earn an increasing proportion of the joint income. Even when men become financially dependent on their wives, they do not necessarily adopt the role of dependent spouse as would be predicted by Resource Theory (Brines, 1994; Greenstein, 2000; Bittman et al., 2003).
The latter findings may be explained by gender norms and gender display (Greenstein, 1996). Deviation from traditional gender norms may result in attempts to compensate for this deviation in other dimensions. For example, wives who earn more than their husband might increase their efforts to perform traditionally female tasks, such as housework, in order to appear less deviant from the male-breadwinner model. This mode of ‘doing gender’ (West and Zimmerman, 1987) is called ‘gender display’ (Brines, 1994), or ‘gender-deviance neutralization’ (Atkinson and Boles, 1984; Greenstein, 2000). Support for this theory comes from quantitative research that finds a curvilinear rather than linear relationship between women’s relative income and the sharing of housework (Evertsson and Nermo, 2004, 2007; Aassve, Fuochi and Mencarini, 2014). However, it is possible that the evidence supporting gender display in housework may be due to a failure to account for the woman’s absolute income (Gupta, 2007; Killewald and Gough, 2010; Sullivan, 2011).
Nevertheless, Tichenor (2005b) finds evidence based on qualitative data that women married to an economically dependent husband may try to compensate for this deviance from traditional gender roles by leaving decision-making up to their husband. The normative expectations concerning men’s and women’s roles may be stronger than the implications of relative income (Tichenor, 2005b), at least in some social contexts.
Therefore, we expect that the likelihood of wives’ making the important financial decisions would not increase monotonically as their relative income increases, but rather that there would be a turning point after which the probability of their being the decision maker would begin to decline due to the gender display, while husbands’ probability of making the important financial decisions would increase for the same reason. This turning point is expected to be at higher levels of relative earnings, at which the wife earns significantly more than her husband. The latter situation is contrary to the norm that the husband should be the main provider, while couples in which both partners contribute approximately the same income are in line with the current dual-earner model (Chesley, 2011; Cherlin, 2016). Based on the gender ideology argument, our second hypothesis contains two parts, a and b:
Hypothesis 2: The positive relationship between women’s relative earnings and decision making is non-linear; the relative earnings of the wife are positively associated with the probability that she rather than her husband would make decisions only up to a certain point: among couples in which the wife earns (almost) all of the couple’s income (a) the probability that she would make decisions is lower, while (b) the probability that he would mainly make these decisions is higher, compared with couples in which she earns more but the husband also provides a substantial part of the joint income.
Relative and Absolute Level of Education
In addition to relative earnings, relative education is likely to be an important factor associated with the couple’s financial decision-making. First, education is a key factor in human capital, so the partner with the higher education may be in a stronger bargaining position with regard to decision-making. Second, an individual’s choice of a partner in terms of education may be correlated with egalitarian attitudes. Women with more traditional values may be less inclined to ‘marry down’, as it is contrary to the traditional expectation that women should ‘marry up’ (Schwartz, 2013; Van Bavel, Schwartz and Esteve, 2018). Conversely, men who choose an equally or more highly educated female partner may possess a more egalitarian attitude. Accordingly, women who have more education than their partner may have more influence in making major financial decisions, compared with more traditional couples. Among other things, partner choice based on education and earnings raises a question of selectivity as partners with lower education or earnings may self-select themselves into unions in which they have a lesser role in decision-making. However, our data do not permit taking selective union formation into account.
Based on these considerations, we may expect that women’s relative education and decision-making would be positively correlated, even after adjusting for relative earnings:
Hypothesis 3: The higher a woman’s level of education relative to the level of education of her male partner, the more likely it is that she is the primary decision maker.
However, a recent study suggests that women who marry down in terms of education are more likely to marry a man with higher relative earnings; this is interpreted as a function of gender display (Qian, 2017). When relative resources in terms of education dominate, we would expect Hypothesis 3 as formulated above to hold true. An alternative hypothesis is that gender display may also play a role with respect to relative education, so that when a woman has more education than her partner, she compensates by leaving the decision-making role to him.
Data and Analysis Plan
We use the 2010 round of EU-SILC. This dataset includes an ad-hoc module on intra-household sharing of resources, targeting households with at least two persons aged 16 years and above. In this module, the questions about decision-making were asked of individuals and they are intended to capture the respondent’s subjective point of view.
Both partners were interviewed separately, but there is no guarantee that the responses of one partner did not influence those of the other, so we cannot treat the responses given by male and female partners as independent samples of information. In order to determine the extent to which the responses of the female and male partners differ, we carried out parallel analyses on both samples separately. In this paper, we focus on the analyses based on women’s responses and report those based on the male sample in Supplementary Part C. Given that both samples produce the same basic patterns, we will discuss the men’s results only when they differ in important ways from the women’s.
Questions pertaining to decision-making were asked only of individuals older than 16 years of age who were living in a household with a partner who was also older than 16 years (n = 129,797). We eliminated observations in which the partner ID was missing (n = 2,062). In order to be able to compute the woman’s share of the contribution to the couple’s joint earnings, we also had to exclude cases in which neither of the partners had any employment income during the income reference period (n = 33,453) or who reported having no household income (n = 146). Couples with the level of education missing for either partner were also left out (n = 2,442). We then dropped observations in which any of the decision-making questions were not answered (n = 4,193). As our focus is on employment income, the study sample is mostly restricted to the population of working age, so we excluded respondents below the age of 20 years and over the age of 60 years. Finally, after preliminary analysis, we omitted two countries (Bulgaria and Latvia, n = 5,860) because of an extremely large percentage of missing values for the financial decision-making variable (31 per cent and 54 per cent, respectively). After these steps, the analytical sample consists of 72,638 couples in 27 countries, representing the year of 2009 during the financial crisis.
Outcome Variables
Our dependent variable on financial decision-making is based on three questions in the ad-hoc module (Eurostat, 2010).1 These questions concern which of the partners would be more likely to make decisions regarding (i) purchasing expensive consumer durables and furniture (durables); (ii) borrowing money, including mortgages and loans (borrowing); and (iii) using savings (savings). The module also contains a question about who makes the everyday shopping decisions (shopping) and a question about who in general has the final word when making important decisions (general). We analyse financial decision-making using the questions regarding consumer durables, borrowing money, and savings, aggregated in a single variable.
When asked about who is likely to make decisions, respondents were given the choices ‘more likely me’, ‘balanced’, and ‘more likely my partner’. They could also say that they had never had to make a particular decision. For instance, the couple did not have any (common) savings or had never considered borrowing money. In order to combine the three finances related questions into a single score, we first converted the answers to each question into numeric values, with 0 representing a balanced outcome, and –1 and +1 indicating, respectively, that the female or male partner is reported to be the decision maker. If a particular decision had never been made by the couple, we coded the answer as missing. The numeric values of the three questions were then added up, creating a total score of –3 to +3. In order to avoid basing the score on a single question, we decided that at least two of the three questions could not be missing. Thus, if a respondent had answered only one or none of the three questions, the score variable is coded as missing (5.1 per cent of all observations, n = 3,702).
A high proportion of women report that the decision-making is balanced, resulting in a total score of zero for our outcome variable for over three-quarters of our sample (Table 2). This reflects at least to some extent the survey guidelines, which state explicitly that if certain decisions are sometimes made by the other partner, the respondent should choose ‘balanced’. This implies that on average, there is a bias towards ‘balanced’ because respondents are encouraged to choose this category if they have doubts. As a result, classification according to the two remaining categories (in which either the man or the woman is reported to be the sole decision maker) becomes more stringent.
Values below zero indicate that the female partner is the main decision maker; these values constitute 9.2 per cent of the analytical sample. Men as the principal decision makers (scores above zero) comprise 10.2 per cent of the sample. In the analysis, we distinguish between cases in which decisions are more often made mainly by the female partner (the total score is negative), by the male partner (positive score), or whether the decision-making is balanced (score is zero). Table 2 shows how single questions and the dependent variable, ‘major outlays’, are distributed among the different levels of relative education.
The second dependent variable, which we model separately, is solely based on the question about who in general has the final say in important decisions (general). This variable results in 80 per cent of the respondents reporting balanced decision-making, 11 per cent report that the female partner is more likely to have the final say and 9 per cent report that the male partner is in general making important decisions. For 0.2 per cent (n = 141) of our sample, the answer is missing for this question.
Explanatory and Control Variables
The two main explanatory variables in this study are the woman’s earnings and education, relative to her partner. Relative earnings, ranging from 0 to 100 per cent, denotes the percentage that the woman contributes to the couple’s joint gross earnings during the income reference period, which is the year preceding the survey. We consider only earnings from employment and self-employment. Various transfers and benefits that are meant to compensate for the lack of employment income are not included. Although transfers and benefits do differ across countries, the effect on the measure of relative earnings is negligible (see Klesment and Van Bavel, 2017). Gross earnings were selected over net earnings because the latter are only available for a limited number of countries in the sample. Table 1 presents the mean and standard deviation of women’s relative earnings in each country.
