Women in the United States have made significant socioeconomic advances over the last generation. The second generation of post-1965 immigrants came of age during this “gender revolution.” However, assimilation theories focus mainly on racial/ethnic trajectories. Do gendered trajectories between and within groups better capture mobility patterns? Using the 1980 decennial census and the 2003–2007 Current Population Survey (CPS), we observe the socioeconomic status of Latino and Asian immigrant parents and their second-generation children 25 years later. We compare the educational, occupational, and earnings attainment of second-generation daughters and sons with that of their immigrant mothers and fathers. We simultaneously compare those socioeconomic trajectories with a U.S.-born white, non-Latino reference group. We find that second-generation women experience greater status attainment than both their mothers and their male counterparts, but the earnings of second-generation women lag behind those of men. However, because white mainstream women experienced similar intergenerational mobility, many gaps between the second generation and the mainstream remain. These patterns remain even after we control for parenthood status. With feminized intergenerational mobility occurring similarly across race, the racial/ethnic gaps observed in 1980 narrow but persist into the next generation for many outcomes. Both gender and race shape mobility trajectories, so ignoring either leads to an incomplete picture of assimilation.
Following the passage of immigration reform in 1965, migration to the United States shifted from Europeans to Latin Americans and Asians. As a result, the United States is more racially diverse now than any other time in its history (Lee and Bean 2010). As this “new second generation”1 of the post-1965 era of immigration is becoming an increasingly larger share of the U.S. population, there is some disagreement about how much intergenerational mobility they have experienced and how their progress might be structured (Alba et al. 2011; Haller et al. 2011).
Additionally, while the new second generation was entering adulthood, women overall in the United States experienced marked gains in educational and occupational attainment. If we do not account for this parallel transformation, we are at risk of confounding changes in how race structures socioeconomic mobility with expanded educational and labor market opportunities largely for women. Theories of immigrant assimilation then miss the role of the gender revolution in accounting for mobility. Therefore, studies of generational immigrant assimilation require examining the mobility specific to second-generation women.
In this article, we analyze the intergenerational mobility of post-1965 immigrant children in comparison with their parents’ generation by race/ethnicity and gender as well as in comparison with a rising mainstream standard. We find that intergenerational mobility does not necessarily result in closing the gap between racial/ethnic groups. However, second-generation women, particularly Latinas, made greater progress toward closing those gaps than did second-generation men.
Immigrant Assimilation and Divergent Paths
From the earliest inception of the Chicago School’s assimilation theory and its variants (Gordon 1964; Park et al. 1925; Warner and Srole 1945), the assimilation of immigrant groups was evaluated by how successfully they closed the racial/ethnic gap with a white “mainstream,” or U.S.-born residents of three or more generations. The relatively low socioeconomic status (SES) of immigrants and a growing economy centered on industrialization provided the context for robust upward mobility for the second generation and the mainstream. For the latter half of the twentieth century, many immigration scholars argued that the basic tenets necessary for the classic assimilation model no longer exist. The 1965 Immigration and Nationality Act gave preference to highly skilled immigrants, which contributed to the rapid diversification of the immigrant population during a time of tremendous social and economic change. Post-1965 immigration significantly contributed to the racial/ethnic as well as the socioeconomic diversification of the United States, with immigrants from different racial/ethnic groups entering different sectors of the economy. Many Asian immigrants were recruited to fill the high-skilled labor needs of a growing technology sector, and Latino immigrants (largely from Mexico) were recruited to fill the increasing demand for low-skilled occupations. The migration flows of Latinos increased with the rapid growth of the service sector in the 1970s and 1980s. Thus, different immigrant groups entered the United States at different rungs of the labor force, perhaps leading to divergent trajectories: white-collar occupational groups experienced increasing earnings over time, and working-class groups experienced stagnant earnings.
These diverging patterns within the occupational structure subsequently set up what we might see with regard to the socioeconomic assimilation of the children of these immigrants. Evidence suggests that the particular selectivity of Asian and Latino immigrant groups has produced self-sustaining dynamics in which the higher human capital among Asians has produced a positive group effect on second-generation human capital, irrespective of individual parental human capital (Borjas 1992; Feliciano 2006). Conversely, many first-generation Latinos, such as the Mexican immigrants in Telles and Ortiz’s (2008) study, were largely relegated to low-skilled industrial jobs, which did not provide the basis of upward mobility for their second-generation children.
With the growing bifurcation of the economy and racial/ethnic diversification, immigrant assimilation in the post-1965 immigration was commonly measured as a comparison of the attainment of immigrant generations with that of U.S.-born coethnics, or those belonging to the same race group (Kasinitz et al. 2008; Lee and Zhou 2014). Many argue that the non-Latino white “mainstream” no longer resonates as the appropriate reference group. On the other hand, Alba and Nee (2003) described how the American mainstream may be constantly changing and historically contingent, but that it serves as the societal context and structures of opportunity in which immigrant generations are striving for upward mobility. The measurement of immigrant assimilation that includes a changing mainstream helps to distinguish immigrant intergenerational mobility from that of a societal “rising tide that lifts all boats.”
Park and Myers (2010) incorporated a changing mainstream in their approach to immigrant intergenerational mobility. They suggested two dimensions of change that need to be measured in order to more fully capture socioeconomic assimilation: (1) the immigrant intergenerational mobility in status attainment from the first generation to their children’s generation; and (2) the tracking of mainstream changes in the same period because it represents the secular trends that set the context for immigrant mobility. This approach allows for the simultaneous measurement of mobility across generation and the socioeconomic gap between an immigrant generation and the mainstream. With this approach, Park and Myers found that while second-generation Latinos in the United States made significant improvements in socioeconomic indicators compared with the first generation, they did not reach parity with the mainstream. Thus, in essence, intergenerational mobility is diverging along racial/ethnic lines, with less evidence of a narrowing nativity gap between Latinos and non-Latino whites but with more evidence of narrowing gaps between Asians and whites. Previous comparisons of the second generation with the white mainstream have presented a more pessimistic picture with regards to continuing stratification. Social mobility scholars have found less evidence that the second generation is converging with white third-generation Americans, with Thernstrom (1973:257) concluding that “there are definite rigidities in the occupation structure,” and Bodnar (1985:173) concluding that “the status of the father foretold the social positioning of a son.” More current studies of intergenerational mobility have also come to similar conclusions (see Telles and Ortiz 2008). Perlmann and Waldinger (1997) stated that some of the second generation who have lower levels of attainment may share a similar fate with the working class more broadly. Therefore, a model to most appropriately capture intergenerational mobility and/or assimilation of immigrant generations must also include the changes experienced by the reference group in order to distinguish immigrant mobility from secular trends in status attainment.
