Durable Disadvantage: Gender and the Mark of Unauthorized Status in Immigrants’ Occupational Trajectories

Abstract

Adverse life course events associated with unemployment can negatively affect individuals’ future labor market prospects. Unauthorized status, and subsequent unauthorized employment, may operate similarly, marring immigrants’ labor market prospects even after they change legal status. However, it is unclear how and why any durable disadvantage associated with prior unauthorized status operates differently by gender. This is an important shortcoming given that legal status and gender overlap to influence both migration and stratification. Using longitudinal data from a nationally-representative sample of lawful permanent residents, we find durable disadvantage associated with prior exposure to unauthorized status, especially among women. Men with prior exposure to unauthorized status experience persistent occupational disadvantage over time relative to men who were never unauthorized. However, women with exposure to unauthorized status experience widening occupational disadvantage over time relative to women who were never unauthorized. Human capital and legal processes help to explain this pattern.

Introduction

Adverse events which trigger unemployment are associated with durable labor market disadvantage (DiPrete 2002; Gangl 2006). Individuals who are unemployed following involuntary layoff or incarceration have greater difficulty finding employment again (Pager 2003, 2008; Pedulla 2018). Migration scholars have begun to argue that unauthorized status—a legal classification which formally denies immigrants the right to work—is a similar adverse life course event that could mar immigrants’ labor market prospects even after they change legal status, as well as the educational attainment of immigrants’ children (Bean et al. 2011; Gleeson and Gonzales 2012; Gonzales 2017; Sisk 2014). Yet unlike job loss or incarceration, unauthorized status is not associated with unemployment. Rather, the vulnerability associated with lacking formal legal status and subsequent work authorization sorts immigrants into precarious, underpaid, and dangerous jobs (Durand, Massey, and Pren 2016; Hall, Greenman, and Farkas 2010). Moreover, unauthorized employment often grants immigrants limited opportunities for promotions or to increase human capital or skills, so immigrants who change status may still struggle to achieve the same occupational status in the primary labor market over time as immigrants who were never unauthorized (Donato and Sisk 2012; Hamilton, Patler, and Savinar 2020; Kossoudji and Cobb-Clark 2002).

While ample research focuses on the experiences of immigrants in the labor market, migration and labor market stratification are highly gendered processes, and there is significant gender inequality in labor market sorting and occupational mobility among immigrants (Curran et al. 2006; Mahler and Pessar 2006; Myers and Cranford 1998). Despite recent evidence on the durable labor market disadvantage associated with unauthorized status, we know little about how and why prior exposure to unauthorized status is associated with different labor market trajectories for female and male migrants. Legal status and gender overlap to influence both migration and labor market stratification (Anthias and Yuval Davis 1983). Indeed, wide gender inequalities exist in occupational sorting and mobility among the general population, immigrants, and presently unauthorized individuals (Donato et al. 2008; Gonzales and Burciaga 2018; Guinea-Martin, Mora, and Ruiz-Castillo 2018; Hegewisch et al. 2010). Understanding how and why prior exposure to unauthorized status and subsequent unauthorized employment might be associated with divergent labor market trajectories for men and women is also important in light of workers’ precarity amid increasingly segmented labor markets. Deindustrialization and the decline of labor unions have led to organizational changes among employers, including the rise of unstable, low-security, and low-wage jobs held by those working in the secondary labor market (Kalleberg 2009; Kronberg 2013). Given that both women and unauthorized individuals are likely to be working in unstable employment situations, gender and unauthorized status may further exacerbate workers’ increasing labor market uncertainty.

Using longitudinal data from a nationally-representative sample of lawful permanent residents, we ask: (1) How is the intersection of gender and prior exposure to unauthorized status associated with both initial occupational sorting and any durable disadvantage over time among immigrant men and women? and (2) What explains variation in occupational disadvantage among women and men over time? Defining exposure to unauthorized status as entering the United States without a valid visa or overstaying a tourist visa, we find that men with prior exposure to unauthorized status experience persistent and stable occupational disadvantage over time relative to men without that exposure. But women with exposure to unauthorized status experience widening occupational disadvantage over time relative to women without exposure. Among both men and women, human capital and legal pathways partially explain the association between exposure to unauthorized status and occupational disadvantage.

Background

Unauthorized Employment and Durable Labor Market Disadvantage

Working while unauthorized might operate like other disruptive life course events, inhibiting immigrants’ employment prospects even after they change legal status (Gleeson and Gonzales 2012; Gonzales 2017; Sisk 2014). Unauthorized status is not associated with a period of labor market absence (despite mandating it legally): 8 million unauthorized immigrants without any legal status are currently employed (Pew Research Center 2018). However, the vulnerability attached to lacking lawful work authorization means that immigrants are sorted into precarious and dangerous jobs for which they can either present false paperwork or receive wages in cash so their employers do not have to report those earnings (Donato and Sisk 2012; Gleeson 2010; Powers, Seltzer, and Shi 1998).

Beyond sorting immigrants into less desirable jobs, unauthorized employment may also be negatively associated with immigrants’ labor market trajectories even after they change legal status. For example, the Immigration Reform and Control Act (IRCA) of 1986, the first policy effort to regulate unauthorized migration, legalized three million immigrants already present in the U.S. and created employer sanctions to deter future ‘unlawful’ immigrant hiring (Donato 1993). Employers passed on those costs to unauthorized workers, continuing to hire them but paying them even lower wages (Phillips and Massey 1999). Workers who legalized following IRCA, or received lawful permanent residence, experienced increased wages but were also less likely to obtain higher-status jobs because of few opportunities to increase cross-firm or industry capital (Amuedo-Dorantes, Bansak, and Raphael, 2007; Becker 1993; Donato and Massey 1991; Rivera-Batiz, 1999; Sisk 2014). Working while unauthorized limits opportunities for upward mobility, especially if the firms in which unauthorized individuals are employed offer few labor protections or do not follow equal labor laws (Scott, Hale, and Padilla 2021). As a result, immigrants who had previous unauthorized employment may struggle to catch up occupationally to immigrants who were never unauthorized (Donato and Sisk 2012; Kossoudji and Cobb-Clark 2002). Although other pathways for immigrants without legal status exist beyond the formal legalization program to receive lawful permanent residence, we know much less about the labor market disadvantages associated with unauthorized experience via other pathways (see Hamilton, Patler, and Savinar 2020; Sisk 2014 and Lofstrom, Hill, and Hayes 2013 for exceptions).

Gender Differences in Durable Labor Market Disadvantage

Despite evidence on the initial sorting and longitudinal disadvantages associated with unauthorized employment, most research focuses solely on unauthorized workers who never changed legal status; pools male and female workers who changed status together; or focuses exclusively on men (Donato and Armenta 2006; Kreisberg 2019). Less research examines how and why prior exposure to unauthorized status and subsequent unauthorized employment relates to sorting and occupational trajectories over time among men versus women. But migration and labor market stratification are highly gendered processes (Curran et al. 2006; Mahler and Pessar 2006). Migration has become feminized in the past one hundred years, for example (Massey, Durand, and Malone 2002). In turn, the differential process and timing by which men and women migrate shapes when immigrants enter the labor market and in what jobs (Parrado and Flippen 2005). As a result of these mutually constitutive processes, there is significant gender inequality in labor market sorting and occupational trajectories among immigrants (Myers and Cranford 1998). In addition, prior unauthorized status among mothers, more so than fathers, appears especially consequential for children’s educational attainment (Bean et al. 2011). Although exposure to unauthorized status in and outside the labor market might exert long-lasting and intergenerational effects on socioeconomic attainment, scant research examines whether unauthorized status and subsequent unauthorized work experience differentially affects women and men in the labor market.

