Intra-couple wealth inequality: What's demographics got to do with it?

Rehm, Miriam; Schneebaum, Alyssa; Schuster, Barbara Working Paper Intra-couple wealth inequality: What's demographics got to do with it? ifso working paper, No. 22 Provided in Cooperation with: University of Duisburg-Essen, Institute for Socioeconomics (ifso) Suggested Citation: Rehm, Miriam; Schneebaum, Alyssa; Schuster, Barbara (2022) : Intra-couple wealth inequality: What's demographics got to do with it?, ifso working paper, No. 22, University of Duisburg-Essen, Institute for Socio-Economics (ifso), Duisburg This Version is available at: https://hdl.handle.net/10419/262357 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ uni-due.de/soziooekonomie/wp ifso working paper 2019 no.5 ifso working paper Miriam Rehm Alyssa Schneebaum Barbara Schuster 2022 no.22 Intra-Couple Wealth Inequality: What‘s Demographics Got to Do With It? uni-due.de/soziooekonomie/ Intra-Couple Wealth Inequality: What’s Demographics Got To Do With It? Miriam Rehm∗Alyssa Schneebaum†Barbara Schuster‡ ∗ Institute of Socio-Economics, University of Duisburg-Essen. Lotharstrasse 20-22, 47057 Duisburg, Germany; and Department of Socio-Economics, Vienna University of Economics and Business. Email: [email protected] † Department of Economics, Vienna University of Economics and Business. Welthandelsplatz 1, 1020 Vienna, Austria. Email: [email protected] . Schneebaum gratefully acknowledges support for this research from the Austrian Science Fund (FWF) under grant number P 32709-G. ‡ The New School for Social Research. 66 West 12th Street, New York, NY 10011, United States. Email: [email protected] 1 Abstract Existing literature shows that on average and across countries, men have higher levels of wealth than women. However, very little is known about the gender-specific wealth gap within couples. This paper studies this phenomenon. The particular focus of the paper is on the relationship between the demographic characteristics of the couple and the couple’s gender wealth gap. We focus on how age, education, marital status, fertility, and migration background are related to the wealth gap within a couple. In both univariate and multivariate analyses, we find that the strongest demographic predictors of an intra-couple wealth gap are the age gap between the members of the couple; the highest level of education; and the composition of migration background in the couple. The gender wealth gap is particularly high in couples with a native-born man and a foreign-born woman. 1 Gender and Wealth Research on wealth inequality has boomed in the last decade, dramatically improving our understanding of how wealth is distributed across households. With the development of this literature, social scientists have also gained new insights into the gender dimensions of wealth inequality. Despite our growing understanding of wealth inequality across households, we have little empirical knowledge about wealth inequality within households. This paper uses unique data on the intra-household distribution of wealth ownership to study gendered wealth inequality within couples. Most empirical studies looking at the gender wealth gap assess the difference in net wealth of households headed by men versus women, find that men have higher wealth holdings than women in both the raw data and in multivariate analysis (Deere & Doss, 2006; M. L. Chang, 2010; Ruel & Hauser, 2013; Schneebaum, Rehm, Mader, & Hollan, 2018). There is also a significant minority of studies that find no gender-specific wealth gap for the subgroup of young households (Schmidt & Sevak, 2006) and for the marriage wealth premium (Lersch, 2017) in the full models in these analyses. Despite the contributions of these studies, the intra-household distribution of wealth – that is, the way in which wealth is distributed within a household – has largely remained a black box. Much of the reason why is because almost all existing data sets collect information on wealth at the household, not person, level. 1 This paper is one of the few to address this hole in the literature. We use data from the second wave (data collected in 2014-2015) of the Household Finance and Consumption Survey (HFCS), which has been a major contributor to the boom in analyses of the distribution of wealth in Europe (Household Finance and Consumption Network (HFCN), 2019). These are the first data to make it possible to investigate the demographic determinants of the gender wealth gap at the personal level in Austria. To the best of our knowledge, Austria is only the third high-income country for which nationally representative person-level wealth data are available; the other two are Germany, where Sierminska, Frick, & Grabka (2010), Grabka, Marcus, & Sierminska (2015), Lersch (2017), & Sierminska, Piazzalunga, & Grabka (2018) have done extensive research using the Socio-Economic Panel (SoEP), and France, where the French HFCS has been analysed by Frémeaux & Leturcq (2020). Austria is an especially interesting case, because the distribution of its household wealth is highly unequal in international comparison (Balestra & Tonkin, 2018), and the question remains open whether intra-household dynamics play a role in this. In studying the wealth gap within households, the unique contribution of the paper is its focus on the ways in which couples’ demographic characteristics – their age, education, marital status, fertility, and migration background – relate to their gender-specific distribution of wealth. In particular, we assess (both theoretically and empirically) the ways in which these five demographic characteristics may be related to the unequal distribution of wealth within couples. Our variables to help explain the couple-level gender wealth gap are also at the level of the couple: the couple’s age difference, the composition of their countries of origin, and the highest level of education in the couple are examples. By structuring the analysis in this way, we can assess how these characteristics relate (or do not relate) to intra-couple wealth inequality. Empirically, we present the relationship between the couple-level demographic characteristics and the intra-couple gender wealth gap in both univariate and multivariate analyses. The latter employs OLS analyses to assess the correlation between the demographics and the mean wealth gap while controlling for other demographic and economic characteristics. 