A major shortcoming of earlier models is that they have ignored gender differences. In the first half of the twentieth century, assimilation theory was originally formulated during a historical period when ethnicity and race were the primary structures of inequality (Gordon 1964; Park et al. 1925; Warner and Srole 1945), but gender was not a consideration (Beller 2009). As Bodnar (1985:173) wrote, “[W]hile instances of immigrant children attaining occupational ranks above those held by their parents existed, however, progress was neither inevitable nor simply a function of time. Women, of course, usually had no long-lasting career outside the home to measure” (emphasis ours). Ignoring gender in models of assimilation and mobility is untenable in an era with rapid increases in women’s own socioeconomic attainments and greater labor force participation. Starting from the latter half of the twentieth century, gender must be included in studies of nativity inequality and assimilation.
Gender Revolution and Second-Generation Women
With economic restructuring and cultural shifts in gender norms of the 1970s through the 1990s, women in the United States experienced increases in educational and occupational attainment (Amott and Matthaei 1996; Cornelia and Northway 2001; Fischer and Hout 2006). Women outpaced men in earning college degrees, becoming the majority of colleges graduates by 1990 (DiPrete and Buchmann 2006; Freeman 2004). Women have lower high school drop-out rates than men in the same racial/ethnic group (Blair and Northway 2001). Women’s labor force participation also increased rapidly (Amott and Matthaei 1996). By the end of the twentieth century, the majority of women were in the paid labor force, and 90 % would enter it sometime during their lifetime (Bianchi and Dye 2001). But despite these gains in educational and occupational attainment, women’s earnings in the United States overall are lower than men’s (Blau and Kahn 2004; Tomaskovic-Devey and Skaggs 2002).
Boyd (1984) conducted one of the earliest studies of differences in immigrant women’s socioeconomic assimilation compared with immigrant men’s advances, with parallel comparisons with the mainstream. She found that immigrant women in Canada faced a “double disadvantage”: their gender and foreign-born status interacted to produce lower earnings and occupational attainment compared with native-born Canadians and men from their same country of origin. She specifically found that the double disadvantage applied to women from origin countries that were considered “less desirable” in Canadian immigration policy—specifically, Asian and Eastern European countries.
Boyd’s (1984) analysis and others who tested the double disadvantage thesis (DeJong and Madamba 2001) included only first-generation immigrants. Evidence suggests, however, that this same dynamic will not apply to the second generation. Survey research has consistently found that second-generation girls perform better in school and have higher academic aspirations than do second-generation boys (Feliciano and Rumbaut 2005; Fuligni 1997; Kao and Tienda 1995; Portes and Rumbaut 2001). Immigrant-specific structural changes have also opened new opportunities for second-generation women that did not exist for the first generation in Boyd’s (1984) study. Increases in service sector occupations and the retracting manufacturing sector have provided new opportunities for 1.5- and second-generation Latinas (who are more likely to work in service sector jobs than the first generation; Myers and Cranford 1998). Tienda et al.’s (1984) work also suggested that economic restructuring in service and blue-collar occupations starting in the 1970s expanded employment opportunities for immigrant women beyond the growth rate of the immigrant workforce itself. Feliciano’s (2008) research indicated that Mexican immigrant women have come to the United States with higher educational levels than Mexican immigrant men in recent decades, suggesting that economic restructuring may be providing better employment opportunities for women.
Despite gains made in their educational and occupational attainment, most research has indicated that immigrant women’s earnings still lag behind mainstream women’s earnings and the earnings of all men. However, results are mixed as to what effect gender has compared with race/ethnicity or nativity. Many scholars have found that gender disparities in earnings are consistently larger than racial or nativity disparities (Avalos 1996; Greenman and Xie 2008; Nawyn and Gjokaj 2014; Sullivan 1984). However, in her study of earnings inequality for highly skilled immigrant women, Lopez (2012) found that nativity status explains more of the earnings gap with U.S.-born men than does gender status.
The presence of children is an important factor when considering gender differences in earnings. This is particularly important when measuring immigrant intergenerational mobility, given that differences between the second generation and their first-generation parents (who are, by definition, mothers or fathers) may be shaped by the presence or absence of children among the second generation. The general pattern in the work and family literature indicates that women with children experience a “motherhood penalty,” in which they earn less than similar women without children (England 2000). This penalty is only partially accounted for by interruptions in labor market experience (Waldfogel 1997) and is largely unrelated to job characteristics, such as level of effort or skill required (Budig and England 2001). The motherhood penalty is not consistent across all women. White women tend to experience larger penalties than black women (Glauber 2007; Hill 1979; Keil and Christie-Mizell 2008; Neumark and Korenman 1994; Waldfogel 1997), with some research indicating that this larger gap is true for only those mothers with more than two children (Budig and England 2001). Non-Latina white mothers also experience larger penalties than Latina mothers (Budig and England 2001; Glauber 2007; Keil and Christie-Mizell 2008). Although little work exists that makes explicit motherhood penalty comparisons between Asian and non-Asian women, research has found that high-skilled Asian women are less likely to reduce their work hours than similarly skilled white women (Greenman 2011). Zhou and Lee (2013) found no motherhood penalty among Asian women.
Conversely, men with children generally experience a “fatherhood premium,” earning more than men without children. Like the motherhood penalty, this premium persists even after controlling for factors, such as hours worked and effort required (Glauber 2008; Lundberg and Rose 2002). The distribution of the premium also is uneven by race, with white fathers receiving a larger premium than black and Latino fathers (Glauber 2008; Hodges and Budig 2010). A range of causal explanations has been explored for both the motherhood penalty and fatherhood premium; for our purposes, we merely present these phenomena as important considerations when determining whether intergenerational mobility is gendered.
