Parental Origins, Mixed Unions, and the Labor Supply of Second-Generation Women in the United States

Abstract

Introduction

In the last two decades, millions of children born in the United States to foreign-born parents have come of age. Compared with their immigrant parents, children of the second generation are upwardly mobile, better-educated, and more likely to hold high status jobs (Park and Myers 2010; Park et al. 2015; Tran and Valdez 2017). They are much more integrated into the mainstream: less likely to live in ethnic enclaves (Iceland 2014) and more likely to marry exogamously (Alba and Foner 2015; Kalmijn and Van Tubergen 2010). Yet, among second-generation women, economic incorporation is uneven (England et al. 2004; Read and Cohen 2007). Conventional models of women’s labor force participation center on human capital, family context, and labor market characteristics. This study considers cultural influences on women’s economic behavior. We investigate the joint effects of parental origins and endogamy—partnering within the same national-origin group—on the labor force participation of second-generation women.

We estimate the effects of parental origins using the epidemiological approach to cultural transmission (Fernández 2011). The epidemiological approach exploits differences in cultural origins among groups that share the same institutional environment in order to identify the influence of culture on behavioral outcomes.¹ Previous studies of cultural transmission in economics and sociology have used the epidemiological approach to show that aggregate indicators of women’s status in the parental country of origin—such as female labor force participation rates, total fertility rates, and secondary enrollment rates—are significant predictors of second-generation women’s employment, especially among married women (Antecol 2000; Blau et al. 2013; Fernández and Fogli 2009; Finseraas and Kotsadam 2017; Polavieja 2015). These results can be interpreted as evidence of cultural transmission of gender norms—that is, the persistence of cultural ideas about gender and family that are shaped in one institutional context, carried to another, and transmitted from parent to child within immigrant families. Findings from studies using the epidemiological approach have echoed results from the childhood socialization literature on the positive intergenerational links between mothers and daughters with respect to gender attitudes (Moen et al. 1997; Starrels 1992), housework (Cunningham 2001; Treas and Tai 2012), and employment behavior (Greene et al. 2013; Johnston et al. 2014; van Putten et al. 2008). Both literatures posit families as the primary engines for the transmission of gender attitudes and behaviors, but studies using the epidemiological approach typically lack any direct evidence of family socialization.

We also investigate a proximate family mechanism for cultural effects on women’s labor force behavior: partner choice. Assortative mating by ethnicity or ancestry among U.S.-born daughters of immigrants reflects women’s agency in ways that reinforce or undermine cultural repertoires inherited from the first generation. For example, women who share the same cultural origins as their partner may be more vulnerable to demands for domestic labor from extended immigrant family networks. Following Kalmijn and Van Tubergen (2010), we compare women who partner with a first- or second-generation immigrant of the same national origins (endogamy) with those in mixed unions (exogamy or intermarriage), whether with a U.S.-born son of two U.S.-born parents (native-mixed union) or a first- or second-generation immigrant with different national origins (mixed-origin union).

Our study makes three contributions to the literature. First, we extend the literature on marital assimilation by examining the labor market consequences of mixed unions among second-generation immigrants. Previous research on intermarriage and women’s employment looked only to the first generation (Blau et al. 2011; Furtado and Theodoropoulos 2010; Furtado and Trejo 2013; Meng and Gregory 2005). Research on exogamy in the second generation has centered on parental pressures (Kasinitz et al. 2008; Zhou and Bankston 1998), identity formation and belonging (Alba and Foner 2015; Kibria 1997), and intermarriage or mixed union propensities (Alba and Foner 2015; Kalmijn and Van Tubergen 2010; Qian et al. 2012). This study is the first to use nationally representative data to investigate marital assimilation and the labor force outcomes of second-generation women. Second, we compare the effects of parental origins on women in marital unions and nonmarital unions. Cohabitation is especially salient given the high rates of marriage among first-generation immigrants (Brown et al. 2008), but prior studies of cultural transmission among U.S. women have either excluded cohabiters or classified them with other unmarried women (Antecol 2000; Blau et al. 2011, 2013).

Third, we advance the theoretical literature on marriage, intermarriage, and immigrant incorporation by using large-scale quantitative evidence to test an explanation for labor market outcomes based on the “dynamic interplay between structure, culture and agency” in immigrant families (Foner 1997:961). We build on recent empirical studies of second-generation achievement that see immigrant families as the locus for the transmission of cultural frames and repertoires (Feliciano and Lanuza 2017; Lee and Zhou 2015), but rather than achievement or success frames, we propose that immigrant groups vary in the extent to which they maintain a gendered domesticity frame that puts a premium on women’s place in the home as good wives and mothers (Read and Oselin 2008:299).