2 The main findings of the paper show that indeed, some of the demographic variables studied here are strongly related to intra-couple wealth gaps. The difference in the ages of the two members of the couple is a particularly powerful characteristic related to the gender wealth gap. Moreover, couples in which the man is native-born and the woman is an immigrant have a particularly high gender wealth gap. 2 Institutional Background: Wealth and Gender Relations in Austria 2.1 Literature on the Gender Wealth Gap Most studies of the gender wealth gap include some reference to (socio-)demographic characteristics; however, the focus of the analysis often lies on other factors, such as labor market characteristics, so the demographic variables function largely as controls (Schmidt & Sevak, 2006; Neelakantan & Y. Chang, 2010; Grabka et al., 2015; Ruel & Hauser, 2013). The main exception is Yamokoski & Keister (2006), who investigate the effect of education, marriage, and fertility on the gender wealth gap, but do not focus on migration background. Other contributions focus on education (Sierminska, Frick, et al., 2010), marital status (Sierminska, Piazzalunga, et al., 2018; Frémeaux & Leturcq, 2020), or migration (Bauer, Cobb-Clark, Hildebrand, & Sinning, 2011). Although the theoretical literature emphasizes that household resources (both income and wealth) cannot be assumed to be pooled and shared equally (Ponthieux & Meurs, 2015), the fact that wealth data are typically collected at the household level has limited the number of studies that assess the wealth gap between men and women within the household. Many existing studies have therefore been restricted in analyzing wealth differentials by gender to comparing single-adult households (Schmidt & Sevak, 2006; Schneebaum et al., 2018), defining the household through a representative member (Ruel & Hauser, 2013), or assessing the gender wealth gap in particular components of wealth sometimes reported at the individual level, such as pensions (Neelakantan & Y. Chang, 2010). The important exceptions to the literature’s reliance on household-level data are 3 based on the wealth module of the German Socio-economic Panel (Sierminska, Frick, et al., 2010; Grabka et al., 2015; Sierminska, Piazzalunga, et al., 2018), and the French Household Finance and Consumption Survey (HFCS) and its national precursor, the Life History and Wealth Survey (Frémeaux & Leturcq, 2020). In addition to Germany and France, seminal papers investigate the gender wealth gap in the U.S. (Schmidt & Sevak, 2006; Ruel & Hauser, 2013; Yamokoski & Keister, 2006). Schneebaum et al. (2018) analyze the gap in several European countries. This paper expands the literature on the gender wealth gap by providing evidence of the determinants of intra-household gender wealth differences in Austria based on the HFCS, looking especially at the role of demographic characteristics on the gap. The rest of this chapter discusses the institutions framing the relevance of these demographic characteristics, and develops hypotheses as to the direction of their effect. 2.2 Age The life-cycle hypothesis (Modigliani, 1966) predicts that resources are accumulated during the economically active years are spent down in retirement. Since men are typically older than women in couples and have thus had more time to accumulate wealth, this should lead to a positive correlation of the age difference within the couple with the gender wealth gap. Furthermore, social and cultural norms may have changed over time, which may lead to variation of the gender wealth gap within older couples relative to younger couples. For instance, a more equal sharing of wealth within couples may have become more common as women generally gained more even footing with men. This suggests that the average age of the couple may contribute to explaining the gender wealth gap; since the data used in the analysis are a cross-section from 2014, the average age of the couple captures information about the institutional conditions of the birth cohort of the people in the couple. 4 2.3 Education We incorporate education as a demographic variable in our analysis, following others in the fields of demography (Lutz, 2010) or psychology (Goldberg, Sweeney, Merenda, & Hughes Jr, 1998; Kravitz, 2004). Lutz (2010) even claims that “education will be at the heart of 21st century demography”, suggesting that the level of educational attainment, besides age and sex, should be among the dimensions routinely addressed in standard demographic analysis. Like age, the level of education is typically positively correlated with wealth (Pfeffer, 2018). Possible channels may either be the link of education to work income, or between education and financial literacy, and thus higher capital income (Cupák, Fessler, Schneebaum, & Silgoner, 2018). At the couple level, if there is a difference in education that favors men, then this would be another possible explanation of the gender wealth gap. Moreover, a gap in education (or indeed age) between partners could correlate with an imbalance in bargaining power within the couple, which could translate to a wealth gap. Finally, assortative mating – the preference for partners with the same or similar level of education – may raise wealth inequality overall (since random partner choice would more often match high wealth individuals to low wealth individuals, which leads to lower average wealth inequality at the household level), but, if anything, should reduce the gender wealth gap. The reason is that if the capacities for wealth accumulation are similar for both partners, then there are presumably lower returns to specialization on market versus non-market work, and there will also be less of an imbalance in bargaining power. This in turn may lead more similar wealth levels of the partners. 