The Economic Assimilation of Second-Generation Men: An Analysis of Earnings Trajectories Using Administrative Records

Abstract

Previous research on the economic assimilation of recent U.S.-born children of immigrants who form the new second generation has disproportionately focused on their educational attainment and other early-life outcomes. In this study, we examine the earnings trajectories of second-generation men through a large part of their adult lives using a unique dataset that links respondents from more than two decades of the Current Population Survey to their longitudinal tax records. This longitudinal information allows us to compare the progress second-generation men of different race and ethnicity make in narrowing the earnings gaps with later generations. We consider the extent to which differences in educational attainment and in early occupational placement affect the earnings trajectories of second-generation men. New second-generation men as a whole experience considerable earnings mobility during their lifetimes. However, we also find large differences by race and ethnicity that cannot be fully explained by educational attainment. Second-generation Hispanic men in particular begin their careers with an earnings deficit relative to later-generation White men and fall further behind. Thus, the stalling or reversal in Hispanic economic assimilation appears to begin during the course of the second generation rather than in later generations as previously thought.

Introduction

Immigrant assimilation has long been recognized as a multigenerational process (Bean et al. 2015; Brubaker 2001; Gordon 1964; Warner and Srole 1945; White and Glick 2009). Despite the difficulties they encounter upon arrival, first-generation immigrants make considerable progress narrowing the socioeconomic gap with the native-born during the course of their working lives. Foreign-born men, for example, have been shown to experience rapid earnings growth, often exceeding that of similarly qualified natives (Lubotsky 2007; Villarreal and Tamborini 2018). However, because immigrants who arrive as adults cannot easily make up for their lower levels of education or the disadvantage of a foreign degree, they often face a limit to their upward mobility. It is instead left to their children and later generations to reach socioeconomic parity.

Researchers have therefore frequently turned their attention to the progress made by the children of immigrants, who make up the second generation. Recent studies have been especially concerned with the fate of the so-called “new second generation” made up of U.S.-born children of immigrants who arrived primarily from non-European countries after 1965 (Farley and Alba 2002; Haller et al. 2011a; Kasinitz et al. 2008; Luthra et al. 2018; Portes and Rumbaut 2001). This new generation is thought to face greater challenges in narrowing the socioeconomic gap with later-generation natives than have previous immigrant cohorts because they confront a U.S. economy that provides fewer opportunities for workers with lower levels of education, and because they encounter discrimination and other forms of disadvantage as members of racial and ethnic minorities (Portes and Rumbaut 2014).

Despite the growing interest in the progress of the new second generation, we know surprisingly little about how members of this generation are faring relative to their parents and over the course of their own lives with regards to earnings. Research on the new second generation has instead focused on other socioeconomic conditions, such as educational attainment. Yet earnings are a key measure of assimilation because they significantly affect many other outcomes used to assess the integration of immigrants and their descendants, including marital and residential patterns (Owens 2016; Schwartz 2013), health and well-being (Chetty et al. 2016; Hummer and Hamilton 2019), and the educational attainment of children (Brooks-Gunn and Duncan 1997; Duncan et al. 2014).

One reason for the relative scarcity of research on the earnings of the new second generation is that until recently children of this generation were too young to be observed beyond their initial incorporation into the labor force (Haller et al. 2011b; Kasinitz et al. 2008; Portes and Rumbaut 2001). Early work on new second-generation individuals therefore tended to focus on their educational experiences during adolescence (e.g., Kao and Tienda 1995; Portes and Hao 2004; Portes and Rumbaut 2001). A key difference between the study of educational and earnings assimilation is that the latter requires an analysis of the progress made by workers throughout their adult lives. Because most individuals complete their formal schooling by young adulthood, educational attainment is almost entirely shaped by early-life experiences. By contrast, workers may continue to face disadvantages related to discrimination and unequal treatment in the labor market during their entire careers.

Theories of assimilation have likewise emphasized the obstacles faced by the new second generation during their childhood and adolescence. A leading perspective known as segmented assimilation theory warns of the risks of downward assimilation for second-generation minority youths (Portes and Rumbaut 2001; Portes and Zhou 1993). The unsuccessful incorporation of their immigrant parents and children's own rapid acculturation and exposure to adversarial attitudes toward school are considered key risk factors preventing second-generation children's educational advancement (Portes and Rumbaut 2001). Yet second-generation individuals may continue to fall behind economically during the course of their working lives in ways that cannot be fully explained by differences in education.

Past research on the new second generation's earnings trajectories has been hampered by data limitations. Longitudinal information is required to follow individuals' earnings over the life course, yet it is rarely available for a representative sample of the second generation. Instead, studies often rely on cross-sectional surveys to compare the earnings of the new second generation to those of a reference group (Cadena et al. 2015; Duncan and Trejo 2018; Waters and Pineau 2015). Such studies must approximate life trajectories by using information from many individuals at different points in their lives.

