Do Employer-Sponsored Immigrants Fare Better in Labor Markets Than Family-Sponsored Immigrants? =============================================================================================== * Julia Gelatt ## Abstract Debates about revising U.S. legal immigration policies tend to question the economic value of immigrants sponsored by family members rather than by employers. To date, little evidence has been cited. This article uses the latest data to measure legal immigrants’ characteristics and economic outcomes by class of entry, comparing employment rates, self-employment rates, and occupational outcomes of family-sponsored immigrants, humanitarian migrants, and diversity visa immigrants with those of employer-sponsored immigrants. It finds that most legal, permanent immigrants to the United States show high employment rates relative to the overall U.S. population after several years in the country, but that employment-sponsored immigrants and their spouses bring the highest education and English proficiency and work in the most highly skilled occupations both initially and over time. * immigration * immigration policy * visas * employment * occupations Over the last decade, debates on reforming the U.S. legal immigration system have centered on proposals to increase the skill level of immigrants to the country by cutting certain family-sponsored immigration categories and increasing the number of green cards for higher-skilled workers. For example, the proposed 2017 RAISE Act would have slashed legal immigration in half, by eliminating all family-sponsored immigration categories other than spouses and minor children, eliminating the diversity visa program, and capping refugee admissions each year. Similarly, in 2006, 2007, and 2013, comprehensive immigration reform debates included proposals to eliminate several family-sponsored categories and to increase the size of the skills-based immigration system. These proposals rest on the assumption that immigrants selected by employers or proposed merit-based points systems would bring greater human capital and greater economic benefit to the United States than those sponsored by their family members or arriving though humanitarian channels. Remarkably little evidence is available to test that assumption, however. Most analyses of the costs and contributions of immigrants rely on data from the U.S. Census Bureau, which does not reveal immigrants’ legal status, much less the admissions category through which those who are legal secured their legal status. This article draws on data from the New Immigrant Survey, which followed a cohort of new lawful permanent residents (LPRs) from 2003, when they obtained their green cards for LPR status, through 2007–2009. The survey provides a rare look at the detailed characteristics of legal immigrants, by class of entry, and of their trajectories over their first few years in the United States. The survey captured information on immigrants’ English ability, educational attainment, employment, self-employment, and occupations. I examine how human capital characteristics of immigrants compare at the time of getting their green card. Then, I investigate how the labor-market trajectories of various categories of family-sponsored immigrants compare with those of employer-sponsored immigrant and those entering through the diversity visa lottery or humanitarian channels. Although these data represent an earlier period in U.S. immigration history, the policies shaping permanent immigration to the United States have not substantially changed since 2003 or in fact since 1990. Further, many of the major legal migration streams to the United States in 2003 are similar to those operating today. The world regions of origin of family-sponsored and employer-sponsored immigrants were similar in fiscal year (FY) 2003 and in FY 2018, the latest year for which data are available, except for a smaller share arriving from European countries among family-sponsored categories. On the other hand, some smaller visa categories have seen substantial change in national origins—diversity visa recipients were more likely to be Asian and less likely to be European in 2018 than in 2003, and refugees and asylees were more likely to be Latin American or Asian (including Middle Eastern) and less likely to be from Europe in 2018 compared to 2003 (OIS 2019, 2004). In this work, I find that new lawful permanent residents from 2003 had high educational attainment when they obtained their green cards—a higher proportion having completed college than among the overall U.S. population. At the same time, substantial shares of new immigrants had limited English proficiency. Employer-sponsored immigrants and their spouses had the highest education and English proficiency on getting a green card, but diversity visa holders also had relatively high education. In terms of labor-market outcomes, many family-sponsored immigrants and diversity visa holders had relatively low employment rates on obtaining a green card, but saw substantial increases in employment by the second survey wave. By the second survey, new LPR women and men had employment-to-population ratios that exceeded those of U.S. male and female workers overall. Nevertheless, employment rates of family-sponsored immigrants and refugees and asylees remained significantly lower than those of employer-sponsored immigrants