2.4 Marital Status The empirical literature consistently documents a marriage wealth premium relative to singles or cohabiting couples (Keister, 2003; Sierminska, Frick, et al., 2010; Vespa & M. A. Painter, 2011; Addo & Lichter, 2013; M. Painter, Frech, & Williams, 2015; Lersch, 2017; Kapelle & Lersch, 2020). This may be due to a longer planning horizon and increased trust due to the higher commitment level of married couples, which may in turn increase 5 specialization, total work hours of the couple, or investment. While the level of wealth thus rises with marriage, it is much less clear whether marriage is also linked to a higher difference in the wealth levels of partners; that is, whether the gender wealth gap differs by marital status (Lersch, 2017). The theoretical expectations for the effect of marital status on the gender wealth gap are not clear a priori. On the one hand, the above-mentioned increased commitment arising from the socio-cultural institution of the marriage pact may lead to more equality in the intra-marital distribution of assets, and thus to a lower gender wealth gap. On the other hand, increased specialization may lead to a weaker labor market attachment of the partner specializing in the household and child care, which may raise the gender wealth gap. Finally, legal questions surrounding asset ownership in marriage might play an enhancing or diminishing role for the gender wealth gap, in particular community versus separate wealth ownership of couples (see also Frémeaux & Leturcq, forthcoming). Regarding the legal institutions, the General Civil Code of Austria of 1811 – which still forms the basis for Austrian civil law, and determined family law until a major family law reform in the 1970s – defined the male partner as the “head of the family”, and the wife as legally subordinate with regard to wealth management (Floßmann, 2006), although the segregation of property was the standard case. 1 Since the General Civil Code assumed the wife by default to have entrusted the husband with managing the wealth which she had brought into the marriage, and since property acquired during the marriage was automatically assumed to be acquired by, and thus owned by, the husband (thus bestowing management and use rights upon the husband) (Lehner, 1987), the literature considers the Austrian legal system until the 1970s as “presumed administrative community” and “disguised communal property” (Floßmann, 2008, p.95). Assets owned by children were also managed exclusively by the father, and the mother was not permitted to manage them even if the father was unable to do so; instead a legal guardian had to be instated (Lehner, 1987, p.22). While the wife was obliged by the General Civil Code to aid the husband in his gainful employment, she did not participate in the ownership of the assets thus acquired (Lehner, 1987). 1Deviation by marriage contract was possible. 6 Table 3: Descriptive statistics for women and men in couple households Females Males Average Age 50.9 53.6 Education: Primary/lower secondary 20.1 10.7 Education: Upper secondary 65.4 65.1 Education: Tertiary 14.4 24.2 Percent migrant 11.3 11.0 Married 92.8 92.8 Legally single 4.6 4.7 Divorced 2.0 2.2 Widowed 0.4 0.3 Share with children in household 30.3 Share with children outside household 24.8 N observations 1,503 1,503 Notes: Authors’ calculations on 2014 HFCS data. Finally, figure 1 illustrates the distribution of the raw gender wealth gap, along the distribution of couples’ net wealth. In absolute terms, the average gender wealth gap rises across the unconditional distribution of couples’ net wealth percentiles. The gap is generally higher the higher the level of wealth is. Further, as shown by figure A1 in the appendix, the gender wealth gap is also somewhat higher for wealthier households when the gap is measured as a percent of household wealth. 13 Figure 1: The raw gender wealth gap between women and men in couple households Notes: Weights and multiple imputations taken into account. No values for percentiles 11, 55, 58, 82, 83 due to varying sets of implicates. Gender wealth gap is the difference between a man’s and woman’s net wealth. Authors’ calculations on 2014 HFCS data. In section 4, we explore the relationship between the five demographic variables of interest and the gender wealth gap for men and women in couple households. We use these univariate analyses to get a sense of how these personand household-level characteristics are related to wealth holdings for men and women to better understand how they will matter for the wealth gap within households. For this part of the analysis, the outcome variable of interest is net wealth for men and women. In section 5, we turn to a multivariate analysis of the demographic determinants of the intra-household wealth gap. That is, we shift our outcome variable of interest from the average or median net wealth of all women or all men in couple households to the gender wealth gap within individual households. We approach the multivariate analysis of intra-couple wealth inequality via OLS, that is, we predict the wealth gap within each couple based on the couple’s demographic and other characteristics. Since there are zero and negative values for net wealth in the data, we use the inverse hyperbolic sine (IHS) transformation of net wealth as the outcome variable. 3 We include further controls in our model to predict the wealth gap within 3The transformation applied is N W =asinh(N W ) = ln(N W +√N W 2+ 1). 14 households: for both people, the employment status (employee, self-employed, employer, unemployed, not in labor force, or retired); the hours worked (full-time or part-time); labor market attachment (the number of years worked divided by potential work years, i.e. age minus 18); and a dummy variable indicating whether the household previously received an inheritance. The multivariate results are presented in section 5 below. 