In the present study, we examine the earnings trajectories of second-generation men. Our analysis relies on a unique dataset that links respondents from over two decades of the Current Population Survey (CPS) to their longitudinal tax records. This longitudinal information allows us to measure new second-generation men's progress in reducing the earnings gap with men of later generations during the course of their working lives. We examine ethnoracial differences in second-generation men's earnings trajectories and assess the extent to which these differences may be explained by educational disparities and early occupational placement in the labor market.

Immigrant Socioeconomic Assimilation

One of the earliest theoretical perspectives on immigrant assimilation, known as the straight-line or classical model, posits the eventual disappearance of social and cultural differences between immigrants and their descendants and the rest of society (Park and Burgess 1921/1969; Thomas et al. 1921/1971; Warner and Srole 1945). Although the initial formulation of the straight-line model referred to the incorporation of immigrants into mainstream society and the cultural convergence between immigrants and natives, researchers have more recently interpreted it to also imply a convergence in socioeconomic status (Greenman and Xie 2008; Villarreal and Tamborini 2018; White and Glick 2009). According to the straight-line assimilation model, each immigrant generation makes consistent progress in narrowing the socioeconomic gap with later-generation native-born individuals during the course of their lives and creates greater opportunities for their children. After several generations, the descendants of immigrants are expected to essentially reach full socioeconomic parity.

The straight-line assimilation model was originally inspired by the experiences of immigrants who arrived during the great wave of European migration in the late nineteenth and early twentieth centuries. Critics of this model argued that it fails to capture the socioeconomic trajectories of more recent immigrants arriving from non-European countries (Portes 1997; Portes and Rumbaut 2001). These new immigrants are thought to face diminished opportunities because they encounter discrimination and other structural impediments in the labor market. Bean et al. (2015) referred to this alternative to straight-line assimilation as the ethnic disadvantage model (see also Bean and Stevens 2003). Although the ethnic distinctiveness of new immigrants hampers the assimilation of new arrivals according to the ethnic disadvantage model, it leaves open the possibility that immigrants and their descendants may eventually reach economic parity with later-generation Whites. Other researchers see a hardening of distinctions between the descendants of immigrants and the rest of society into racial boundaries that more permanently limit socioeconomic assimilation. In their multigenerational analysis of Mexican Americans, for example, Telles and Ortiz (2008) found evidence of a lack of progress in educational attainment beyond the second generation, which they attributed to a process of racialization.

One of the most influential critiques of straight-line assimilation, known as segmented assimilation theory, incorporates elements of the ethnic disadvantage model (Portes and Fernández-Kelly 2008; Portes and Zhou 1993; Zhou 1997). According to segmented assimilation theory, immigrants belonging to racial and ethnic minority groups follow bifurcated paths toward socioeconomic integration. While some children of immigrants experience full socioeconomic incorporation into the middle class, others are at risk of downward assimilation. Whether second-generation children experience upward or downward assimilation depends on a complex set of factors that include the way in which their immigrant parents are incorporated upon arrival, the pace of acculturation of children relative to their parents, and the economic and cultural barriers children encounter (Portes and Fernández-Kelly 2008; Portes and Rumbaut 2001). Because many children of immigrants grow up in disadvantaged settings, they are at risk of exposure to “norms of behavior inimical to upward mobility as well as an adversarial stance that justifies these behaviors” (Portes and Rumbaut 2001:61).

Segmented assimilation theorists and other early researchers examining the socioeconomic conditions of the new second generation focused especially on children's educational attainment and other early-life outcomes. This is understandable given the young age of second-generation children at the time the research was conducted. For example, 90% of second-generation individuals in Portes and Rumbaut's (2001) classic study were under the age of 16.¹ However, in its original form, segmented assimilation theory left undertheorized the kinds of disadvantages minority second-generation children face in the labor market later in life. In other words, it did not sufficiently explore intragenerational mobility among the second generation.

In contrast to theories of intergenerational mobility, which focus on parental background and early-life experiences as predictors of children's outcomes later in life (Blau and Duncan 1967; Haller and Portes 1973; Sewell et al. 1969), theories of intragenerational mobility emphasize the continued importance of individual-level factors during adulthood, as well as exposure to new structural forces (Kalleberg and Mouw 2018; Rosenfeld 1992). Discrimination can continue to play an important role in reducing second-generation minority workers' upward mobility within the labor market (Pomer 1986). Structurally, some accounts of intragenerational mobility focus on the operation of internal labor markets (Althauser and Kalleberg 1981; Doeringer and Piore 1971; Rosenfeld 1992). Mechanisms of social closure allow jobs clustered in some groups of firms or occupations to provide greater opportunities for advancement and skill acquisition. Minority immigrants and their children may experience lower intragenerational mobility if they are left out of these parts of the economy. The bifurcated paths of assimilation for children of minority immigrants proposed by segmented assimilation theory may continue and further diverge in adulthood independently of education and early-life experiences if second-generation workers are relegated to jobs with lower earnings growth. In our analysis, we explicitly examine whether differences in the earnings trajectories of the new second generation can be explained by their educational attainment and by their exclusion from occupations with high earnings growth early in their careers.