. | . | Women’s relative education per cent . | Relative earnings . | Major outlays per cent . | General decision-making per cent . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | N . | Lower . | Equal . | Higher . | Mean . | SD . | Woman . | Both . | Man . | NA . | Woman . | Both . | Man . | NA . |
Austria | 2,202 | 28 | 60 | 12 | 34 | 29 | 7 | 78 | 13 | 2 | *16 | 69 | 15 | 0 |
Belgium | 993 | 21 | 55 | *24 | 39 | 30 | *11 | 79 | 9 | 1 | *12 | 75 | 12 | 0 |
Cyprus | 1,891 | 14 | 62 | *24 | 32 | 26 | 6 | 77 | 12 | 5 | 3 | 86 | 11 | 0 |
Czech Rep. | 3,693 | 15 | 75 | 10 | 34 | 27 | 6 | 82 | 7 | 5 | *3 | 94 | 3 | 0 |
Denmark | 1,306 | 16 | 59 | *25 | 43 | 23 | 6 | 82 | 8 | 4 | *4 | 92 | 4 | 0 |
Estonia | 2,218 | 12 | 60 | *28 | 40 | 30 | *6 | 81 | 4 | 9 | *12 | 81 | 7 | 0 |
Finland | 2,374 | 14 | 57 | *29 | 43 | 27 | 9 | 70 | 16 | 5 | *21 | 60 | 18 | 1 |
France | 3,938 | 19 | 55 | *26 | 40 | 28 | *15 | 67 | 12 | 6 | *24 | 59 | 16 | 1 |
Germany | 4,846 | 28 | 59 | 13 | 33 | 30 | 5 | 79 | 10 | 6 | *8 | 89 | 3 | 0 |
Greece | 2,786 | 17 | 64 | *19 | 32 | 32 | 10 | 63 | 13 | 14 | 4 | 81 | 15 | 0 |
Hungary | 4,062 | 16 | 67 | *17 | 40 | 31 | *4 | 87 | 3 | 6 | *4 | 94 | 2 | 0 |
Ireland | 1,026 | 13 | 59 | *28 | *52 | 31 | *26 | 57 | 15 | 2 | *17 | 75 | 8 | 0 |
Iceland | 705 | 21 | 50 | *29 | 37 | 22 | 16 | 64 | 20 | 0 | *32 | 51 | 17 | 0 |
Italy | 7,487 | 17 | 59 | *24 | 30 | 30 | 7 | 72 | 17 | 4 | 10 | 78 | 12 | 0 |
Lithuania | 2,118 | 10 | 64 | *26 | 45 | 33 | *10 | 76 | 6 | 8 | *4 | 93 | 2 | 1 |
Luxembourg | 2,438 | 20 | 63 | 17 | 33 | 29 | *16 | 65 | 16 | 3 | *16 | 70 | 14 | 0 |
Malta | 1,501 | 21 | 65 | 14 | 21 | 27 | *7 | 86 | 5 | 2 | *2 | 96 | 2 | 0 |
Netherlands | 2,324 | 23 | 54 | *23 | 32 | 25 | 7 | 78 | 13 | 2 | *18 | 73 | 8 | 0 |
Norway | 1,088 | 19 | 56 | *25 | 38 | 23 | 7 | 82 | 9 | 2 | *12 | 83 | 5 | 0 |
Poland | 5,495 | 9 | 75 | *16 | 33 | 32 | *6 | 77 | 6 | 11 | *21 | 71 | 8 | 0 |
Portugal | 1,923 | 8 | 73 | *19 | 38 | 29 | *8 | 79 | 8 | 5 | 8 | 76 | 16 | 0 |
Romania | 2,760 | 17 | 75 | 8 | 32 | 30 | 7 | 79 | 8 | 6 | 8 | 76 | 16 | 0 |
Slovakia | 2,814 | 12 | 76 | *12 | 39 | 27 | *17 | 69 | 12 | 2 | *7 | 86 | 6 | 0 |
Slovenia | 2,041 | 17 | 59 | *24 | 44 | 29 | *6 | 79 | 6 | 9 | *15 | 73 | 12 | 0 |
Spain | 5,487 | 18 | 59 | *23 | 36 | 31 | *11 | 83 | 5 | 1 | *3 | 96 | 1 | 0 |
Sweden | 1,082 | 14 | 56 | *30 | 40 | 24 | *15 | 70 | 12 | 3 | *31 | 59 | 8 | 2 |
UK | 2,040 | 15 | 62 | *23 | 37 | 29 | 18 | 54 | 24 | 4 | *13 | 77 | 10 | 0 |
Total | 72,638 | 19 | 62 | *19 | 35 | 30 | 9 | 76 | 10 | 5 | *11 | 80 | 9 | 0 |
. | . | Women’s relative education per cent . | Relative earnings . | Major outlays per cent . | General decision-making per cent . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | N . | Lower . | Equal . | Higher . | Mean . | SD . | Woman . | Both . | Man . | NA . | Woman . | Both . | Man . | NA . |
Austria | 2,202 | 28 | 60 | 12 | 34 | 29 | 7 | 78 | 13 | 2 | *16 | 69 | 15 | 0 |
Belgium | 993 | 21 | 55 | *24 | 39 | 30 | *11 | 79 | 9 | 1 | *12 | 75 | 12 | 0 |
Cyprus | 1,891 | 14 | 62 | *24 | 32 | 26 | 6 | 77 | 12 | 5 | 3 | 86 | 11 | 0 |
Czech Rep. | 3,693 | 15 | 75 | 10 | 34 | 27 | 6 | 82 | 7 | 5 | *3 | 94 | 3 | 0 |
Denmark | 1,306 | 16 | 59 | *25 | 43 | 23 | 6 | 82 | 8 | 4 | *4 | 92 | 4 | 0 |
Estonia | 2,218 | 12 | 60 | *28 | 40 | 30 | *6 | 81 | 4 | 9 | *12 | 81 | 7 | 0 |
Finland | 2,374 | 14 | 57 | *29 | 43 | 27 | 9 | 70 | 16 | 5 | *21 | 60 | 18 | 1 |
France | 3,938 | 19 | 55 | *26 | 40 | 28 | *15 | 67 | 12 | 6 | *24 | 59 | 16 | 1 |
Germany | 4,846 | 28 | 59 | 13 | 33 | 30 | 5 | 79 | 10 | 6 | *8 | 89 | 3 | 0 |
Greece | 2,786 | 17 | 64 | *19 | 32 | 32 | 10 | 63 | 13 | 14 | 4 | 81 | 15 | 0 |
Hungary | 4,062 | 16 | 67 | *17 | 40 | 31 | *4 | 87 | 3 | 6 | *4 | 94 | 2 | 0 |
Ireland | 1,026 | 13 | 59 | *28 | *52 | 31 | *26 | 57 | 15 | 2 | *17 | 75 | 8 | 0 |
Iceland | 705 | 21 | 50 | *29 | 37 | 22 | 16 | 64 | 20 | 0 | *32 | 51 | 17 | 0 |
Italy | 7,487 | 17 | 59 | *24 | 30 | 30 | 7 | 72 | 17 | 4 | 10 | 78 | 12 | 0 |
Lithuania | 2,118 | 10 | 64 | *26 | 45 | 33 | *10 | 76 | 6 | 8 | *4 | 93 | 2 | 1 |
Luxembourg | 2,438 | 20 | 63 | 17 | 33 | 29 | *16 | 65 | 16 | 3 | *16 | 70 | 14 | 0 |
Malta | 1,501 | 21 | 65 | 14 | 21 | 27 | *7 | 86 | 5 | 2 | *2 | 96 | 2 | 0 |
Netherlands | 2,324 | 23 | 54 | *23 | 32 | 25 | 7 | 78 | 13 | 2 | *18 | 73 | 8 | 0 |
Norway | 1,088 | 19 | 56 | *25 | 38 | 23 | 7 | 82 | 9 | 2 | *12 | 83 | 5 | 0 |
Poland | 5,495 | 9 | 75 | *16 | 33 | 32 | *6 | 77 | 6 | 11 | *21 | 71 | 8 | 0 |
Portugal | 1,923 | 8 | 73 | *19 | 38 | 29 | *8 | 79 | 8 | 5 | 8 | 76 | 16 | 0 |
Romania | 2,760 | 17 | 75 | 8 | 32 | 30 | 7 | 79 | 8 | 6 | 8 | 76 | 16 | 0 |
Slovakia | 2,814 | 12 | 76 | *12 | 39 | 27 | *17 | 69 | 12 | 2 | *7 | 86 | 6 | 0 |
Slovenia | 2,041 | 17 | 59 | *24 | 44 | 29 | *6 | 79 | 6 | 9 | *15 | 73 | 12 | 0 |
Spain | 5,487 | 18 | 59 | *23 | 36 | 31 | *11 | 83 | 5 | 1 | *3 | 96 | 1 | 0 |
Sweden | 1,082 | 14 | 56 | *30 | 40 | 24 | *15 | 70 | 12 | 3 | *31 | 59 | 8 | 2 |
UK | 2,040 | 15 | 62 | *23 | 37 | 29 | 18 | 54 | 24 | 4 | *13 | 77 | 10 | 0 |
Total | 72,638 | 19 | 62 | *19 | 35 | 30 | 9 | 76 | 10 | 5 | *11 | 80 | 9 | 0 |
Source: EU-SILC 2010, authors’ own calculations, sampling weights applied to the calculation of distributions.