4 Descriptive Results This section presents the co-variation of the gender wealth gap with our demographic characteristics of interest, that is, age, education, marital status, the presence of children, and migration background. It focuses on the mean and median of these covariates. A multivariate analysis is undertaken in section 5. 4.1 Age As discussed in section 2, we expect the age difference of the two partners to correlate with the intra-couple gender wealth gap – the older partner (more often the man) will likely have more wealth. The rows in table 4 show the age difference of couples, and its columns the concomitant mean and median wealth levels of men and women in these couples, as well as the gender wealth gap. As expected, the men are older than the women in our sample; this is the case for roughly 73% of couples. There is also a clear preference for similar age in couples. Both when the men and when the women are older, the most common household type among these groups has a rather small age gap of less than five years. The gap increases in both median and mean with the age difference when the men are older, from about 13% when the man is less than five years older to 43% when he is more than 10 years older. In couples with a small negative age gap – where women are less than five years older than men – the gender wealth gap is positive, in favor of the man. Only when the women in the couples are more than five years older than their partners, the gender wealth gap is inverted – although the very high values for a gap over ten years should be treated with caution due to a limited number of observations. 15 The descriptive evidence thus supports the predictions of the life-cycle hypothesis – the age gap in couples is positively correlated with the gender wealth gap in our data. However, this finding does not hold equally for both genders. Women need to be considerably older for the gender wealth gap to be in their favor. Given that the life-cycle hypothesis also predicts that women should accumulate more wealth than men during their active phase due to their longer life expectancy, it is likely that additional explanatory factors play a role for the gender wealth gap. Table 4: Net wealth and wealth gaps by age difference (in EUR) Sample Share Mean Mean gap Median Median gap Same age (∆=0) Women 132 8.6% 190,777 -0% 107,941 6% Men 190,455 115,442 Woman is younger ∆<5 years Women 699 45.3% 159,205 7% 68,833 13% Men 171,657 78,890 ∆5-10 years Women 345 23.3% 131,432 21% 66,335 24% Men 165,483 87,184 ∆>10 years Women 69 4.6% 94,936 25% 36,132 43% Men 125,856 63,860 Woman is older ∆<5 years Women 192 13.2% 141,220 71% 68,973 17% Men 483,521 82,658 ∆5-10 years Women 51 3.7% 112,130 -16% 74,476 -10% Men 96,575 67,729 ∆>10 years Women 15 1.2% 222,021 -105% -590% Men 108,488 31,141 Notes: Weights and multiple imputations taken into account. Sample size indicates the number of individuals in the respective subgroup of the sample. Shares are the respective share of each subgroup. “Gender wealth gap” is defined as the difference between a man’s and woman’s net wealth compared to the man’s net wealth. Authors’ calculations on 2014 HFCS data. 4.2 Education As with age, we expect the relative education level between members of a couple to contribute to a gender wealth gap within the couple. Figure 2 compares the relative education level of women and men in couples to their mean net wealth levels. The three 16 panels show the highest obtained level of education of the male partner, and the bars are sorted by the female partner’s education level. The population share of each group is also indicated on the left-hand side axis. Figure 2: Comparison of women’s and men’s education levels in couples with regard to their average net wealth Notes: Weights and multiple imputations taken into account. Values in brackets refer to the shares of the respective couples. Authors’ calculations on 2014 HFCS data. Regardless of their level of education, and their education relative to men’s, Women own on average less wealth. The only combination in which the gap virtually disappears is when both women and men in couples have completed at most lower secondary education (F=M in the top panel in figure 2). This is also the group of men with the lowest average net wealth of all groups of men considered here (orange bars). If women with more education are partnered with a man with at most lower secondary education (F>M), the gender wealth gap is small (about € 16,000, or 11%) – but indeed still positive, favoring men. For couples in which the man completed secondary education (middle panel), the gender wealth gap is 10% (more than € 13,300) when women also completed secondary education – a group which makes up almost half the population. The mean gap is 12% 17 (about € 26,500) if women completed tertiary education, and 30% (about € 29,000) if they have at most lower secondary education (F<M in the middle panel of figure 2). If men have tertiary education (the bottom panel), the gender wealth gap amounts to 6% or less than € 14,000 if the woman has less education, and a whopping 60% or over € 500,000 if the woman also holds a tertiary degree. 4 This very large gap in figure 2 is thus driven by outlying households with very high levels of wealth (and very large gender wealth gaps). In sum, figure 2 thus shows that education does not close the gender wealth gap in these data. Even when women are more highly educated than men, a gender wealth gap persists. 