by wave two. Among those employed, refugees and asylees showed the highest rates of operating businesses that employ other workers. Diversity visa holders and refugees and asylees saw the greatest gains in the skill level of their occupations in the four to six years after obtaining their green card. All the same, employer-sponsored immigrants were in more-skilled occupations than all other groups in wave two. ### U.S. Permanent Immigration Policy Each year around one million immigrants obtain green cards for lawful permanent resident status in the United States. Following a complex series of quotas, floors, and ceilings, green cards are allocated among several main categories. Over the past five years, 13 percent of those obtaining green cards each year were sponsored by employers (though about half of these visas went to family members of employer-sponsored principal applicants), about two-thirds through family sponsorship, 14 percent through humanitarian channels as asylees or refugees, 5 percent through the diversity visa lottery, and a small share through some other channel (see figure 1). Of those securing green cards each year, many have been already been living and working in the United States on some sort of temporary visa or as unauthorized immigrants. In fiscal year 2018, 52 percent were adjusting their status from within the United States. For employer-sponsored green card holders, this share was 80 percent (OIS 2019). ![Figure 1.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F1.medium.gif) [Figure 1.](http://www.rsfjournal.org/content/6/3/70/F1) **Figure 1.** Major Class of Entry for Those Securing Green Cards, FY 2014–2018 Average *Source*: OIS 2019. Claims that employer-sponsored immigrants contribute more to the country than family-sponsored immigrants generally point to the higher education of employer-sponsored immigrants. A wide body of evidence shows that more highly educated immigrants contribute more to economic growth of the country and to government coffers (National Academies of Sciences, Engineering, and Medicine 2017). U.S. immigrant selection criteria do mean that most employer-sponsored immigrants are highly educated. Most employer-sponsored green cards are preserved for workers who have at least a college education. Only five thousand employer-sponsored visas are available each year for low-skilled workers with fewer than two years of training. In contrast, family migrants are eligible to immigrate regardless of skills and education if they have a family sponsor. LPRs may sponsor spouses, minor children, and unmarried adult children, and U.S. citizens may also sponsor parents, siblings, and married adult children. Refugees and asylees are likewise not selected on the basis of skills. Instead, they may be eligible for U.S. admission if they demonstrate a well-founded fear of persecution in their home country for reasons of race, religion, nationality, political opinion, or membership in a particular social group, and they pass thorough screenings and background checks. Refugees are vetted and selected abroad; asylees apply for recognition from inside the United States. Diversity visa holders are selected through a lottery and can apply if they are from a country that sends few migrants to the United States. They are also required to have a high school diploma or equivalent education, and to have at least two years of experience in a job that requires at least two years of training. ## RELATIONSHIP BETWEEN CLASS OF ENTRY, SKILL, AND LABOR-MARKET OUTCOMES Being selected on the basis of skills does not guarantee that employer-sponsored immigrants will fare better in the labor market than those selected through family ties, humanitarian claims, or the diversity visa lottery. Those selected for reasons other than skill levels may nevertheless bring strong education and skills to the United States. Further, those bringing the highest skills may not always experience the greatest labor-market success. U.S. immigrants overall are more likely to be college educated than U.S.-born residents and have been since at least the 1990s (Batalova and Fix 2017), suggesting that substantial numbers of all legal immigrants, not just those who are employer sponsored, arrive in the country with strong educational backgrounds. Further, immigrants may fare best in U.S. labor markets when their skills complement rather than compete with those of U.S. workers. This may mean that technical and trade skills benefit immigrant workers as much as a college or graduate education. Immigrants are also, on average, more entrepreneurial than U.S.