Earnings Disparities Among the New Second Generation

Despite the concerns of segmented assimilation theorists and others, empirical studies have demonstrated that children of immigrants generally perform quite well and often exceed their later-generation White peers in educational attainment (Alba and Holdaway 2013; Crosnoe and López Turley 2011; Lee and Zhou 2015; Waters and Pineau 2015; White and Glick 2009). Using cross-sectional information from the CPS, for example, Duncan and Trejo (2018) found that second-generation men and women overall have nearly identical years of education as later-generation Whites, and similar high school dropout and college completion rates. However, the high overall educational attainment among the second generation obscures important differences by race and ethnicity and by national origin. Whereas second-generation Asians often attain higher levels of education than later-generation Whites (Lee and Zhou 2015; White and Glick 2009), second-generation Hispanics, and particularly Hispanics of Mexican and Central American ancestry, lag behind (Duncan and Trejo 2018; Portes and Hao 2004). Second-generation Mexican American men have approximately one year less of education, more than twice the high school dropout rate, and half the college completion rate as later-generation Whites (Cadena et al. 2015).

Comparatively less research has examined the earnings of second-generation individuals rather than their educational attainment. Yet available evidence suggests that the ethnoracial disparities in the earnings assimilation of second-generation men parallel those for educational attainment. Second-generation Asian men have been found to have slightly higher earnings than later-generation White men, while second-generation Hispanic men have lower earnings (Duncan and Trejo 2018). Waters and Pineau (2015:271–272) reported that the weekly earnings of first-generation Hispanic men are 54.4% lower than those of contemporary third-plus-generation White men, while those of second- and third-plus-generation Hispanic men are 21.6% and 22.1% lower, respectively. Overall, their results suggest a stagnation in the intergenerational earnings mobility of Hispanics beyond the second generation that parallels that found for educational attainment.

An important limitation of these estimates of ethnoracial disparities in earnings among the second generation is that they are based on static differences evaluated by grouping all individuals in a broad age range (e.g., those aged 25–59) (Cadena et al. 2015; Duncan and Trejo 2018; Waters and Pineau 2015). Studies using this approach ignore the earnings growth experienced by the second generation during the course of their lives and how that growth may contribute to a narrowing or widening of ethnoracial disparities (Tomaskovic-Devey et al. 2005). The earnings trajectories of second-generation individuals may vary by race and ethnicity for several reasons. First, as already noted, previous research has found significant differences in educational attainment among the second generation by race and ethnicity (Waters and Pineau 2015). Education has in turn been found to affect earnings growth over time (Cheng et al. 2019; Villarreal and Tamborini 2018). The lower education of second-generation Hispanics in particular may result in slower earnings growth later in life.

Second, as discussed in the preceding section, theories of intragenerational mobility often emphasize the role of internal labor markets (Althauser and Kalleberg 1981; Doeringer and Piore 1971; Rosenfeld 1992). Children of immigrants from disadvantaged ethnoracial groups may be excluded from occupations that provide greater opportunities for advancement and skill acquisition, resulting in slower earnings growth. Tomaskovic-Devey et al. (2005) demonstrated that ethnoracial discrimination may influence earnings inequality over the life course by depriving Black and Hispanic men of employment opportunities that allow for greater accumulation of human capital. A similar pattern of exclusion from jobs characterized by higher skill acquisition and earnings growth may disadvantage second-generation Black and Hispanic men. In the following analysis, we will specifically test whether differences in education and in the occupation in which individuals start their careers explain the observed disparities in earnings growth among the second generation.

Ethnoracial differences in earnings trajectories among the second generation may also be the result of an intergenerational process that leads to greater opportunities for some children of immigrant parents. For example, prior research has found that Black and Asian immigrants have higher levels of education compared with others in their countries of origin, and sometimes even compared with the U.S. population as a whole (Hamilton 2019; Lee and Zhou 2015). As Lee and Zhou (2015) noted, this “hyperselectivity” in educational attainment among Asian immigrants improves the educational outcomes of second-generation children because it encourages the formation of cultural institutions that foster high academic standards for their children. By contrast, they describe the condition of Mexican immigrants as one of “hyposelectivity,” based on their low college completion rates. The degree of selectivity of immigrant parents may also influence the earnings trajectories of the second generation not only through its effect on educational attainment but also by shaping children's career aspirations.