Unauthorized exposure may intersect with gender among adults in the labor market in several ways, both with respect to initial occupational sorting and occupational trajectories. At a given point in time, there is a larger wage gap between current unauthorized and authorized men than women, with unauthorized status explaining more of the wage gap among men than women (Hall, Greenman, and Farkas 2010; Lofstrom et al. 2013). Unauthorized status may therefore be a stronger sorting mechanism for men. Female immigrants are likely to be occupationally segregated in domestic labor and other care work, regardless of legal status (Hondagneu-Sotelo 2011), suggesting that the occupational sorting penalty attached to unauthorized status may be weaker than the penalty attached to being female. It may therefore be the case that unauthorized status operates as a stronger initial occupational sorting mechanism for men than for women.

H1a. Men will experience a larger initial occupational sorting penalty associated with exposure to unauthorized status than women.

Evidence on how the intersection of gender and unauthorized exposure relates to immigrants’ occupational trajectories is more mixed. Recent evidence suggests that previously unauthorized men and women experienced comparable wage increases after receiving lawful permanent residence, despite both groups’ wages remaining below those of continuously authorized immigrants (Lofstrom et al. 2013; Sisk 2014). However, among current unauthorized workers, women are occupationally segregated from and earn less than men (Donato and Sisk 2012; Bernhardt, Spiller, and Polson 2013). Over time, unauthorized women also have lower occupational mobility than men (Hall, Greenman, and Yi 2019). Finally, women who legalized from IRCA experienced lower wage gains than men (Kossoudji and Cobb-Clark2002; Amuedo-Dorantes, Bansak, and Raphael 2007). Little research examines how any gendered penalty associated with exposure to unauthorized status and work experience changes after immigrants’ legal status changes outside the IRCA context. However, female immigrants are particularly vulnerable in the labor market, as gender, nativity, and unauthorized status all independently increase the chances that women are sorted into precarious occupations—such as in cleaning and care work—that have fewer opportunities for later upward mobility in those or other industries than men (Brown and Misra 2003; de la Luz Ibarra 2000; Hondagneu-Sotelo 2011). It is therefore possible that exposure to unauthorized status might hamper women’s occupational trajectories more than men’s.

H1b. While both men and women with exposure to unauthorized status will have worse outcomes in the labor market, the association between previously unauthorized status and occupational status will become more negative for women than men over time.

Explaining Variation in Durable Disadvantage of Unauthorized Status among Men and Women

We suspect that a gendered pattern of durable disadvantage might be explained by factors related to unauthorized status as well as unauthorized employment. First, a number of factors may explain a larger penalty attached to prior unauthorized exposure among women, including the vulnerability of living unauthorized inside the household, and the vulnerability of working unauthorized in precarious conditions. Meanwhile, human capital may be especially salient in explaining occupational patterns among previously unauthorized men. Finally, legal pathways to permanent residence may explain occupational patterns among both men and women.

Household dynamics.

Women have historically migrated later than men, bringing their families with them or taking care of family members already in the United States (Donato, Brandon, and Evelyn 2008). However, since the 1970s, unauthorized women who migrate no longer migrate solely for family: a large majority also work upon entering the U.S. (Curran et al. 2006). Despite working and much like women without exposure to unauthorized status, women also continue to maintain most of the household responsibilities, in part because they consider male partners’ work to be more difficult (Parrado and Flippen 2005). But unlike women without exposure to unauthorized status, unauthorized women who reaped the most benefits from legalization in 1986 through IRCA tended to be married with husbands who had dense social networks (Hagan 1998). Such gendered household dynamics, including care work or marital status, may explain why unauthorized women had few gains after legalizing from IRCA (Powers and Seltzer 1998). Being married and having household obligations may therefore help explain both the initial penalty and durable disadvantage over time for women rather than men. We hypothesize that:

H2a. Marital status, children present in the home, and household size will help explain both the initial occupational disadvantage, and flatter occupational trajectories, among women with prior exposure to unauthorized status relative to women without prior exposure.

Work arrangements.

Occupational tenure and contingent work arrangements may also explain variation in sorting and trajectories among women rather than men. Occupational tenure is typically associated with higher wages (Chiswick 1978). Although immigrants tend to have shorter job tenures than native-born workers (McDonald and Worswick 1998), dual labor market theory asserts that job tenure may not increase immigrants’ labor market outcomes if they are sorted into lower-wage jobs with fewer chances for promotion (Bonacich 1972; Piore 1975; Portes and Stepick 1985). In fact, a longer tenure may even increase disadvantage over time if immigrants only stay in exploitative jobs because they fear their unauthorized status will be exposed elsewhere (Hall, Greenman, and Farkas 2010). Female migrants have lower occupational returns to a longer job tenure than males (Zhou and Logan 1989; Salzinger 1991). In addition, even in similar occupations, women working unauthorized may be more likely to enter contingent work arrangements than both men and women without exposure, and women in unauthorized employment also more often face unstable and unpredictable working hours (Catanzarite 2002). These kinds of contingent work arrangements may explain the initial sorting penalty and durable disadvantage over time among women. Specifically, we separate years in the job and hours worked on the job from exposure to unauthorized status, hypothesizing that:

H2b. A longer duration in a given occupation, and fewer hours worked on the job, will help explain both the initial occupational sorting penalty, as well as occupational trajectories, among women with prior exposure to unauthorized status.

Human capital.

While household dynamics and work arrangements may contribute disproportionately to the labor market experiences of previous unauthorized immigrant women, differences in human capital may explain occupational sorting and trajectories among men. Unauthorized immigrants are typically negatively selected on, and have lower returns to, human capital compared to immigrants without exposure to unauthorized status (Hall, Greenman, Farkas 2010). Similarly, women are often more positively selected into migration based on pre-migration human capital factors than men (Feliciano 2008). Any initial occupational sorting penalty associated with prior exposure to unauthorized status may therefore be better explained by human capital among men. In addition, immigrants with unauthorized employment often struggle to translate their existing human capital into more upwardly mobile jobs over time (Calavita 1992; Borjas and Tienda 1993). After legalizing through IRCA, however, men were better able than women to maximize their human capital by moving more freely through the labor market (Kossoudji and Cobb-Clark 2002; Pan 2012). Little research examines whether variation in human capital explains occupational patterns among men beyond the IRCA context. This is in part because many studies lack accurate measures of pre-migration human capital, including pre-migration educational attainment, socioeconomic status, and occupational prestige. Using several measures of pre-migration human capital to account for processes of selection, we hypothesize that:

H2c. Human capital will explain the initial occupational sorting penalty and occupational trajectories among men with prior exposure to unauthorized status.

Legal pathways.

Finally, pathways to legalization may help explain occupational sorting and trajectories for men and women. Immigrants today receive lawful permanent residence through a number of lawful starting points, including employment visas, family reunification, the diversity visa—which operates on a lottery system—humanitarian visas, and formal legalization programs. These starting points are not merely administrative; they unequally allocate resources vital for occupational mobility (Gleeson and Gonzales 2012), and this unequal resource allocation process helps explain variation in occupational sorting and mobility over time (Kreisberg 2019). Prior exposure to unauthorized status often restricts which lawful starting pathways immigrants have access to, as unauthorized immigrants are frequently denied employment visas, the diversity visa, and humanitarian visas (Hamlin and Wolgin 2012; Wardle 2005). These process may operate in similar ways for men and women:

H2d. Immigrants’ legal pathways to permanent residence will help explain the initial sorting penalty and occupational trajectories over time associated with prior exposure to unauthorized status for men and women.