4.3 Marital Status The legal institutional analysis in section 2 indicated that we would expect a higher gender wealth gap among married couples, especially older ones. Table 5 shows the sample size, the share in the population, and the level and relative gender gap at the mean and median for married and unmarried but cohabitating couples in our sample. A large majority of couples, almost 93%, are married, and our descriptive evidence confirms the marriage wealth premium with married women owning roughly € 155,000 on average compared to non-married women’s € 85,000. For men, the marital premium is even higher, at a mean of about €215,000 versus €93,000 for men in unmarried couples. There is a positive gender wealth gap both at the mean and the median for both married and unmarried couples. However, the difference between the median and the mean shows that the gender wealth gap is right-skewed in married couples, and left-skewed in unmarried couples. That is, in couples with higher levels of wealth, the gender wealth gap tends to be larger. Women in married couples might therefore indeed be accumulating less wealth, which might be due to their weaker labor market attachment. In particular, rearing children is one way in which women’s labor market attachment can be weakened, which the next section investigates. 4 Figure A2 in the appendix reproduced the same figure using the median, not mean, level of wealth, and the wealth gap in this category is not as large when using the median. 18 Table 5: Net wealth by marital status (in EUR) Sample Share (%) Mean Mean wealth gap Median Median wealth gap Couple households Married Women 1,399 92.8 154,034 29% 73,661 16% Men 216,343 87,290 Not married Women 104 7.2 84,792 9% 21,010 30% Men 92,833 30,030 Notes: Weights and multiple imputations taken into account. Sample size indicates the number of individuals in the respective subgroup of the sample. Shares are the respective share of women and men in each subgroup. “Wealth gap” is defined as the difference between a man’s and woman’s net wealth compared to the man’s net wealth in couple or single households. Authors’ calculations on 2014 HFCS data. 4.4 Fertility Theory provides arguments both for a higher and for a lower gender wealth gap due to children present in the household (see section 2). The data show a u-shaped pattern of the average gender wealth gap with regard to the number of children under 16 present in the household. The gender wealth gap is largest, at about 32%, in couples without children; it declines to 13% and 12% in couples with one or two children; and it rises again to about 26% for couples with more than two children. At the median, the gender wealth gap rises with the number of children from about 12% to about 26%. That the gender wealth gap is right-skewed in the group of couples without children suggests that there are wealthy childless couples with a larger gender wealth gap in our data. Apart from these households, the gender wealth gap appears to rise with the number of children. As suggested by the women’s labor market-attachment hypothesis, higher fertility is associated with a higher gender wealth gap in this descriptive evidence. 4.5 Migration As discussed in section 2, we hypothesize partners with a migration background (that is, those who are born abroad) to have lower wealth than their native partners; and that the gender wealth gap is smaller in couples in which both people have the same migration background status (migrant or native). Table 7 shows (1) couples in which both partners are natives, (2) couples in which both partners are migrants, (3) couples in which only 19 Table 6: Net wealth and wealth gaps by number of children (in EUR) Sample Share Mean Mean gap Median Median gap No children Women 1,061 69.7% 154,976 32% 80,165 12% Men 229,015 90,968 One child Women 207 13.6% 116,567 13% 46,960 18% Men 134,277 57,191 Two children Women 180 12.1% 169,643 12% 71,306 21% Men 192,193 90,645 More than two children Women 55 4.6% 101,511 26 % 23,636 26% Men 137,408 31,929 Notes: Weights and multiple imputations taken into account. Sample size indicates the number of individuals in the respective subgroup of the sample. Shares are the respective share of each subgroup. “Gender wealth gap” is defined as the average difference between a man’s and woman’s net wealth compared to the man’s net wealth. Authors’ calculations on 2014 HFCS data. the female partner has a migration background, and (4) couples in which only the male partner has a migration background. Unsurprisingly, couples in which both partners are native-born make up the majority of the population, at almost 85%. They have the highest mean gender wealth gap, and it is right-skewed (the gap is only 13% at the median). Couples comprised of two migrants have much lower net wealth at the mean and in the median, and their gender wealth gap is inverse and minimal. In couples where only the woman has a migration background, the gender wealth gap re-emerges, at about 23% on average and 41% at the median. If only men in the couple has a migration background, then women’s wealth catches up, and the gender wealth gap on average disappears. These findings suggest that migration background does indeed work in mixed couples as hypothesized; women in couples can “make up” for the gender wealth gap through being native-born. However, this does not explain the larger gender wealth gap within native couples, except to the extent that the richest households in the data comprise two native-born Austrians and that these households have a high gender wealth gap. 20 Table 7: Net wealth and wealth gaps by migration background (in EUR) Sample Share Mean Mean gap Median Median gap Neither partner Women 1,253 84.2% 152,013 31% 84,364 13% Men 219,981 96,793 Both partners Women 102 6.5% 53,961 -3% 9,770 -9% Men 52,305 8,940 Female partner only Women 75 4.8% 102,255 23% 51,106 41% Men 133,376 86,204 Male partner only Women 73 4.5 280,321 -1% 46,152 29% Men 276,417 64,963 Notes: Weights and multiple imputations taken into account. Sample size indicates the number of individuals in the respective subgroup of the sample. Shares are the respective share of each subgroup. “Gender wealth gap” is defined as the average difference between a man’s and woman’s net wealth compared to the man’s net wealth. Authors’ calculations on 2014 HFCS data. 5 