-born workers, which may help them secure greater economic and labor-market mobility, even with when they have less education. Even if employer-sponsored immigrants bring greater skills to the country, they may not find jobs commensurate with their education and training. As many as a quarter of immigrants with college degrees in the United States experience “brain waste”—either lacking employment or working in low-skilled jobs (Batalova, Fix, and Bachmeier 2016). Canada began to shift its immigration policies in recent years after studies showed that substantial numbers of highly skilled workers selected through their points system were working in jobs that did not match their education level (Vincenza Desiderio and Hooper 2016). In addition, although employer-sponsored immigrants may come to the United States as pioneers, moving to a country where they have few ties, family-sponsored immigrants, by definition, have family networks in the country. The social capital embedded in these networks may be very helpful in locating and applying for open jobs, and could assist family-sponsored immigrants in finding jobs commensurate with their education, skills, and training. Further, given that the federal safety net for low-income families is mostly unavailable to new legal immigrants in their first five years, family-sponsored migrants may rely on their family’s support as they adapt to the United States, using their family support as a launching pad to faster success in the country. ## LITERATURE A few, mostly decades-old, studies have looked at how immigrants entering through different migration streams fare within the United States. Reflecting the long-standing nature of concerns about the productivity of family-sponsored migrants, in the 1990s, Harriet Duleep and Mark Regets (1996) explore how immigrants from a country admitted through family versus employer sponsorship shaped initial wages and wage trajectories. Using 1980 census data, they find that immigrants from countries that sent more family migrants had lower initial earning but faster earnings growth than their counterparts from countries sending mainly employer-sponsored migrants. This earnings growth was fast enough that they predicted family-sponsored migrants would reach wage parity with employer-sponsored immigrants in eleven to eighteen years (Duleep and Regets 1996). However, these analyses control for education, one of the strongest predictors of earnings—so they compare earnings of employer-sponsored and family-sponsored immigrants *with the same education level.* For policy purposes, it would be better to compare outcomes without this education control, given that the real question for policy is whether employer-sponsored streams are more highly educated than family-sponsored immigrants, and if so, whether that higher education brings substantially improved labor-market outcomes. Research on men from the 1977 cohort of new legal immigrants offers similar findings—immigrants sponsored by spouses had lower-skilled jobs when getting their green cards than employer-sponsored immigrants but relatively rapid occupational upgrading, at the same time that employer-sponsored immigrants saw slight occupational downgrading. Despite this, the data suggest that occupational standing would never converge between the two groups (Jasso and Rosenzweig 1995). Looking at the first wave only of the New Immigrant Survey, Jeanne Batalova and Michael Fix (2008) find that family-, diversity-, and humanitarian-based LPRs experience occupational downgrading from their last occupation abroad to their first U.S. occupation, but that employer-sponsored immigrants do not. This downgrading was worst for diversity immigrants, who are least likely to have family ties in the United States. This observed downgrading on arrival suggests that their subsequent occupational upgrading, observed between their first U.S. occupation and their occupation at the time of securing a green card, may represent a partial return to premigration occupational standing. Employer-sponsored immigrants saw no such upgrading, perhaps in part because they had not experienced downgrading in the first place. ## RESEARCH QUESTIONS In this article, I update the literature by exploring the labor-market trajectories of the 2003 cohort of new legal immigrants four to six years after they secure a green card. I ask two basic sets of questions: > How do the skills and characteristics of family-sponsored, diversity visa, and humanitarian LPRs compare with those of employer-sponsored LPRs on getting a green card? How do those characteristics compare with the overall U.S. population? > > How do the labor-market trajectories of family-sponsored, diversity visa, and humanitarian LPRs compare with those of employer-sponsored LPRs several years after obtaining a green card? How do these immigrants compare in terms of employment rates, self-employment rates, and occupational skill level several years after obtaining green cards? Do these differences persist after controlling for factors unrelated to how immigrants are selected for migration to the United States? These questions are inherently descriptive. The goal of this article is to inform changes to legal immigration policies by describing what the immigrants who come through existing U.S. immigration channels look like and how they fare in the United States. The goal is not to isolate a causal effect of arriving through a particular channel on outcomes in the United States, net of other characteristics. Therefore, selection by immigrants of different characteristics into particular U.S. immigration channels is part of the story this article aims to portray. ## DATA: NEW IMMIGRANT SURVEY I answer these questions using the New Immigrant Survey (NIS). The NIS is a longitudinal, representative survey of people who became lawful permanent residents in 2003. It covers topics such as family demographics, schooling, English ability and use, health, labor-force participation, occupations, public benefits use, and religion. Respondents were sampled from the top eighty-five metropolitan statistical areas in the United States and the top thirty-eight