Previous research using both contemporary and historical data has found higher rates of intergenerational mobility between first- and second-generation individuals than between parents and children of nonimmigrants (Abramitsky et al. 2021; Chetty et al. 2020). While research on intergenerational mobility also adopts a static view by evaluating parents' and children's earnings at a particular age rather than considering their respective earnings growth over time, the socioeconomic attainment of parents may also be associated with the earnings trajectories of children.

In our analysis, we examine the intragenerational earnings mobility of new second-generation individuals. Our unique longitudinal dataset linking survey respondents to their annual tax earnings allows us to overcome many of the methodological limitations of previous studies. Specifically, we are able to follow the earnings trajectories of second-generation men of different race and ethnicity through a large part of their adult lives to compare the progress they make in narrowing the earnings gap with later-generation Whites and with later-generation men of their same race and ethnicity. We explore differences in the earnings trajectories within ethnoracial groups by national ancestry. We also consider the extent to which differences in educational attainment and in early occupational placement affect the earnings trajectories of second-generation men.

Data and Methods

Sample and Measures

Data for our analysis are drawn from the March CPS files for the years 1996 to 2018.² Respondents of the CPS have been matched with longitudinal administrative records compiled by the Social Security Administration (SSA). Because the CPS is a rotating panel survey in which individuals are sampled eight times over the span of 16 months, some respondents may be sampled in two consecutive March waves. To avoid duplicate cases we keep only individuals living in households who are in their last four of the eight interviews, except in 2018. Individuals residing in all households are kept in that last year since there is no risk of duplication.

We restrict our sample to men born between 1970 and 1979 to approximate the new second-generation cohort. To ensure, insofar as possible, that the men selected have completed their formal education by the time they are surveyed by the CPS, we further restrict our sample to respondents who were at least 25 years of age at the time they were interviewed. Our analysis is further limited to noninstitutionalized men at the time of the CPS interview. We exclude women from our analysis because complex interactions among nativity, work, and family require a different analytic strategy to examine women's earnings. Among other things, an analysis of second-generation women's earnings trajectories would require us to account for their selective participation in the labor force in response to life events such as marriage and childbearing.

Respondents from the CPS in each year have been merged with their annual earnings contained in SSA Detailed Earnings Records (DER) based on agreements between the SSA and the Census Bureau (Davies and Fisher 2009; Duleep and Dowhan 2002).³ Once CPS respondents are matched with these records, we can obtain their annual earnings for the years 1980 to 2019 to create a person–age file we can use to examine their long-term earnings trajectories from ages 25 to 45. The dependent variable in our statistical models is individuals' logged annual earnings. The DER file provides CPS respondents' total annual earnings for all jobs as reported on employer-provided W-2 forms submitted to the Internal Revenue Service (IRS). Total earnings in the DER file include salaries, wages, and other compensation from all jobs. Because self-employment earnings are obtained from a different data source, we include as a predictor a dummy variable indicating whether an individual receives any earnings from self-employment.⁴ All earnings are converted into 2020 dollars using the CPI-W series. We exclude earnings from workers who are only marginally attached to the labor market by including only individuals whose annual earnings are $5,000 or more in a given year (3.7% of all person-years). Similarly, to reduce the effect of extreme outliers at the top of the income distribution, we cap earnings at the 99.5 percentile (0.7% of all person-years).⁵

Linked administrative records provide significant advantages for the study of earnings assimilation. First, administrative records contain repeated measures of the same individuals' earnings over their adult lives, allowing us to analyze their earnings trajectories. Second, administrative records have been shown to yield more accurate estimates of individuals' earnings than self-reported values from national household surveys (Kim and Tamborini 2014). Tax earnings are not affected by proxy reporting and nonresponse (Tamborini and Kim 2013). Finally, earnings obtained from administrative records contain few missing values and no sample attrition over a long period of time.

However, one potential limitation of using earnings from administrative records linked with surveys such as the CPS is the exclusion of respondents who cannot be matched with their earnings records. Fortunately, the rate of successful matches is high for all CPS years. The overall match rate for our sample is 85.4%. Previous research using data matching survey respondents with their tax earnings shows that the risk of bias due to differences in match rates across sociodemographic groups is low (Czajka et al. 2008; Davis and Mazumder 2011). Nevertheless, following previous studies, we created modified weights to correct for differences in the rate at which individuals are matched with their earnings files (Cheng et al. 2019; Villarreal and Tamborini 2018). To construct these weights we first estimated a logistic regression model of a match using individuals' characteristics as predictors for each CPS year. We then multiplied the original person weight provided by the CPS by the inverse of the probability of a match given an individual's characteristics.⁶ As shown in Table 1, once the adjusted weights are used, the social and demographic characteristics of second-generation men of each race and ethnicity in the matched sample are remarkably close to those of men in the full sample, which includes nonmatched cases. The mean and median earnings from the CPS for the year in which the men were interviewed are also similar in the matched and full samples.