Note: Relative education is based on the three-level educational attainment variable. Relative earnings is the average percentage that women contribute to the couple’s earnings. The asterisk denotes that the women’s percentage in the variable is equal to or higher than the men’s respective value.
. | . | Women’s relative education per cent . | Relative earnings . | Major outlays per cent . | General decision-making per cent . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | N . | Lower . | Equal . | Higher . | Mean . | SD . | Woman . | Both . | Man . | NA . | Woman . | Both . | Man . | NA . |
Austria | 2,202 | 28 | 60 | 12 | 34 | 29 | 7 | 78 | 13 | 2 | *16 | 69 | 15 | 0 |
Belgium | 993 | 21 | 55 | *24 | 39 | 30 | *11 | 79 | 9 | 1 | *12 | 75 | 12 | 0 |
Cyprus | 1,891 | 14 | 62 | *24 | 32 | 26 | 6 | 77 | 12 | 5 | 3 | 86 | 11 | 0 |
Czech Rep. | 3,693 | 15 | 75 | 10 | 34 | 27 | 6 | 82 | 7 | 5 | *3 | 94 | 3 | 0 |
Denmark | 1,306 | 16 | 59 | *25 | 43 | 23 | 6 | 82 | 8 | 4 | *4 | 92 | 4 | 0 |
Estonia | 2,218 | 12 | 60 | *28 | 40 | 30 | *6 | 81 | 4 | 9 | *12 | 81 | 7 | 0 |
Finland | 2,374 | 14 | 57 | *29 | 43 | 27 | 9 | 70 | 16 | 5 | *21 | 60 | 18 | 1 |
France | 3,938 | 19 | 55 | *26 | 40 | 28 | *15 | 67 | 12 | 6 | *24 | 59 | 16 | 1 |
Germany | 4,846 | 28 | 59 | 13 | 33 | 30 | 5 | 79 | 10 | 6 | *8 | 89 | 3 | 0 |
Greece | 2,786 | 17 | 64 | *19 | 32 | 32 | 10 | 63 | 13 | 14 | 4 | 81 | 15 | 0 |
Hungary | 4,062 | 16 | 67 | *17 | 40 | 31 | *4 | 87 | 3 | 6 | *4 | 94 | 2 | 0 |
Ireland | 1,026 | 13 | 59 | *28 | *52 | 31 | *26 | 57 | 15 | 2 | *17 | 75 | 8 | 0 |
Iceland | 705 | 21 | 50 | *29 | 37 | 22 | 16 | 64 | 20 | 0 | *32 | 51 | 17 | 0 |
Italy | 7,487 | 17 | 59 | *24 | 30 | 30 | 7 | 72 | 17 | 4 | 10 | 78 | 12 | 0 |
Lithuania | 2,118 | 10 | 64 | *26 | 45 | 33 | *10 | 76 | 6 | 8 | *4 | 93 | 2 | 1 |
Luxembourg | 2,438 | 20 | 63 | 17 | 33 | 29 | *16 | 65 | 16 | 3 | *16 | 70 | 14 | 0 |
Malta | 1,501 | 21 | 65 | 14 | 21 | 27 | *7 | 86 | 5 | 2 | *2 | 96 | 2 | 0 |
Netherlands | 2,324 | 23 | 54 | *23 | 32 | 25 | 7 | 78 | 13 | 2 | *18 | 73 | 8 | 0 |
Norway | 1,088 | 19 | 56 | *25 | 38 | 23 | 7 | 82 | 9 | 2 | *12 | 83 | 5 | 0 |
Poland | 5,495 | 9 | 75 | *16 | 33 | 32 | *6 | 77 | 6 | 11 | *21 | 71 | 8 | 0 |
Portugal | 1,923 | 8 | 73 | *19 | 38 | 29 | *8 | 79 | 8 | 5 | 8 | 76 | 16 | 0 |
Romania | 2,760 | 17 | 75 | 8 | 32 | 30 | 7 | 79 | 8 | 6 | 8 | 76 | 16 | 0 |
Slovakia | 2,814 | 12 | 76 | *12 | 39 | 27 | *17 | 69 | 12 | 2 | *7 | 86 | 6 | 0 |
Slovenia | 2,041 | 17 | 59 | *24 | 44 | 29 | *6 | 79 | 6 | 9 | *15 | 73 | 12 | 0 |
Spain | 5,487 | 18 | 59 | *23 | 36 | 31 | *11 | 83 | 5 | 1 | *3 | 96 | 1 | 0 |
Sweden | 1,082 | 14 | 56 | *30 | 40 | 24 | *15 | 70 | 12 | 3 | *31 | 59 | 8 | 2 |
UK | 2,040 | 15 | 62 | *23 | 37 | 29 | 18 | 54 | 24 | 4 | *13 | 77 | 10 | 0 |
Total | 72,638 | 19 | 62 | *19 | 35 | 30 | 9 | 76 | 10 | 5 | *11 | 80 | 9 | 0 |
. | . | Women’s relative education per cent . | Relative earnings . | Major outlays per cent . | General decision-making per cent . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | N . | Lower . | Equal . | Higher . | Mean . | SD . | Woman . | Both . | Man . | NA . | Woman . | Both . | Man . | NA . |
Austria | 2,202 | 28 | 60 | 12 | 34 | 29 | 7 | 78 | 13 | 2 | *16 | 69 | 15 | 0 |
Belgium | 993 | 21 | 55 | *24 | 39 | 30 | *11 | 79 | 9 | 1 | *12 | 75 | 12 | 0 |
Cyprus | 1,891 | 14 | 62 | *24 | 32 | 26 | 6 | 77 | 12 | 5 | 3 | 86 | 11 | 0 |
Czech Rep. | 3,693 | 15 | 75 | 10 | 34 | 27 | 6 | 82 | 7 | 5 | *3 | 94 | 3 | 0 |
Denmark | 1,306 | 16 | 59 | *25 | 43 | 23 | 6 | 82 | 8 | 4 | *4 | 92 | 4 | 0 |
Estonia | 2,218 | 12 | 60 | *28 | 40 | 30 | *6 | 81 | 4 | 9 | *12 | 81 | 7 | 0 |
Finland | 2,374 | 14 | 57 | *29 | 43 | 27 | 9 | 70 | 16 | 5 | *21 | 60 | 18 | 1 |
France | 3,938 | 19 | 55 | *26 | 40 | 28 | *15 | 67 | 12 | 6 | *24 | 59 | 16 | 1 |
Germany | 4,846 | 28 | 59 | 13 | 33 | 30 | 5 | 79 | 10 | 6 | *8 | 89 | 3 | 0 |
Greece | 2,786 | 17 | 64 | *19 | 32 | 32 | 10 | 63 | 13 | 14 | 4 | 81 | 15 | 0 |
Hungary | 4,062 | 16 | 67 | *17 | 40 | 31 | *4 | 87 | 3 | 6 | *4 | 94 | 2 | 0 |
Ireland | 1,026 | 13 | 59 | *28 | *52 | 31 | *26 | 57 | 15 | 2 | *17 | 75 | 8 | 0 |
Iceland | 705 | 21 | 50 | *29 | 37 | 22 | 16 | 64 | 20 | 0 | *32 | 51 | 17 | 0 |
Italy | 7,487 | 17 | 59 | *24 | 30 | 30 | 7 | 72 | 17 | 4 | 10 | 78 | 12 | 0 |
Lithuania | 2,118 | 10 | 64 | *26 | 45 | 33 | *10 | 76 | 6 | 8 | *4 | 93 | 2 | 1 |
Luxembourg | 2,438 | 20 | 63 | 17 | 33 | 29 | *16 | 65 | 16 | 3 | *16 | 70 | 14 | 0 |
Malta | 1,501 | 21 | 65 | 14 | 21 | 27 | *7 | 86 | 5 | 2 | *2 | 96 | 2 | 0 |
Netherlands | 2,324 | 23 | 54 | *23 | 32 | 25 | 7 | 78 | 13 | 2 | *18 | 73 | 8 | 0 |
Norway | 1,088 | 19 | 56 | *25 | 38 | 23 | 7 | 82 | 9 | 2 | *12 | 83 | 5 | 0 |
Poland | 5,495 | 9 | 75 | *16 | 33 | 32 | *6 | 77 | 6 | 11 | *21 | 71 | 8 | 0 |
Portugal | 1,923 | 8 | 73 | *19 | 38 | 29 | *8 | 79 | 8 | 5 | 8 | 76 | 16 | 0 |
Romania | 2,760 | 17 | 75 | 8 | 32 | 30 | 7 | 79 | 8 | 6 | 8 | 76 | 16 | 0 |
Slovakia | 2,814 | 12 | 76 | *12 | 39 | 27 | *17 | 69 | 12 | 2 | *7 | 86 | 6 | 0 |
Slovenia | 2,041 | 17 | 59 | *24 | 44 | 29 | *6 | 79 | 6 | 9 | *15 | 73 | 12 | 0 |
Spain | 5,487 | 18 | 59 | *23 | 36 | 31 | *11 | 83 | 5 | 1 | *3 | 96 | 1 | 0 |
Sweden | 1,082 | 14 | 56 | *30 | 40 | 24 | *15 | 70 | 12 | 3 | *31 | 59 | 8 | 2 |
UK | 2,040 | 15 | 62 | *23 | 37 | 29 | 18 | 54 | 24 | 4 | *13 | 77 | 10 | 0 |
Total | 72,638 | 19 | 62 | *19 | 35 | 30 | 9 | 76 | 10 | 5 | *11 | 80 | 9 | 0 |
Source: EU-SILC 2010, authors’ own calculations, sampling weights applied to the calculation of distributions.
