Gender and the Economic Consequences of Divorce in the United States: Variation by Race and Ethnicity

Abstract

Gender differences in the economic consequences of divorce are well established and reveal how a traditional gender-based division of paid and unpaid labor can render women economically vulnerable when marriages dissolve. Guided by intersectional approaches that recognize systemic racism and entrenched gender inequality, we assess how race/ethnicity and gender intersect to pattern the economic consequences of divorce. Drawing on 28 waves of the National Longitudinal Survey of Youth 1979 (NLSY79), we conduct a descriptive analysis of the short-term economic impact of marital disruption for non-Hispanic Black women and men, Hispanic women and men, and non-Hispanic white women and men. Our bivariate and multivariable results indicate that the economic consequences of marital disruption vary substantially on the basis of race/ethnicity and gender. All groups of women fare worse than men in postdissolution economic wellbeing and in changes in economic status. Black and Hispanic men and the three groups of women fare worse than white men, with Black women experiencing the highest levels of economic precarity.

Women’s economic wellbeing outside of marriage is of longstanding interest to social scientists and policy makers. Catalyzed by newly available longitudinal data in the 1970s, researchers began tracing the impact of divorce on women’s and men’s economic wellbeing (Corcoran, 1979; Duncan & Hoffman, 1985; Hoffman, 1977; Hoffman & Duncan, 1988; Holden & Smock, 1991; Peterson, 1996). The conclusion is clear: divorce has a substantial negative impact on women’s economic situations, with results indicating a less severe impact for men. Despite advances in women’s educational and occupational attainment since the 1970s, studies continue to document the economic costs of divorce for women (Avellar & Smock, 2005; Bianchi et al., 1999; Brüggmann & Kreyenfeld, 2023; De Vaus et al., 2017; Leopold, 2018; Lin & Brown, 2021; McKeever & Wolfinger, 2001; Smock, 1993; Smock et al., 1999; Tach & Eads, 2015).

Thus, gender disparities have been the focus of this literature either explicitly or implicitly, consistent with rising interest in the “feminization of poverty” (Pearce, 1978). Some studies make explicit empirical comparisons between men and women (e.g., Burkhauser et al., 1991; Lin & Brown, 2021; Leopold, 2018; Smock, 1994) while others focus solely on women (e.g., McKeever & Wolfinger, 2001; Tach & Eads, 2015) and at least one solely on men (McManus & DiPrete, 2001). Even in studies focused on single-sex samples, theories of gender inequality often are used to frame or motivate the research (e.g., Tach & Eads, 2015).

Missing from this literature is sustained and explicit attention to race and ethnicity in addition to gender. Our study advances understanding of gender inequality in the economic consequences of divorce¹ by assessing whether and how race/ethnicity and gender intersect to affect economic wellbeing in a large nationally representative sample of U.S. adults born in the late 1950s through early 1960s. Prior studies sometimes provide descriptive contrasts but do not focus on them, or relegate race/ethnicity to a control variable (Avellar & Smock, 2005; Lin & Brown, 2021; McKeever & Wolfinger, 2001; Smock, 1994). Guided by intersectional approaches that recognize systemic racism and gender inequality, we propose that holding race constant is not sufficient to understand the complex ways that race/ethnicity and gender may intersect to shape adults’ economic wellbeing in the wake of divorce. Drawing on 28 waves of the National Longitudinal Survey of Youth 1979 (NLSY79), we address this gap by assessing the short-run economic consequences of separation and divorce for six groups: Hispanic men and women, non-Hispanic white men and women, and non-Hispanic Black men and women. Understanding the economic impacts of divorce across diverse population subgroups is an important query, as this may be a critical yet neglected mechanism contributing to the well-documented economic precarity of children in single-parent families as well as Black and Hispanic women and men aging without a spouse (e.g., Carr, 2020; McLanahan & Percheski, 2008; Wolfe & Thomeer, 2021).

BACKGROUND

The protective effects of marriage and deleterious effects of marital dissolution for women’s economic wellbeing are well-documented (Avellar & Smock, 2005; Bianchi et al., 1999; Edin & Lein, 1997; Sidel, 1986; White & Rogers, 2000). Extensive research confirms that the economic toll of marital disruption is worse for women than men, with the gap attributed largely to a gender-typed division of labor in the home and the labor market (e.g., Brüggmann & Kreyenfeld; Burkhauser et al., 1991; Duncan & Hoffman, 1985; Lin & Brown, 2021; Peterson, 1996; Smock, 1993). Men lose less family income than women upon dissolution due to the persistent gender gap in wages and married women’s tendency to reduce their labor supply (and consequently, personal earnings) in response to family responsibilities. Men also typically experience postdissolution gains in economic measures that account for household size (e.g., per capita income; income-to-needs ratio) because children live primarily with their mothers after divorce (Berman & Daneback, 2022).

Women’s labor force attachment and men’s involvement in domestic labor have increased throughout the late 20^th and early 21^st centuries, yet the lion’s share of unpaid household work is still performed by women, including caregiving for both children and aging parents, household management, emotional and cognitive labor, and housework (Carlson et al., 2021; Ciciolla & Luthar, 2019; Daminger, 2019; Hochschild & Machung, 2012; Hook, 2017; Mattingly & Sayer, 2006; Rose & Hartmann, 2020). Upon parenthood, women take time off or reduce their employment hours more so than men, even among women strongly attached to the labor market (e.g., Blair-Loy, 2000).

Gender differences in the economic consequences of divorce and the mechanisms implicated in these differences are well-established, although few studies have explicitly examined whether and how these patterns vary on the basis of race and ethnicity (e.g., Hoffman & Duncan, 1988; Harkness, 2022; Tach & Eads, 2015). While some studies present race/ethnic differences in descriptive statistics (e.g., Brown & Lin, 2020; Duncan & Hoffman, 1985; Hoffman, 1977; Holden & Kuo, 1996; Smock, 1993, 1994) or control statistically for race/ethnicity in multivariable analyses (e.g., Bianchi, et al., 1999; Drewianka & Meder, 2020; McKeever & Wolfinger, 2001; Smock, 1993, 1994), we know of no studies that directly evaluate whether and how race/ethnicity and sex intersect in patterning the economic consequences of divorce.