counties. The first survey reached 8,573 adults—68.6 percent of the adult sampling frame. The survey oversampled spouses of U.S. citizens, employment-sponsored immigrants, and immigrants with diversity visas, and applied sampling weights to adjust for this sampling strategy. Respondents were first interviewed, on average, seventeen weeks after obtaining their green card (Massey, Jasso, and Espinoza 2017). About 60 percent of baseline interviews were conducted by telephone, and 40 percent in person (Massey 2011). The survey was conducted in respondents’ language of choice. Respondents were interviewed again between 2007 and 2009. Locating respondents for this second wave was more difficult than anticipated, because attitudes toward immigrants became more hostile over this period, and the U.S. Citizenship and Immigration Services (USCIS) did not provide the study directors with updated contact information for families (Massey, Jasso, and Espinoza 2017). The follow-up survey had a lower response rate of 46 percent of the baseline sample, reaching 3,902 respondents. To correct for survey nonresponse, which was observed to be nonrandom across several observed characteristics, the survey developers generated nonresponse weights. These weights assume that LPRs who did not respond to the survey were still be in the United States, which is a reasonable assumption given that U.S. emigration rates are relatively low. Jonanthan Schwabish (2009) looks at immigrants with earnings tracked by the Social Security Administration and finds that those in their first four years in the United States had an annual emigration rate of 2.7 percent. ## METHODS I use both waves of the New Immigrant Survey to examine longitudinal labor-market outcomes. Specifically, I observe employment status during the first and second waves, rates of being both self-employed and employing other workers, and occupational skill level among those who are employed at each survey wave. Employment status reflects the employment-to-population ratio. I measure occupational skill level by matching occupations with job zones, developed by the Department of Labor, which classify all occupations into five categories ranging from 1, occupations that need little or no preparation to 5, occupations that need extensive preparation. Job zone 1 jobs generally require a high school diploma or less education; job zone 5 occupations generally require a graduate degree (for more, see table A1). I examine these outcomes descriptively, then run multivariate regression models predicting wave two outcomes, controlling for a range of factors. I use information provided in the NIS to classify adult LPRs into eight groups by class of entry: spouses of LPRs and U.S. citizens, parents of U.S. citizens, siblings of U.S. citizens, employer-sponsored principal visa applicants, spouses of employer-sponsored immigrants, diversity visa holders, refugees and asylees, and other legal immigrants outside these groups. The other legal immigrant category includes adult unmarried children of LPRs and adult married and unmarried children of U.S. citizens, among others. I focus on these eight groups to demonstrate how the occupational trajectories of those favored by immigration reform proposals—employer-sponsored immigrants—and those maintained by immigration reform proposals—spouses of LPRs and U.S. citizens—compare with categories often slated to be reduced or eliminated by reform proposals: siblings, parents, adult children, diversity visa holders, and refugees and asylees. Before looking at these outcomes among each immigrant group, I compare the background and demographic characteristics of each immigrant group before showing descriptive statistics of employment outcomes for those respondents who made it to the second survey wave. Last, I complete multivariate regressions predicting employment in wave two; having a business that employs others in wave two, among those employed in that wave; occupational skill level in wave two, among those employed in that wave; and person-level change in occupational skill level between wave one and wave two, among those employed in both waves. I look at rates of being both self-employed and employing other workers rather than simply at self-employment, because self-employment can represent a survival strategy for struggling workers, rather than a sign of economic success. People who are both self-employed and managing other employees are more likely to be operating a successful business, and are increasing employment opportunities in the United States. For the first two outcomes, I use a linear probability model. For occupational skill level, I use an ordinary least squares (OLS) regression model because the distribution of wave two occupational rankings and of the change in occupational rankings both approximate a normal distribution. In these models, I introduce two sets of controls. The first set controls for factors that vary between the eight groups I study but are unrelated to whether someone is eligible to obtain a green card through any particular channel: gender, age (and age-squared), the year of interview, the person’s years of U.S. residence in the second survey wave, whether they adjusted status to a green card within the United States, and their country or region of birth. Years of U.S. residence are based on when the person last came to the