Note: Relative education is based on the three-level educational attainment variable. Relative earnings is the average percentage that women contribute to the couple’s earnings. The asterisk denotes that the women’s percentage in the variable is equal to or higher than the men’s respective value.
Relative education . | Woman . | Man . | Balanced . | Missing . | Total . |
---|---|---|---|---|---|
Durables | |||||
Homogamy | 7.9 | 6.0 | 84.6 | 1.5 | 100 |
Hypogamy | 10.5 | 5.6 | 82.6 | 1.3 | 100 |
Hypergamy | 6.3 | 7.0 | 85.2 | 1.5 | 100 |
Borrowing | |||||
Homogamy | 3.9 | 8.6 | 73.1 | 14.3 | 100 |
Hypogamy | 6.0 | 6.6 | 75.0 | 12.4 | 100 |
Hypergamy | 2.4 | 11.2 | 72.1 | 14.4 | 100 |
Savings | |||||
Homogamy | 4.6 | 5.0 | 74.5 | 15.9 | 100 |
Hypogamy | 6.2 | 3.7 | 72.9 | 17.3 | 100 |
Hypergamy | 3.4 | 6.6 | 75.4 | 14.5 | 100 |
Major outlays (durables + borrowing + savings) | |||||
Homogamy | 9.0 | 11.4 | 74.1 | 5.5 | 100 |
Hypogamy | 12.7 | 9.2 | 73.1 | 5.0 | 100 |
Hypergamy | 6.9 | 14.2 | 73.6 | 5.3 | 100 |
General decision-making | |||||
Homogamy | 11.3 | 8.8 | 79.7 | 0.2 | 100 |
Hypogamy | 15.7 | 7.1 | 76.9 | 0.3 | 100 |
Hypergamy | 9.6 | 10.5 | 79.8 | 0.1 | 100 |
Everyday shopping | |||||
Homogamy | 54.3 | 5.1 | 40.5 | 0.01 | 100 |
Hypogamy | 54.7 | 4.8 | 40.5 | 0.00 | 100 |
Hypergamy | 51.3 | 4.6 | 44.1 | 0.03 | 100 |
Relative education . | Woman . | Man . | Balanced . | Missing . | Total . |
---|---|---|---|---|---|
Durables | |||||
Homogamy | 7.9 | 6.0 | 84.6 | 1.5 | 100 |
Hypogamy | 10.5 | 5.6 | 82.6 | 1.3 | 100 |
Hypergamy | 6.3 | 7.0 | 85.2 | 1.5 | 100 |
Borrowing | |||||
Homogamy | 3.9 | 8.6 | 73.1 | 14.3 | 100 |
Hypogamy | 6.0 | 6.6 | 75.0 | 12.4 | 100 |
Hypergamy | 2.4 | 11.2 | 72.1 | 14.4 | 100 |
Savings | |||||
Homogamy | 4.6 | 5.0 | 74.5 | 15.9 | 100 |
Hypogamy | 6.2 | 3.7 | 72.9 | 17.3 | 100 |
Hypergamy | 3.4 | 6.6 | 75.4 | 14.5 | 100 |
Major outlays (durables + borrowing + savings) | |||||
Homogamy | 9.0 | 11.4 | 74.1 | 5.5 | 100 |
Hypogamy | 12.7 | 9.2 | 73.1 | 5.0 | 100 |
Hypergamy | 6.9 | 14.2 | 73.6 | 5.3 | 100 |
General decision-making | |||||
Homogamy | 11.3 | 8.8 | 79.7 | 0.2 | 100 |
Hypogamy | 15.7 | 7.1 | 76.9 | 0.3 | 100 |
Hypergamy | 9.6 | 10.5 | 79.8 | 0.1 | 100 |
Everyday shopping | |||||
Homogamy | 54.3 | 5.1 | 40.5 | 0.01 | 100 |
Hypogamy | 54.7 | 4.8 | 40.5 | 0.00 | 100 |
Hypergamy | 51.3 | 4.6 | 44.1 | 0.03 | 100 |
Source: EU-SILC 2010, sampling weights applied, authors’ own calculations.
Relative education . | Woman . | Man . | Balanced . | Missing . | Total . |
---|---|---|---|---|---|
Durables | |||||
Homogamy | 7.9 | 6.0 | 84.6 | 1.5 | 100 |
Hypogamy | 10.5 | 5.6 | 82.6 | 1.3 | 100 |
Hypergamy | 6.3 | 7.0 | 85.2 | 1.5 | 100 |
Borrowing | |||||
Homogamy | 3.9 | 8.6 | 73.1 | 14.3 | 100 |
Hypogamy | 6.0 | 6.6 | 75.0 | 12.4 | 100 |
Hypergamy | 2.4 | 11.2 | 72.1 | 14.4 | 100 |
Savings | |||||
Homogamy | 4.6 | 5.0 | 74.5 | 15.9 | 100 |
Hypogamy | 6.2 | 3.7 | 72.9 | 17.3 | 100 |
Hypergamy | 3.4 | 6.6 | 75.4 | 14.5 | 100 |
Major outlays (durables + borrowing + savings) | |||||
Homogamy | 9.0 | 11.4 | 74.1 | 5.5 | 100 |
Hypogamy | 12.7 | 9.2 | 73.1 | 5.0 | 100 |
Hypergamy | 6.9 | 14.2 | 73.6 | 5.3 | 100 |
General decision-making | |||||
Homogamy | 11.3 | 8.8 | 79.7 | 0.2 | 100 |
Hypogamy | 15.7 | 7.1 | 76.9 | 0.3 | 100 |
Hypergamy | 9.6 | 10.5 | 79.8 | 0.1 | 100 |
Everyday shopping | |||||
Homogamy | 54.3 | 5.1 | 40.5 | 0.01 | 100 |
Hypogamy | 54.7 | 4.8 | 40.5 | 0.00 | 100 |
Hypergamy | 51.3 | 4.6 | 44.1 | 0.03 | 100 |
Relative education . | Woman . | Man . | Balanced . | Missing . | Total . |
---|---|---|---|---|---|
Durables | |||||
Homogamy | 7.9 | 6.0 | 84.6 | 1.5 | 100 |
Hypogamy | 10.5 | 5.6 | 82.6 | 1.3 | 100 |
Hypergamy | 6.3 | 7.0 | 85.2 | 1.5 | 100 |
Borrowing | |||||
Homogamy | 3.9 | 8.6 | 73.1 | 14.3 | 100 |
Hypogamy | 6.0 | 6.6 | 75.0 | 12.4 | 100 |
Hypergamy | 2.4 | 11.2 | 72.1 | 14.4 | 100 |
Savings | |||||
Homogamy | 4.6 | 5.0 | 74.5 | 15.9 | 100 |
Hypogamy | 6.2 | 3.7 | 72.9 | 17.3 | 100 |
Hypergamy | 3.4 | 6.6 | 75.4 | 14.5 | 100 |
Major outlays (durables + borrowing + savings) | |||||
Homogamy | 9.0 | 11.4 | 74.1 | 5.5 | 100 |
Hypogamy | 12.7 | 9.2 | 73.1 | 5.0 | 100 |
Hypergamy | 6.9 | 14.2 | 73.6 | 5.3 | 100 |
General decision-making | |||||
Homogamy | 11.3 | 8.8 | 79.7 | 0.2 | 100 |
Hypogamy | 15.7 | 7.1 | 76.9 | 0.3 | 100 |
Hypergamy | 9.6 | 10.5 | 79.8 | 0.1 | 100 |
Everyday shopping | |||||
Homogamy | 54.3 | 5.1 | 40.5 | 0.01 | 100 |
Hypogamy | 54.7 | 4.8 | 40.5 | 0.00 | 100 |
Hypergamy | 51.3 | 4.6 | 44.1 | 0.03 | 100 |
Source: EU-SILC 2010, sampling weights applied, authors’ own calculations.