This limited attention to race or ethnic differences may reflect that marriage, and consequently marital dissolution, has become less common among Black and Hispanic adults in the United States.² Researchers have documented a contemporary “retreat” from marriage, highlighting the widening Black-White gap in marriage rates (Lichter et al., 1992; Raley, 1996; Raley et al., 2015; Smock & Schwartz, 2020). If people are becoming less likely to marry in the first place, investigating the economic consequences of divorce may seem less compelling to social scientists.

Yet marriage remains the most common relationship context for current cohorts of midlife adults across all racial and ethnic groups. For members of the NLSY79 cohort, men and women born between 1957 and 1964, 68% of Blacks, 90% of whites, and 85% of Hispanics had married by age 46 (Aughinbaugh et al., 2013; see also Cohen, 2022). Divorce rates are remarkably similar across racial and ethnic groups. Among members of the NLSY79 cohort, the proportion of ever-married women who had divorced by age 46 was 48% among Blacks, 44% among whites, and 47% among Hispanics (Aughinbaugh et al., 2013).

A second possible reason for limited attention to racial and ethnic disparities is that traditional conceptual frameworks used to understand the gendered economic consequences of divorce are based implicitly on a white, middle-class model of gendered allocation of labor in the home and market (e.g., Wight et al., 2013). For example, “relative resource” frameworks propose that the relative economic positions of men and women shape how much time they invest in household versus market labor. Classic sociological approaches typically emphasize social power within the marital dyad, such that the partner with higher levels of education and income can translate these resources into power, which can be used to evade domestic chores (Blood & Wolf, 1960; Brines, 1994). Even in dual-earner couples, women tend to be economically dependent on their spouses and do not have the power to negotiate their level of domestic work (e.g., Greenstein, 1996).

Classic economic perspectives similarly emphasize each partner’s “comparative advantage” in the dyad (Becker, 1965, 1985). Households are believed to divide labor to maximize efficiency by having one partner specialize in market work and the other in domestic work. Because men typically earn more in the labor market, couples allocate household labor to women and market labor to men to maximize “efficient” household operations (Killewald & Gough, 2013; Lundberg & Pollak, 2003). Therefore, marital dissolution would have a dramatic effect on the economic wellbeing of women, particularly those who specialized in household labor. The loss of spousal income and the erosion of one’s labor market experience and earnings potential, as well as the gender wage gap (e.g., Cha & Weeden 2014), are key mechanisms accounting for women’s economic precarity after marital dissolution.

THE PRESENT STUDY

Our goal is to document the extent to which well-established gender differences, whereby women experience lower levels and larger reductions in economic wellbeing than men, vary on the basis of race and ethnicity. Intersectional approaches emphasize that social categorizations such as race, ethnicity, and gender create overlapping and interdependent systems of structural advantage and disadvantage (Cho et al., 2013; Collins, 1993; McCall, 2005).

We evaluate both absolute levels of economic wellbeing as well as economic changes individuals experience when marriages end, as both are important for understanding the economic toll of separation and divorce. Our analyses use four income-based measures: personal income, family income, income-to-needs, and poverty risk. The first largely reflects one’s own earnings although it also includes non-earnings income such as public assistance or child support. The second is also earnings-based but will decline upon a change from two earners to one, as ex-spouses typically reside apart postdissolution. Income-to-needs, and its derivative poverty risk, consider family size as well as income.

Personal income is highly stratified by both gender and by race/ethnicity. Women earn less than men across the board, due to factors noted above, discrimination, and women’s disproportionate responsibility for caring and domestic labor (Perry-Jenkins & Gerstel, 2020; Rose & Hartmann, 2020). In addition, Black and Hispanic men and women earn less than their white counterparts. Among year-round, full-time workers in 2019, white women earned $51,324 whereas white men earned $65,208. Hispanic women earned the least ($36,110), largely due to their concentration in low-paying service sector jobs, and Black women earned $41,098. Hispanic and Black men earned in the low to mid-40’s (Institute for Women’s Policy Research, 2020).

Racial/ethnic income disparities stem from several factors. A history of systemic racism in the U.S. has created a context in which Black men and women have both lower levels of education and smaller economic returns per year of schooling, relative to whites (e.g., Cheng et al., 2019; Hout, 2012). Higher rates of poverty, and exposure to racism and discrimination have limited Black men’s and women’s labor market opportunities and suppressed their earnings. Hispanics’ lower earnings relative to whites is also due to systemic factors including lower levels of formal education, linguistic isolation, discrimination, and xenophobia, although the magnitude of these gaps varies on the basis of country of origin (Guzman, 2020).

In terms of changes in the wake of marital disruption, we expect possible increases in women’s personal incomes because they may not have been employed or employed only part-time or part-year while married; they may thus increase their labor market engagement upon marital disruption. In 2019, only 57% of white, 63% of Black, and 54% of Hispanic married women were employed (Bureau of Labor Statistics, 2021). Overall, however, women’s post-divorce personal incomes be less than men’s, and Black and Hispanic men will earn less than white men.

However, we anticipate different patterns for family income. Past studies indicate that both men’s and women’s family incomes decline when divorce occurs due to the loss of the former spouse’s income (Duncan & Hoffman, 1985; Hoffman, 1977; McManus & DiPrete, 2001; Smock, 1993). Women typically experience steeper declines than men due to the gender wage gap and married women’s lesser labor market involvement. It is possible that Black women will experience more modest declines than white women, and Black men will experience larger declines than white men. Black husbands and wives have greater earnings parity relative to their white counterparts, leading to a more modest loss for Black women and more substantial loss for Black men (Winslow-Bowe, 2009). Black couples also tend to maintain a more equitable division of household labor and hold more egalitarian attitudes compared to whites or Hispanics (Dow 2016, 2019; John & Shelton, 1997; Kane, 2000; Sayer & Fine, 2011; Small, 2023). Hispanic women may experience the steepest declines; a study based on data from 1989 and 1990 indicates that Hispanic wives’ earnings comprise the smallest portion of family income relative to whites and Blacks (23% compared to 28% for whites and 36% for Blacks), thus they stand to lose the most upon divorce (Choi, 1999; see also Hayghe, 1993).