Multivariate OLS Results We now turn to a multivariate analysis of the relationship between demographic characteristics and the intra-household gender wealth gap. The outcome variable of interest is the gender-specific wealth gap within the couple; the control variables are all coupleand household-specific. The value added of the multivariate analysis is that it allows us to assess the relationship between our demographic characteristics of interest and the gender wealth gap, while holding all other characteristics constant across households. Thus, we are able to disentangle the role of any one demographic from the other covariates, which include the other demographic variables, as well as a battery of other couple-level controls. Each model also includes an indicator of whether the survey respondent was female; as we will see, this variable consistently indicates that households with a female survey respondent have, on average, a lower gender wealth gap than couples with a male respondent. This phenomenon is likely because households with a female respondent have, by definition, named the female as the “financially most knowledgeable person” in the household; in such couples, the woman likely has more bargaining power than the women in couples whose male partners are the “financially most knowledgeable” member of the household. Given the focus of the analysis on demographics, the specifications in tables 8-9 have 21 age, education, fertility, marital status, and migration as their main explanatory variables. First, age is measured as the difference in age (in years) between the two members of the couple as well as the average age of the couple (the latter is to capture potential cohort effects). Second, education is measured as the highest education level in the couple and as the difference in education classes among the two members of the couple (recall from section 3 that educational classes are identified as primary, secondary, and tertiary). Third, fertility is captured via three indicators: a dummy variable indicating whether there are children still living in the household; a dummy variable indicating whether either partner has an older child living outside of the household; and a variable interacting the couple’s average age with the presence of children. The latter variable is meant to capture differences in the effect of the presence of children across birth cohorts. Fourth, marital status is represented by a dummy variable indicating that the couple is currently married; the alternative is that the members of the couple state that they are divorced, widowed, or legally unmarried, or any combination of these. We also include a dummy variable indicating that the couple is married and “older” – that is, born before 1958 – to account for the institutional changes around gender equality for married couples that occurred in 1978, as described in section 2. Finally, migration is captured via three mutually exclusive dummy variables: only the female is a migrant; only the male is a migrant; or both partners are immigrants. The control group is that both partners are native-born Austrians. Along with the demographic characteristics, the models include what we call “labor controls” and “wealth controls.” The former include indicators of the labor market situation in the couple: mutually exclusive categories of whether the male only, female only, or both partners are employers, employees, unemployed, self-employed, or not in the labor force, as well as an indicator of the difference in the work histories (number of years worked) of the members of the couple. The wealth controls are a dummy variable indicating that the household has received an inheritance or gift and the inverse hyperbolic sine transformed level of wealth owned by the couple. The different columns of the results tables represent specifications that include different combinations of these control variables. Table 8 presents the baseline results. In this and all remaining OLS tables, the first column shows the results of models that control only for the couple’s demographics; 22 Table 11: OLS results: Demographic determinants of the intra-household gender wealth gap. Only couples without a migrant and those reporting intra-couple wealth inequality. (1) (2) (3) (4) Female Respondent -5.904∗∗∗ -5.849∗∗∗ -5.775∗∗∗ -5.671∗∗∗ (1.363) (1.510) (1.303) (1.430) ∆Age 0.304∗∗∗ 0.246∗∗ 0.346∗∗∗ 0.292∗∗ (0.103) (0.120) (0.099) (0.122) Avg. Age of Couple 0.051 0.047 0.033 0.045 (0.074) (0.073) (0.082) (0.078) Highest Education in Couple -0.977 -1.513 -1.732 -2.125 (1.257) (1.299) (1.269) (1.311) ∆Education 1.539 1.558 1.961 1.926 (1.276) (1.273) (1.252) (1.296) Children 8.684 8.718 9.584∗9.476∗ (5.794) (6.166) (5.496) (5.665) Children out of Household -3.172∗∗ -2.277 -2.427∗-1.693 (1.292) (1.425) (1.328) (1.465) Children Present * Average Age -0.145 -0.162 -0.180 -0.189 (0.142) (0.144) (0.134) (0.132) Married 2.748 1.282 2.584 1.133 (2.210) (2.625) (2.254) (2.643) Married, born before 1958 -2.419 -0.136 -2.765 -0.472 (2.243) (2.738) (2.301) (2.854) Constant 2.082 1.093 1.004 -0.697 (4.597) (5.846) (4.541) (5.683) Demographic characteristics X X X X Labor controls X X Wealth controls X X N 338 338 338 338 R20.133 0.192 0.161 0.214 Notes: This table predicts the demographic determinants of the mean intra-household gender wealth gap in couples, where the gap is the IHS transformed difference between the male’s and the female’s net wealth. The sample comprises only households in which both members of the couple are native-born Austrians, and among them, only those households who indicated an unequal distribution within the couple. ∆indicates the difference between the man’s and the woman’s variable value. The variables included in the wealth and labor market controls are described in the text. Standard errors in parentheses. ∗p < 0.1,∗∗ p < 0.05,∗∗∗ p < 0.01. Authors’ calculations on 2014 HFCS data. 29 6 Discussion and Conclusion One key dimension of gender inequality is the unequal distribution