In summary, we expect household dynamics and contingent work arrangements to help explain the initial sorting penalty and occupational trajectories over time associated with exposure to unauthorized status for women. Human capital may play a more salient explanatory role among men, and legal pathways to permanent residence may explain occupational patterns among both men and women.

Data and Methods

Data

Data come from the New Immigrant Survey (NIS), a publicly available, nationally-representative (n=8,573), longitudinal survey of immigrants who received LPR in 2003. The sampling frame consists of immigrants aged 18 and older who received LPR between May and November of 2003. The survey and analytic sample included people who immediately received LPR in 2003 (“new arrivals”), as well as people who adjusted their status to LPR (“adjustees”) because they were already living in the U.S. The survey conducted its second wave from 2007–2009, with a median time between waves of five years (henceforth this wave is therefore referred to by the year 2008). Data were collected through in-person and telephone interviews in 15 languages.

NIS data are representative of all immigrants in the United States who received LPR in 2003. Though the data do not represent all immigrants in the U.S., they are ideal among existing data sources that measure legal status because they include information on pre-migration experiences, occupations pre- and post-migration, and legal status upon entry into the United States, including whether and what type of unauthorized experience and employment immigrants had prior to receiving LPR (Bachmeier, Van Hook, and Bean 2014). Moreover, the NIS is the only survey that follows immigrants longitudinally after they receive LPR.

From the original sample of 8,573 respondents, we restrict analyses to respondents present in both 2003 and 2008 (n=4,363). Despite attrition between the two waves (a decline in the response rate from 69 to 46 percent), there is no selective attrition based on any one particular variable. In addition, there is no evidence of differential attrition by prior unauthorized experience (Massey, Jasso, and Espinosa 2017; Kreisberg 2019; León-Pérez, Patterson, and Coelho 2021). As an additional check on this possibility, we test the sensitivity of our results to alternative sample restrictions, including maximum likelihood estimators (MLE), which allow us to retain all respondents in the analysis even if they are missing in one or more waves, as well as regardless of whether they are employed and report an occupation or not in one or more waves. The results (available upon request) are similar to those presented here.

To understand occupational trajectories over time, we follow prior research by further restricting the sample to those who retrospectively reported information about their first job after migrating to the U.S., which we refer to as wave 0 (Akresh 2008). Wave 0 employment for new arrivals is possible because they either took prior trips to the U.S. or because for those without prior trips, there was a time lag of about two months between arriving and being interviewed for the New Immigrant Survey. New arrivals therefore retrospectively report on their first job upon moving to the United States to live. Then we restrict the sample to those who report being employed in 2003 (referred to as wave one) and in 2008 (referred to as wave two), yielding a final analytic sample of 2,187 people or 5,042 person-waves. For most new arrivals, wave 0 employment tended to be a very short time interval before wave 1, although the results (available upon request) are robust to restricting the sample to just adjustees. All missing data is imputed using multiple imputation techniques. Because immigrants may be positively selected on these characteristics, the results may not be generalizable to all immigrants in the labor force—only to those employed after migrating to the United States.

Measures

Occupational sorting and trajectories.

We test for both levels of, and changes in, occupational prestige at three time points: at the first U.S. job (wave zero), in 2003 (wave one), and 2008 (wave two). Respondents report retrospectively in wave one about the first job they had upon entering the U.S. They then answer what job they currently have at both waves one and two. We therefore interpret the wave zero job as an indicator of initial occupational sorting, and we interpret the change from the wave zero job to jobs at waves one and two as an indicator of occupational trajectories over time. Those jobs are matched to occupational prestige scores measured by the Socioeconomic Index (Duncan 1961), ranging from 1 to 100. Although it would be informative to include multiple jobs in the U.S. before wave 1, the New Immigrant Survey only asked about the first U.S. job. This grants valuable insight into occupational trajectories across immigrants’ exposure to the United States labor market. In addition, although there are multiple indicators of occupational quality, the Socioeconomic Index is a long-standing and reliable quantitative measure of the health of an individual’s employment, as well as the state of that individuals’ occupational standing in the social structure (Kalleberg 2011). While it would be ideal to have additional measures of wages as an alternative indicator of socioeconomic status, high levels of missing data on wages in the New Immigrant Survey make this infeasible. Examining occupational status is nonetheless valuable for comprehensively representing socioeconomic status, and it usefully complements existing research on wage trajectories after IRCA (Kossoudji and Cobb-Clark2002; Amuedo-Dorantes, Bansak, and Raphael 2007).

Unauthorized status.

To measure exposure to unauthorized status upon entering the U.S., prior to receiving LPR in 2003, we use survey questions and administrative records to construct a variable with two categories: exposure to unauthorized status and no exposure, or continuously authorized. To operationalize exposure to unauthorized status, we generate a summary variable of anyone who had any previous unauthorized spell, even if ultimately, they received a pathway to LPR. This was possible in the NIS because respondents reported about each and every trip they took to the U.S., as well as which visa (or lack thereof) was associated with each trip. To create this summary variable, by trip, we combine those who: report not entering with any visa, those who get LPR from a legalization program (for which only unauthorized immigrants were eligible), those whose administrative records report entering without inspection or those with an administrative entry code of unknown or who have no code, and those who overstay a tourist visa by more than six years (Jasso 2011; Kreisberg 2019). If on any trip, a respondent met any one of those conditions, they were coded as having unauthorized exposure. This means that immigrants who ultimately receive a humanitarian or family reunification visa, or enter into a formal legalization program, may have still had some exposure to unauthorized status (Jasso et al. 2008).

By contrast, individuals who never had any unauthorized spell on previous trips were coded as having no exposure to unauthorized status. Specifically, this meant that, on each trip, those who report having no temporary visa overstay and who report entering with a lawful visa pathway (whether that be family reunification, humanitarian, employment, or diversity), and whose administrative reports confirm they entered with that visa, are coded as having no exposure to unauthorized status.

Because we are also interested in the ramifications of unauthorized employment, and not just general unauthorized status, we also constructed a separate variable indicating whether individuals’ first U.S. jobs were on authorized or unauthorized trips. This variable was highly correlated, though not perfectly correlated, with exposure to unauthorized status because individuals with unauthorized status could have worked before or after that exposure, on an authorized trip. The results (explained in more detail and presented in Appendix B) are similar to those presented in the main text.

Household dynamics.

We operationalize several household dynamics reported at wave one. Marital status is a dichotomous variable indicating whether the respondent is married. We also include a dichotomous indicator of whether the respondent has children. Finally, we measure the number of people in the household.

Work arrangements.

To operationalize job tenure, we construct a measure of years on the job at each wave. This is calculated by subtracting the year at which the respondent left the job from the year the respondent first reported entering the job at each wave. At all three waves, we also include a measure of contingent work arrangements, operationalized as hours worked per week on the job.

Human capital.

We include several measures of pre-migration human and financial capital: occupational prestige, operationalized continuously (0 for unemployed, and 1 indicating lowest prestige up to 100 indicating highest prestige); childhood income, measured by whether the respondent’s family’s income was below, average, or above average compared to other households in the country at 16; childhood environment, measured by whether the respondent lived in a rural area when he or she was 10; and years of education received outside the U.S.—all reported in wave one. We include pre-migration investments, which we operationalize with a dichotomous indicator of whether respondents gave money or time to a number of social groups before migrating. This is also an important indicator of social capital prior to migration (Curran et al. 2005; Glick-Schiller 1999). We also measure two post-migration human capital characteristics: any education in the U.S. and English language proficiency, measured by whether the respondent speaks English very well, well, not well, or not at all. These were also reported in wave one.

Legal pathways.