The income-to-needs ratio and poverty risk gauge one’s standard of living by accounting for income relative to the number of people that income must support. The gendered division of domestic labor and caregiving results in separated and divorced women retaining primary physical custody of minor children more often than men (Grall, 2020). Consistent with past research, we expect women to fare more poorly than men across the three racial/ethnic groups, both in postdisruption economic precarity and changes in wellbeing. While we expect that whites will fare better than Hispanics or Blacks in absolute terms, we expect men’s income-to-needs to rise upon divorce and women’s to decrease, consistent with past studies (e.g., Bianchi et al., 1999). Similarly, we expect women’s postdisruption poverty risk, a binary indicator of whether one’s income-to-needs is less than 1.0, to be substantially higher than men’s across the three racial/ethnic groups, and for Black and Hispanic women to fare the worst.

The next section describes our data, measures, and analyses. We then present results in two parts. First, we contrast pre- and postdisruption economic status as well as changes in economic status catalyzed by marital disruption for the six gender-racial/ethnic groups. Second, we estimate multivariable quantile regression models at the median of two indicators of postdisruption economic wellbeing - personal income and income-to-needs - as a second step in identifying the existence and direction of intersectionality. They also evaluate the extent to which intersectionality persists after adjusting for human capital and family characteristics that are well-established correlates of post-divorce economic wellbeing (Bianchi et al. 1999). Human capital theories broadly posit that rewards from market work depend on one’s “investments,” such as years of education and work experience (Becker, 1985). We use four measures of human capital (i.e., educational attainment, work experience, the number of employment interruptions, and predisruption employment status). We also adjust for number of children in the postdivorce household, which is particularly salient for women in this cohort; they were far more likely than their ex-spouses to have primary physical custody of minor children. Finally, we adjust for social background and adolescent health, as they are associated with both the risk of dissolution and economic wellbeing (Conger et al., 2010; Sbarra, 2015).

METHODS

Data and Sample

Data are from the National Longitudinal Survey of Youth 1979 (NLSY79), a nationally representative sample of 12,676 non-institutionalized men and women born between 1957 and 1964 in the United States, including oversamples of Blacks, Hispanics and economically disadvantaged whites. Interviews were conducted annually from 1979–94 and biennially thereafter. This study uses the 1979–2018 waves. Retention rates are high with 69% of the original sample participating in 2018. Across waves, attrition is fairly random with respect to marital status and sociodemographic characteristics (Aughinbaugh et al., 2017).

The large diverse sample and longitudinal nature of the NLSY79 make it ideally suited for a prospective intersectional analysis of the short-term economic consequences of divorce. Most notably, the survey collects repeated measures of marital status, labor force participation, income from all sources, and information on spouses and other household members, enabling us to prospectively trace the impact of marital status changes on subsequent economic changes.

We focus on the short-term ramifications of marital disruption because, at least for women, the initial economic shock of marital disruption is often prolonged unless they remarry (Brüggmann & Kreyenfeld, 2023; Duncan & Hoffman, 1985; Holden & Smock, 1991; Leopold, 2018). A focus on short-term effects also reduces the likelihood that results are biased by race and gender differences in repartnering shortly after dissolution, and sample attrition. We define marital disruption as either a legal divorce or separation due to discord; economically disadvantaged persons tend to delay or forgo legal divorce due to the costly legal fees (McCarthy, 1978; Tumin et al., 2015). Overall, most disruptions are from first marriages (92%). We include higher-order marriages (n =232) to maximize our sample size; of these, nearly all (n=210) are second marriages.³

When a respondent reports being married in a particular survey wave and separated or divorced in a subsequent survey wave, we define this change as marital disruption and identify the time of disruption as the interview year in which separation or divorce is first reported. “Predisruption” refers to the final year of marriage and “postdisruption” refers to the survey wave immediately following the interview in which separation or divorce is reported. In 1994, the NLSY transitioned from annual to biennial interviews. For marital disruptions occurring between 1980 and 1993, the predisruption observation is the prior interview year and the postdisruption observation is the subsequent year. For divorces reported in 1994, the predisruption observation is 1993 and the postdisruption observation is 1996 because data were not collected in 1995. For respondents with marital disruptions occurring in 1996 or thereafter, pre- and postdisruption observations are captured roughly two calendar years before and after the disruption was reported, respectively.

We performed sensitivity analyses to assess the potential impact of the change in NLSY periodicity. We re-estimated all analyses for the entire period (1979–2018) as if we had only biennial data throughout. Although results based on biennial data revealed slightly higher levels of post-divorce income and slightly smaller impacts of dissolution, the direction and magnitude of subgroup differences were similar (results available from authors). In multivariable analyses, we include a binary variable indicating whether the respondent’s postdisruption interview occurred after the switch to biennial interviews.

Our analytic sample is limited to respondents reporting separations or divorces between 1980 and 2016 because those reporting a disruption at the 1979 interview have no predisruption observation whereas those reporting a disruption in 2018 have no postdisruption observation. We also exclude respondents who are repartnered (i.e., cohabiting or remarried) at the postdisruption observation (n = 408). In addition, to ensure that respondents did not marry and divorce more than once between the pre- and postdisruption observations, we examined individual respondent’s reported changes in marital status and excluded a small number of cases (n = 13) with more than one divorce, or cases in which we could not determine a respondent’s actual marital history (n = 30).⁴ Finally, we excluded 531 respondents with missing earnings pre- or post-dissolution (or spouse earnings for the predisruption year) although zeros are valid and considered non-missing.

Our final analytic sample consists of 2,891 individuals in different-sex marriages.⁵ This includes 732 non-Hispanic white, 352 non-Hispanic Black, and 230 Hispanic men and 906 non-Hispanic white, 384 non-Hispanic Black, and 287 Hispanic women. Most respondents initially married in their early to mid-20s (M = 25 for men and 23 for women). Mean age of divorce was 34 (range 19–59) for men and 32 (range 16–59) for women.