The CPS questionnaire asks respondents for their country of birth and the country of birth of both their mother and father. We define as first generation any individual who was born outside the United States and its territories and is not a child of a U.S. citizen. Individuals born in Puerto Rico are not considered immigrants. We define as second generation any individual who was born in the United States with at least one parent born abroad (Cadena et al. 2015; Duncan and Trejo 2018; Farley and Alba 2002; Waters and Pineau 2015).⁷ All individuals who are neither first nor second generation are defined as third-plus generation and are used as the reference category against which the second generation's earnings trajectories are compared in our analysis of intragenerational mobility. Our sample excludes first-generation immigrants, since we are interested in comparing the earnings trajectories of second-generation men to those of later generations.

We distinguish four pan-ethnoracial categories: non-Hispanic White, non-Hispanic Black, non-Hispanic Asian or Pacific Islander, and Hispanic (of any race).⁸ We use these pan-ethnoracial categories in most of our analyses for comparability with previous studies and because these categories allow us to explore how the process of economic assimilation of immigrant generations helps reproduce existing ethnoracial disparities in the United States, which are often estimated using these broad categories (e.g., Smelser et al. 2001; Tomaskovic-Devey et al. 2005). Nevertheless, we also test models separating Asian and Hispanic second-generation men into more detailed ethnic or national ancestry categories.⁹

To classify second-generation Asian and Hispanic men into their specific ethnic or national ancestry groups, we relied on information from their parents' country of origin. We classified an individual as belonging to a specific ethnic or national ancestry group if they were classified in the corresponding broader pan-ethnoracial category and either parent was born in a corresponding country of ancestry. Thus, a second-generation individual who identified as Asian American and who had a mother or father born in Korea was classified as Korean American. Because an individual may have parents belonging to different ethnic or national ancestry groups, it is possible for the same individual to be classified in two different ethnic or national ancestry groups. That is, the categories are not fully mutually exclusive. This approach is preferable to choosing only one ethnic category for these multiethnic individuals.

Because some specific ethnic or national ancestry categories were too small to disaggregate in our analysis, we limit our models examining differences among specific groups to the four largest Asian and Hispanic groups. Specifically, our model for second-generation Asian men is limited to those classified as Chinese, Korean, Indian, or Filipino. Together these four categories constitute 74.1% of all second-generation Asian men in our sample. Similarly, our analysis for second-generation Hispanic men is limited to those classified as Mexican, Central American, or Cuban. These three Hispanic categories encompass 84.3% of all second-generation Hispanic men in our sample.

We use other variables from the CPS record in our statistical models. Educational attainment is coded using three categories: did not complete high school, completed high school, or completed college or more. As noted in the following, we test models including these educational categories as predictors. We also test separate models for men with and without a college degree. To examine the effect of individuals' occupation at the start of their careers, we coded occupations into 364 occupational categories based on the 1990 census classification system. We harmonized occupations across CPS years using information from the Integrated Public Use Microdata Series (IPUMS) (Flood et al. 2022).

Modeling Strategy

We model individual's earnings over time using random-effects models of the form

where $yit$ are the earnings for individual i in year t; $ageit$ is a set of seven binary variables corresponding to age intervals (25–27, 28–30, 31–33, 34–36, 37–39, 40–42, and 43–45)¹⁰; the age dummy variables are fully interacted with the ethnoracial and immigrant generation categories omitting third-plus-generation Whites as the reference category (i.e., second-generation Hispanic, third-plus-generation Hispanic, etc.); $Xit$ are control variables that include an indicator for self-employment described earlier, a dummy variable for recession years to adjust for the effect of macroeconomic conditions, and dummy variables for each of the CPS years from which individuals in our sample were extracted; $νi$ is an individual-level error term; and $εit$ is an individual-time-specific error term. This specification allows us to model men's earnings over their life course without imposing a specific functional form such as a polynomial function. The interactions between age and generation–race allow the trajectories to fully vary for each group in our sample. We avoid including stand-alone predictors for generation–race by interacting all age categories with each ethnoracial and immigrant generation category.¹¹

To examine whether the differences in earnings trajectories for second-generation men are explained by the occupation in which they begin their working lives, we test separate models for men who are only 25–35 years of age at the time of the CPS. We control for their occupation using 364 occupational categories based on the 1990 census occupational classification system. We limit this portion of our analysis to a comparison between second-generation Hispanics and third-plus-generation Whites because we have an insufficient number of cases of other second-generation ethnoracial categories of men when the sample is reduced by selecting only those interviewed at younger ages.