Measuring Gendered Immigrant Intergenerational Mobility
On the whole, more research is needed that makes comparisons across generations at different points in time, examines gender dynamics, and uses the mainstream as a referent to truly extricate the different effects of gender, race/ethnicity, and nativity on educational, occupational, and earnings advancement of the second generation. Park and Myers (2010) have started to explore gendered paths of intergenerational mobility, and we build on that work here. Their methodological innovation and findings on gendered paths of mobility are expanded and improved in this article in several critical ways. First, we expand the method to test differences by gender within groups, not just across groups. Second, we also examine personal earnings as part of socioeconomic attainment because previous studies have shown that the progress made by women in education and occupation does not translate to closing the gender gap in earnings. Because Park and Myers (2010) examined only educational and occupational attainment, they missed the aspects of socioeconomic attainment in which women do not catch up to men. Third, the literature illustrates the importance of including the presence of children in analyses of socioeconomic attainment by gender. Therefore, we also include parental status in our analyses to determine whether it works in the same way for second-generation women and men. Our explicit attention to gender in these ways reveals the full dynamics of feminized mobility across immigrant generations.
This study builds on Boyd’s (1984) findings of a double disadvantage, examining whether such a disadvantage exists for second-generation women. Ample evidence suggests that second-generation women will fare better than their immigrant mothers. Is this because of assimilation, in which second-generation women are closing the gap with mainstream women? Or is it due to economic restructuring that has impacted all women, so that the trajectories between first- and second-generation women run parallel to those of mainstream women and their daughters without closing the racial/ethnic gap? Are the observed socioeconomic gains made by daughters largely explained by differing fertility choices from that of their mothers, or are they reflective of actual rising SES? Last, will those gains allow them to catch up to or even surpass second-generation men, such that there is a process of feminized assimilation in which second-generation women are experiencing greater mobility than second-generation men?
We answer these questions by comparing first-generation immigrant mothers with their U.S.-born second-generation daughters, and then comparing these immigrant trajectories with non-Latino whites, or mainstream, third- or later-generation women across different outcomes, with identical comparisons between and with men. In this way, we can ascertain whether intergenerational mobility of immigrant women is due to economic restructuring that benefited all women or true assimilation, and also whether that assimilation looks different than men’s experiences.
For the measurement of SES attainment of both the parents and children at approximately the same ages, they must be observed at different points in time. To accomplish this, we use a data set constructed from the 1980 decennial census 5 % Public Use Microdata Samples (PUMS) and the pooled data from the Current Population Survey (CPS) March Annual Social and Economic Supplement. Because of a relatively small sample size for the second generation in the CPS, we pool 2003, 2005, and 2007 data. Because of the nature of the sampling, taking CPS data from every other year, which hereafter will be referred to as “2005,” allows us to have a larger sample without replicating or throwing out any cases. Also, selecting cases from these years not only allows for observing the second generation in adulthood (conventionally, a 25- to 30-year gap between generations) but also ensures that the attainment of the second generation is not depressed by the impacts of the Great Recession that began in 2008.
The sample is drawn from the entire United States and is based on observing the second generation.2 For 1980, the second generation is observed at ages 0–16 so that we can identify and select their coresiding immigrant parents—the first generation to be in the analyses. For comparability with the second generation, the sample of parents is restricted to ages 25–44. This sample of immigrant parents is then compared with a separate sample of the second-generation adult children in 2005 to measure intergenerational mobility. Only those between the ages of 25 to 41 are selected among the second generation because they are now 25 years older and are approximately the same age that their parents were in 1980. By definition, everyone in the first generation is necessarily a parent. We do not place a parent restriction on the second generation because it may skew the assessment of their overall achievement. Therefore, we present and test the degree of intergenerational mobility separately for those in the second generation with children and those without. Finally, although we select from a cohort of actual parents and children, the data are structured around repeated observations of the second generation (once in 1980 when they were children, and again in 2005 as adults) and are not actually tracing kin between generations.
Because the majority of post-1965 immigrants are from Latin America and Asia, we limit our analysis to these two immigrant groups and their second generation. The SES attainments of Latino and non-Hispanic Asian first and second generations determine immigrant intergenerational mobility, but there is now the issue of defining the American “mainstream” for tracking the changing social standards during the same period. Although a third-generation comparison would be ideal for following intergenerational progress, there was a gap in immigrant generation data after the 1970 decennial census. It was not until the 1994 CPS that the immigrant generation questions appeared again in a large census data set. In addition, a broader definition of the mainstream or comparison group may be in order for the purposes of our analysis because they will serve to proxy societal standards. Therefore, we select U.S.-born, non-Latino whites to reflect the “mainstream” in both 1980 and 2005. Whites are also restricted to the same ages as the immigrant generation in each year. Because our defined mainstream is the majority of the U.S. population, the sample size in both 1980 and 2005 is extremely large and far outweighs the immigrant generations in statistical analyses. Therefore, we take a 2.5 % random sample of U.S.-born, non-Hispanic whites in 1980 from the PUMS data and a 10 % random sample in 2005 from the CPS data, which have many fewer cases. The sample is described in Table 1.
SES attainment is measured by select outcome variables in the following ways: educational attainment is measured by either high school completion or higher, or bachelor’s degree completion or higher. Labor force participation is divided into three categories: not in the labor force, part-time worker, and full-time, full-year worker. Full-time worker is defined as those working 35 or more hours per week. Full-year is defined as those working 45 or more weeks in the previous year. Those who worked less than full-time, full-year are categorized as part-time workers. Occupational attainment is measured as the percentage of workers who are in upper white-collar occupations (those in professional and managerial occupations). Income attainment is measured by the total mean annual personal earnings, with negative earnings coded as missing values. The earnings in 1980 are adjusted for inflation to 2005 dollars using the Consumer Price Index. Occupational and earnings attainment are presented separately for part-time and full-time workers in the descriptive results, while the multivariate analyses present results for only full-time workers.