Measures

Dependent Variables

We focus on four dimensions of economic wellbeing before and after marital disruption: personal income, family income, income-to-needs-ratio, and poverty status. The NLSY79 collects information at every wave on income from wages/earnings, farms, businesses, unemployment or veterans’ benefits, alimony or child support, income from food stamps and other public assistance programs, and other sources. We include all sources in our measures. At each wave, all income is reported for the prior calendar year, thus the pre- and postdisruption interviews capture a full year in which the respondent was married or separated/divorced.

Personal income includes respondent’s income from all sources, excluding the income of the spouse and any co-residing family members. Family income is the sum of income from all sources received by respondents, spouses (in the predisruption observation only), and any co-residing adult family members. Income-to-needs is the ratio of family income to the federal poverty line (FPL as determined by family size). Poverty status indicates whether an individual’s income-to-needs is below 1.0, indicating an individual or family is living beneath the FPL. We present income amounts in 2022 dollars, adjusted for inflation using CPI-U-RS multiplier.

Independent Variables

Our focal predictor is a six-category indicator of one’s gender and race/ethnicity. We classified respondents based on self-reported gender (male or female), and race/ethnicity (non-Hispanic white, non-Hispanic Black, or Hispanic).

Human capital covariates include educational attainment and employment characteristics. Educational attainment, measured at the predisruption interview, refers to the highest level of completed schooling: high school or less (reference group), some college, and bachelor’s degree or more. We also include total years of work experience, the number of employment interruptions, and employment status at the predisruption wave (not employed is reference group). Total years of work experience is calculated as the number of years a respondent was employed, either part-time or full-time, from 1979 through the predisruption interview. Consistent with Bureau of Labor Statistics (BLS) guidelines, we consider individuals working 50 weeks or more per year as working a full year. Total number of employment interruptions prior to divorce or separation is the total number of transitions from employment or active military duty to unemployment or out-of-the-labor force status.

Number of children living with the respondent at the postdisruption interview ranges from 0 to 10. Because the NLSY79 does not obtain direct measures of legal or physical custody arrangements, we carried out supplemental bivariate analyses and results approximate national data on gender differences in child custody and living arrangements. Across all race/ethnic groups, men report fewer children in the household at the postdisruption interview compared to women (0.3 and 1.25 children, respectively).^6,7 Social background is operationalized as mother’s education (high school or less, some college, or bachelor’s or higher). Health is measured with a baseline (1979) indicator of whether respondents reported a health concern that could affect their ability to work. As noted earlier, multivariable analyses adjust for whether a marital disruption occurred after the change to a biennial interview schedule.

Analytic Plan

We first carry out bivariate analyses, contrasting race/sex groups with respect to the four economic wellbeing indicators. We present median personal income, family income and income-to-needs, rather than means, because mean values are affected by skewness and thus are biased by extreme values. To test for statistically significant differences across subgroups, we use the Mann-Whitney test which evaluates whether two samples are drawn from the same population.⁸ For poverty risk, we present and contrast the percentage of respondents living below the federal poverty line.

We also calculate the median percentage change experienced by individuals.⁹ Although median predisruption and postdisruption incomes levels reveal the absolute level of economic resources before and after divorce, they do not describe the typical change, whether positive or negative, that individuals experience. For each respondent, we calculate the percentage change between pre- and postdivorce personal income, family income, and the income-to-needs ratio. We then take the median of these values.

Second, we carry out multivariable analyses to evaluate the extent to which race/sex differences persist net of covariates. We chose two of the four measures as dependent variables; one to reflect respondent’s labor market position and the other to tap familial context, both important for gauging economic wellbeing (Curtis, 1986).¹⁰ For the former, we rely on personal income and the latter, the income-to-needs ratio. We use personal income rather than family income because it directly measures individual economic standing. Although both poverty and income-to-needs account for family size, income-to-needs is more nuanced because it captures the full range of economic status.

Preliminary analyses showed subgroup differences in skewness of the dependent variables, and we thus estimate quantile regressions at the median (Hao & Naiman, 2007). We know of no formal test for groups differences in skewness, yet we take distributional differences among race/ethnic and gender groups to be real and chose not to normalize them by using log transformations.¹¹ Interpretation of coefficients is similar to ordinary least squares (OLS) regression although the models predict conditional medians rather than means. We present two models for each dependent variable. The baseline model (Model 1) includes only race/ethnicity and gender. The fully adjusted Model 2 incorporates all covariates. To capture intersectionality, it is important to examine every race/ethnic and gender contrast; we thus estimate models altering reference groups for each dependent variable.

RESULTS

Descriptive Statistics

Table 1 shows that the sample median postdisruption personal income is roughly $44,000 and the mean is $55,000. Median income-to-needs after divorce is 2.5 with a mean of about 3.5. Whites account for roughly 80% of the sample, Blacks account for 13%, and Hispanics 7%; this distribution is comparable to Census data for members of the NLSY79 cohort (Rothstein et al., 2019). Sixteen percent have completed a B.A. or more and 63% have a high school diploma or less. Median and mean years of work experience are 8 and nearly 10 years, respectively, with an average of just one work interruption (range 0–6). The mean number of children in the postdisruption household is 0.77. Only 5% of respondents’ mothers completed college, with 82% having a high school diploma or less. Very few respondents report a health-limiting condition at baseline (5%) and 40% of the sample separated or divorced after the change to biennial interviews in 1994.¹²

Table 1.

Descriptive Statistics for Measures Used in Multivariable Analysis NLSY79, 1979–2018

	Median/Proportion	Mean	SD
Dependent variables
Postdisruption personal income	$43,530	$54,666	$53,584
Postdisruption income-to-needs	2.49	3.46	3.77
Independent variables
Race/ethnicity and gender group
White men	0.37
Black men	0.06
Hispanic men	0.03
White women	0.44
Black women	0.07
Hispanic women	0.04
Covariates
Highest level of education completed
High school or less	0.63
Some college	0.21
Bachelor’s degree or more	0.16
Years of work experience	8.19	9.96	7.84
Number of work interruptions	1.00	0.98	1.1
Respondent is employed	0.87
Number of children living in household at the postdisruption interview	0.00	0.77	1.04
Mother’s highest level of education completed
High school or less	0.82
Some college	0.10
Bachelor’s degree or more	0.08
Respondent has a work-limiting health condition	0.05
Postdisruption interview took place after 1994	0.40
N of cases	2,981

Descriptive statistics are weighted with the NLSY79 baseline sample weight. Sample is restricted to currently separated or divorced men and women who are not remarried or cohabiting at the postdisruption interview. The dependent variables and children are reported at the postdisruption interview, human capital measures from baseline through the predisruption interview, and other covariates were ascertained at the baseline interview.