EU-SILC uses the ISCED-97 scale to index the highest level of education attained (UNESCO, 2003); however, the user database combines all levels of tertiary (college or university level) education in one category. In our analysis, we distinguish between three groups: lower (ISCED levels 0–2, up to the second stage of basic education, equivalent to the seventh to ninth grades in the United States), medium (ISCED 3–4, secondary or post-secondary but not tertiary education; up to the 12th grade, vocational, junior and community college education) and higher education (includes ISCED 5–6, university level, Bachelors, Masters and PhD degrees). To be sure that our results are not affected by the aggregation of the education variable, we also checked our models using the original education variable (not shown in the paper).
Our relative education variable is based on the level of education of both partners, as described above, and categorizes each couple’s educational status as homogamous (both partners equally educated), hypergamous (the man is more educated), or hypogamous (the woman is more educated). As shown in Table 1, homogamous couples constitute the majority in all countries. However, if the partners are not equally educated, it is more often the woman who is more educated than the man. In 21 out of 27 countries, the proportion of hypogamous couples equals or exceeds the proportion of hypergamous couples, which is in line with recent findings about the reversal of the gender gap in education (Esteve et al., 2016; De Hauw, Grow and Van Bavel, 2017).
In the regression analyses, we control for a number of individual- and couple-level characteristics. Marital status distinguishes between those who are currently married, never married, or previously married (divorced or widowed). Given that we only include women who are living with a male partner, the latter two status categories imply unmarried cohabitation. We also include a control for how long the woman has been living with her current partner. Next, since gender roles tend to become more traditional after having children (Sanchez and Thomson, 1997), we control for whether there are children living in the household.
Other controls include the woman’s age, partner’s unemployment, and a measure of the couple’s total income. It has been found that the proportion of female breadwinners is generally higher among low-income families and female breadwinners generally earn less than male breadwinners (Winslow-Bowe, 2006; Vitali and Arpino, 2016; Klesment and Van Bavel, 2017; Kowalewska and Vitali, 2021). Because there is a large variation in income across countries, the absolute level of family income is included as a country-specific 20-quantile of the couple’s joint earnings. This is entered in the model as a continuous variable and we also add its quadratic term. Unemployment of the male partner directly affects the couple’s relative earnings. We control for the number of months the man was unemployed during the income reference period.
Methods
We use multinomial logistic regression to model the probability that either the woman or the man is more likely reported as the main financial decision maker as opposed to balanced decision-making by both partners (the modal outcome). Observations with missing values for the dependent variable are omitted. In order to ensure that missing values for the major outlays variable are random with respect to our main explanatory variables, we test the correlation using a binary logistic regression model (included in Supplementary Part B). Neither relative earnings nor educational pairing predicts missing values in the dependent variable. We have tested other specifications of our major outlays variable as alternatives to the three-category version used below (see also Supplementary Part D). The results were consistent with the findings discussed below.
Our models include fixed effects to account for the cross-country variability in decision-making patterns. Some countries are represented by a relatively small sample (Table 1) and to minimize the chances that our results are biased by a small part of the study sample, we estimate our models with a bootstrap resampling technique using 200 replications. Although country background may have important implications for patterns of family life (e.g. Knudsen and Wærness, 2008), international comparisons are beyond the scope of this paper. However, in a preliminary analysis, we also fitted our models for each country separately to check that the results are not driven by a few influential countries.
We test Hypotheses 1–3 by estimating the association between the likelihood of either partner’s decision-making and the woman’s relative earnings and education. The woman’s relative earnings are entered in the regression model as a six-category factor (0–10 per cent, 11–25 per cent, 26–50 per cent, 51–75 per cent, 76–90 per cent, or 91–100 per cent of the joint income) to allow for non-linear relationships with the dependent variable without any pre-specified functional form.
Results
Table 2 presents decision-making according to relative education. The upper three blocks show row-wise percentages of responses to each question included in the construction of our first dependent variable. The fourth block gives the percentages for the summary dependent variable, major outlays. For all three constituent questions, respondents most often reported that decision-making was balanced in their household. As of decision-making by either the woman or the man alone, in hypogamous unions, the proportion of women’s decision-making exceeds that of men’s, except for the question concerning borrowing money. In homogamous unions, the proportion of couples in which the woman makes the decisions about consumer durables is also higher than that in which the man makes such decisions. Men are more likely to be reported as the principal decision maker in hypergamous unions. After aggregating all three questions into the three-category major outlays variable, women’s proportion of decision-making is higher in hypogamous unions, whereas men are more often reported to be the decision maker among homogamous and hypergamous couples.
The fifth and the sixth block in Table 2 show general and everyday shopping decision-making by relative education. General decision-making, which is our second dependent variable, is most often balanced in all relative education categories. When it is not balanced, the proportion of women’s decision-making is higher than that of men’s if she is at least as highly educated as the male partner. The proportion of men’s general decision-making exceeds that of women’s only in hypergamous couples. The distribution of general decision-making is similar to major outlays as both show a high proportion of balanced decisions. In contrast to that, everyday shopping decision-making is quite different from major outlays and general decisions: women are most often reported as primary decision makers in all relative education groups. Balanced decisions are the second biggest category for these decisions, while men are least likely to be reported as being in charge of daily shopping decisions.
We then show how the two dependent variables are distributed according to six different levels of relative earnings. Figure 1 shows that, both for major outlays (left panel) and general decision-making (right panel), balanced decision-making is more common among dual-earner couples in which the woman earns a considerable share of the joint income but not more than the man (26–50 per cent of joint earnings). In the case of major outlays, the proportion in the balanced decision-making group is lower among women who earn less than 25 per cent of the total income, as well as among those who earn more than their male partner. The same pattern is present also in general decision-making.
Decision-making mainly by the woman or by the man exhibits a clear gradient across relative earnings: for both of our dependent variables shown in Figure 1, a propensity for the woman to be the principal decision maker is higher when the woman earns more than her partner. This result is consistent with Hypothesis 1 concerning the association between relative earnings and decision-making. Men are more often reported to be the main financial decision maker when women earn less than a quarter of the joint earnings. When the woman earns at least a quarter of the joint earnings, the likelihood of the man being the sole decision maker drops considerably. A similar gradient holds for general decision-making, but here the proportions of female decision-making are slightly higher and the proportions of male decision-making are lower compared with major outlays.
In Figure 1, we include missing values for both of our dependent variables. In the case of major outlays, the proportion of missing values is lower in the groups in which the woman earns at least 10 per cent but not more than 75 per cent of the joint income. In the remaining groups, the proportion of missing values is relatively higher. We have, however, checked that the relative earnings variable is not a statistically significant predictor of missing values when other individual characteristics are included in the model (Supplementary Part B). General decision-making has very few missing values that are relatively evenly distributed across relative earnings groups.
Regression Results: Major Outlay Decision-Making
Table 3 shows the results of our multinomial regression model, using balanced decision-making as the reference category. The coefficients represent the association of the explanatory variables with decision-making primarily by the wife (‘Woman’ in Table 3) or by the husband (‘Man’ in Table 3). Relative earnings are included as an explanatory variable (using the same six categories as in Figure 1); couples in which the woman earns 26–50 per cent of the joint income constitute the reference category. Coefficients in the left column of the table can be interpreted as relative log-odds of female decision-making. For example, if she earns 0–10 per cent of the joint income, the log-odds of female decision-making over balanced decision-making increase by 0.102 compared with the situation in which she earns 26–50 per cent. For a more complete picture of how the three outcomes are shaped by relative earnings, we compute predictive margins (probabilities of each outcome).