This article uses and extends the immigrant generation cohort method, or the Park-Myers method (Park and Myers 2010) to examine the socioeconomic advancement for the new second-generation Latinos and Asians, separately for men and for women. The unique foundation for this “immigrant generation cohort” method is the data selection to represent generations. The data essentially contain four groups of people, which include the first and second generations of an immigrant group as well as the mainstream in both 1980 and 2005. These groups are pooled in the data set, and we use two key temporal variables to distinguish them in the model. We have generational status (G), with which first-generation parents from the 1980 sample and second-generation grown children from the 2005 sample are coded G = 1; native-born, non-Hispanic whites from both 1980 and 2005 are coded G = 0. The changes in societal standards and between the generations are represented by Year and its interactions. The main effect of Year represents period change in outcomes for the mainstream reference group between 1980 (Y = 0) and 2005 (Y = 1). The differential effect of passage between immigrant generations is represented by Year × Generation.
Last, age is measured in exact years, center-coded to age 35. The handling of age warrants some discussion here. It is well recorded in the analysis of lifetime socioeconomic attainment; most individuals experience an increase in attainment, such as earnings. Therefore, we have restricted each immigrant generation and the mainstream to approximately the same age range in both 1980 and 2005 so that we are comparing everyone at the same stage in the life cycle. Furthermore, we also include age in the regression model to further control for any age effects that a skewed age distribution might have on the observed attainment levels for any of the four pooled groups.
The two sets of models used in this article deviate from the basic structure of the immigrant generation cohort method to centrally situate gender in the analysis of intergenerational mobility. The first set of regression models—logistic regression for educational and occupational attainment, and ordinary least squares (OLS) regression for mean earnings—offers a more simplified version of the immigrant generation cohort method in that immigrant generations (one for Latinos and another for Asians) are analyzed separately from the mainstream. Instead of comparing immigrant generations with the mainstream, the status and advancement of female immigrant generations are compared with the male counterparts of the same immigrant group. Therefore, the generation variable (G) is replaced with a Women variable (see upcoming Table 3). This model structure teases out the difference in women’s intergenerational mobility compared with that of men in the same immigrant group. Because both women and men of the second generation are achieving their SES during a time when societal standards are also rising, this same model is also shown for the mainstream for the sake of comparison.
The second set of models introduces a more nuanced inclusion of gender into the immigrant generation cohort method. We use the original immigrant generation cohort model to gauge overall intergenerational mobility separately by gender. Conceptually, these two sets of models operationalize the intersectional approach to understanding intergenerational mobility as a process that includes both race/ethnicity and gender. Furthermore, the assessment of socioeconomic attainment separately by gender necessitates the inclusion of parental status because previous studies have shown that it impacts women and men differently.
For all the socioeconomic attainment outcomes, we first present graphs that chart the attainment levels of immigrant generations and the mainstream by gender. The graphic layout allows for the simultaneous assessment of both intergenerational mobility and gaps. Figure 1 compares the educational attainment (high school completion and bachelor’s degree completion) and labor force participation (percentage not in the labor force and percentage full-time workers) of Latino and Asian immigrant generations to that of the mainstream.
For Latinos, we see that the first generation in 1980 had extremely low rates of high school completion. Much of this is explained by the educational system from which they came, where the completion of the eighth grade was considered compulsory rather than being high-school dropouts in the United States. Second-generation Latinos closed much of the gap by 2005, with very little difference by gender. However, the story is very different for attaining a bachelor’s degree. Among the mainstream, women had lower rates of college completion in 1980 than men. By 2005, though, as well documented in the literature, more young-adult women than men had a college degree. Therefore, the societal standard for bachelor’s degree attainment has risen differently by gender. Within this context, the educational attainment trajectory of second-generation Latinas follows more closely to that of mainstream women than Latino men. Like mainstream women, first-generation Latinas in 1980 had lower rates of college completion (5.7 %) than Latinos (8.2 %); but by 2005, there were more second-generation Latinas with a bachelor’s degree (23.3 %) than second-generation Latino men (20.2 %). However, because mainstream women also experienced the same upward mobility from their mothers’ generation and greater mobility than mainstream men, the gap in college achievement between second-generation Latinas and mainstream women remains. Second-generation Latino men closed the gap slightly with mainstream men but only because mainstream men improved so little; while second-generation Latino men experienced greater intergenerational mobility than mainstream men in their same generation, the slope of their improvement was not as dramatic as that of Latina women.
In sum, with high school completion, we observe marked intergenerational mobility regardless of gender and also a substantial closing of the high school attainment racial/ethnic gap with the mainstream. Therefore, with high school completion, we see evidence of both intergenerational mobility and assimilation, with Latino/a men and women closing the racial/ethnic gap with mainstream men and women. However, with college completion, we see a different picture. There is evidence of intergenerational mobility within racial/ethnic groups, with Latinos achieving college degrees at much higher rates than their parents. The college completion gap between Latino and mainstream men became smaller, although it did not completely disappear, indicating that some assimilation has taken place. However, mainstream women experienced similar intergenerational mobility with college completion. Therefore, there is little evidence that the racial/ethnic gap in college completion is closing between Latinas and mainstream women, and thus less evidence of assimilation. Instead, between Latinas and mainstream women, we see intergenerational mobility without assimilation.
As a result of immigration policy, many Asian immigrants arrived with high levels of college degree attainment3 that surpass those of the mainstream. Therefore, racial/ethnic convergence would indicate downward assimilation for the Asian second generation. For education, there is no indication of downward assimilation, but we do see a pattern of intergenerational trajectories that are similar to Latinos but with a higher starting point. Among first-generation Asians in 1980, 48.8 % of women and 60 % of men had a college degree, and almost one-half those respective values for mainstream women and men. In 2005, second-generation Asian women continued to make strides in educational attainment with trajectories very similar to those of mainstream women, albeit at higher levels. At the same time, both second-generation Asian men and mainstream men have not attained much more education than their fathers’ generation. So, as Asian immigrants enter with higher educational attainment than the mainstream, their children continue to have an educational advantage, albeit a smaller one as the mainstream made some headway in catching up. Meanwhile, women across the board closed the gender gap in education within their respective racial/ethnic groups.