We operationalize legal pathways as reported in wave one with a five-category variable measuring whether the respondent ultimately received LPR through an employment visa, family reunification, a diversity visa, a humanitarian visa, or a formal legalization program. The humanitarian visa program category collapses immigrants who received refugee status from abroad, asylum status from within the U.S., and parole status. Although these three statuses undoubtedly receive different resources once within the U.S., their resource-set is still largely differentiated from the resources of employment visa holders, family reunification, and diversity visa holders, given that they are eligible for government and non-governmental services. The legalization pathway is perfectly correlated with unauthorized experience, as only individuals with such experience qualified for legalization under certain programs, and similarly, no individuals without exposure to unauthorized experience were included in the legalization pathway. We therefore make this category the reference group to understand how the other pathways compare to legalization.

Gender, demographic characteristics, and geographic context.

We measure binary gender (self-reported as “male” or “female”). Additionally, after gaining LPR in 2003, the NIS reports what region respondents live in: Northeast, South, Midwest, or West. We measure self-reported race: white, black, Asian or other. We also measure region of origin: Latin, Central, South America and the Caribbean; Africa; South and East Asia; Europe and Central Asia; and the Middle East. Because immigrants with unauthorized experience are likely to come from Latin American countries, we also run sensitivity analyses that restrict the total sample to those born in Latin America. Results are similar to those presented here.

The question wording for all variables reported in this analysis (excluding those relying on administrative records, defined above) is included in Appendix A.

Methods

Our approach involves careful description of gender differences in occupational prestige associated with prior exposure to unauthorized status at the three waves. To answer our first question—is the intersection of gender and prior exposure to unauthorized status associated with occupational sorting and any disadvantage over time?—we compute predicted occupational prestige values from growth curve models stratified by gender. Growth curve models allow both starting levels and rates of change to vary both within and across individuals over time (Bryk and Raudenbush 1987). The first level, within individuals, measures changes in each individual’s own occupational trajectory over time. The second, between-individual level measures differences among individuals in occupational prestige growth trajectories. This approach is useful for examining both occupational sorting (or the intercept in the growth curve model) as well as trajectories (or the slopes in the growth curve model). We measure sorting and trajectories in occupational quality as a linear function based on previous work treating occupational indexes continuously (Akresh 2008).

To answer the second question—what explains variation in occupational trajectories among women and men over time?—we successively incorporate four groups of factors into our models: human capital, work arrangements, household dynamics, and legal pathways. For men and women, we specify the most inclusive equation:

$$y_{ti}={\alpha }_0+D_i+D_i{T}_{ti}+T_{ti}+X_i+{HD}_i+W_i+W_i{T}_{ti}+{HC}_i+P_i+u_i{+e}_{ti}$$

Where D_i is an indicator of whether the individual, i, had exposure to unauthorized status; $D_i{T}_{ti}$ is a cross-level interaction between exposure to unauthorized status and time; $T_{ti}$ is the time variable, where time is measured by wave for each person, i; W_i is a vector of the covariates that vary over time (hours worked per week at each occupation and job tenure) and W_iT_ti are those time-varying covariates and their cross-level interactions with each wave; X_i is a vector of time-invariant covariates (e.g. region of origin); u_i are the individual-specific residuals; and e_ti are residuals at the measurement occasion level. The first model predicts occupational prestige as a function of exposure to unauthorized status initially and over time, net of a vector of demographic characteristics. Examining the intercepts will allow us to test Hypothesis 1B: that intercepts for unauthorized status are larger for men than women. Examining the slopes will afford a test of Hypothesis 1B: that slopes are larger for women than for men.

In the second model, we also account for household dynamics (measured at wave one), HD_i, in order to test Hypothesis 2A: that controlling for household dynamics will explain intercepts (D_i) and slopes (D_iT_ti) associated with unauthorized experience among women.

In the third model, we add work arrangements (measured across waves), W_i, consisting of years on the job and hours worked per week squared, and a cross-level interaction of work arrangements with wave, W_iT_ti. Hypothesis 2B predicts that account for work arrangements will explain intercepts and slopes associated with unauthorized experience among women.

In the fourth model, we incorporate human capital characteristics (measured at wave one), HC_i. We do not expect accounting for human capital characteristics to explain either the initial sorting gap or disadvantage over time for women. However, Hypothesis 2C predicts that controlling for human capital characteristics will reduce the size and significance of the coefficient indicating the initial occupational sorting penalty attached to unauthorized status, D_i, as well as any disadvantage associated with prior exposure to unauthorized status over time, D_iT_ti, for men.

The fifth and final model incorporates legal pathways (measured at wave one), P_i. Hypothesis 2D predicts that adding in the pathways to legalization will explain intercepts and slopes associated with unauthorized experience among both men and women.

We compute predicted occupational prestige values from the most inclusive female and male models, and we hold all values other than gender and unauthorized status at their group-specific means. An important assumption in these models is that there are no unobserved differences between respondents with and without unauthorized experience, or between women and men, that are correlated with occupational position. To relax this core assumption, we also estimate a Mundlak-Chamberlain correction model which allows for covariance between the error term and the regressors. The results, available upon request, are consistent with those presented here. Despite this sensitivity analysis and the fact that we measure a rich set of observed characteristics, it is important to emphasize that our models provide descriptive estimates of the relationship among unauthorized status, gender, and occupational sorting and trajectories.

Testing Differences between Men and Women

Our hypotheses and approach are organized largely around explaining variation in sorting and trajectories among men and women. However, it is also possible that trajectories—and the factors explaining them—will differ significantly across men and women. Although we do not formulate explicit hypotheses for how divergent trajectories will shape inequality between gender, we do conduct statistical tests to determine whether any differences exist in sorting, trajectories, and the explanatory power of each group of factors between men and women, in addition to among men and women. Specifically, we estimate a model that interacts gender with all other variables. For ease of interpretation of the results, we present superscripts (A) in Tables 3 and 4 where differences between men and women are statistically significant at the p<.05 level.

Table 3.