. | Main financial decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | 0.102** | (0.035) | 0.583*** | (0.036) |
11–25 | 0.031 | (0.052) | 0.398*** | (0.047) |
51–75 | 0.286*** | (0.038) | –0.267*** | (0.046) |
76–90 | 0.513*** | (0.079) | –0.209 | (0.120) |
91–100 | 0.469*** | (0.054) | 0.074 | (0.064) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.317*** | (0.036) | –0.106** | (0.037) |
Hypergamy | –0.292*** | (0.043) | 0.130*** | (0.036) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.188*** | (0.039) | 0.117** | (0.040) |
High | –0.252*** | (0.038) | 0.026 | (0.033) |
Woman’s age | 0.010*** | (0.003) | –0.003 | (0.002) |
Woman’s age squared | –0.000 | (0.000) | 0.001*** | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.295*** | (0.049) | 0.179*** | (0.046) |
Separated, widowed | 0.590*** | (0.074) | 0.090 | (0.084) |
Number of children in household | 0.017 | (0.014) | 0.047** | (0.014) |
Joint earnings 20—quantile | 0.009 | (0.012) | –0.035** | (0.011) |
Joint earnings 20—quantile squared | –0.000 | (0.000) | 0.003*** | (0.000) |
Length of cohabitation in years | –0.002 | (0.003) | –0.010*** | (0.002) |
Man’s unemployment in months | 0.013* | (0.006) | –0.026*** | (0.008) |
Constant | –2.693*** | (0.126) | –1.988*** | (0.097) |
N | 68,827 |
. | Main financial decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | 0.102** | (0.035) | 0.583*** | (0.036) |
11–25 | 0.031 | (0.052) | 0.398*** | (0.047) |
51–75 | 0.286*** | (0.038) | –0.267*** | (0.046) |
76–90 | 0.513*** | (0.079) | –0.209 | (0.120) |
91–100 | 0.469*** | (0.054) | 0.074 | (0.064) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.317*** | (0.036) | –0.106** | (0.037) |
Hypergamy | –0.292*** | (0.043) | 0.130*** | (0.036) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.188*** | (0.039) | 0.117** | (0.040) |
High | –0.252*** | (0.038) | 0.026 | (0.033) |
Woman’s age | 0.010*** | (0.003) | –0.003 | (0.002) |
Woman’s age squared | –0.000 | (0.000) | 0.001*** | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.295*** | (0.049) | 0.179*** | (0.046) |
Separated, widowed | 0.590*** | (0.074) | 0.090 | (0.084) |
Number of children in household | 0.017 | (0.014) | 0.047** | (0.014) |
Joint earnings 20—quantile | 0.009 | (0.012) | –0.035** | (0.011) |
Joint earnings 20—quantile squared | –0.000 | (0.000) | 0.003*** | (0.000) |
Length of cohabitation in years | –0.002 | (0.003) | –0.010*** | (0.002) |
Man’s unemployment in months | 0.013* | (0.006) | –0.026*** | (0.008) |
Constant | –2.693*** | (0.126) | –1.988*** | (0.097) |
N | 68,827 |
Source: EU-SILC 2010, authors’ own estimates.
Note: Country dummy variables omitted from the table. Bootstrapped model estimation using 200 replications.
P < 0.05,
P < 0.01,
P < 0.001, two-tailed test.
. | Main financial decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | 0.102** | (0.035) | 0.583*** | (0.036) |
11–25 | 0.031 | (0.052) | 0.398*** | (0.047) |
51–75 | 0.286*** | (0.038) | –0.267*** | (0.046) |
76–90 | 0.513*** | (0.079) | –0.209 | (0.120) |
91–100 | 0.469*** | (0.054) | 0.074 | (0.064) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.317*** | (0.036) | –0.106** | (0.037) |
Hypergamy | –0.292*** | (0.043) | 0.130*** | (0.036) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.188*** | (0.039) | 0.117** | (0.040) |
High | –0.252*** | (0.038) | 0.026 | (0.033) |
Woman’s age | 0.010*** | (0.003) | –0.003 | (0.002) |
Woman’s age squared | –0.000 | (0.000) | 0.001*** | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.295*** | (0.049) | 0.179*** | (0.046) |
Separated, widowed | 0.590*** | (0.074) | 0.090 | (0.084) |
Number of children in household | 0.017 | (0.014) | 0.047** | (0.014) |
Joint earnings 20—quantile | 0.009 | (0.012) | –0.035** | (0.011) |
Joint earnings 20—quantile squared | –0.000 | (0.000) | 0.003*** | (0.000) |
Length of cohabitation in years | –0.002 | (0.003) | –0.010*** | (0.002) |
Man’s unemployment in months | 0.013* | (0.006) | –0.026*** | (0.008) |
Constant | –2.693*** | (0.126) | –1.988*** | (0.097) |
N | 68,827 |
. | Main financial decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | 0.102** | (0.035) | 0.583*** | (0.036) |
11–25 | 0.031 | (0.052) | 0.398*** | (0.047) |
51–75 | 0.286*** | (0.038) | –0.267*** | (0.046) |
76–90 | 0.513*** | (0.079) | –0.209 | (0.120) |
91–100 | 0.469*** | (0.054) | 0.074 | (0.064) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.317*** | (0.036) | –0.106** | (0.037) |
Hypergamy | –0.292*** | (0.043) | 0.130*** | (0.036) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.188*** | (0.039) | 0.117** | (0.040) |
High | –0.252*** | (0.038) | 0.026 | (0.033) |
Woman’s age | 0.010*** | (0.003) | –0.003 | (0.002) |
Woman’s age squared | –0.000 | (0.000) | 0.001*** | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.295*** | (0.049) | 0.179*** | (0.046) |
Separated, widowed | 0.590*** | (0.074) | 0.090 | (0.084) |
Number of children in household | 0.017 | (0.014) | 0.047** | (0.014) |
Joint earnings 20—quantile | 0.009 | (0.012) | –0.035** | (0.011) |
Joint earnings 20—quantile squared | –0.000 | (0.000) | 0.003*** | (0.000) |
Length of cohabitation in years | –0.002 | (0.003) | –0.010*** | (0.002) |
Man’s unemployment in months | 0.013* | (0.006) | –0.026*** | (0.008) |
Constant | –2.693*** | (0.126) | –1.988*** | (0.097) |
N | 68,827 |
Source: EU-SILC 2010, authors’ own estimates.
Note: Country dummy variables omitted from the table. Bootstrapped model estimation using 200 replications.
P < 0.05,
P < 0.01,
P < 0.001, two-tailed test.
Predictive margins, displayed in Figure 2, show that the probability of the woman being the principal financial decision maker is less than 10 per cent if she earns less than half of the joint income, but it exceeds 10 per cent if she earns more than her male partner. When she earns more than half of the joint income, the probability increases by about 3–4 percentage points relative to the reference category 26–50 per cent. The probability of male decision-making is negatively related to the woman’s relative earnings, at least if she earns up to 75 per cent of income. Beyond that, the association reverses and becomes positively associated with relative earnings. Consequently, our Hypothesis 1 is supported: higher relative earnings are associated with an increase in the probability of being the main decision-maker. This holds both for women and men, with the exception that for men the association reverses if women earn over 75 per cent of the joint income.
Hypothesis 2 posits that gender display reduces the probability of female decision-making when she earns all or almost all of the joint income, and increases the probability of male decision-making. In Table 3, the coefficients of the relative earnings variable for female decision-making are highest when she earns 76–90 per cent (0.513***). The coefficient is smaller when she earns 91–100 per cent of the joint income (0.469***). As these coefficients are relative log-odds, this does not necessarily imply that the woman is less likely to be the main decision maker in the latter case. In order to determine that, we need to examine the predictive margins, plotted in Figure 2. These indicate a small decrease in the probability of female decision-making among the 76–90 and 91–100 categories, but the confidence intervals are too large for this decrease to be statistically significant.2 Therefore, these results do not support Part (a) of Hypothesis 2: There is no strong evidence that gender display plays a role in determining the likelihood that the woman would be the family’s principal financial decision maker.
Furthermore, according to the gender display argument and contrary to expectations based solely on relative resources, Part (b) of Hypothesis 2 predicts that the male partner would more often be the principal decision maker when the woman earns (almost) all of the income. Our data do support this Part (b). The overall probability of male decision-making is predicted to be lowest for the 51–75 per cent category (coefficient –0.267*** in Table 3). The probability is higher than that in the category 91–100 per cent, for which the coefficient is not significantly different from the reference category 26–50 per cent (0.074 in Table 3). The probability of male decision-making is higher when the woman earns over 90 per cent of the joint income, compared with the situation when she earns 51–75 per cent of the joint income. This is also illustrated in Figure 2. It is important to note that these results are based on the sample of female respondents. Similar results are obtained if we use the male sample.3
Thus far, we have established that relative earnings positively associate with the probability that major outlay decisions are reported to be made by women. Hypothesis 3 posits that the relative education of the woman also may increase this probability. The results shown in Table 3 indicate that educational hypogamy is associated with higher relative log-odds for female decisions than those for balanced decisions (0.317***), when compared with homogamous unions. In contrast, hypergamy has the opposite effect: the relative log-odds of the same outcome are smaller (–0.292***). In order to show the effects of hypogamy and hypergamy on each outcome, we estimated the marginal effects of relative education, as displayed in Table 4. Compared with homogamy, hypogamy increases the probability of female decision-making by 3.1 percentage points, and hypergamy reduces this probability by 2.3 percentage points. These findings support Hypothesis 3.