Figure 2 shows intergenerational mobility for the percentage in upper white-collar occupations (namely, managerial and professional occupations) and mean personal earnings separately for part-time and full-time workers. For labor force participation rates, we see the same general pattern as with educational attainment, with women participating in full-time white-collar employment at higher rates in the second generation than men. These patterns are consistent with other studies, which have observed a narrowing of the gender gap in formal labor market participation in the United States (e.g., Fischer and Hout 2006). Most notable for our analysis, the slope between the first and second generations is steeper for women than men among both second-generation Latinas and Asians. The end result is that occupational gaps remain between Latinos and the mainstream, although that gap shrinks between the first and second generation, indicating that some assimilation in occupational attainment is taking place. Conversely, Asian men and women in the second generation maintain their parents’ occupational advantage over the mainstream. Both Latina and Asian second-generation women mirror the occupational attainment of mainstream women, so that all three groups of women in 2005 have higher full-time white-collar occupational rates than men in their respective racial/ethnic groups.
Surpassing men in occupational attainment in 2005 did not translate into women surpassing men’s earnings. Among full-time workers, mainstream women between 1980 and 2005 gained twice the increase in earnings as men (an increase of $10,618 compared with $5,169 for mainstream men). Although mainstream women have made stronger progress in the past 25 years, their mean earnings were still $14,574 less than that of mainstream men. This is especially noteworthy given that there is now a higher share of women in high-status occupations than men. Within this mainstream gendered context, we again see that Latina and Asian second-generation women are tracking more closely to the trajectories of mainstream women than the trajectories of men in their racial/ethnic group. Second-generation Latino and Asian men experienced more modest gains in earnings similar to that of mainstream men. In fact, both groups of part-time male workers experienced a decline in mean earnings in the past few decades, while the women experienced modest gains. This may be evidence that men who are more weakly tied to the labor market are in increasingly precarious positions, and yet this decline has now closed the gender earnings gap among part-time workers.
Latina second-generation women working part-time in 2005 earned slightly less than mainstream women working part-time and had slightly less intergenerational mobility than mainstream women. However, full-time Latina workers in 2005 experienced greater intergenerational mobility and earnings that nearly converged with those of mainstream women. By contrast, Latino second-generation men working full-time in 2005 had higher earnings than their fathers’ generation in 1980 but made little progress in closing the earnings gap with mainstream men. This indicates that for Latinos, there was intergenerational mobility but assimilation only for women working full-time.
For Asians, there were few changes in the earnings of part-time women workers between 1980 and 2005, and Asian men’s part-time earnings declined between 1980 and 2005. Full-time workers fared better: both Asian men and women in the second generation had earnings increases over their parents’ generation, with steeper slopes than mainstream men and women during the same time period. Thus, second-generation Asian men and women both experienced intergenerational mobility with an expansion of their earnings advantage over the mainstream.
The graphic displays of intergenerational mobility in Figs. 1 and 2 group those in 2005 who do not have coresiding children with those who do. The parental status of the second generation is important to our comparisons because the first generation are by definition all parents, and thus comparisons between the first generation and the second need to take into consideration the parental status of the second generation because it is associated with significant differences in socioeconomic outcomes. Therefore, we present the different attainment levels for those with children in 2005 separately from those without in Table 2. For most second-generation women, those without children have higher status attainments than those with children. These findings are consistent with the motherhood penalty hypothesis and mirror the patterns found among mainstream women. The exception is Asian mothers, who have higher college completion rates and higher earnings than childless Asian women. For men, mobility patterns are more complicated. Asian and Latino men without children have a higher proportion of college completion and white-collar occupational attainment than Asian and Latino fathers. However, for second-generation men, fathers are more likely to be working full-time and have higher earnings than childless men. Again, these descriptive results show some evidence that the “fatherhood bonus” may also apply to second-generation men, at least with regard to labor force participation and earnings.
The descriptive results reveal several interesting patterns in intergenerational mobility and differences by gender and immigrant group. In all instances, regardless of race/ethnicity, second-generation women experienced greater gains in socioeconomic attainment compared with their mothers than second-generation men experienced compared with their fathers. In educational and occupational attainment in 2005, second-generation Latina and Asian women—like mainstream women—achieved more than men. In earnings, all women have made significant gains compared with their mothers’ generation but still lag considerably behind their male counterparts. Women with children had lower attainment levels than women without children, but we observed a different trend for men. Overall, Latina and Asian second-generation women are making socioeconomic strides that are more similar to mainstream women than to their second-generation male counterparts, with all women’s SES rising at similar slopes between generations. Thus, evidence of intergenerational mobility is prevalent, but evidence of assimilation is more mixed as the gaps between the immigrant generations and the mainstream remain virtually unchanged in the past 25 years across many outcomes. This may be an indication of an overall secular rise in socioeconomic attainments along the lines of gender combined with persistent racial/ethnic stratification. This results in second-generation Latinos having lower socioeconomic levels than the mainstream, while second-generation Asians maintain their socioeconomic advantages. The descriptive results provide a clear and detailed profile of intergenerational mobility as well as mainstream changes. However, they do not provide statistical analyses of whether these observed changes or differences are statistically significant.
The first set of models test the differences in intergenerational mobility by gender separately for Latinos, Asians, and the mainstream. Table 3 presents the results for educational, occupational, and earnings attainment. The Year × Women coefficient equals the difference in the intergenerational slope for women compared with men.
For Latinos, there was a large and significant increase in high school completion from the first to second generation (as noted by the Year coefficient of 2.492). First-generation Latinas in 1980 had virtually the same high school completion as men. But by 2005, second-generation Latinas had even greater mobility compared with Latinos (as noted by the Year × Women coefficient of 0.375). Asian women show similar patterns in greater intergenerational mobility compared with Asian men. The same is true for college degree attainment, with Latina, Asian, and mainstream women in 2005 exhibiting greater intergenerational mobility than men of their same generation.