Background

Shifts in the U.S. economic landscape over the last half-century produced profound changes in marriage and family life. The decline in young men’s economic prospects, continued expansion of women’s economic opportunities, lower fertility, the rising cost of children, and the increase in marital instability form the backdrop for the movement away from specialization in the gender division of labor (Becker 1991) toward increased symmetry in the time spent on paid work by marital partners (Goldscheider and Waite 1991; Oppenheimer 1994, 1997; Spain and Bianchi 1996). The shift from the breadwinner–homemaker marriage model to dual-earner marriage as the social norm has reinforced the American cultural norm of gender egalitarianism (Goldscheider et al. 2015), arguably much more so than in most immigrant-sending countries. In addition, studies of the labor force outcomes of first- and second- generation women have found unexplained differences in the employment of national-origin groups from Latin America (England et al. 2004) and Asia (Read and Cohen 2007; Yamanaka and McClelland 1994). These findings motivate an inquiry into cultural explanations.

Patterson (2001:208) defined culture as “a repertoire of socially transmitted and intra-generationally generated ideas about how to live and make judgements, both in general terms and in regard to specific domains of life.” For Patterson, culture interacts with structural forces to constrain or motivate individual behavior, facilitating both cultural and structural change. Two implications follow that make this a useful theoretical framework for interpreting the literature on culture transmission to the second generation. First, both aggregate and individual indicators of behavioral outcomes can be interpreted as evidence of agency in response to the interaction between structure and culture. Second, although we should expect change in cultural models—especially as they are transmitted to new environments, with new structural constraints, opportunities, and social norms—the second generation still inherits cultural models from the preceding generation through socialization. The heterogeneity in parental origins and the associated differences in cultural models provide a basis for measuring cultural transmission.

Large-scale studies using the epidemiological approach have shown that family and gender patterns in origin countries are reproduced among immigrants in destination countries, including living arrangements (Alesina and Giuliano 2010), fertility (Blau 1992), domestic labor (Frank and Hou 2015), female labor force participation (Blau and Kahn 2007; Blau et al. 2011; Frank and Hou 2015; Greenlees and Sáenz 1999; van Tubergen et al. 2004), and the gender gap in employment (Antecol 2000). The findings demonstrate the persistence of cultural models of work, family, and gender among first-generation immigrants in the new structural environment. Antecol (2000) also compared first-generation immigrants with U.S.-born individuals who reported their ancestry in the 1990 census and found that the coefficient on the gender gap in employment was substantially smaller for the U.S.-born, as would be expected with assimilation.

Research using the epidemiological approach to study cultural transmission in the second generation produces more compelling evidence for the persistence of culture because the economic and institutional conditions of the country of ancestry are not experienced firsthand by the second generation. Instead, “preferences and beliefs embodied in these variables may still matter if parents and/or neighborhood transmitted them to the next generation” (Fernández and Fogli 2009:148). Fernández and Fogli (2009) found that lagged measures of fertility and female labor force participation from father’s birthplace were significant predictors of the fertility and labor supply of second-generation women in the 1970 U.S. Census. Studies of more recent immigrant cohorts found that family attitudes (Alesina and Giuliano 2010) and labor force gender ratios (Blau et al. 2013) are significant predictors of the labor force behavior of U.S.-born daughters.

Blau et al.’s (2013) primary interest was in the intergenerational transmission of fertility, education, and labor supply. They compared outcomes for immigrants, lagged to approximate the “parent” generation, with outcomes for children of immigrants from the same country of origin. Their findings of significant intergenerational associations across all three domains lend credence to the idea that cultural transmission occurs through socialization and social learning within families and in immigrant communities. More telling are the disparate findings in the domains of education and labor supply. Patterns of educational transmission for second-generation men and women were similar, but women’s labor supply was correlated only with the labor supply of women in the immigrant generation, and men’s labor supply was correlated only with the labor supply of men in the immigrant generation. We see this as evidence of the persistence of gendered cultural models of work and family life across immigrant generations, in line with past findings showing that support for women’s domestic role in families is associated with lower levels of female labor force participation, both in the United States and cross-nationally (Davis and Greenstein 2009; Fortin 2005).

Endogamy and Mixed Unions

Although culturally shaped ideas about gender can be altered by economic contingencies in immigrant families, or discarded by children of immigrants in processes of selective acculturation during the transition to adulthood (Portes and Rumbaut 2014), ideas and dispositions regarding women’s domesticity in family life may be especially resilient because they are reinforced within the private sphere. Case studies have documented the ways that gendered cultural repertoires are preserved and maintained within immigrant families (Lim 1997; Parrado and Flippen 2005; Read 2004), and studies of intermarriage and employment have provided counterfactual evidence of immigrant women’s employment in the absence of shared cultural origins. As an example, marriage reduces the labor supply of immigrant women more so than native women (Donato et al. 2014; Read and Cohen 2007), yet the labor supply of immigrant women who intermarry with U.S.-born men more closely resembles the employment patterns of U.S.-born women (Blau et al. 2011). Read (2004) found that regardless of their personal gender ideology, Arab American wives are more likely to participate in the labor force if their husbands are not Arab American.