Results

Descriptive Results

Figure 1 shows the median annual earnings of second-generation men in our sample using three-year intervals from ages 25 to 45. The median earnings of third-plus-generation Whites are also shown as a reference category. We see substantial ethnoracial differences in earnings trajectories. Second-generation White and Asian men experience greater earnings growth during the course of their working lives. The median earnings of these two groups in fact exceed those of third-plus-generation Whites for the entire age range. Second-generation Black men also experience rapid growth but they begin their careers with a greater disadvantage. By contrast, second-generation Hispanic men experience lower earnings growth and fall further behind during the course of their lives.

Multivariate Results

Table 2 shows the results of our random-effects model comparing the earnings of all men in our sample to those of the reference category made up of third-plus-generation Whites at all age categories. The coefficients in the top panel compare the earnings of second-generation men of each race and ethnicity, while the bottom panel compares the earnings of third-plus-generation men. The coefficients indicate significant ethnoracial differences in the earnings trajectories of second-generation men. Black and Hispanic second-generation men begin their working lives with earnings that are 17.4% and 11.0% lower than those of third-plus-generation Whites, respectively (equivalent to $6,387 and $4,030). However, while second-generation Black men reduce the earnings gap with third-plus-generation Whites during the course of their working lives, Hispanic men fall further behind such that by age 43–45 they earn 14.5% less (equivalent to $5,308). The lower earnings growth of second-generation Hispanic men is important because it indicates that the stalling or reversal in the socioeconomic assimilation of Hispanics relative to Whites begins during the course of the second generation rather than in the third and later generations as suggested by previous studies (Smith 2003; Telles and Ortiz 2008; Waters and Pineau 2015).¹² By contrast, second-generation White and Asian men have higher earnings than their third-plus-generation White counterparts throughout most age intervals. They also experience greater earnings growth such that by age 43–45, White and Asian second-generation men earn 14.6% and 20.4% more, respectively (equivalent to $9,552 and $13,353).

Figures 2a and 2b, respectively, plot the predicted ratio of earnings of second-generation men of each ethnoracial category relative to third-plus-generation Whites and relative to third-plus-generation men of the same race and ethnicity using the results of the random-effects model.¹³ A ratio of one, shown as a dashed line, indicates full parity with the corresponding reference category. A comparison of the earnings trajectories of second-generation men of all ethnoracial groups relative to third-plus-generation Whites allows us to better visualize the progress that each group makes in relation to a common baseline, while a comparison of the earnings trajectories relative to third-plus-generation men of the same race and ethnicity allows us to assess whether second-generation men are experiencing a process of assimilation into disadvantaged minorities consistent with the ethnic disadvantage and segmented assimilation models.

Figure 2a shows the pattern in earnings trajectories relative to third-plus-generation Whites already discussed: White and Asian second-generation men experience greater earnings growth throughout their adult lives. Black and Hispanic second-generation men begin their working lives at a disadvantage. However, whereas Black second-generation men make progress in catching up with their third-plus-generation White counterparts, second-generation Hispanics fall further behind.

Figure 2b shows that all second-generation men have significantly higher earnings than their third-generation co-ethnoracial counterparts. In particular, second-generation Black and Hispanic men have significantly higher earnings than third-generation Black and Hispanic men, respectively. The earnings growth of second-generation Black men far exceeds that of third-plus-generation Blacks, such that by ages 43–45 they earn 38.9% more than their later-generation counterparts. By contrast, although second-generation Hispanic men's earnings are 9.9% higher than those of third-plus generation Hispanics at ages 25–27, their earnings growth is essentially flat during the entire age range. Rather than a process of segmented assimilation whereby second-generation Black and Hispanic men reach parity with their later-generation co-ethnoracial counterparts, the earnings trajectories of second-generation Black and Hispanic men in Figure 2b suggest a pattern of second-generation superachievement relative to their co-ethnoracials. But the earnings trajectory of second-generation Hispanics also reflects a loss of momentum since they make no further progress during their prime working years. It is important to note that second-generation superachievement relative to co-ethnoracials may not only signal the progress made by the second generation, but it may also imply a third-plus-generation decline because the latter form the baseline against which the second generation's earnings are being compared.

Differences by Education

We begin our examination of the effect of educational differences on the earnings trajectories of second-generation men by testing a model that simply controls for educational attainment. As shown in Figure 3, controlling for educational attainment substantially reduces the earnings advantage of second-generation Asian and White men, and the disadvantage of second-generation Hispanic men. This suggests that differences in educational attainment account for a large part of the observed ethnoracial disparities among second-generation men. Yet even after controlling for educational attainment, Asian and White men have statistically significantly higher earnings compared to third-plus-generation Whites beginning at ages 37–39. Similarly, second-generation Hispanic men continue to have significantly lower earnings than the same reference category beginning at ages 34–36.