. | . | dy/dx . | SE . | P . | Lower 95 per cent CI . | Upper 95 per cent CI . |
---|---|---|---|---|---|---|
Outcome | Relative education | |||||
Balanced | ||||||
Hypogamy | –0.018 | 0.005 | 0.000 | –0.027 | –0.009 | |
Hypergamy | 0.007 | 0.004 | 0.082 | –0.001 | 0.015 | |
Woman | ||||||
Hypogamy | 0.031 | 0.004 | 0.000 | 0.024 | 0.038 | |
Hypergamy | –0.023 | 0.003 | 0.000 | –0.028 | –0.017 | |
Man | ||||||
Hypogamy | –0.013 | 0.003 | 0.000 | –0.019 | –0.007 | |
Hypergamy | 0.016 | 0.003 | 0.045 | 0.009 | 0.022 |
. | . | dy/dx . | SE . | P . | Lower 95 per cent CI . | Upper 95 per cent CI . |
---|---|---|---|---|---|---|
Outcome | Relative education | |||||
Balanced | ||||||
Hypogamy | –0.018 | 0.005 | 0.000 | –0.027 | –0.009 | |
Hypergamy | 0.007 | 0.004 | 0.082 | –0.001 | 0.015 | |
Woman | ||||||
Hypogamy | 0.031 | 0.004 | 0.000 | 0.024 | 0.038 | |
Hypergamy | –0.023 | 0.003 | 0.000 | –0.028 | –0.017 | |
Man | ||||||
Hypogamy | –0.013 | 0.003 | 0.000 | –0.019 | –0.007 | |
Hypergamy | 0.016 | 0.003 | 0.045 | 0.009 | 0.022 |
Source: Model in Table 3, authors’ own estimates.
Note: The reference category is Homogamy.
. | . | dy/dx . | SE . | P . | Lower 95 per cent CI . | Upper 95 per cent CI . |
---|---|---|---|---|---|---|
Outcome | Relative education | |||||
Balanced | ||||||
Hypogamy | –0.018 | 0.005 | 0.000 | –0.027 | –0.009 | |
Hypergamy | 0.007 | 0.004 | 0.082 | –0.001 | 0.015 | |
Woman | ||||||
Hypogamy | 0.031 | 0.004 | 0.000 | 0.024 | 0.038 | |
Hypergamy | –0.023 | 0.003 | 0.000 | –0.028 | –0.017 | |
Man | ||||||
Hypogamy | –0.013 | 0.003 | 0.000 | –0.019 | –0.007 | |
Hypergamy | 0.016 | 0.003 | 0.045 | 0.009 | 0.022 |
. | . | dy/dx . | SE . | P . | Lower 95 per cent CI . | Upper 95 per cent CI . |
---|---|---|---|---|---|---|
Outcome | Relative education | |||||
Balanced | ||||||
Hypogamy | –0.018 | 0.005 | 0.000 | –0.027 | –0.009 | |
Hypergamy | 0.007 | 0.004 | 0.082 | –0.001 | 0.015 | |
Woman | ||||||
Hypogamy | 0.031 | 0.004 | 0.000 | 0.024 | 0.038 | |
Hypergamy | –0.023 | 0.003 | 0.000 | –0.028 | –0.017 | |
Man | ||||||
Hypogamy | –0.013 | 0.003 | 0.000 | –0.019 | –0.007 | |
Hypergamy | 0.016 | 0.003 | 0.045 | 0.009 | 0.022 |
Source: Model in Table 3, authors’ own estimates.
Note: The reference category is Homogamy.
Regression Results: General Decision-Making
To see how our results on major financial outlays compare with general decision-making, we estimate our multinomial regression model with general decision-making as the dependent variable. Balanced decision-making is the reference category and all predictor and control variables remain the same as in the model of major outlay decisions. We comment mainly on how the results differ from the results on major outlay decisions. Shown in Table 5, the coefficients for female decision-making increase and the coefficients for male decision-making decrease as the woman’s relative earnings rise. A more complete picture of this, showing predicted probabilities of each outcome, can be seen in Figure 3. Thus, Hypothesis 1 finds support also when general decision-making is considered.
. | Main general decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | –0.066 | (0.036) | 0.305*** | (0.036) |
11–25 | 0.027 | (0.045) | 0.179*** | (0.051) |
51–75 | 0.196*** | (0.036) | –0.288*** | (0.052) |
76–90 | 0.373*** | (0.075) | –0.244* | (0.118) |
91–100 | 0.237*** | (0.051) | –0.172* | (0.068) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.383*** | (0.035) | 0.009 | (0.043) |
Hypergamy | –0.233*** | (0.043) | 0.040 | (0.041) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.217*** | (0.040) | 0.329*** | (0.043) |
High | –0.344*** | (0.034) | –0.228*** | (0.040) |
Woman’s age | –0.001 | (0.002) | –0.007* | (0.003) |
Woman’s age squared | 0.000 | (0.000) | 0.000* | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.163*** | (0.042) | –0.011 | (0.054) |
Separated, widowed | 0.321*** | (0.071) | –0.030 | (0.094) |
Number of children in household | 0.039** | (0.012) | 0.005 | (0.014) |
Joint earnings 20—quantile | 0.022* | (0.010) | –0.049*** | (0.012) |
Joint earnings 20—quantile squared | –0.002*** | (0.000) | 0.002*** | (0.001) |
Length of cohabitation in years | –0.001 | (0.002) | 0.006* | (0.003) |
Man’s unemployment in months | 0.017** | (0.006) | –0.006 | (0.007) |
Constant | –1.506*** | (0.094) | –1.545*** | (0.111) |
N | 72,386 |
. | Main general decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | –0.066 | (0.036) | 0.305*** | (0.036) |
11–25 | 0.027 | (0.045) | 0.179*** | (0.051) |
51–75 | 0.196*** | (0.036) | –0.288*** | (0.052) |
76–90 | 0.373*** | (0.075) | –0.244* | (0.118) |
91–100 | 0.237*** | (0.051) | –0.172* | (0.068) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.383*** | (0.035) | 0.009 | (0.043) |
Hypergamy | –0.233*** | (0.043) | 0.040 | (0.041) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.217*** | (0.040) | 0.329*** | (0.043) |
High | –0.344*** | (0.034) | –0.228*** | (0.040) |
Woman’s age | –0.001 | (0.002) | –0.007* | (0.003) |
Woman’s age squared | 0.000 | (0.000) | 0.000* | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.163*** | (0.042) | –0.011 | (0.054) |
Separated, widowed | 0.321*** | (0.071) | –0.030 | (0.094) |
Number of children in household | 0.039** | (0.012) | 0.005 | (0.014) |
Joint earnings 20—quantile | 0.022* | (0.010) | –0.049*** | (0.012) |
Joint earnings 20—quantile squared | –0.002*** | (0.000) | 0.002*** | (0.001) |
Length of cohabitation in years | –0.001 | (0.002) | 0.006* | (0.003) |
Man’s unemployment in months | 0.017** | (0.006) | –0.006 | (0.007) |
Constant | –1.506*** | (0.094) | –1.545*** | (0.111) |
N | 72,386 |
Source: EU-SILC 2010, authors’ own estimates.
Note: Country dummy variables omitted from the table. Bootstrapped model estimation using 200 replications.
P < 0.05,
P < 0.01,
P < 0.001, two-tailed test.