As with educational attainment, second-generation women also experienced more intergenerational mobility than their male counterparts with upper white-collar occupations and earnings. Even with earnings, Latina and Asian second-generation women experienced greater intergenerational mobility than second-generation men in their racial/ethnic group; mainstream women did not. However, the gains may be bigger for second-generation women even though their attainment level in 2005 may be lower than that of men because their advancement is only in addition to the level at which their mothers started in 1980. For example, second-generation Latino men experienced an $8,375 increase in earnings from what their fathers’ cohort earned. Relative to that increase, second-generation Latinas earned an additional $4,464 in earnings. At first glance, it may seem that second-generation Latinas have higher earnings that their male counterparts. However, to calculate the actual mean earnings of second-generation Latinas, the base mean earnings start with what the first-generation mothers’ cohort earned (which is $12,962 lower than that of the first-generation men). As seen in the descriptive results, even though second-generation women have socioeconomically advanced past men in education and occupation, along with the mainstream women, men still earn more than women.
Overall, women of both the second generation and mainstream experienced greater socioeconomic gains than their male counterparts. This is clear evidence that intergenerational mobility is a gendered process. Table 4 presents regression results that gauge the progress of immigrant generations compared to that of the mainstream distinguished by gender in the same model. Examining intergenerational mobility separately by gender also allows us to observe the differential impact of parenthood for men and women. As stated earlier, the Year variable in these models is a measure of progress from 1980 to 2005, and the generation variable represents the immigrant generations. The interaction term, Year × Generation, represents the additional progress between the first and second generations, net of the mainstream progress in the same time interval. For example, second-generation Latinas experienced significant gains in high school completion (Year × Generation coefficient of 1.111), above and beyond that of mainstream women (Year coefficient of 1.234). As displayed in Fig. 1, these results statistically confirm the observed progress of second-generation Latinas toward closing the high school completion gap with non-Latino white women.
The additional advancement of the second generation is consistent across all outcome indicators for both Latinas and Latinos. However, this additional progress of the second generation (the interaction variable of Year × Generation) net of the mainstream progress does not mean that the second generation is necessarily closing the gap with non-Latino whites. Much of the observed additional second-generation progress is tempered by the relatively low starting point of their immigrant parents. For example, only 5.7 % of immigrant Latinas had completed a bachelor’s degree in 1980 (denoted by a generation coefficient of –1.415). This large and negative coefficient places immigrant Latinas at an even lower attainment level than the low starting point of mainstream women in 1980 (denoted by an intercept of –1.973). Therefore, the positive and significant Year × Generation coefficient of 0.748 is partly making up for the lower starting point of their mothers as well as indicating intergenerational mobility. And given that these are logistic regression coefficients, increments of change at the lower and upper bounds are noted with larger log-odds ratios than for those closer to a base probability of .5. Additional calculations from these regression results are needed, and are therefore presented later in Table 5, to determine whether the second generation’s intergenerational mobility is enough to close the gap with the mainstream.
How does parenthood enhance or detract from these attainments? Motherhood does not affect second-generation Latinas’ high school completion. However, for all other outcomes, second-generation Latinas and mainstream women without children are likely to have higher attainment levels than mothers in 2005. This parenthood effect for men is much different—either insignificant or with the opposite sign of women (see Table 4). For second-generation Latino and mainstream men, educational and occupational attainments do not differ significantly between fathers and men without children. In the case of earnings, there is a fatherhood premium effect, with fathers earning almost $10,000 more than men without children.
Intergenerational progress for Asians follows dynamics similar to those of Latinos, but our estimations begin from much higher statuses in the first generation. Therefore, the models for Asians in Table 4 will consistently yield a positive coefficient for generation. For high school completion and occupational attainment, second-generation Asian women did not progress significantly more than white women. This is also true for Asian men compared with white men. In college completion, second-generation Asian women and men actually experienced lower gains than their white counterparts. Again, this cannot be interpreted as downward mobility because second-generation advancement is relative to the high starting point of their immigrant parents. In other words, the negative Year × Generation coefficients indicate that the mainstream experienced greater mobility than second-generation Asians and are catching up to Asians. In mean earnings, Asian women and men experienced greater gains than did non-Latino whites, thus enlarging the gap they already enjoyed relative to whites.4 Similar to what we observed in the models for Latinos, women without children generally fared better than mothers in 2005, but men without children earned less than fathers.
Returning to the question of how much the gap between the second generation and mainstream has been closed, we calculated the expected probabilities for each group in 1980 and 2005 from the model results in Table 4. This allows us to determine whether second-generation Latinos have narrowed the gap experienced by their immigrant parents from the mainstream and also whether second-generation Asians have lost their advantage over the mainstream. (Because the main goal is to examine racial/ethnic gaps, we do not show the results separately for parents and nonparents.) These socioeconomic gaps by racial/ethnic group and gender are presented in Table 5. In 1980, immigrant Latina mothers’ high school completion was 46.9 percentage points lower than for the mainstream. By 2005, second-generation Latinas have narrowed the gap with non-Latino whites to 9 percentage points (similar results are seen for Latino men). Because the expected values used in this table include only statistically significant coefficients, any narrowing of the gap is statistically significant, as in the case of second-generation Latinas’ college completion. This narrowing of the gap is also evident for occupational attainment and earnings. Similarly, second-generation Latino men across all outcome measures have experienced a narrowing of the gap with non-Latino white men.
As previously shown, Asian immigrants and their second-generation children have higher SES than non-Latino whites across all outcome indicators. Therefore, the calculated gaps between these two groups reveal whether status attainments have become more similar between them. For high school completion, there is a small gap between groups, but white mainstream women and men narrowed the gap in 2005. The sizable gaps in bachelor’s degree attainment decreased for both women and men. Unlike educational attainment, the gaps in occupational attainment slightly increased, but Asian immigrant generations considerably widened the mean earnings gap. Second-generation Asian men more than doubled the gap with white mainstream men, but the increase in the gap was smaller among women. This indicates that the earnings of Asian second-generation men and women may be boosted by their high educational attainment without the depressive effects of being foreign-born that their first-generation parents experienced. Overall, Table 5 shows that although there is considerable intergenerational mobility among the second generation and particularly among women, many gaps between the second generation and the mainstream remain, with Latinos below and Asians above the mainstream.