Because education is a time-invariant predictor for all men in our sample, its introduction as a predictor affects only the intercept in our models and shifts the trajectories for second-generation men of different race and ethnicity up or down depending on the educational composition of the group. To further examine how educational attainment may change the shape of the earnings trajectories later in life, we tested separate models for men with low and high levels of education, defined as those with less than completed college and those with a college degree or more, respectively. Figures 4a and 4b show the predicted ratio of earnings of second-generation men of each race and ethnicity relative to the earnings of third-plus-generation Whites with similar levels of education. The higher educational attainment of second-generation Black and Asian men appears to partly explain their overall earnings growth relative to third-plus-generation Whites. Black men with a college degree experience very little earnings growth relative to later-generation Whites with the same level of education throughout the entire age span considered, while second-generation Asian men without a college degree have earnings that do not exceed those of later-generation Whites.

Educational attainment also plays an important role in shaping the earnings trajectories of second-generation Hispanic individuals. Second-generation Hispanic men without a college degree have earnings that are not significantly different from those of third-plus-generation Whites by ages 40–42. By contrast, second-generation Hispanics with a college degree or more start with significantly lower earnings and fall further behind their third-plus-generation White peers during the course of their working lives. By ages 43–45, their earnings are 17.2% lower than that of their third-plus-generation counterparts with similar levels of education. The lower overall earnings trajectories for second-generation Hispanics are therefore not entirely explained by their level of education.

Differences by National Ancestry

To examine differences in the earnings trajectories within broad pan-ethnoracial categories, we tested the same random-effects models separating second-generation Asian and Hispanic men into specific ethnic or national ancestry groups while controlling for educational attainment. The results in Figure 5a show considerable variation in earnings trajectories across second-generation Asian groups even after controlling for education. However, with the exception of Filipino men, all other second-generation Asian men have significantly higher earnings than third-plus-generation Whites by their middle to late 30s. These results are therefore consistent with those presented in Figure 3 for second-generation Asian men as a whole. The lower earnings of second-generation Filipino men are consistent with findings from previous studies that document lower socioeconomic outcomes such as educational attainment for that group relative to other Asians (e.g., Feliciano and Lanuza 2017).

The results in Figure 5b comparing the earnings growth of second-generation Hispanic men in each specific ethnic or national ancestry group are also generally consistent with those grouping all second-generation Hispanic men into a single category. The earnings of second-generation Mexican men are significantly lower than those of third-plus-generation Whites beginning at ages 34–36, while the earnings for second-generation Central American men are significantly lower before that age category. Only second-generation Cuban Americans have earnings that are not significantly different from the reference group in any age category after controlling for education. The higher earnings of second-generation Cuban men are consistent with previous findings showing better socioeconomic outcomes for this group (e.g., Feliciano and Lanuza 2017; Haller et al. 2011a). Notably, the lower earnings growth over time found for the broad Hispanic category of men as a whole in earlier models appears to be driven mostly by second-generation Mexican men.

The Effect of Early Occupational Insertion in the Labor Market

As noted in the conceptual framework, differences in the earnings trajectories of second-generation men may reflect their incorporation or exclusion from occupations that are associated with higher earnings growth. To examine how differences in the occupation in which men are first incorporated into the labor market affect their later earnings trajectories, we tested models using the CPS-reported occupation at age 25–35 by limiting our sample to individuals who were interviewed by the CPS in that age range. We further limited this portion of our analysis to a comparison between second-generation Hispanic men and third-plus-generation Whites because there is an insufficient number of cases of other second-generation ethnoracial groups when only men interviewed at ages 25–35 are selected.

Figure 6 shows the ratio of predicted earnings between second-generation Hispanic men and their third-plus-generation counterparts in our baseline model, which controls for educational attainment, and in a model controlling for both educational attainment and occupation at age 25–35. The results indicate that education explains a large part of the disparities in the earnings trajectories of second-generation Hispanic men and third-plus-generation Whites. Further controlling for occupation appears to explain the remaining differences. After controlling for both education and early occupation, the earnings differences between second-generation Hispanics and third-plus-generation Whites are statistically nonsignificant at every age category.¹⁴

Conclusions

In this study we used a unique dataset linking respondents from multiple years of the CPS to their longitudinal tax records to examine the earnings assimilation of second-generation men. Our restricted-use data allowed us to overcome key methodological limitations of previous studies. In particular, repeated measures of individuals' earnings throughout their adult lives allowed us to better estimate their intragenerational mobility. We were able to compare the earnings trajectories of second-generation men of different race and ethnicity instead of using only snapshots at particular points in their lives.