. | Main general decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | –0.066 | (0.036) | 0.305*** | (0.036) |
11–25 | 0.027 | (0.045) | 0.179*** | (0.051) |
51–75 | 0.196*** | (0.036) | –0.288*** | (0.052) |
76–90 | 0.373*** | (0.075) | –0.244* | (0.118) |
91–100 | 0.237*** | (0.051) | –0.172* | (0.068) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.383*** | (0.035) | 0.009 | (0.043) |
Hypergamy | –0.233*** | (0.043) | 0.040 | (0.041) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.217*** | (0.040) | 0.329*** | (0.043) |
High | –0.344*** | (0.034) | –0.228*** | (0.040) |
Woman’s age | –0.001 | (0.002) | –0.007* | (0.003) |
Woman’s age squared | 0.000 | (0.000) | 0.000* | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.163*** | (0.042) | –0.011 | (0.054) |
Separated, widowed | 0.321*** | (0.071) | –0.030 | (0.094) |
Number of children in household | 0.039** | (0.012) | 0.005 | (0.014) |
Joint earnings 20—quantile | 0.022* | (0.010) | –0.049*** | (0.012) |
Joint earnings 20—quantile squared | –0.002*** | (0.000) | 0.002*** | (0.001) |
Length of cohabitation in years | –0.001 | (0.002) | 0.006* | (0.003) |
Man’s unemployment in months | 0.017** | (0.006) | –0.006 | (0.007) |
Constant | –1.506*** | (0.094) | –1.545*** | (0.111) |
N | 72,386 |
. | Main general decision maker . | |||
---|---|---|---|---|
. | Woman . | Man . | ||
. | Coef. . | SE . | Coef. . | SE . |
Relative earnings per cent (ref = 26–50) | ||||
0–10 | –0.066 | (0.036) | 0.305*** | (0.036) |
11–25 | 0.027 | (0.045) | 0.179*** | (0.051) |
51–75 | 0.196*** | (0.036) | –0.288*** | (0.052) |
76–90 | 0.373*** | (0.075) | –0.244* | (0.118) |
91–100 | 0.237*** | (0.051) | –0.172* | (0.068) |
Woman’s relative education (ref=Homogamy) | ||||
Hypogamy | 0.383*** | (0.035) | 0.009 | (0.043) |
Hypergamy | –0.233*** | (0.043) | 0.040 | (0.041) |
Woman’s level of education (ref=Medium) | ||||
Low | 0.217*** | (0.040) | 0.329*** | (0.043) |
High | –0.344*** | (0.034) | –0.228*** | (0.040) |
Woman’s age | –0.001 | (0.002) | –0.007* | (0.003) |
Woman’s age squared | 0.000 | (0.000) | 0.000* | (0.000) |
Woman’s marital status (ref=Married) | ||||
Never married | 0.163*** | (0.042) | –0.011 | (0.054) |
Separated, widowed | 0.321*** | (0.071) | –0.030 | (0.094) |
Number of children in household | 0.039** | (0.012) | 0.005 | (0.014) |
Joint earnings 20—quantile | 0.022* | (0.010) | –0.049*** | (0.012) |
Joint earnings 20—quantile squared | –0.002*** | (0.000) | 0.002*** | (0.001) |
Length of cohabitation in years | –0.001 | (0.002) | 0.006* | (0.003) |
Man’s unemployment in months | 0.017** | (0.006) | –0.006 | (0.007) |
Constant | –1.506*** | (0.094) | –1.545*** | (0.111) |
N | 72,386 |
Source: EU-SILC 2010, authors’ own estimates.
Note: Country dummy variables omitted from the table. Bootstrapped model estimation using 200 replications.
P < 0.05,
P < 0.01,
P < 0.001, two-tailed test.
With regard to our Hypothesis 2, we find only a slight hint of gender display as both female and male decision-making reverse their overall association with relative earnings in the 91–100 per cent relative earnings category. However, for both partner’s decision-making probabilities, the confidence intervals are too wide to confirm a significant reduction (for the woman) or increase (for the man) in the 91–100 per cent category compared with the 76–90 per cent or the 51–75 per cent category. As a result, neither Part (a) nor Part (b) of Hypothesis 2 finds support in the case of general decision-making.
Based on Table 5, Hypothesis 3 does find support. The coefficient for hypogamy increases the log-odds of female decision-making. When the woman is less educated than the man, this reduces the chances that she is reported as the one who generally has the final say. This is in line with the results based on major outlays.
Control Variables
The models in Tables 3 and 5 controlled for various characteristics of individuals and couples. Some of the control variables produce similar results in both models. It may come as a surprise that the woman’s absolute level of education is negatively associated with both her and his decision-making probabilities. This would imply that in the same relative income setting, highly educated women are more likely to be reported as making joint decisions together with the spouse than less educated women (see also Treas and Tai, 2012). Another explanation to this result is our model specification as we also include her relative education in the model.4 Being married implies a higher likelihood of balanced decision-making, while women in non-marital cohabitation have a higher probability of deciding on their own. The husband’s number of months of unemployment is associated with a higher probability of the female decision-making and with a lower probability of male decision-making, when compared with the balanced category.
With regard to absolute earnings, measured by joint earnings 20-quantiles, male decision-making displays a U-shaped association as both the linear and the quadratic term are statistically significant. The inverse relationship is true for the woman’s side in general decision-making model. Parameter estimates for the woman’s age, number of children in the household, and the length of the union are not consistent in the two models. For example, having children is associated with an increased male decision-making in the major outlays model and female decision-making in the general decision-making model.
Conclusions
Shifting patterns of assortative mating have enhanced women’s resources relative to their male partner’s in the context of the family. In order to gain more insight into the role of relative education and earnings, we analysed how they are associated with the balance of decision-making in the couple, which is an important aspect of equality in the family (Rosenbluth, Steil and Whitcomb, 1998; Kirchler et al., 2001). Focusing on major financial and other generally important decisions, we hypothesized that women’s relative earnings are positively associated with the probability that she is reported as the household decision maker. The results of multinomial regression modelling suggest that this is indeed the case. The woman’s relative earnings are positively associated with the probability that she would be reported to be the chief decision maker.
Our second hypothesis was that norms and attitudes about gender roles would distort the association between relative earnings and decision-making. Specifically, we hypothesized that if the woman earned all or almost all of the couple’s income, this would either (i) reduce the probability that she would be reported to be the decision maker or (ii) increase the man’s respective probability. In the case of financial decision-making, the second part of this hypothesis is formally supported, but a non-linear pattern is found concerning the reduction in women’s decision-making. Our results indicate that when the man has very little labour income compared with the woman, the likelihood that he is reported to be the decision maker in financial issues is higher, when compared with the man whose earnings are relatively lower, but more similar to that of the woman. These results run against predictions that would be based solely on a relative resources perspective. They rather lend support to the argument that non-traditional couples may in some ways display traditional gender roles. Tichenor (2005a) theorized that women in non-traditional couples leave decision-making to the men in order to portray their family as more traditional than it actually is. Our findings pertaining to financial decision-making are consistent with this view. Regarding decision-making on other important questions, however, we do not find support for the hypothesis about gender display.
We also hypothesized that the relative education of the woman would have a positive association with her making the decisions. The findings indicate that the higher relative education of the woman increases the odds that she would make the major financial or other important decisions.
Our findings suggest that both relative education and relative earnings may have important implications for the decision-making balance in the family. An increase in hypogamous marriages and female breadwinning may alter spending and consumption patterns, and also have an effect on childbearing intentions and, consequently, on fertility (Van Bavel, 2012).
This study has limitations and offers several avenues of further improvement. The data only pertain to 2009, which is characterized by a weaker male employment situation in many countries. Also, we only consider income from labour and have no information about individual wealth, which could influence the balance of decision-making. Due to the cross-sectional data, selection into union based on education and earnings is an issue that we have not been able to address but which may have implications for the balance of decision-making. Finally, our results present a European average5 as we have not addressed international differences in decision-making patterns.
Funding
The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013)/ERC Grant Agreement no. 312290 for the GENDERBALL project. This research also benefited from support by Estonian Research Council (research theme PRG71).
Martin Klesment (1976) is a researcher at the Estonian Institute for Population Studies, Tallinn University. Currently, he is also visiting professor of contemporary history at Tartu University. His research focuses on family and fertility, historical demography, and economic history.
Jan Van Bavel (1971) is full professor of demography and sociology at the University of Leuven (KU Leuven), where he is currently programme director of sociology. His research addresses long-term trends in fertility, family formation, and union dissolution, from the 19th into the 21st century.
Footnotes
All questions used in the current analysis and the relevant options given in the EU-SILC guidelines are included in the Supplementary Part A.
In order to test the difference formally, we re-estimated the model by changing the reference category of the relative earnings variable to 91–100 per cent (model not shown here). The predicted average marginal effects (not shown) confirmed that neither categories 51–75 per cent nor 76–90 per cent are associated with a statistically significant higher probability of female decision-making when compared with the 91–100 per cent category.
The complete results for major outlay decisions based on the male sample are provided in Supplementary Part C.
Fitting models, so that they exclude relative education and include the man’s absolute education instead, results in a positive educational gradient in decision-making for both partners (see Supplementary Part F), as would be expected based on the resource theory. Since we are more interested in the relative education variable, we keep it in our main models. Another sensitivity check using both partner’s absolute education in the same model, but omitting relative education, was carried out with the original, finer grade ISCED variable instead of the aggregated three-level variable. This did not change our findings about relative earnings and balance of decision-making.
We have, however, checked during preliminary analysis that our results are not driven by a few outlier countries.