A woman’s choice of intimate partners reflects both the structural opportunities to find a mate and her own agency. Consistent with Oppenheimer’s (1988) extended premarital sorting theory of marriage timing, second-generation women tend to delay marriage as they pursue educational and employment opportunities (Brown et al. 2008). These strategies increase the economic viability of a future marital union, and they can weaken gender traditionalism (Davis and Greenstein 2009). Women’s investments in their own human capital also expand their social networks, increase their number of potential partners, and make them more attractive on the marriage market and potentially more open to a partner outside their own national-origin group (Sweeney 2002). Many second-generation women avoid partners from their national-origin group in order to escape the gendered cultural models of their family or ethnic community (Kasinitz et al. 2008). As Min and Kim (2009:456) wrote, “American-born Asian American men are likely to be less patriarchal than their immigrant parents. But they are slower than women in adopting more egalitarian marital relations.”

Still, second-generation women are more likely to find first- or second-generation partners from the same national-origin group than would happen by chance alone, and an additional unknown number are third-generation immigrants from that same group (Kalmijn and Van Tubergen 2010). A woman with close relationships to kin may prefer a partner who shares her cultural origins, who might then reproduce culturally normative gender expectations. Endogamy expands coethnic family ties, multiplying the social interactions that reinforce gendered cultural frames of separate spheres (Parrado and Flippen 2005; Read 2004). We expect that endogamy will be associated with a reduction in women’s labor force activity, whether the partner is foreign-born or a U.S.-born child of immigrants. Although we believe that endogamy is an important mechanism for cultural transmission in the second generation, we cannot fully rule out unobserved factors associated with both self-selection into exogamous unions and increased labor supply. We explore these possibilities in supplementary analyses.

We also probe differences among women in mixed unions. Despite the long-standing interest in mixed ancestry and panethnic intermarriage in the immigration literature (Alba and Nee 2003; Gordon 1964; Qian and Lichter 2007; Qian et al. 2012), studies of marital assimilation and women’s economic outcomes have centered on the dichotomy either between foreign-born and U.S.-born partners (Blau et al. 2011; Greenlees and Sáenz 1999) or across panethnic groups (Read 2004; Read and Cohen 2007). We are agnostic about differences in the labor market activity among women in different types of mixed unions. Women who partner with men who have a long family history in the United States may have greater exposure to dominant American gender norms that are supportive of women’s employment, yet women who partner with men from a different national-origin group may have greater freedom within the public sphere because they are reaching across boundaries and possibly forging new identities and norms.

Educational Attainment

Education is likely to act as a mediator that weakens parentally transmitted models of female domesticity and reduces the probability of forming an endogamous union. The education gradient for women’s labor supply has steepened over time in the overall population (Montez et al. 2014), and we expect the same for second-generation women. This is not to say that the academic experiences of children in immigrant families are much the same as those of natives. Immigrant parents often encourage academic achievement while expecting their daughters to submit to gendered social controls (Kasinitz et al. 2008; Lee and Zhou 2015): many parents insist that their daughter attend college locally because living at home keeps her close and saves money (Borgen and Rumbaut 2011). As students advance through the educational system, however, they are exposed to ideas that result in a decline in support for gender traditionalism (Davis and Greenstein 2009). Higher education shifts social interaction away from culture-affirming family and community groups toward a diverse and expanded range of social ties, increasing the probability of contact, friendship, and intimacy across group lines (Massey and Denton 1993). This contact, in turn, increases the likelihood that a woman will enter a mixed union. The prevalence of panethnic, interethnic, and interracial unions among both first-generation and second-generation immigrants increases with educational attainment (Qian and Lichter 2007; Qian et al. 2012).

Endogamy is more common among less-educated women, but highly educated women may privilege cultural homogamy over educational homogamy when relatively few potential partners meet both criteria. Endogamous unions can reinforce cultural transmission and counter the weakening of coethnic ties and cultural dispositions that occurs as women progress through the educational system and gain work experience. As with Read and Oselin’s findings for well-educated Arab American women, a college degree can be recast as a domestic resource to “ensure the proper socialization of children, solidarity of the family, and ultimately the maintenance of ethnic identity” (Read and Oselin 2008:296). Endogamy can also come with caretaking responsibilities for family-dependent extended kin. When immigrant family members require care, women who are nearby may unexpectedly become “prisoners of love,” taking on the burden of kinship care (Folbre 2012). In short, assortative mating by parental origins can serve as a mutually reinforcing embrace of cultural values that promote domesticity across the educational spectrum. We expect endogamy to weaken women’s attachment to the labor market, especially among less-educated women.