Discussion and Conclusion
The new second generation of the post-1965 immigration era has come of age during a very dynamic period in U.S. history. In the past few decades, all women—of immigrant origin or not—have experienced substantial gains in higher educational and occupational attainment as well as earnings. The rapidly changing context has been highly gendered, and it would seem fitting that second-generation women would also experience different opportunity structures in education and the labor market. In this article, we have graphically and statistically shown how much intergenerational mobility occurred for second-generation women and men relative to their immigrant parents. We also analyzed whether Latino/a and Asian second-generation women and men closed the gap with non-Latino whites, using multiple outcome indicators to determine whether socioeconomic assimilation is occurring by gender within racial/ethnic groups.
Our findings paint a comprehensive picture of the pattern of feminized intergenerational mobility that has emerged in recent decades and then how this feminized mobility, albeit in some outcomes dramatic, has not produced racial/ethnic assimilation between the Asian and Latino second generation and the mainstream. First, the second generation certainly has achieved higher status than their immigrant parents across all outcome indicators. Therefore, we can clearly state that both second-generation Latinos and Asian Americans, of both genders, have experienced intergenerational mobility. Second, similar to mainstream non-Latino white women in the past few decades, both second-generation Latinas and Asian American women have surpassed the educational and occupational attainments of their male counterparts. Although second-generation women may have experienced challenges or opportunities in the home, school, and the labor market that are distinct from those of mainstream white women, our findings show that the overall rise for mainstream white women applies to second-generation women as well. In other words, both mainstream and second-generation women have successfully closed the gender gap in education and occupation that once prevailed between their mothers and fathers. However, neither mainstream nor second-generation women have completely eliminated the earnings gender gap. All groups of women have only reduced the considerable earnings gender gap that prevailed in their parents’ generation. Nonetheless, the socioeconomic advancement of the second generation from the status of their immigrant parents has really been structured along gender lines.
Third, the differing impacts of parenthood on socioeconomic attainment for mainstream women and men impact the second-generation in very similar ways. Second-generation women face a motherhood penalty for most outcomes, while their male counterparts enjoy a fatherhood premium in earnings. The central focus of this article is not on the effect of parenthood by gender, but its inclusion further substantiates the importance of examining intergenerational mobility patterns by gender. The same motherhood penalty and fatherhood premium patterns observed for the mainstream are also evident for the second generation. In the past few decades, gender has continued to structure socioeconomic opportunities in the United States. To omit the gendered effects of parenthood in assessing the attainments of the second generation would be incomplete at best or inaccurate at worst. Its inclusion in our analysis further confirms that the new second-generation women experience upward mobility regardless of their parental status. Their higher status attainment is not because second-generation women are forgoing motherhood with greater frequency than their immigrant mothers; rather, they are experiencing the same mobility as their mainstream counterparts regardless of parenthood status.
Finally, within this feminized intergenerational mobility, the other most striking finding is that the strong progress of second-generation Latinos/as merely narrows the gap with the non-Latino white mainstream along the lines of either gender or race. For certain outcome indicators, such as high school completion, upper white-collar occupations, and mean earnings, at least one-half of the gaps between Latinos and the mainstream are narrowed. But for bachelor’s degree attainment, second-generation Latinas experienced virtually no closure of the same gap with non-Latino white women that their immigrant mothers experienced decades earlier. Second-generation Latino men closed almost one-half of the educational attainment gap experienced by their immigrant fathers. In contrast, very little convergence with the mainstream was experienced by Asian immigrant generations because they largely maintained their parents’ advantage relative to the white mainstream. There is some narrowing of the race gap in bachelor’s degree attainment but virtually no convergence in occupational attainment. In fact, the earnings race gap between Asians and the mainstream has increased for both men and women.
Overall, every racial/ethnic group has experienced somewhat similar secular advancement in socioeconomic attainments; and because of that, the racial/ethnic gaps continue to exist for some socioeconomic indicators. Although the second generation is not falling behind the mainstream, as might have been predicted by proponents of downward or segmented assimilation, there is limited evidence of socioeconomic convergence as might be expected by the new assimilation theory (Alba and Nee 2003). Contrary to what Thernstrom (1973) and Bodnar (1985) suggested, the starting point of the parents does not completely limit the mobility of their children, but the low starting point of Latinos makes their significant mobility insufficient to eliminate the gaps with the mainstream. Further, many of the gains made by Latinas across generations are offset by similar intergenerational gains made by mainstream women, leading to the continuation of racial/ethnic gaps in the second generation. However, the mobility of Asians follows more closely with Telles and Ortiz’s (2008) assertion that mobility may be more salient for those with higher SES. The higher SES of Asian immigrants and their children will continue to challenge current conceptions of assimilation and intergenerational mobility. It can no longer be assumed that immigrants necessarily start at the socioeconomic bottom. However, their relatively high SES cannot be equated to social acceptance or acculturation with the white mainstream (Kim 1999). Certainly, the case of Asian Americans introduces the possible delineation of socioeconomic attainment from social acceptance, which reemphasizes the multiple dimensions of racial/ethnic stratification (Kim 1999).5
Our findings diverge from Boyd’s theory of double disadvantage. Contrary to her findings on first-generation immigrant women, second-generation Asian and Latina women outperform second-generation men from their respective racial/ethnic groups. Further, our study indicates that gender stratification, while still present, may be more malleable than racial/ethnic stratification. Educational and occupational attainment in particular have been more open to women’s advancement, while racial/ethnic stratification within these sectors remains largely intact. However, increasing gender equality in educational and occupational attainment has not translated to the closing of the gender earnings gap, reflecting the well-documented stalled gender revolution in the 1990s (Cotter et al. 2011; England 2010). Whether women will eventually close the earnings gap or if their advancement will plateau remains to be seen.