of wealth between men and women. This topic is still under-explored in the literature, and this paper contributes to the discussion by considering how age, education, marital status, fertility, and migration background are related to the intra-household gender wealth gap in Austria. A key take-away is that the demographic characteristics are indeed core determinants of the intra-household gender wealth gap. In particular, we show in bivariate analysis that wealth rises for both men and women with age and education, whereas a migration background is negatively correlated with wealth. However, the gender-specific wealth gap persists beyond the mitigating factors of age and education: women need to be considerably older for the gender wealth gap to become negative, and the wealth gap persists even when women are more educated than men. In contrast, being native-born appears to enable women to “catch up” regarding wealth ownership. Furthermore, we find some descriptive evidence that women in married couples may be accumulating less wealth than married men, on average, and that higher fertility correlates with a larger gender wealth gap. These results lend support to the labor market attachment hypothesis. Moreover, we used multivariate analysis to investigate these findings in more detail. OLS regressions show that intra-couple age differences, the education level of the couple, and the couple’s composition of migration background do, in fact, play a key role in explaining the gender wealth gap. Of the three, the most important determinant of the gender wealth gap is the migration background. When looking only at native-born couples, though, we find that the age difference within a couple, and to a lesser extent the education of the couple, are significant determinants of the intra-household wealth gap. The results of this paper provide important insights into the role of intersectionality in the existence and size of a gender wealth gap. Intersectionality is the idea that identity matters in social and economic outcomes in multidimensional ways: there are not just wealth differences between men and women, for example, but there are even larger gaps in the wealth holding of immigrant women and native-born men. Economic disadvantages are 30 thus multi-dimensional; the analysis in this paper helps to identify the aspects of identity that are related to wealth inequality within the couple. Since this is the first investigation of demographic-specific explanations of the gender wealth gap within households, many research questions remain open. First and foremost, our results beg the question of whether they apply to other countries. Second, a more indepth analysis of the exact conditions of migration and the disadvantages facing immigrant women in Austria would help to explain the strong results regarding migration and the gender wealth gap. Third, other data could potentially move past a major question mark in this study. The gender wealth gap as measured in our data depends on people and households acknowledging to an interviewer that their household resources are held unequally. Register data and interviews could potentially help provide more information about the existence and extent of intra-household wealth inequality. 31 Bibliography Addo, Fenaba R. & Daniel T. Lichter (2013). “Marriage, Marital History, and Black – White Wealth Differentials Among Older Women”. In: Journal of Marriage and Family 75.2, pp. 342–362. Balestra, Carlotta & Richard Tonkin (2018). Inequalities in household wealth across OECD countries: Evidence from the OECD Wealth Distribution Database. Tech. rep. 88. OECD Working Paper. Bauer, Thomas K., Deborah A. Cobb-Clark, Vincent A. Hildebrand, & Mathias G. Sinning (2011). “A Comparative Analysis of the Nativity Wealth Gap”. In: Economic Inquiry 49.4, pp. 989–1007. doi:10.1111/j.1465-7295.2009.00221.x. Chang, Mariko Lin (2010). Shortchanged: Why Women Have Less Wealth and What Can Be Done About It. Oxford University Press. Cupák, Andrej, Pirmin Fessler, Alyssa Schneebaum, & Maria Silgoner (2018). “Decomposing gender gaps in financial literacy: New international evidence”. In: Economics Letters 168, pp. 102–106. doi:https://doi.org/10.1016/j.econlet.2018.04.004. Deere, Carmin Diana & Cheryl R. Doss (2006). “The gender asset gap: What do we know and why does it matter?” In: Feminist Economics 12.1-2, pp. 1–50. Floßmann, Ursula (2006). Frauenrechtsgeschichte. Linzer Schriften zur Frauenforschung. Floßmann, Ursula (2008). Österreichische Privatrechtsgeschichte. Springer. Frémeaux, Nicolas & Marion Leturcq (2020). “Inequalities and the individualization of wealth”. In: Journal of Public Economics 184, pp. 104–145. doi: https://doi.org/ 10.1016/j.jpubeco.2020.104145. Frémeaux, Nicolas & Marion Leturcq (forthcoming). “Wealth accumulation across couples in France”. In: European Journal of Population Studies. Gobillon, Laurent & Matthieu Solignac (2019). “Homeownership of immigrants in France: selection effects related to international migration flows”. In: Journal of Economic Geography 20.2, pp. 355–396. doi:10.1093/jeg/lbz014. Goldberg, Lewis R, Dennis Sweeney, Peter F Merenda, & John Edward Hughes Jr (1998). “Demographic variables and personality: The effects of gender, age, education, and ethnic/racial status on self-descriptions of personality attributes”. In: Personality and Individual differences 24.3, pp. 393–403. Grabka, Markus, Jan Marcus, & Eva Sierminska (2015). “Wealth distribution within couples”. In: Review of the Economics of the Household 13.2, pp. 459–486. doi: https: //doi.org/10.1007/s11150-013-9229-2. Grinstein-Weiss, Michal, Yeong Hun Yeo, Min Zhan, & Pajarita Charles (2008). “Asset holding and net worth among households with children: Differences by household type”. In: Children and Youth Services Review 30.1, pp. 62–78. Hansen, Randall (2003). “Migration to Europe since 1945: Its History and its Lessons”. In: The Political Quarterly 74.s1, pp. 25–38. Household Finance and Consumption Network (HFCN) (2017). Household finance and consumption network: About the survey.url: http://www.ecb.eu/home/html/ researcher_hfcn.en.html. Household Finance and Consumption Network (HFCN) (2019). Report about the Research Activities of the HFCN since the Release of the Wave 2 Dataset.url: https://www. ecb.europa.eu/home/pdf/research/hfcn/201903_Research_Report_HFCS.pdf? 4f69e44965732d62284015b274bab529. 