Marital Unions Versus Cohabitation

Immigration law, material hardship, and origin-country cultural norms all contribute to high rates of marriage among immigrants in the United States, but cohabitation rates among young women double from the first to the second generation (Brown et al. 2008). Widespread acceptance of cohabitation in the United States is likely to weaken cultural models of female domesticity for several reasons. First, cohabiting unions in the broader U.S. population are characterized by more gender equality than marital unions on measures of employment, earnings, and domestic labor (Brines and Joyner 1999); and cohabiting second-generation women may be acculturated into egalitarian cultural repertoires that are widely shared among their U.S.-born peers. Second, although consensual unions are common in some parts of Latin America, cohabitation is still taboo in many conservative societies (Brown et al. 2008). Premarital cohabitation can signal a rejection of traditional norms on family formation and create distance in the relations between immigrant parents and their daughters (Borgen and Rumbaut 2011), reducing emotional and financial support. Third, the cultural meaning of cohabitation in the United States remains distinct from the cultural meaning of marriage. Cohabiting unions in the broader U.S. population are characterized by lower relationship commitment and greater fragility than marital unions (Brines and Joyner 1999), and cohabiting couples are more likely to be heterogamous on race, ethnicity, and religion than married couples (Blackwell and Lichter 2004). Moreover, the fragility of cohabitation makes domesticity a risky prospect. We expect that the effects of both cultural origins and endogamy will be strongest among married women and muted among women in cohabiting unions.

Data, Measures, and Methods

We use harmonized data from the 1996–2016 IPUMS Current Population Survey’s Annual Social and Economic Supplement (CPS-ASEC), a nationally representative survey of the U.S. population collected each March as a supplement to the basic monthly survey (Flood et al. 2015). The CPS includes questions on each person’s birthplace and the birthplace of each parent. We select women aged 25–49 in heterosexual marital and cohabiting unions who were born in the United States and had at least one parent born outside the United States or in U.S. outlying areas. To avoid duplicate observations resulting from the rotational design of the CPS, we restrict the sample to the first observation on each woman.² We exclude observations for which the birthplaces of both parents are unknown or the birthplaces of partner and both of partner’s parents are unspecified. Our analysis sample includes observations on married or cohabiting women born to parents who had immigrated to the United States from one of 131 countries and territories (online appendix, Table A1).

Dependent Variables

The dependent variable is labor supply, calculated as the number of weeks worked for pay in the previous calendar year multiplied by the usual number of hours worked per week. We also include a dichotomous measure of workforce participation, coded as 1 for all women who engaged in at least 50 hours of paid work in the past year.

Mixed Unions and Marital Status

Our key predictors are measures of endogamy and exogamy based on national-origin group. If either immigrant parent of the woman shared the same birthplace as her partner, his mother, or his father (in that order), we assign her to that national-origin group and classify the union as endogamous. Women in mixed (exogamous) unions are assigned to the mother’s birthplace if the mother was born abroad, and to the father’s birthplace otherwise.³ We classify mixed unions with a male partner who is U.S.-born with two U.S.-born parents as native-mixed unions. We classify mixed unions with a male partner who is a first- or second-generation immigrant from a national-origin group other than her own as mixed-origin unions. We include an indicator for women in cohabiting unions.

Cultural Transmission Measures

We follow previous epidemiological studies of culture in using female labor force participation (FLFP) among women aged 15–64 in the parental origin country (Alesina and Giuliano 2010; Fernández and Fogli 2009), and we use a two-decade lag to approximate a generation.⁴ The data are compiled from the World Bank Database and country-specific sources; we fill missing values using linear interpolation. FLFP in the parental origin country ranges from 9.7 % to 89.1 %. We center FLFP on FLFP in the United States; second-generation women overwhelmingly (86 %) originate in countries with lower FLFP rates than in the United States.

Controls

We control for structural predictors, including racial/ethnic group⁵ as an indicator of context of reception (Portes and Rumbaut 2014) and individual and household characteristics associated with women’s labor supply: age and age-squared, education, number of own children and own children under age 5 in the household, and a second-order polynomial to control for the total amount of constant dollar household income from sources other than her labor income (England et al. 2004; Kahn and Whittington 1996; Lehrer and Nerlove 1986; Mincer 1962; Stier and Tienda 1992; Yamanaka and McClelland 1994). To control for family of origin differences, we include dummy variables to indicate which parent is foreign-born.