Growth Curve Models Predicting Occupational Prestige Among Women

	Model 1: Basic Model	Model 2: Household Dynamics	Model 3: Work Arrangements	Model 4: Human Capital	Model 5: With Legal Pathways
Exposure to Unauthorized Status	−10.166***	−9.462***	−9.047***	−3.599†	−3.759†
	(2.290)	(2.260)	(2.217)	(2.028)	(2.218)
Wave 1	4.265***	4.274***	−3.129	−1.438	−5.875
	(1.113)	(1.112)	(4.888)	(4.907)	(5.849)
Wave 2	9.421***	9.370***	3.889	5.931	4.161
	(1.043)	(1.041)	(6.107)	(6.036)	(6.407)
Exposure to Unauthorized Status*Wave 1	−1.986	−1.988	−3.064	−2.950	−1.429
	(1.889)	(1.888)	(1.959)	(1.938)	(2.379)
Exposure to Unauthorized Status*Wave 2	−3.860*	−3.909*	−4.226*	−4.084*	−3.824† A
	(1.799)	(1.796)	(1.891)	(1.853)	(2.286)
Household Dynamics
Married		5.289***	5.277***	3.545**	3.385*
		(1.543)	(1.524)	(1.334)	(1.314)
Has Children		−8.954***	−8.759***	−3.296*	−2.814†
		(1.653)	(1.631)	(1.453)	(1.440)
Household Size		−0.314	−0.308	−0.003	0.011
		(0.404)	(0.399)	(0.361)	(0.356)
Occupational Tenure
Years in Job			0.444	0.324	0.126
			(0.519)	(0.502)	(0.505)
Years in Job Squared			−0.011	−0.008	0.001
			(0.026)	(0.025)	(0.025)
Wave 1*Years in Job			0.835	0.834	0.918
			(0.628)	(0.610)	(0.625)
Wave 2*Years in Job			0.068	0.019	0.137
			(0.827)	(0.797)	(0.798)
Wave 1*Years in Job Squared			−0.023	−0.024	−0.025
			(0.033)	(0.032)	(0.032)
Wave 2*Years in Job Squared			0.015	0.015	0.009
			(0.035)	(0.034)	(0.034)
Contingent Work Arrangements
Hours Worked per Week			0.366*	0.389*	0.344*
			(0.151)	(0.148)	(0.148)
Hours Worked per Week Sq.			−0.004**	−0.004**	−0.004**
			(0.001)	(0.001)	(0.001)
Wave 1*Hours Worked per Week			0.110	0.051	0.051
			(0.181)	(0.180)	(0.182)
Wave 2*Hours Worked per Week			−0.111	−0.149	−0.097
			(0.195)	(0.188)	(0.188)
Wave 1*Hours Worked per Week Sq.			0.001	0.001	0.001
			(0.002)	(0.002)	(0.002)
Wave 2*Hours Worked per Week Sq.			0.004*	0.004*	0.003* A
			(0.002)	(0.002)	(0.002)
Human Capital
Years of Education Outside the U.S.				1.058***	0.980***
				(0.178)	(0.176)
Pre-Migration Occupational Prestige				0.021	0.017
				(0.021)	(0.021)
Childhood Income (Below Average Omitted)
Average Childhood Income				0.860	1.061
				(1.443)	(1.439)
Above Average Childhood Income				2.641	2.619
				(1.790)	(1.786)
Lived in Rural Environment as Child				−2.249†	−2.513*
				(1.212)	(1.198)
Gave to Social Groups Before U.S.				0.971	0.259
				(1.221)	(1.213)
Any Education in the U.S.				7.055***	7.257***
				(1.545)	(1.517)
English Language Fluency (Not at All Omitted)
Not Well				−2.428	−1.854
				(1.854)	(1.838)
Well				5.269*	3.936†
				(2.246)	(2.211)
Very Well				12.971***	10.528***
				(2.329)	(2.341)
Legal Pathways (Legalization Omitted)
Family Reunification					−0.143
					(2.903)
Employment					6.369† A
					(3.438)
Diversity					−7.512†
					(3.977)
Humanitarian					−2.414
					(3.950)
Constant	30.179**	20.137*	18.765†	−6.035	−1.444
	(9.703)	(9.750)	(10.323)	(9.767)	(10.102)
N of Individuals	863
N of Person-Waves	1,958

^†p<.10

^*p<.05

^**p<.01

^***p<.001

Notes: Superscript A refers to significant differences between male and female models at the p<.05 level. All models contain demographic characteristics, including self-identified race, region of origin, age, age squared, U.S. region, and the final wave interview year. The final model includes a cross-level interaction between the pathways and wave.

Table 4.

Growth Curve Models Predicting Occupational Prestige Among Men

	Model 1: Basic Model	Model 2: Household Dynamics	Model 3: Work Arrangements	Model 4: Human Capital	Model 5: Legal Pathways
Exposure to Unauthorized Status	−16.353***	−16.098***	−15.958***	−10.297***	−8.102***
	(2.011)	(2.009)	(2.024)	(1.841)	(1.909)
Wave 1	2.117*	2.110*	−8.588	−8.980+	−8.042
	(0.875)	(0.875)	(5.355)	(5.298)	(5.982)
Wave 2	4.403***	4.396***	−6.827	−4.922	−2.505
	(0.823)	(0.822)	(4.932)	(4.758)	(5.183)
Exposure to Unauthorized Status*Wave 1	4.457**	4.452**	4.291**	4.389**	4.184*
	(1.458)	(1.458)	(1.568)	(1.557)	(1.868)
Exposure to Unauthorized Status*Wave 2	5.100***	5.149***	4.206**	4.316**	3.215† A
	(1.427)	(1.428)	(1.570)	(1.539)	(1.802)
Household Dynamics
Married		4.226*	3.607*	2.137	1.620
		(1.790)	(1.760)	(1.504)	(1.445)
Has Children		−5.303***	−5.279***	−3.627**	−3.728**
		(1.572)	(1.541)	(1.300)	(1.248)
Household Size		−0.777*	−0.827*	−0.217	−0.031
		(0.375)	(0.369)	(0.310)	(0.297)
Occupational Tenure
Years in Job			1.643***	1.366**	0.845†
			(0.445)	(0.436)	(0.479)
Years in Job Squared			−0.078**	−0.057*	−0.031
			(0.029)	(0.029)	(0.030)
Wave 1*Years in Job			−0.882†	−0.709	−0.410
			(0.452)	(0.445)	(0.495)
Wave 2*Years in Job			−0.574	−0.495	−0.180
			(0.472)	(0.464)	(0.526)
Wave 1*Years in Job Squared			0.057†	0.041	0.020
			(0.029)	(0.029)	(0.031)
Wave 2*Years in Job Squared			0.060*	0.045	0.024 A
			(0.029)	(0.029)	(0.031)
Contingent Work Arrangements
Hours Worked per Week			−0.009	0.009	0.011
			(0.151)	(0.148)	(0.143)
Hours Worked per Week Squared			0.001	0.001	0.000
			(0.002)	(0.002)	(0.002)
Wave 1*Hours Worked per Week			0.412†	0.427*	0.403†
			(0.214)	(0.211)	(0.211)
Wave 2*Hours Worked per Week			0.258	0.218	0.214
			(0.185)	(0.180)	(0.175)
Wave 1*Hours Worked per Week Sq.			−0.003	−0.003	−0.003
			(0.002)	(0.002)	(0.002)
Wave 2*Hours Worked per Week Sq.			−0.002	−0.002	−0.002
			(0.002)	(0.002)	(0.002)
Human Capital
Years of Education Outside the U.S.				0.999***	0.876***
				(0.121)	(0.117)
Pre-Migration Occupational Prestige				0.081***	0.069***
				(0.019)	(0.018)
Childhood Income (Below Average Omitted)
Average Childhood Income				0.429	0.581
				(1.218)	(1.165)
Above Average Childhood Income				4.316**	3.585*
				(1.524)	(1.469)
Lived in Rural Environment as Child				−1.955†	−1.759†
				(1.052)	(1.010)
Gave to Social Groups Before U.S.				2.396*	1.375
				(1.025)	(0.988)
Any Education in the U.S.				8.471***	7.785***
				(1.323)	(1.273)
English Language Fluency (Speaks English Not at All Omitted)
Not Well				0.077	0.915
				(2.141)	(2.042)
Well				8.219***	7.645***
				(2.273)	(2.172)
Very Well				16.692***	14.354***
				(2.497)	(2.346)
Legal Pathways (Legalization Omitted)
Family Reunification					1.194
					(2.648)
Employment					14.489*** A
					(2.663)
Diversity					−3.146
					(3.263)
Humanitarian					−5.239
					(3.265)
Constant	16.550*	15.655*	18.062*	6.594	10.336
	(7.501)	(7.808)	(8.430)	(7.539)	(7.909)
Number of Persons	1,324
Number of Person-Waves	3,084

^†p<.10

^*p<.05

^**p<.01

^***p<.001

Notes: Superscript A refers to significant differences between male and female models at the p<.05 level. All models contain demographic characteristics, including self-identified race, region of origin, age, age squared, years in the U.S., U.S. region, and the final wave interview year. The final model includes a cross-level interaction between the pathways and wave.