32 Kapelle, Nicole & Philipp M. Lersch (Feb. 2020). “The Accumulation of Wealth in Marriage: Over-Time Change and Within-Couple Inequalities”. In: European Sociological Review 36.4, pp. 580–593. doi:10.1093/esr/jcaa006. Keister, Lisa A. (2003). “Sharing the wealth: The effect of sibling on adults’ wealth ownership”. In: Demography 40. Kleven, Henrik, Camille Landais, Johanna Posch, Andreas Steinhauer, & Josef Zweimüller (May 2019). “Child Penalties across Countries: Evidence and Explanations”. In: AEA Papers and Proceedings 109, pp. 122–26. doi:10.1257/pandp.20191078. Kravitz, David A (2004). “Affirmative action”. In: Encyclopedia of applied psychology 1, pp. 65–77. Lehner, Oskar (1987). Familie - Recht - Politik. Springer. Lersch, Philipp M. (2017). “The Marriage Wealth Premium Revisited: Gender Disparities and Within-Individual Changes in Personal Wealth in Germany”. In: Demography 3. Lutz, Wolfgang (2010). “Education will be at the heart of 21st century demography”. In: Vienna Yearbook of Population Research 8, pp. 9–16. Maroto, Michelle (2018). “Saving, Sharing, or Spending? The Wealth Consequences of Raising Children”. In: Demography (55), pp. 2257–2282. Modigliani, Franco (1966). “The life cycle hypothesis of saving, the demand for wealth and the supply of capital”. In: Social Research, pp. 160–217. Muckenhuber, Mattias, Miriam Rehm, & Matthias Schnetzer (2021). The Migrant Wealth Gap at the Household Level: Evidence From RIF Regressions for Austria. Tech. rep. 13. IfSO Working Paper. Neelakantan, Urvi & Yunhee Chang (2010). “Gender differences in wealth at retirement.” In: American Economic Review: Papers and Proceedings 100, pp. 362–367. Painter, Matthew, Adrienne Frech, & Kristi Williams (2015). “Nonmarital Fertility, Union History, and Women’s Wealth”. In: Demography 1. Pfeffer, Fabian T. (2018). “Growing Wealth Gaps in Education”. In: Demography 55.3, pp. 1033–1068. doi:10.1007/s13524-018-0666-7. Ponthieux, Sophie & Dominique Meurs (2015). “Gender Inequality”. In: Handbook of Income Distribution. Ed. by Anthony B. Atkinson & François Bourguignon. Vol. 2. North Holland: Elsevier. Chap. 12, pp. 981–1146. Ruel, Erin & Robert M. Hauser (2013). “Explaining the gender wealth gap”. In: Demography 50.4, pp. 1155–1176. Schmidt, Lucie & Purvi Sevak (2006). “Gender, marriage, and asset accumulation in the United States.” In: Feminist Economics 12.1-2, pp. 139–166. Schneebaum, Alyssa, Miriam Rehm, Katharina Mader, & Katarina Hollan (2018). “The gender wealth gap across European countries”. In: Review of Income and Wealth 64.2, pp. 295–331. Sierminska, Eva, Joachim Frick, & Markus Grabka (2010). “Examining the gender wealth gap”. In: Oxford Economic Papers 62 (4), pp. 669–690. Sierminska, Eva, Daniela Piazzalunga, & Markus Grabka (2018). Transitioning towards more equality? Wealth gender differences and the changing role of explanatory factors over time. Tech. rep. 2018-18. LISER Working Paper Series. Vespa, Jonathan & Matthew A. Painter (2011). “Cohabitation History, Marriage, and Wealth Accumulation”. In: Demography 48.3, pp. 983–1004. Yamokoski, Alexis & Lisa A. Keister (2006). “The wealth of single women: Marital status and parenthood in the asset accumulation of young baby boomers in the United States.” In: Feminist Economics 12.1-2, pp. 167–194. 33 Appendix Figure A1: Gender wealth gap within couples by percentiles (in %) Notes: Weights and multiple imputations taken into account. No values for percentiles 11, 55, 58, 82, 83 due to varying sets of implicates. Gender wealth gap is defined as the difference between a man’s and woman’s net wealth compared to the couple’s total net wealth. Authors’ calculations on 2014 HFCS data. A1 Table A1: Further descriptive statistics Females Males Share Employee 47.9 51.6 Share Employer 1.9 4.9 Share self-employed 3.7 4.3 Share unemployed 2.0 1.9 Share not in LF 17.1 1.1 Share retired 27.5 36.1 Part-time share 45.2 5.7 Full-time share 54.8 94.3 Work attachment history 0.70 0.88 Years worked 21.2 29.7 Share with employee income 52.9 55.0 Average value employee income 21,440 35,473 Share with self-employment income 7.4 10.8 Average value self-employment income 16,829 39,626 Share with other income 31.2 39.5 Average value other income 13,368 23,411 Total income 16,757 33,029 Share of households with inheritance 31.2 N observations 1,503 1,503 Notes: Authors’ calculations on 2014 HFCS data. Figure A2: Comparison of women’s and men’s education levels in couples with their median net wealth Notes: Weights and multiple imputations taken into account. Values in brackets refer to the shares of the respective couples. Authors’ calculations on 2014 HFCS data. A2 uni-due.de/soziooekonomie/wp Institute for Socio-Economics University of Duisburg-Essen Lotharstr. 65 47057 Duisburg Germany uni-due.de/soziooekonomie [email protected] ifso working paper ifso working papers are preliminary scholarly papers emerging from research at and around the Institute for Socio-Economics at the University of Duisburg-Essen. All ifso working papers at uni-due.de/soziooekonomie/wp ISSN 2699-7207 This work is licensed under a Creative Commons Attribution 4.0 International License Institute for Socio-Economics University of Duisburg-Essen Lotharstr. 65 47057 Duisburg Germany uni-due.de/soziooekonomie [email protected] ifso working paper ifso working papers are preliminary scholarly papers emerging from research at and around the Institute for Socio-Economics at the University of Duisburg-Essen. All ifso working papers at uni-due.de/soziooekonomie/wp ISSN 2699-7207 This work is licensed under a Creative Commons Attribution 4.0 International License