This article answers many questions about immigrant assimilation but also raises additional questions that warrant further research. First, we have combined all Latino and Asian groups as if they were monolithic racial/ethnic groups. Both Latinos and Asians are heterogeneous groups with varying circumstances of migration, differing economic and community resources, and residence in different regional labor markets. Although there are data limitations due to small sample sizes or lack of Asian ethnicity distinction in the case of the CPS, replication of these analyses for specific ethnic groups will build our knowledge about how intergenerational mobility works. For instance, Park et al. (2014) showed that the Mexican-origin second generation experienced significant intergenerational mobility even though their parents arrived with relatively low socioeconomic attainment. This would be especially important for investigating trends for Latino and Asian immigrant groups that arrive with different levels of socioeconomic attainments than what might be assumed for those groups.
The recognition of specific ethnic groups within panethnic Latino and Asian groups also raises the issue of undercount for some groups more than for others and its impact on intergenerational mobility estimations. For example, Ibarraran and Lubotsky (2007) used both the 2000 Mexican and U.S. census data to determine the SES of Mexican migrants. They found that Mexican migrants captured in the U.S. census are better educated than migrants in the Mexican census. They speculated that this may be partially due to an undercount of young undocumented, lower-skilled Mexicans in the U.S. census. It is difficult to pinpoint the exact magnitude or nature of undercounting for each ethnic group—especially for historical data—but it must be acknowledged that generally, those with the lowest socioeconomic attainments are the most likely to be undercounted. Therefore, if anything, our estimations of intergenerational mobility are underestimated or conservative if lower-skilled migrants are more likely to be undercounted.
Second, we have applied our model to a limited set of outcome indicators. Other indicators of assimilation, such as intermarriage or racial/ethnic identification, can be examined to extend and to connect socioeconomic mobility with social integration and identity. For example, Emeka and Vallejo (2011) as well as Duncan and Trejo (2011) have documented that U.S.-born persons of Mexican descent with the highest socioeconomic prospects are less likely to identify themselves or their children as ethnically Hispanic than are Mexican immigrants. Future research can examine whether this issue of “ethnic attrition” for those with the highest SES exists for groups other than Mexican Americans. For our analyses, ethnic attrition would again suggest a possible underestimation of intergenerational mobility but would not change the overall findings.
Third, some scholars argue that mobility between the first and second generation is inevitable across all groups but that socioeconomic stagnation really occurs for the third generation. Given that the second generation of the post-1965 immigration era just entered adulthood in the past few decades, it stands to reason that their third-generation children have yet to reach adulthood. Future research will be needed to determine the long-term trajectories for the descendants of immigrants who entered the United States in the 1960s and 1970s.
Fourth, linking actual family members is critical to determining direct first- to second- generation transmission. This would allow scholars to measure the kinds of parental resources that are available to different groups of the second generation. A vast literature has examined the varying economic as well as social capital or community resources that might be available for some children but not for others. These data do not allow such linking, so longitudinal data sets are essential for measuring the mechanisms of intergenerational transmission. Such data sets should include other outcome indicators of social or cultural acceptance, which may be an area where Asians continue to experience a disadvantage compared with the mainstream.
Finally, the observed feminized mobility and persistent gender or racial/ethnic gaps are historically and contextually specific. Although the United States has a large immigrant population, immigrants are an even higher share of several other countries, such as Canada and Australia. Further, the existence and extent of a gender revolution varies greatly across the globe, as does the racial and ethnic diversity. Policies and their impact to address immigration or social inequalities also differ greatly by country (Donato et al. 2008). Therefore, the dynamics of immigrant assimilation and racial/ethnic stratification in the United States may differ from those in other countries (Boyd 2002). Comparisons of similar migration flows and settlement processes in different country contexts are useful analytic tools to determine how cultural and sociopolitical contexts shape migrant integration (Bloemraad 2013). Thus, future studies that include a cross-national comparison would better determine whether the processes of intergenerational mobility and stratification in the United States are consistent with those found in other countries or whether they reflect the uniquely American historical, social, and economic context.
Ultimately, our findings provide definitive answers to selected questions about general patterns in mobility and socioeconomic assimilation of Latino/a and Asian second-generation immigrants in the United States. We demonstrate that intergenerational mobility is now feminized, with greater mobility for women, but this does not translate into complete assimilation within racial groups for second-generation immigrants. These findings illustrate the critical importance of explicitly examining gender differences in intergenerational mobility rather than combining men and women together or merely using gender as a control variable. Future research should include measures of feminized intergenerational mobility, and studies of the third generation should examine whether the greater mobility experienced by Latinas and Asian second-generation women compared with their male counterparts carries forward into the next generation.
An earlier version of this article was presented at the 2011 meetings of the American Sociological Association. We thank our anonymous reviewers, Min Zhou, and Reeve Vanneman for their helpful comments and suggestions.
We use the word “new” to distinguish the second-generation of the post-1965 immigration era from the second generation of the turn-of-the-century immigrants or other previous waves of immigrants (Portes and Zhou 1993).
Following the method Park and Myers (2010) used and for the purposes of this article, we define the second generation as a U.S.-born person who has two foreign-born parents. Those with only one foreign-born parent (referred to as the “2.5 generation”) have significantly different attainment levels than the true second generation (Ramakrishnan 2004). For an in-depth discussion of alternative definitions of the second generation, see Oropesa and Landale (1997).
During the 1960s and 1970s, many Asian immigrants were admitted to the United States as high-skilled workers needed to fill the growing labor demands in science and technology. This positive selection of the highly skilled from Asia has continued for decades with a marked increase in the past decade (Pew Research Center 2013).
We tested a separate model for Asian women only to clarify the earlier anomaly found in the descriptive results of second-generation Asian mothers earning more than those without children. The difference in earnings for second-generation Asian mothers and nonmothers is not statistically significant.
Kim’s (1999) theory of racial triangulation articulates the racial positioning of Asian Americans on two axes: relative valorization, and civic ostracism. Although Asian Americans may be valorized over other racial groups, they are treated “as immutably foreign and unassimilable with Whites on cultural and/or racial grounds” (Kim 1999:107).