Bivariate Results

Table 2 shows the four economic status measures before and after marital disruption by gender and race/ethnicity: median annual family income, personal income, and income-to-needs ratio, and percent in poverty.

Table 2:

Economic Wellbeing Before and After Marital Disruption, by Gender and Race/Ethnicity, NLSY79, 1979–2018

	Men		Women
	Predisruption	Postdisruption	Predisruption	Postdisruption
Median Personal Income
Black	$42,459	$41,062	$28,567	$31,588
Hispanic	$44,810	$49,390	$27,663	$31,911
White	$54,750	$54,850	$26,258	$36,934
Median Family Income
Black	$62,086	$43,367	$61,803	$32,508
Hispanic	$67,786	$53,004	$68,258	$35,184
White	$80,031	$58,370	$81,915	$40,411
Median Income-to-Needs
Black	2.56	2.74	2.30	1.40
Hispanic	2.54	3.09	2.52	1.57
White	3.27	3.65	3.36	2.08
Percent in Poverty
Black	13%	15%	14%	35%
Hispanic	12%	16%	15%	32%
White	8%	9%	7%	23%

Income is presented in 2022 dollars. Predisruption represents the wave prior to separation or divorce. Postdisruption represents the wave after separation or divorce. The sample is restricted to divorced or separated men and women who are not remarried or cohabiting at the postdisruption wave. Sample size includes 732 white, 352 Black, and 230 Hispanic men, and 906 white, 384 Black, and 287 Hispanic women. Data are weighted using NLSY79 baseline sample weights.

Personal Income

Prior to divorce, men’s personal income is significantly higher than women’s across racial/ethnic groups (see Table 5). We also detect a racialized pattern of racial/ethnic inequality, wherein white men fare better than Black and Hispanic men, with median values of $54,750, $42,459, and $44,810, respectively. The difference between white and Black men’s baseline personal income is statistically significant, although we do not detect statistically significant differences between Hispanic men and either white or Black men. Patterns are similar for men’s postdivorce personal incomes with levels differing only slightly from predisruption personal income: white men report the highest incomes, followed by Hispanic and Black men ($54,850, $49,390, and $41,062, respectively). Only the Black-white difference is statistically significant.

Table 5.

Statistically Significant Differences in Medians of Economic Wellbeing by Race/Ethnicity and Gender, NLSY79, 1979–2018

	Hispanic Women (A)	Hispanic Men (B)	White Women (C)	White Men (D)	Black Women (E)	Black Men (F)
Predisruption
Personal income	B*** D*** F***	A*** C*** E***	B*** D*** E** F***	A*** C*** E*** F**	B*** C** D*** F***	A*** C*** D** E***
Family income	E**	C†	B† E*** F**	E** F**	A** C*** D**	C** D**
Income-to-needs	C*** D**	C** D**	A*** B** E*** F***	A** B** E*** F**	C*** D*** F†	C*** E† D**
Postdisruption
Personal income	B*** D*** F***	A*** C*** E***	B*** D*** F***	A*** C*** E*** F***	B*** D*** F***	A*** C*** D*** E***
Family income	B** D*** F***	A*** C*** E*** F**	B*** D*** E** F***	A*** C*** E*** F***	B*** C** D*** F***	A*** B** C*** D*** E***
Income-to-needs	B*** C** D*** F***	A*** C*** D** E***	A** B*** D*** E*** F***	A*** B** C*** E*** F***	B*** C*** D*** F***	A*** C*** D*** E***

Significance is denoted as †p < 0.10. **p < 0.05. ***p < 0.001. Data are unweighted. The Mann-Whitney test was used for the continuous measures.

Women’s median personal incomes predisruption are much lower than men’s and the range is extremely narrow ($26,258 to $28,567). Among women in our sample, Blacks have the highest personal incomes and whites the lowest; this difference is statistically significant (Table 5). Although women uniformly show increases in personal income postdisruption, these levels are still significantly lower than their male counterparts across all three racial/ethnic subgroups. As Table 2 shows, postdisruption median personal incomes for Black and Hispanic women are somewhat higher than predisruption levels (about $3,000 to $4,000 higher) and white women’s are substantially higher by over $10,000.

Figure 1 displays subgroup differences in the median percentage change in personal income. Men typically experience negligible change (less than 3%) upon separation or divorce. In stark contrast, women’s personal incomes increase by 16%, 27%, and 37% for Blacks, Hispanics, and whites, respectively. These changes are due in part to changes in women’s employment status, with the largest changes among whites. The proportion of women working full time increased from 61% to 72% among whites, 67% to 72% among blacks, and 61% to 66% percent among Hispanics pre- and postdisruption (not in table). Among men, by contrast, levels of full-time employment remained relatively stable and high. Yet, as Table 1 suggests, women’s efforts to increase their labor supply immediately following dissolution are not sufficient to attain income parity with their male counterparts.

Fig. 1.

Median percent change in personal income from the predisruption to postdisruption observation.

Family Income

Whereas predisruption personal income varied dramatically by gender, predisruption family income shows patterns of stark racial inequality. Black married men and women reported the lowest family incomes predisruption, whites the highest, and Hispanic men and women in between. Black married men’s median annual family income before divorce is $62,086 compared to $67,786 and $80,031 for Hispanic and white men, respectively. Median predivorce family incomes are nearly the same for their female counterparts ($61,803, $68,258, and $81,915). For both men and women, Black-white differences are statistically significant. Among women, Hispanics also have significantly higher predisruption median family income relative to Blacks.