Results

Descriptive Statistics

Tables 1 and 2 show the means and standard deviations of time-constant and time-varying characteristics, respectively. Table 1 shows that the average man with exposure to unauthorized status has been in the United States about 12 years prior to receiving a green card in 2003, is largely from Latin America, and is likely to be married with children. By contrast, continuously authorized men have less exposure to the United States by 2003, living in the U.S. only 4 years. In addition, they are more heterogeneous by region of origin. Women with and without exposure to unauthorized status also differ along these key dimensions. In terms of occupational prestige, men with exposure to unauthorized status begin with a prestige score of 22.5 in wave zero, increasing in average prestige to 31.7 by wave two. By contrast, men without unauthorized experience start with higher prestige scores but have less occupational mobility. Continuously authorized women have slightly more mobility than women with exposure to unauthorized status, progressing from 35.6 in wave zero to a 45.3 in wave two.

Table 1.

Means and Standard Deviations for Characteristics Measured at 2003–2004 Wave

	Men				Women
	Undoc. Experience		Contin. Authorized		Undoc. Experience		Contin. Authorized
	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Demographic Characteristics
Years in the U.S.	12.16	5.84	4.16	4.92	11.38	5.94	3.89	5.10
Region of Origin
Mexico, Central, Latin, South America, Carib.	0.79	--	0.16	--	0.79	--	0.22	--
Africa	0.03	--	0.14	--	0.01	--	0.08	--
South and East Asia	0.08	--	0.38	--	0.09	--	0.41	--
Europe and Central Asia	0.08	--	0.26	--	0.10	--	0.25	--
Middle East	0.02	--	0.05	--	0.01	--	0.03	--
Self-Reported Race
White	0.78	--	0.42	--	0.81	--	0.47	--
Black	0.07	--	0.15	--	0.05	--	0.11	--
Asian	0.08	--	0.39	--	0.09	--	0.40	--
Other	0.07	--	0.04	--	0.05	--	0.02	--
Age	37.17	9.70	36.60	9.32	37.32	9.58	36.38	8.82
U.S. Region at 2003
Northeast	0.19	--	0.33	--	0.17	--	0.28	--
Midwest	0.07	--	0.20	--	0.05	--	0.18	--
South	0.20	--	0.27	--	0.21	--	0.27	--
West	0.54	--	0.21	--	0.57	--	0.27	--
Year Interviewed at 2007–2009
2007	0.62	--	0.70	--	0.61	--	0.70	--
2008	0.14	--	0.13	--	0.15	--	0.11	--
2009	0.24	--	0.18	--	0.24	--	0.20	--
Household Dynamics
Married	0.82	--	0.75	--	0.67	--	0.70	--
Has Children	0.70	--	0.58	--	0.75	--	0.57	--
Household Size	4.20	1.64	3.81	1.69	3.82	1.61	3.79	1.77
Human Capital
Years Education Outside U.S.	9.77	4.77	14.71	4.48	9.14	5.11	13.96	3.64
Childhood Income
Below Average Childhood Income	0.44	--	0.20	--	0.43	--	0.17	--
Average Childhood Income	0.43	--	0.56	--	0.42	--	0.59	--
Above Average Childhood Income	0.13	--	0.24	--	0.14	--	0.24	--
Lived in Rural Environment as Child	0.48	--	0.30	--	0.46	--	0.35	--
Gave to Social Groups Before U.S.	0.26	--	0.45	--	0.25	--	0.46	--
Any Education Inside the U.S.	0.28	--	0.22	--	0.29	--	0.24	--
English Language Proficiency
Speaks English Not at All	0.09	--	0.08	--	0.17	--	0.11	--
Speaks English Not Well	0.38	--	0.25	--	0.39	--	0.26	--
Speaks English Well	0.32	--	0.31	--	0.25	--	0.30	--
Speaks English Very Well	0.22	--	0.37	--	0.18	--	0.33	--
Legal Pathways
Family Reunification	0.37	--	0.24	--	0.48	--	0.37	--
Employment	0.18	--	0.42	--	0.10	--	0.36	--
Diversity	0.02	--	0.27	--	0.03	--	0.20	--
Humanitarian	0.07	--	0.08	--	0.04	--	0.08	--
Legalization	0.36	--	0.00	--	0.35	--	0.00	--
N of Individuals	405		919		269		594

Table 2.

Means and Standard Deviations for Time-Varying Characteristics

	Unauthorized Experience			No Unauthorized Experience
	First U.S. Job (Wave 0)	2003 (Wave 1)	2008 (Wave 2)	First U.S. Job (Wave 0)	2003 (Wave 1)	2008 (Wave 2)
Men
Occupational Prestige
SEI Score	22.5	28.7	31.7	44.3	46.4	47.9
	(18.9)	(21.3)	(23.1)	(26.5)	(26.2)	(25.8)
Occupational Tenure
Years in the Job	3.8	5.9	10.1	2.0	3.1	7.1
	(4.3)	(5.9)	(5.4)	(2.5)	(3.6)	(3.8)
Contingent Work
Hours per Week	41.6	43	44.5	40.8	42.5	43.4
	(13.7)	(11.5)	(13.2)	(12.9)	(11.2)	(43.4)
N of Individuals	405			919
Women
Occupational Prestige
SEI Score	23.6	25.6	29.8	35.6	40	45.3
	(17.3)	(19.3)	(22.0)	(23.8)	(24.4)	(23.8)
Occupational Tenure
Years in the Job	3.6	4.7	8.9	1.6	2.6	6.8
	(4.6)	(4.5)	(4.6)	(2.5)	(3.5)	(3.6)
Contingent Work
Hours per Week	37.5	36.6	36.6	36.2	37.7	38.1
	(14.2)	(12)	(13.0)	(12.4)	(11.8)	(12.1)
N of Individuals	269			594

Gender Differences between Men and Women in Durable Disadvantage

Figure 1 shows predicted occupational prestige levels and trajectories at the three time points for women and men, holding all values other than unauthorized status at their means. The results demonstrate durable disadvantage associated with unauthorized exposure, but this pattern varies by gender.¹ The left side of Figure 1 illustrates that there is an initial occupational sorting gap associated with prior exposure to unauthorized status among women. The predicted gap in occupational prestige between women with and without exposure to unauthorized status widens to 14 points by the time both groups attain LPR, and this larger gap persists at 15 points five years after gaining LPR. In 2008, five years after LPR, previously unauthorized women have an occupational prestige score of 30 points, well behind the 45-point score of continuously authorized women (p<.001). In addition, this widening is statistically different from men at the p<.05 level.

Figure 1. Predicted Occupational Prestige Trajectories for Women (Left) and Men (Right).

Note: Predictions generated from final models of Table 3 (left) and Table 4 (right). Error bars plus/minus the standard error.

Men with exposure to unauthorized status, on the other hand, start with a larger within-gender sorting penalty immediately upon migrating than women. The occupational gap persists even after both groups of men receive LPR, although it converges slightly five years after gaining LPR, with previously unauthorized men scoring 32 and continuously authorized men scoring 48 (p<.001). Consistent with Hypothesis 1a, we find that men have a larger initial sorting penalty attached to prior exposure to unauthorized status than women. In addition, consistent with Hypothesis 1b, while both men and women experience persistent disadvantage associated with exposure to unauthorized status over time, that disadvantage widens for women as it persists for men.

Explaining Variation in Durable Disadvantage of Unauthorized Status among Women

Table 3 presents the factors that may explain differences in occupational sorting and trajectories among immigrant women. Accounting for demographic characteristics in Model 1, women have an initial sorting penalty (the intercept) attached to unauthorized status of 10.17 points (p<.001). Over time (the slope), between waves one and three, that disadvantage widens by 3.86 points (p<.05).