Analytical Approach

Results

Descriptive statistics are presented in Table 1. More than one-half (55.7 %) of the second-generation women studied are in mixed unions with a U.S.-born child of U.S.-born parents, but nearly 3 in 10 married women and more than 2 in 10 cohabiting women are in endogamous unions with men from the same national-origin group. The remaining women are in mixed-origin unions with men who are themselves immigrants or children of immigrants. Compared with women in mixed unions, women who partner endogamously are more likely to have two foreign-born parents. They are also younger, much less educated, less affluent, and far more likely to identify as nonwhite (especially Hispanic or Asian) than women in mixed unions. On average, endogamously partnered second-generation women clocked 1,292 hours in the previous year, compared with 1,424 hours for women in mixed unions. Women in cohabiting unions are significantly less likely than married women to be in a same-origin union, consistent with the general characterization of cohabiting unions as more heterogamous than marital unions. They are also significantly younger and less educated, and they have lower income and fewer children than married women.

Cultural Transmission

Because gender repertoires can affect women’s labor supply through their influence on women’s choices with respect to geographical residence, education, and especially family formation, we use a series of nested models to demonstrate cultural transmission and the factors that mediate origin effects. The first model in Table 2 presents baseline origin-group variation in labor supply using only period controls. The intercept represents average work hours in 2005, the comparison year in these models. The second model shows the positive association between the lag of female labor force participation in the parental birthplace and the labor supply of second-generation women. The cultural transmission measure provides a significant improvement of model fit (⁠$χ12$ = 13.4, p < .01) and reduces origin-group variation by 12.5 % compared with the baseline model. Because this measure is centered on the U.S. measure, the intercept in Model 2 represents the average labor supply of women in a country with the same (lagged) participation rates as the United States—higher than the overall average in Model 1 because FLFP is lower in most sending countries. Model 3 shows little change in the lagged FLFP coefficient with the addition of labor market controls and individual-level controls.

Family characteristics mediate the effects of cultural transmission much more so than women’s own human capital, as is evident when marital status and household income are included in Model 4. Adding children to Model 5 suggests that cultural transmission affects women’s labor supply through differential fertility across groups. Still, the coefficient on the lagged measure of FLFP remains statistically significant net of selection into location, education, marriage, and fertility.

Endogamy and Exogamy

Table 3 shows the labor supply of women in mixed unions as opposed to endogamous unions, including all controls. Exogamy is associated with significant differences in women’s labor supply: the results in Model 1 indicate that women in mixed unions average more than 50 additional hours of work each year. Disaggregating mixed unions into native-mixed unions (i.e., the male partner is a U.S.-born child of U.S.-born parents) and mixed-origin unions with first- and second-generation immigrants (Model 2) indicates that both native-mixed unions and mixed-origin unions are important drivers of second-generation women’s labor supply. On average, women in native-mixed unions work one full-time week more, and women in mixed-origin unions work two full-time weeks more, than women who share their partner’s national-origin group.

One alternative explanation for the dampening effect of endogamy on women’s labor supply is that it is driven by foreign-born partners. If the pre-immigration exposure of foreign-born men to norms of female domesticity reduces their expectations for wives’ economic contributions to the household, women who are more traditional in their gender orientations might prefer to partner with immigrants. In this case, we would expect the labor supply of women to be lower in both same-origin and mixed-origin unions with immigrants.

As a check on this source of selectivity, we further disaggregate same-origin and mixed-origin unions by the male partner’s generational status. The results in Model 3 produce no evidence to support the hypothesis of cultural preferences for domesticity in cross-generation mixed unions. Women in mixed unions, whether their partner is an immigrant, a child of immigrants, or a third-generation native, all have a significantly greater labor supply than women in shared-origin unions. In contrast, endogamy is associated with lower labor supply regardless of partner’s nativity, with a small and insignificant difference between shared-origin unions with second-generation as opposed to immigrant partners. Moreover, women in endogamous unions are less likely than women in mixed unions to have partners with higher educational attainment than their own (20 % vs. 24 %); they are more likely to partner “down” (34 % vs. 30 %). We conclude that shared cultural origin is a valued trait that can substitute for partner’s economic potential in marriage markets, and it is shared cultural origin that most inhibits the labor market activity of second-generation women.

To buttress the findings from the labor supply models, we also estimate multilevel logistic regression models for the probability that women were employed for at least one week in the previous calendar year. Figure 1 shows odds ratios for any employment in the previous year compared with women in shared-origin unions with second-generation men. The shaded bars indicate that women in native-mixed unions, mixed-origin unions with second-generation men, and mixed-origin unions with first-generation men are significantly more likely to have worked in the previous year, with increased odds that range from 24 % to 35 %. The hollow bar indicates that the difference between women in endogamous unions with immigrant men and women in endogamous unions with second-generation men is statistically insignificant. Overall, we find that exogamy increases the odds of employment by approximately 22 % (results available in online appendix, Table A2).