After divorce, family income is lower than predisruption levels for all groups, due to loss of spousal earnings, and these differences are especially marked for women due to lower personal incomes relative to men. Among men, median postdisruption family is highest among whites at $58,370, $53,004 among Hispanics, and considerably lower at $43,367 among Blacks. Black men’s income is significantly lower than their white and Hispanic counterparts. Across all three race/ethnic groups, women’s family incomes are significantly lower than men’s following marital disruption. Racial disparities among women follow a pattern similar to that observed among men, albeit at lower levels, with post-divorce family incomes of $40,411 (white), $35,184 (Hispanic), and $32,508 (Black); the Black-white gap is statistically significant (Table 5).

As Figure 2 shows, both men and women experience substantial drops in family income postdissolution yet women evidence drops considerably larger than those experienced by men. The magnitude of these drops also differs slightly on the basis of race/ethnicity. Black men experience declines of 29%, while white and Hispanic men see their family incomes drop by less than a quarter (−24% and −23%). In stark contrast, women experience declines roughly twice that of men’s (46–50%). These declines are similar to those reported in prior research (e.g., Avellar & Smock, 2005; McKeever & Wolfinger, 2001 but see Tach & Eads, 2015 who show more modest declines for women divorcing in the 2000s).

Fig. 2.

Median percent change in family income from the predisruption to postdisruption observation

Income-to-Needs and Poverty Risk

Children are more likely to reside primarily with their mothers than their fathers after a marriage ends, particularly for this cohort. Thus, measures that account for family size provide a critically important indicator of economic precarity, capturing one’s standard of living. Table 2 shows that women’s postdisruption standard of living in terms of income-to-needs and poverty risk is precarious and men’s is less so.

Prior to divorce, white men and women have the highest income-to-needs. Income-to-needs for Black and Hispanic married men and women are roughly 2.3 to 2.6 compared to about 3.3 to 3.4 for white men and women. White men and women fare significantly better than all other subgroups but do not differ significantly from one another (Table 5). After divorce, women’s median income-to-needs ranges from 1.4 to 2.08, while men’s ranges from 2.74 to 3.65.¹³ Significance tests indicate that white men are advantaged relative to the two other groups of men and all groups of women, and Hispanic and Black men fare better than all groups of women. Among women, there is similar evidence of racial inequality, as white women fare significantly better than Hispanic and Black women (Table 5).

Figure 3 illustrates subgroup differences in average change in standards of living following divorce. Income-to-needs increases for men, although the increase is considerably smaller among Black versus Hispanic and white men (4, 11 and 12%, respectively). Women, conversely, experience sizeable declines of more than one-third (−35% to −38%).

Fig. 3.

Median percent change in income-to-needs from the predisruption to postdisruption observation.

Poverty rates follow a similar pattern. As shown in Table 1, prior to divorce rates are low for all groups and lowest for whites; 7% to 8% of whites are living in poverty compared to 12% to 15% for Hispanic and Black men and women. Postdisruption, a gender divide emerges: 35% of Black, 32% of Hispanic, and almost one in four white women live in poverty. Among men, comparable percentages are 15%, 16%, and 9%, respectively. As shown in Figure 4, among women who were not poor prior to marital dissolution, 19–24% became poor afterwards. This is considerably higher than the percent of men becoming poor in the wake of divorce: among non-poor married men, 6% to 11% enter poverty after divorce.

Fig. 4.

Percentage of individuals entering poverty after marital disruption.

Multivariable Results

Tables 3 and 4 show quantile regression results for two measures of economic status, personal income (Table 3) and the income-to-needs ratio (Table 4). Model 1 in each table includes only the six category variable for race/ethnicity and gender and Model 2 includes the covariates from Table 1. Overall, the results indicate that race/ethnicity and gender simultaneously impact postdissolution economic wellbeing.

Table 3:

Quantile Regression of Personal Income in Survey Wave Following Marital Disruption, NLSY79, 1979–2018

	Model 1	Model 2
Race/ethnicity and gender group
White men (reference)
Black men	−8,551.17*** (2,271.76)	−6,237.03** (2,008.20)
Hispanic men	−312.79 (2,647.54)	−4,847.51** (2,335.97)
White women	−16,521.51*** (1,740.65)	−12,248.41*** (1,623.82)
Black women	−18,406.12*** (2,206.92)	−16,275.57*** (2,093.25)
Hispanic women	−17,955.48*** (2,439.30)	−11,784.87*** (2,279.54)
Predisruption human capital
Highest level of education completed
High school or less (reference)
Some college		9,901.52*** (1,464.38)
Bachelor’s degree or more		33,941.49*** (1,929.28)
Years of employment experience		1,123.64*** (113.93)
Number of employment interruptions		−2,488.50*** (597.05)
Respondent is employed		12,441.10*** (1,556.13)
Children
Number of children living in household at the postdisruption interview		2,263.46*** (602.34)
Controls
Mother’s highest level of education completed
High school or less (reference)
Some college		4,617.14** (2,171.52)
Bachelor’s degree or more		697.08 (2,460.94)
Respondent has a work-limiting health condition		861.27 (2,658.90)
Constant	50,032.56	24,405.39
Pseudo R2	0.0311	0.2002

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. We include, but do not show, a dichotomous measure for whether respondents’ postdisruption interview took place after the change to biennial interviews.

Table 4:

Quantile Regression of Income-to-Needs in Survey Wave Following Marital Disruption, NLSY79, 1979–2018

	Model 1	Model 2
Race/ethnicity and gender group
White men (reference)
Black men	−0.57*** (0.16)	−0.42** (0.13)
Hispanic men	−0.36† (0.19)	−0.34** (0.15)
White women	−1.41*** (0.12)	−0.75*** (0.11)
Black women	−1.86*** (0.16)	−1.02*** (0.14)
Hispanic women	−1.76*** (0.17)	−0.79*** (0.15)
Predisruption human capital
Highest level of education completed
High school or less (reference)
Some college		0.58*** (0.10)
Bachelor’s degree or more		2.16*** (0.13)
Years of employment experience		0.06*** (0.01)
Number of employment interruptions		−0.15*** (0.04)
Respondent is employed		0.54*** (0.10)
Children
Number of children living in household at the postdisruption interview		−0.29*** (0.04)
Controls
Mother’s highest level of education completed
High school or less (reference)
Some college		0.13 (0.14)
Bachelor’s degree or more		0.03 (0.16)
Respondent has a work-limiting health condition		0.08 (0.18)
Constant	3.30	2.04
Pseudo R2	0.0513	0.2068

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. We include, but do not show, a dichotomous measure for whether respondents’ postdisruption interview took place after the change to biennial interviews.