Model 2 accounts for household dynamics. Contrary to Hypothesis 2a, the initial sorting gap between women with and without unauthorized experience is only reduced slightly once accounting for household dynamics—by less than one SEI point (10.17 points vs. 9.46 points, both at p<.001). Model 3 shows that work arrangements also explain little of the initial occupational sorting penalty or change over time, in contrast to Hypothesis 2b. However, human capital characteristics (Model 4) explain 65 percent of the initial sorting penalty, with the initial gap of 10.17 points (p<.001) in Model 1 dropping to 3.60 points (p<.1) in Model 4. This reduction is not consistent with Hypothesis 2c, which predicts a larger explanatory role of human capital for men than for women. Examining the slopes demonstrates that the occupational disadvantage associated with unauthorized status over time increases by wave 3 (five years after LPR) after accounting for human capital characteristics, to 7.6-points (−3.599–4.084)—consistent with the widening disadvantage over time observed in Model 1.

Finally, Model 5 suggests that legal pathways do not explain the initial sorting penalty attached to unauthorized experience for women. The coefficient for unauthorized status does not change at wave one, inconsistent with Hypothesis 2d. Overall, these results suggest that human capital is a salient factor in explaining compounding occupational disadvantage for women. Contrary to hypotheses 2a and 2b, household dynamics and contingent work arrangements are not as salient for women as human capital and legal pathways.

To further examine whether other unobserved factors related to human capital or the pathways to legalization explain the relationship between unauthorized status and occupational disadvantage, we estimate a model with individual fixed effects. Because women with unauthorized exposure are likely differently selected in the likelihood of migration than women without exposure, this model helps to account for these stable and potentially unobserved differences. The results, available upon request, show that cumulative disadvantage persists but is insignificant at the p<.05 level—providing further evidence that unobserved selection based on human capital may help explain women’s trajectories over time.

Explaining Variation in Durable Disadvantage of Unauthorized Status among Men

Table 4 presents the results of growth curve models predicting occupational prestige trajectories for men. Model 1 shows that the gap between immigrant men with and without unauthorized experience is 16 points at wave one (the intercept). By wave three, that gap converges slightly by about five points (the slope).

Accounting for household dynamics in Model 2 or work arrangements in Model 3 explains little of the initial penalty attached to unauthorized experience for men, as expected in Hypothesis 2a. However, contrary to Hypothesis 2b, contingent work arrangements do partially explain differences in occupational trajectories, as the slope for unauthorized status decreases (5.15 at p<.001 to 4.21 at p<.01).

Accounting for human capital characteristics in Model 3 reduces the initial occupational sorting penalty for men at wave one by about 35 percent, to ten points (−16.10 at p<.001 vs. −10.30 at p<.001), consistent with Hypothesis 2c. However, differences in human capital do not predict the magnitude or significance of the interaction between unauthorized status and wave. These results are partially consistent with Hypothesis 2c, which predicts that human capital should explain differences in occupational sorting and trajectories among immigrant men.

Finally, consistent with hypothesis 2d, Model 4 shows that legal pathways explain some of the initial sorting and disadvantage over time for men. The occupational sorting gap of 10.30 points (p<.001) in Model 3 declines by about 20 percent, to 8.10 points in Model 4, and the slopes decline in magnitude and significance as well. From wave one to three, for example, the interaction between unauthorized exposure and wave changes from 4.32 (p<.01) to 3.22 (p<.1).

In sum, after accounting for work arrangements, human capital, and legal pathways, previously unauthorized men start out with, and maintain, considerable disadvantage in the labor market relative to continuously authorized men (consistent with Hypothesis 1). Consistent with Hypotheses 2c and 2d, this initial occupational penalty and disadvantage over time is partially explained by human capital and legal pathways. To further examine selection into unauthorized status based on human capital and legal pathways, we estimate additional fixed effects models for men. The results, available upon request, show similar convergence between waves 1 and 2 but a maintained disadvantage from waves 1 to 3.

Conclusion

Despite a rich body of research documenting the experiences of immigrants in the labor market, we know little about how and why prior exposure to unauthorized status and subsequent unauthorized employment is associated with different labor market trajectories for female and male migrants. Using rich longitudinal data from a nationally-representative sample of lawful permanent residents, we examine how gender and prior exposure to unauthorized status interact in association with occupational levels and trajectories in the United States, as well as consider major sources of gendered occupational patterns.

Our results demonstrate persistent occupational disadvantage associated with unauthorized status among men and increasing disadvantage among women, consistent with the idea that unauthorized exposure is associated with durable, gendered disadvantage in the labor market. Although the initial sorting gap between women with and without exposure to unauthorized status is smaller than the initial male gap, as expected under hypothesis 1a, after both groups gain LPR, the female gap widens to produce meaningful differences in women’s occupational trajectories over time, as expected under hypothesis 1b. Among both men and women, the occupational disadvantage associated with unauthorized status is substantially (though not fully) explained by human capital and legal pathways, as outlined in hypotheses 2c and 2d, and consistent with the idea that selection and legal processes manifest in important ways in the labor market. Models with individual fixed effects provide further evidence that selection plays an important role in explaining the divergent occupational trajectories that women and men experience after transitioning their legal status. Contrary to our expectations, patterns of disadvantage for women were not explained by household dynamics or contingent work arrangements. This finding is inconsistent with prior research on wage gains following IRCA legalization programs (Kossoudji and Cobb-Clark2002; Amuedo-Dorantes, Bansak, and Raphael 2007), suggesting that beyond the IRCA context, women’s additional contingent and often precarious labor both inside and outside the home may be less salient than we once thought in shaping occupational attainment over time.

Although the data analyzed here are the best available for investigating the relationship between unauthorized status and occupational attainment over time, we are unable to measure the full population of currently unauthorized immigrants in the United States. Our focus on immigrants who receive LPR means that we may underestimate the disadvantage associated with unauthorized status, especially if those who ultimately receive LPR are more advantaged than their peers who remain unauthorized. Secondly, and relatedly, our findings are not causal. It is possible that there are additional unobserved differences between respondents with and without unauthorized experience, as well as between men and women. For example, one legal pathway leading to permanent residence, the employer visa, is largely reserved for highly educated workers. Because there are fewer immigrants with unauthorized experience who ultimately acquire LPR through the highly skilled visa program, systematic, unobserved differences between immigrants with employer visas and previously unauthorized immigrants could be driving some of the occupational prestige gap, even after controlling for human capital and those legal pathways. Similarly, there could be unobserved, time-varying factors driving women into both employment and distinct career trajectories than men. For example, to the extent that men worked more than women across waves, this could explain why household dynamics did not play as large a role as expected. Although we account for selection empirically, we may be overestimating the association among gender, unauthorized experience, and occupational trajectories. Finally, we rely chiefly on an indication of unauthorized status to account for occupational sorting and trajectories. Although we also estimate the association between unauthorized employment and occupations, it is possible that there are additional instances where unauthorized status and unauthorized work experience operate independently. Future research should continue to examine individuals’ work and legal status trajectories using the New Immigrant Survey and other sources of administrative data.