Our results showed that new second-generation men overall experience considerable earnings mobility from young adulthood to midlife. However, we also found large differences in this pattern across ethnoracial groups. Second-generation White and Asian men experience rapid earnings growth that exceeds that of later-generation Whites. Second-generation Black men also undergo substantial earnings growth even though they are unable to fully make up their initial disadvantage during the course of their working lives. By contrast, second-generation Hispanic men begin their careers with an earnings deficit relative to later-generation Whites and fall further behind later in life. This last finding is important because it signals that the stalling or reversal in Hispanic earnings assimilation begins during the course of the second generation rather than in the third generation as previously thought (Smith 2003; Telles and Ortiz 2008; Waters and Pineau 2015).

Our findings help inform broader debates regarding the economic assimilation of immigrants and their descendants. Although standard theories of immigrant assimilation often emphasize the progress made across immigrant generations, the mobility experienced within each generation is also important for achieving economic parity. The intragenerational mobility experienced by the second generation during the course of their own lives will have a profound effect on the economic outcomes of their children, who constitute the third generation. In this way, intragenerational mobility is intrinsically tied to intergenerational mobility. Despite the progress made by second-generation Hispanic men relative to their immigrant parents, they are experiencing lower earnings growth relative to later-generation Whites in their own lifetimes. The lower intragenerational mobility of second-generation Hispanic men is inconsistent with straight-line assimilation. On the other hand, men of all ethnoracial categories including Hispanics have higher earnings than their respective later-generation co-ethnoracials. The superachievement relative to later-generation men of the same race and ethnicity contradicts the worst predictions of segmented assimilation theory. As noted earlier, however, the progress of second-generation men relative to later generations of the same race and ethnicity should be interpreted with caution because it may also imply an earnings decline among third-plus-generation men against whom the second generation is being compared rather than only an increase in the latter's earnings.

We found substantive heterogeneity in the earnings trajectories of second-generation men by educational attainment, but the ethnoracial disparities we found could not be fully explained by differences in education. For example, second-generation Hispanic men with a college degree experience an even greater disparity in earnings relative to similarly educated third-plus-generation Whites. This suggests that factors beyond early life experiences, such as differential treatment in the labor market, play an important role in the economic assimilation of the new second generation. The downward assimilation process for children of Hispanic immigrants proposed by segmented assimilation theory thus appears to be exacerbated in adulthood.

We also explored the effect of second-generation men's early occupational placement in the labor market on their subsequent earnings growth. We found that controlling for both education and their early occupation accounts for the lower earnings growth among second-generation Hispanic men. Children of Hispanic immigrants appear to be incorporated into occupations that result in lower earnings trajectories later in life. This finding is consistent with Tomaskovic-Devey et al. (2005), who demonstrated that ethnoracial discrimination among Black and Hispanic men may influence earnings inequality over the life course by depriving them of employment opportunities that allow for greater accumulation of human capital. More research is required to identify the precise reasons why second-generation Hispanic men are being incorporated into occupations that result in lower earnings trajectories.

The results of our analysis disaggregating second-generation men into detailed ethnic or national ancestry groups were generally consistent with those using broader pan-ethnoracial categories, but also revealed some important differences. Among second-generation Asians, Filipino men start with lower earnings and experience less earnings growth during their lifetimes. The lower earnings of second-generation Filipino men relative to other second-generation Asians parallel previous findings from studies of educational attainment (Feliciano and Lanuza 2017). Among second-generation Hispanics, Cuban men experience the greatest earnings growth. In fact, second-generation Cuban men have similar earnings trajectories as later-generation Whites. The lower earnings growth of second-generation Hispanics as a whole appears to be driven largely by Mexican American men, who constitute the largest ethnic group in the broader pan-ethnoracial category. As Lee and Zhou (2015) noted, Mexican immigrant parents are “hyposelected” insofar as they are less likely to have a college degree than both Mexicans in Mexico and the U.S. population as a whole (see also Zhou and Gonzales 2019). The lower education level of Mexican immigrant parents may limit their second-generation children's opportunities and influence their career aspirations.

Finally, one potential limitation of our study is our inability to identify the second-generation children of undocumented immigrants. Prior studies have demonstrated that the undocumented immigration status of first-generation parents may adversely affect the opportunities available to the second and later generations (Bean et al. 2015; Bean et al. 2011). It is possible that lower earnings trajectories observed among second-generation Hispanic men, and among those of Mexican origin in particular, may be partly explained by the undocumented status of their parents given the higher rates of undocumented status among those groups. Additional research is required to examine the effect of documentation status of parents on the long-term earnings trajectories of children.

Acknowledgments

The views expressed here are those of the authors and do not represent the views of the Social Security Administration or any federal agency.

Notes

References