The coefficients in the unadjusted model for personal income indicate that, compared to every other race/ethnic and gender category, white men experience a bonus. These disparities persist after adjusting for human capital characteristics and other covariates. While the economic penalties experiences by all other groups are somewhat attenuated in Model 2, the absolute gaps remain substantial – ranging from $6,273 for Black men to more than $16,000 for Black women. Although the coefficient for Hispanic men does not attain statistical significance in the baseline model (Model 1), the fully adjusted model shows that, compared to Hispanic men, white men receive a significant benefit (b = $4,847). Table 4 shows analogous results for postdisruption income-to-needs, such that white men again evidence higher postdisruption economic status relative to the five other groups.

The independent variables in both models generally operate as expected, with higher education, more years of work experience, and current employment is positively associated with postdisruption economic wellbeing. The number of children in the postdisruption household has a net positive correlation with personal income and negative one with income-to-needs. The former is suggestive that children, net of the other covariates, perhaps serves to increase work effort while the latter stems from income-to-needs being based on family size.

Tables 6–10 alternate the reference group to examine all race/ethnic and gender comparisons.¹⁴ The results can be reported succinctly and they tell an intersectional story. In the fully adjusted models, women of any race/ethnicity have significantly lower personal income and income-to-needs relative to Black men, Hispanic men and white men. White men also enjoy higher postdisruption economic wellbeing on both measures than Black men and higher income-to-needs relative to Hispanic men. Black women fare worse in terms of both personal income and the income-to-needs ratio than white women. Further, Black women are disadvantaged in personal income relative to Hispanic women although the coefficient is only marginally significant (p < 0.10).

Table 6:

Quantile Regression of Personal Income and Income-to-Needs in Survey Wave Following Marital Disruption, NLSY79, 1979–2018

	Personal Income	Income-To-Needs
Race/ethnicity and gender group
Black men (reference)
White men	6,237.03** (2,008.20)	0.42** (0.13)
Hispanic men	1,798.57 (2,614.87)	0.07 (0.17)
White women	−6,011.39** (2,030.84)	−0.34** (0.13)
Black women	−10,027.94*** (2,415.11)	−0.61*** (0.16)
Hispanic women	−5,547.84** (2,573.53)	−0.38** (0.17)
Constant	18,168.36	1.63
Pseudo R2	0.2002	0.2068

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. Both models adjust for covariates from Table 1. N=2,891

Table 10:

Quantile Regression of Personal Income in Survey Wave Following Marital Disruption, NLSY79, 1979–2018

	Personal Income	Income-To-Needs
Race/ethnicity and gender group
Hispanic women (reference)
White men	11,784.87*** (2,279.54)	0.79*** (0.15)
Black men	5,547.84** (2,573.53)	0.38** (0.17)
Hispanic men	6,937.36** (2,818.14)	0.44** (0.19)
White women	−463.55 (2,113.70)	0.04 (0.14)
Black women	−4,490.71† (2,402.92)	−0.23 (0.16)
Constant	12,620.52	1.25
Pseudo R2	0.2002	0.2068

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. Both models adjust for covariates from Table 1. N=2,981

DISCUSSION

Research on the economic consequences of divorce has overwhelmingly focused on gender disparities, with race and ethnicity either secondary or invisible (e.g., Bianchi et al., 1999; Duncan & Hoffman, 1985; Hoffman; 1977; McKeever & Wolfinger, 2001; Smock, 1994; Tach & Eads, 2015). Classic economic and sociological frameworks for understanding the gendered economic consequences of divorce are implicitly based on a white middle-class division of household and market labor (e.g., Blood & Wolf, 1960; Becker, 1965). Thus, an important question stands: how does divorce affect the economic well-being of white, Black, and Hispanic men and women?

Our study employed an explicitly intersectional approach that simultaneously documented within-sex race/ethnic differences and within-race/ethnic group gender differences in the consequences of divorce. Although gender is crucial in understanding economic disparities stemming from separation and divorce, our results show substantial race/ethnic differences among both men and women – important disparities revealed through an intersectional approach.

Overall, our bivariate and multivariable results affirm well-established gender differences but also uncover within-gender racial stratification. White men are privileged relative to the five other subgroups in absolute levels of postdisruption economic wellbeing. Additionally, Black men fare better than Black and Hispanic women, although only marginally better than white women in some of our descriptive results. For example, Black men’s median postdisruption family income is similar to white women’s ($43,000 and $40,000, respectively). White women, in turn, are significantly better off than Black women.

In terms of degree of change induced by divorce, the gender dimension is most apparent. For example, women experience drops in family income of 46–50%, nearly double the drops experienced by men. Women also see substantial decreases of 35–38% in income-to-needs, compared to slight increases for men. Yet, there are within-gender race/ethnic differences; increases in income-to-needs, for example, are more modest for Black men (4%, Figure 3) than their white and Hispanic counterparts (11–12%). In addition, although Black and Hispanic men’s postdisruption income-to-needs ratios are above 1 (the poverty threshold) they are not much above 2 (see Table 2). Poverty and policy scholars often use 200% of the poverty threshold to identify low-income households (e.g., Jiang et al., 2017).

The dramatic post-divorce declines in women’s family income (by roughly 50%) and income-to-needs ratios (by roughly one-third), as well as their high poverty risk (see Table 2) reveals the potentially devastating consequences of divorce for women and, in many cases, their children, even in the late Baby Boomer cohort that attained high levels of education relative to prior cohorts of women. Yet, among women, white women fare best and Black women worst.