Partnership Effects by Education Level

We expect education to mediate the intergenerational transmission of culture, such that parental origins would have a stronger impact on the labor force activity of women with less rather than more education. The first two columns of Table 4 show results for second-generation women with no more than a high school diploma, and the last two columns show results for women with at least a four-year college degree. Cultural transmission effects appear to be relatively strong among less-educated women—the coefficient on FLFP is more than one-third larger compared with the full-sample results—while college-educated women show no significant effects of origin country FLFP. The difference between less-educated and highly educated women is statistically significant (p = .02), an affirmation that education mediates the cultural transmission of gender norms.

In contrast, the implications of exogamy for both the most-educated and least-educated daughters of immigrants are strikingly similar, despite different rates of mixed unions. In our sample, 71 % of women with a college degree are in a mixed union, compared with only 59 % of women with no more than a high school diploma. In both groups, the partners in mixed unions are more educated than the partners in endogamous unions. Among less-educated women, nearly two-fifths (38.5 %) of men in mixed unions have at least some college, compared with one-fifth of the men in endogamous unions. Among college-educated women, more than 70 % of men in mixed unions have a college degree or more, compared with 60 % of the men in endogamous unions. In both groups, women in mixed unions work an average of two to three full-time weeks (80–120 hours) more than women who partnered with men from the same national-origin groups (Models 1 and 3). Results from Models 2 and 4 indicate that both native-mixed unions and mixed-origin unions are associated with an increase in women’s labor supply, although the effect of mixed-origin unions among less-educated women does not reach statistical significance. The evidence that second-generation endogamy suppresses women’s economic activity is especially strong for highly educated women, who are partnering (overall) with highly educated men. Their education appears to have left them immune to the effects of cultural transmission through childhood socialization. What is notable, then, is that endogamy has a similar effect across educational strata, reducing the labor supply of college-educated daughters of immigrants and those with a high school education by comparable amounts.

We illustrate the pervasive effects of endogamy on women’s employment using recent ASEC data that include measures of ethnic identification for major Hispanic and Asian groups. Using the subsample of married women who identified as belonging to one of the major ethnic groups and who also had a parent born in that country,⁷ we estimate a labor supply model using a specification that includes interaction terms for mixed union status by ethnic groups (see online appendix, Table A3). Figure 2 plots the results. Despite small sample sizes in some groups, intermarriage is associated with significantly higher labor supply for most ethnic groups, including those with very high average educational attainment (Asian Indians, Taiwanese, and Hong Kong Chinese). In this subsample, only ethnic Cubans, Dominicans, and Filipinas spend fewer hours in the workforce when they marry exogamously.

Cohabitation Versus Marriage

Do cultural origins have the same impact on the economic behavior of women in cohabiting unions and marital unions? Table 5 shows results for each group for comparison. For the baseline analyses, we separate cohabiters into those with children and those without children because in preliminary examinations, we found significant differences in the baseline between child-free cohabiting women and cohabiting women with children. The same was not true for married women.

The first three columns of Table 5 are estimates from baseline models that include only FLFP, the indicator for exogamy and period dummy variables. The baseline model for married women (Model 1) shows strong cultural effects through both parental inheritance and partner choice. Cohabiting mothers (Model 2) also show a positive effect of FLFP, which was not expected. Instead, our expectation that cohabitation would weaken cultural frames of domesticity holds only for nonmothers, who show a negligible association between source-country FLFP and labor supply (Model 3). Neither group of cohabiters shows an effect of mixed unions in the baseline models.

When all individual, family, and labor market controls are included, the results for married women in Model 4 look much the same as the full sample results (Table 3, Model 1). Neither the child-free sample (not shown) nor the cohabiting mothers (Model 5) indicate any significant effects of FLFP or partner choice. When both subsamples are pooled (Model 6), the coefficient on mixed unions is negative and statistically significant, and the effect of parental country FLFP disappears entirely. These findings provide partial support for our expectation that cultural frames of domesticity would be more evident within marital rather than cohabiting unions. However, these results affirm previous research that variation in cohabitation among first- and second-generation immigrants is complex (Brown et al. 2008), and cultural understandings of the meaning of cohabitation may vary within as well as across national-origin groups.

Supplementary Analyses

We performed several robustness checks on our findings. Using proxy measures from father’s country of origin instead of mother’s yield the same substantive findings. We experimented with alternative proxy measures of gendered cultural repertoires in the parent country of origin, including total fertility; the adolescent birth rate; educational gender parity indices (GPIs) for primary school, secondary school, and school life expectancy through secondary school; and the gender labor force ratio; we also included per capita gross domestic product as an indicator of economic development. Some of these analyses indicated systematic differences in labor supply across groups by level of development in the parental origin country. Other measures apparently capture cultural transmission that operates through educational attainment and fertility, but the employment measures remained statistically significant when all individual and family controls are included. Across all models, we found higher labor supply among women in mixed unions.