These limitations notwithstanding, our study provides important descriptive evidence on the relationship between gender and documentation status in the labor market. One possible explanation for our findings is that immigrant women with unauthorized experience are sorted into vastly different occupations than immigrant women who arrive to the U.S. with documentation. Indeed, in an additional analysis, we include a 13-category variable of broad occupational categories. The results suggest that variation is largely due to occupational segregation between, rather than within, occupations. Considering longstanding occupational segregation by both nativity status and gender (Myers and Cranford 1998), these results suggest that unauthorized women are more likely to enter lower-wage jobs with fewer opportunities for upward mobility than men, such as domestic labor, cleaning and care work, and low-wage service sector jobs (Brown and Misra 2003; de la Luz Ibarra 2000; Hondagneu-Sotelo 2011). In addition, women with unauthorized experience are more likely to stay in those occupations, while continuously authorized women either move up into supervisory roles or out into other industries. These results have important implications for studies of labor market segmentation, the accumulation of disadvantage, and gender inequality.

Historically, in the context of labor market restructuring and industrialization, people with the ascribed characteristics of “white” and “male” were more likely to have more prestigious jobs, with strong prospects for upward mobility in the primary labor market; whereas people ascribed as “black” and “female” were more likely to have less prestigious jobs, without strong prospects for upward mobility in the secondary labor market (Doeringer and Piore 1971; Reich, Gordon, and Edwards 1973). Today, however, deindustrialization, increasing ethno-racially diverse immigration post-1965, and contingent work arrangements have added nativity and documentation status to the characteristics by which people are sorted into inferior work (Hudson 2007). We find that, in today’s segmented labor market, gender and documentation status could be important mechanisms by which adults are sorted into jobs.

These results also suggest that unauthorized experience is a salient life course adversity (Menjívar and Kanstroom 2013). Unauthorized status not only affects initial levels of occupational resources; it also compounds to produce widening gaps in those resources over time, consistent with scarring theories and the predictions of cumulative inequality theory (Diprete and Eirich 2006; Merton 1968). Given that one in four immigrants has some exposure to unauthorized status (Zong, Batalova, and Hallock 2018), future studies would benefit from accounting for immigrants’ prior unauthorized status, as it may operate in tandem with other social categories to shape immigrants’ life chances—both in occupational prestige as well as earnings. In addition, while we cannot account for the timing of unauthorized experience in an immigrant’s life course, it would be useful in future research to investigate whether divergent occupational trajectories also depend on the timing of unauthorized experience.

Finally, our results highlight the importance of gender, both in and outside the labor market. Gender theorists and ethnographers have long recognized that gender is an important facet of work and migration (Mahler and Pessar 2006). Bringing gender to the forefront of migration studies, qualitative work notes that women’s experiences upon migrating are vastly different than men’s depending on their race and class (Mahler and Pessar 2006; Pedraza 1991; Pessar and Mahler 2003). Our study suggests that not only is gender a pivotal axis of stratification; it also works with documentation status in ways that have cumulative effects on the social organization of labor. Future qualitative research should seek to understand the process by which disadvantage accumulates for women even after their legal status changes. Given that women experience lower returns to gaining documentation in the present legal context, increasing occupational disadvantage over time could be due to the legalization process itself. For example, married women are often dependent on the legal status of their largely male spouses. Should women receive different or less information than male spouses, the legalization process could intersect with gender and exposure to unauthorized status in ways that structure women’s vastly different occupational trajectories.

Overall, this research suggests that unauthorized status is a salient life course adversity associated with immigrants’ socioeconomic integration, and may shape the labor market experiences of millions of U.S. immigrants, perhaps especially women. Even after immigrants change their legal circumstances, women remain confined to a set of occupational prospects that shape their career mobility for years to come.

Appendices

Appendix A.

Variables Included in Analysis

Variable Name	New Immigrant Survey Question Wording	Survey Section Source
Gender	I need to ask these questions of everyone, are you male or female?	A
SEI	What kind of work do you do in this job?	C*
Years in US	Now I would like to ask you about your most recent move. In what month and year did you most recently move?	K
Education Outside the US	Now, I have a few questions about your education. How many years of schooling in total have you completed? How many of these years in school were spent in the United States?	A
Education Inside the US
Pre-Migration Occupational Prestige	I’m interested in learning more about the kinds of jobs people have had in their lives both in the United States and in foreign countries. Now I would like to ask you questions about some of the jobs you have had since your sixteenth birthday. Before you came to the United States, had you ever worked for pay or as a family worker in a household enterprise?; What kind of work did you do on this job?	B*
Childhood Income	Now I’d like to ask you some questions about when you were a child. Thinking about the time when you were 16 years old, compared with families in the country where you grew up, would you say your family income during that time was far below average, below average, average, above average, or far above average?	A
Childhood Rural	Were you living in a rural area when about age 10?	A
Social Group Dummy	I’m going to ask you now about things that you did before coming to the United States. While living outside of the United States did you give money, time or goods to any of the following organizations outside of the United States? A labor union?; A business or professional organization?; A charitable organization?; A sports or recreational association?; A social club or community group?; An ethnic or national origin association?	J
Linguistic Incorporation	How well would you say you speak English? (1-Very well to 4-Not at all)	J
Region Born	In what country were you born?	A
Race	What race do you consider yourself to be?	N
Age	In what year were you born?	A
US Region		A
Married	Are you now: Married?; Living together in a marriage-like relationship but not married?; Separated?; Divorced?; Widowed?; Never married, not living with someone in a marriage like relationship?	A
Children	Now, we would like to ask about births of children. How many children have you yourself ever given birth to / have you ever fathered?	A
Household Size	Including yourself, how many people are currently living in your household?	A
Years in the Job	In what year did you start working in [occupation]?	C
Hours per Week	How many hours a week do you usually work at this job?	C

Notes:

^*Signifies NIS data merged on occupations with occupational prestige scores from the 2003 American Community Survey. Steven Ruggles, Sarah Flood, Ronald Goeken, Megan Schouweiler and Matthew Sobek. IPUMS USA: Version 12.0 [dataset]. Minneapolis, MN: IPUMS, 2022. https://doi.org/10.18128/D010.V12.0.

Appendix B

Our coding of unauthorized experience is valuable for capturing individuals who had any unauthorized experience in the U.S. However, there could be instances where individuals with unauthorized experience had a trip where they worked with authorization. Similarly, it is possible there are instances where individuals who did not have unauthorized experience had unauthorized employment on a previous trip. We therefore conduct a sensitivity analysis where we construct unauthorized work experience based on working without a valid visa (or working on a visitor’s visa for pleasure) on any given trip. The results (Table B) are comparable to those presented in the main text.

Although outside the scope of the primary aims of our analysis, we also examined the profiles of individuals who were coded as having unauthorized employment but not unauthorized experience and vice versa. The results suggest the majority of individuals who were not unauthorized worked briefly on a visitor’s visa for pleasure prior to getting LPR status. Meanwhile, the largest share of individuals who had authorized work experience but unauthorized experience entered on a range of visa statuses throughout trips and reported first U.S. jobs on those authorized trips, even if they also took unauthorized trips.

Table B.

Replicating Model 1 with Unauthorized Employment Variable

	Men	Women
Exposure to Unauthorized Employment	−13.346***	−9.347***
	(1.1999)	(2.247)
Wave 1	2.598**	3.955***
	(0.838)	(1.029)
Wave 2	5.369***	9.129***
	(0.797)	(0.968)
Exposure to Unauthorized Emp*Wave 1	3.789*	−1.482
	(1.540)	(1.974)
Exposure to Unauthorized Emp*Wave 2	2.908+	−3.958*
	(1.530)	(1.865)
Constant	12.750+	25.399**
	(7.479)	(9.690)
N of Individuals	1,324	863
N of Person-Waves	3,084	1,958

^†p<.10

^*p<.05

^**p<.01

^***p<.001

Notes: Both models contain demographic characteristics, including self-identified race, region of origin, age, age squared, years in the U.S., U.S. region, and the final wave interview year.