Research suggests that unless remarriage occurs, divorced women’s economic status remains low for a protracted period of time, and this situation has been characterized as chronic (Brüggmann & Kreyenfeld, 2023; Leopold, 2018). However, in view of sharp declines in remarriage rates over the past several decades, remarriage is no panacea (Payne, 2018). Although cohabitation has compensated for part of the decline, cohabiting relationships tend to be less stable than marriages (Smock & Schwartz, 2020). Blacks and Hispanics are also less likely to repartner after marital disruption and, given the linkage between positive economic prospects and remarriage, less likely to remarry (McNamee & Raley, 2001; Raley et al., 2015; Song, 2022). A focus on more recent cohorts incorporating the dissolution of cohabiting unions (e.g., Tach & Eads, 2015) is an important direction for future research.

Our results have potentially important implications for economic security in the longer-term. The high levels of economic precarity documented among divorced women relative to men, and especially Black women, bode poorly for their capacity to save, to make contributions to employer-provided pensions (if fortunate enough to receive one) and to amass wealth – a critical resource for economic security as these midlife adults enter their retirement years (Addo & Lichter, 2013; Tamborini & Kim, 2020). Immediate financial needs may prevent divorced women from saving for the future (Carr, 2020). Although Social Security provides a minimum source of income for retirement-age men and women, the program is not sufficient to remedy the economic disadvantages experienced by divorced women who do not subsequently remarry. Divorced women, especially those with relatively low earnings, may opt not to receive their own retiree benefit, preferring to take their former spouse’s benefit. However, divorcees are eligible for only half of their ex-spouse’s benefit, provided their marriages were at least ten years long. Our results suggest that revisions to income security programs are needed to reduce gender differences in the economic well-being of U.S. adults even prior to their retirement transition, and to address the particular vulnerability of divorced Black women.

There are at least three limitations to our study. First, our multivariable models adjust for human capital characteristics including education and work experience, but structural racism and sexism affect both the acquisition of and economic returns to human capital (Bailey et al., 2017; Graetz et al., 2022; Sen & Wasow, 2016). Results from our multivariable models are thus unable to delve into sources giving rise to racial/ethnic and gender inequality in the economic toll of marital disruption. Second, we do not take account of the non-randomness of marital dissolution. Divorce and its outcomes may be influenced by unmeasured selection processes that may vary by subgroup. However, consistent with other research on the economic consequences of divorce, our purpose has been descriptive, focusing on observed economic wellbeing as individuals move from marriage to divorce.

Third, we examined the short-run economic consequences of divorce. Even after the switch to biennial interviews, the window is still relatively short-term. Yet knowledge about the short-term costs is important because what unfolds shortly after marital disruption sets the stage for future recovery or further depletion of resources. If people already have, or are able to find, well-paying jobs with promotion opportunities, pension benefits, and especially for women, workplaces that integrate employment and parental responsibilities, these factors can positively impact their economic security over the longer-term and far beyond to retirement.

Despite these limitations this is the first study we know of to adopt an explicitly intersectional approach to tracing the economic consequences of separation and divorce. Our results shed light on the ways that deeply entrenched race and gender inequality in the labor force can amplify the dire economic consequences of divorce, especially for Black women who may experience “double jeopardy.” More broadly, our results indicate pronounced race and gender stratification in the economic ramifications of divorce with white men in the most privileged position, then Hispanic men, then Black men, white women, Hispanic women, and then Black women.

To date, the prevailing narrative about the economic toll of divorce has implicated families, especially the gendered allocation of domestic and care work, as the main mechanism undermining divorced women’s financial security. Yet, our results suggest that the intersections of race/ethnicity and gender, and not gender alone, are powerful stratifying mechanisms that shape the economic consequences of marital dissolution.

Table 7:

Quantile Regression of Personal Income and Income-to-Needs in Survey Wave Following Marital Disruption, NLSY79, 1979–2018

	Personal Income	Income-To-Needs
Race/ethnicity and gender group
Hispanic men (reference)
White men	4,438.46† (2,335.97)	0.35** (0.15)
Black men	−1,798.57 (2,614.87)	−0.07 (0.17)
White women	−7,809.96** (2,351.07)	−0.41** (0.16)
Black women	−11,826.51*** (2,676.93)	−0.67*** (0.18)
Hispanic women	−7,346.41** (2,818.14)	−0.44** (0.19)
Constant	19,966.93	1.69
Pseudo R2	0.2002	0.2068

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. Both models adjust for covariates from Table 1. N=2,891

Table 8:

Quantile Regression of Personal Income and Income-To-Needs in Survey Wave Following Marital Disruption, NLSY79, 1979–2018 (N=2,891)

	Personal Income	Income-To-Needs
Race/ethnicity and gender group
White women (reference)
White men	12,248.41*** (1,623.82)	0.75*** (0.11)
Black men	6,011.39** (2,030.84)	0.34** (0.13)
Hispanic men	7,809.95** (2,351.07)	0.40** (0.16)
Black women	−4,016.55** (1,903.93)	−0.27** (0.13)
Hispanic women	463.55 (2,113.70)	−0.04 (0.14)
Constant	12,156.97	1.29
Pseudo R2	0.2002	0.2068

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. Both models adjust for covariates from Table 1. N=2,891

Table 9:

Quantile Regression of Personal Income and Income-To-Needs in Survey Wave Following Marital Disruption, NLSY79, 1979–2018

	Personal Income	Income-To-Needs
Race/ethnicity and gender group
Black women (reference)
White men	16,264.97*** (2,093.25)	1.02*** (0.14)
Black men	10,027.94*** (2,415.11)	0.61*** (0.16)
Hispanic men	11,826.51*** (2,676.93)	0.67*** (0.18)
White women	4,016.55** (1,903.93)	0.27** (0.13)
Hispanic women	4,480.10† (2,402.92)	0.23 (0.16)
Constant	8,140.42	1.02
Pseudo R2	0.2002	0.2068

Standard errors in parenthesis; statistical significance is denoted as †p < 0.10 **p < 0.05 ***p < 0.001. Both models adjust for covariates from Table 1. N=2,891

Data Availability

The data supporting these findings are publicly available at the National Longitudinal Surveys website (https://www.nlsinfo.org/content/cohorts/nlsy79).