The data that we used are unbalanced, with some countries contributing few observations and others, notably Mexico, accounting for a large share of second-generation women. As a check on the robustness of our findings, we reestimated all models using national-origin groups with at least 10 observations (95 groups). We repeated this procedure using only those groups with at least 30 observations (66 groups). In both cases, the main findings on assortative mating and the comparisons by educational attainment were substantively the same. We also reestimated models after excluding women with parental origins in Mexico. Again, the main findings were substantively the same, although estimates of differences across types of mixed unions were not statistically significant.

We considered the hypothesis that the moderating effects of education might be explained by cultural transmission that inhibits second-generation women’s educational attainment, and we found no support. We also attempted to address differential selection into marital and cohabiting unions across national-origin groups. Women from national-origin groups that discourage nontraditional unions may be more likely to marry young, less likely to cohabit, and less likely to separate and divorce, resulting in inflated estimates of the causal impact of cultural origin and endogamy. Including women in cohabiting unions in the main analyses addresses this bias to some extent, and we repeated the cohabitation analyses using only national-origin groups that included women in both marital and cohabiting unions. As a further test, we estimated a multinomial probit model to predict marital status (marriage vs. cohabitation or not being in a union) and incorporated the inverse Mills ratios (IMR) into pooled separate labor supply models for married and cohabiting women. Controlling for selection into marriage increases the magnitude and precision of the estimated effects of cultural transmission and endogamy in both the full sample and the subsample of married women. The findings for cultural transmission and endogamy among married women are robust to selectivity. Results from the two-step selection models for cohabiting women showed no significant effects of cultural origins.

Discussion

We investigated two sources of family culture that shape the labor market incorporation of adult daughters of immigrants. Using the tools of the epidemiological approach to culture and well over 100 parental birthplaces, we found evidence confirming findings from previous studies that women whose parents migrated from places with lower FLFP spend significantly fewer hours in paid work, and we showed that much of the baseline association between origin-country FLFP and women’s labor supply is accounted for by women’s choices with respect to marriage, cohabitation, and fertility. We interpret the findings as plausible evidence that cultural frames of female domesticity originating in the parental birthplace have a causal impact on the labor force behavior of U.S.-born daughters of immigrants. Immigrant families are the primary mechanism for cultural transmission in childhood and adolescence, and we expected that education would weaken cultural persistence. Our finding that cultural transmission is mediated by education confirmed our expectation: the acquisition of a college education appears to fully disrupt the relationship between origin-country FLFP and the labor supply of second-generation women.

Our main findings center on assortative mating by cultural origins as a second source of family variation that shapes the labor supply of adult daughters of immigrants. We found that second-generation women who marry within the same national-origin group have lower levels of economic activity, even if the husband is born in the United States. Exogamy is associated with increases in the labor supply of second-generation women, whether the partner is U.S.-born or foreign-born. In contrast with evidence that ascriptive cultural repertoires have no effect on women with a college degree, education does not appear to moderate the impact of endogamy. Although the labor supply of highly educated women is much greater than that of women with no more than a high school diploma, the decline in work hours associated with endogamy is comparable for the least-educated and most-educated women.

Although the findings for married women are robust, our expectation that cohabitation would disrupt cultural transmission was only partially supported. As expected, we found no effect of origin country FLFP on the labor supply of cohabiting nonmothers, who comprise just over one-half the women in our cohabitation subsample. Contrary to expectations, we found strong baseline effects of origin country FLFP on cohabiting mothers. We expected the effects of assortative mating by cultural origins among cohabiters to be muted; and although this was true in the baseline models for both mothers and nonmothers, the pooled model with all controls show a coefficient on exogamy that is statistically significant and negative. These disparate findings highlight the complexity of cohabitation patterns among second-generation immigrants.

Intermarriage has mainly been seen as “an indicator of the degree to which different groups in society accept one another” (Kalmijn and Van Tubergen 2010), and recent literature has advanced our understanding of the patterns and structural bases for assortative mating across immigrant generations and racial/ethnic groups (Lichter et al. 2011; Min and Kim 2009; Qian et al. 2012). The rapid shifts in the pace and origin of immigration flows has appropriately drawn attention to the implications of immigration for intermarriage among second-generation immigrants. A much smaller literature is emerging on the implications of intermarriage for second-generation integration (Alba and Foner 2015). Intermarriage has the potential to diversify the social networks of immigrants, increase their access to knowledge about labor market institutions (Meng and Gregory 2005), and expose them to different cultural norms regarding gender roles (Pyke and Johnson 2003). Our study advances that line of research and adds a correction to an earlier literature on the economic implications of intermarriage for first-generation immigrants. The earlier literature centered on the structural barriers that new immigrants face and the labor market advantages they gain by partnering with native-born individuals in terms of rapid language acquisition, acculturation, instrumental social networks, and tacit knowledge of U.S. labor markets. These are certainly valuable resources, and we do not doubt that our finding on the increased labor supply of mixed unions is due in part to “structural integration” through intermarriage (Brown 2006). However, our evidence that the association between endogamy, exogamy, and labor supply are similar whether partner is U.S.-born or foreign-born strongly implicates culture.

Empirical research on immigrant incorporation frequently falls into disparate camps. Patterson’s (2001) framework for cultural influence is best illustrated by qualitative studies providing rich detail on shifting cultural repertoires as immigrants and their children negotiate the structural environment. These studies are necessarily limited in their generalizability. Conventional large-scale studies of first- and second-generation outcomes have tended to be rich in measures of structural position but have left cultural variation unobserved. The epidemiological approach to culture improves on conventional large-scale models by incorporating both structural and cultural measures in studies of first- and second-generation outcomes. By investigating second-generation women’s agency with respect to partner choice, this study further advances cultural arguments for variation in the economic incorporation of second-generation women. The contrast between less-educated women and more-educated women is especially instructive because education captures both structural location and exposure to egalitarian gender ideologies. We expected education to attenuate the effects of familial origins as proxied by source-country FLFP, and it did. In line with prior studies, we also found the prevalence of mixed unions increased with educational attainment. However, among those women who selected into mixed unions, the associated increase in labor supply was similar regardless of educational attainment, possibly because women who select into mixed unions do so in part to distance themselves from the cultural repertoires that characterized their family of origin. Women’s selection into mixed or endogamous unions reflects opportunity structure, cultural influence, and agency.

Our findings are preliminary in that we cannot pinpoint the specific mechanisms and causal arrows behind the association between endogamy and women’s employment. Unlike country-of-origin measures, we can make no determination of the causal order of partner choice and participation in the labor force. Women who enter the labor force may be more likely to seek out and meet a partner outside their ethnic group who supports their labor force participation, whereas women who are not employed may be limited in their partner choice to members of their same ethnic group. We do, however consider possible explanations. We are inclined to rule out economic necessity, real or perceived, as a confounding factor because women in mixed unions spend more time at work than women in endogamous unions, net of controls for household income and socioeconomic status, whether the partner is foreign-born or U.S.-born, and across ethnic groups with both high and low levels of education. Caregiving responsibilities might also account for the difference. We control for children in the household, and in exploratory analyses, we found no effect of total household size (large households might indicate older family members who require care). We leave a full examination of caregiving in and outside the household for future study. In any case, we see caregiving as consistent with an interpretation of endogamous unions as a vehicle for the maintenance of shared cultural repertoires, including a cultural frame of feminine domesticity.

Studies of adult children of immigrants from the post-1965 era have been hampered by the relative scarcity of data on immigrant groups with smaller U.S. populations, and this study is no exception. The absence of detailed data on ethnicity means that our findings on national origins can only approximate the diversity of ethnic identities in the United States. The CPS-ASEC is the richest source of data on the work and family situation of children of immigrants. Yet, even with a pooled sample over two decades, and despite the increase in cohabitation across all groups, the sample of second-generation women in cohabiting unions proved too diverse for other than tentative conclusions. Another limitation inherent in our study and other epidemiological studies is the reliance on source-country behavioral indicators as cultural proxies for gender norms. It is likely that that these indicators vary cross-nationally in the extent to which they represent cultural ideas about gender and structural opportunities in labor markets.

Despite these limitations, our findings speak to the resilience of cultural repertoires that define women’s work and family lives across generations. Our study also provides insights into shifting trends in gender attitudes in the United States since the 1990s. Cotter et al. (2011) identified a “new cultural frame” for gender attitudes in the United States—one that combines a rejection of male primacy with an essentialist view of domesticity in marriage and motherhood. Children of immigrants from national-origin groups that favor a more traditional gender division of labor may be especially receptive to such a view; indeed, they may have contributed to the emergence of this new cultural frame. As these women age, and the children of younger cohorts of immigrants come of age, second-generation women’s partner choices will have an increasingly important role for both second-generation integration and gender outcomes in the population as a whole.

Acknowledgments

This research was funded in part by the National Science Foundation Grant NSF-SES-1637083 and the Horowitz Foundation for Social Policy. We thank Clem Brooks, Tom DiPrete, Jennifer C. Lee, the Indiana University Immigration Working Group, and the editors and several anonymous reviewers from Demography for helpful suggestions. The ideas expressed herein are those of the authors.

Notes

References