United States to live. I control for the year of the interview to account for the fact that the Great Recession was setting in as the second wave of interviews were occurring. The control for country or region of birth controls for a variety of factors—how well a country’s educational credentials are recognized in the United States, cultural preferences related to employment, and racialized experiences within the United States, among other factors. The second set of controls accounts for human capital, measured through educational attainment; whether any education was earned in the United States, given that U.S. employers may value U.S. education more highly; whether the respondent had limited English proficiency in the wave one survey; and whether they worked before coming to the United States. All these factors are related to selection factors under U.S. immigration policy, or in the case of English proficiency, proposed changes to policy. It does not make much sense from a policy perspective to look at how groups compare, net of their human capital, because human capital is inherent to a person’s ability to migrate through a particular skills-based migration stream. I include these factors in the last iteration of each regression model simply to view how much of variation in labor-market outcomes between groups can be explained by observable human capital differences.1 In examining baseline characteristics of each group, I use wave one survey weights, provided with the NIS data. In looking at employment outcomes, I limit the sample to those who responded to the second survey wave, and use combined wave one and wave two weights, generated by the survey team, which combine wave one sampling weights and wave two nonresponse rates. I impute missing data using Stata 14’s mi impute chained command. ## FINDINGS ### Characteristics by Class of Entry First, I compared the demographic and background characteristics of immigrants, by class of entry. Table 1 presents these descriptive statistics. View this table: [Table 1.](http://www.rsfjournal.org/content/6/3/70/T1) **Table 1.** Baseline Characteristics of LPRs by Class of Entry, 2003 As to gender, a slight majority of new green card holders are women, driven by the fact that two-thirds of immigrants through larger, uncapped categories of spouses of U.S. citizens and of parents of U.S. citizens are women. The majority—66 percent—of employer-sponsored principal applicants are men, while the great majority of spouses of employer-sponsored applicants—78 percent—are women. As to age, unsurprisingly, parents of U.S. citizens are oldest, at a mean of sixty-three. The next oldest group is siblings of U.S. citizens, at forty-eight years old. This reflects the fact that due to per-country caps on green cards, many people face decade or longer backlogs for sibling visas. Employer-sponsored immigrants are already well into their careers, age thirty-seven on average; refugees and asylees are on average forty. Diversity visa holders are younger, at a mean of thirty-three. As a result of these differences, employment outcomes at the first NIS survey wave reflect earlier U.S. career outcomes for some groups and mid-U.S. career outcomes for other groups. As to the rate adjusting status within the United States, the great majority of employer-sponsored immigrants—72 percent—were adjusting to LPR status from some other status in the United States, likely student visas or H-1B temporary worker visas. Spouses also had high rates of adjusting status within the United States; other groups show lower rates. In regard to place of origin, immigrants securing green cards through spouses or parent categories are primarily from Mexico or other parts of Latin America, whereas most siblings and employer-sponsored immigrants are from Asia. Diversity visa holders in 2003 were primarily from Europe, Africa, or the Middle East. Refugees and asylees showed a wide mix of origins. I looked at rates of both limited English proficiency (LEP), defined as speaking English less than very well, on a self-reported measure, and of low LEP, defined as speaking English less than well, on the same measure. Overall rates of LEP are quite high, 71 percent for the total sample. LEP rates are lowest for employer-sponsored immigrants, at 45 percent, and highest for parents of U.S. citizens, siblings of U.S. citizens, and refugees and asylees at 89, 87, and 84 percent, respectively.2 Looking at low LEP rates shows similarly that parents and siblings are most commonly low LEP. Overall educational attainment is relatively high. Forty-six percent of these categories of LPRs have a bachelor’s degree or higher. As a point of comparison, just 27 percent of all Americans, age twenty-five and older, had a bachelor’s degree or higher in 2003 (Stoops 2004). Education is, unsurprisingly, highest among employer-sponsored immigrants, a full 80 percent of employer-sponsored principal applicants having a bachelor’s degree or higher. Diversity visa holders are next, 55 percent having a bachelor’s or higher. Parents have the least education, likely reflecting historically lower education levels in their countries of origin. In sum, the characteristics of new permanent immigrants vary considerably by class of entry. Rates of educational attainment are high—and highest among employer-sponsored immigrants—though a majority of new LPRs have limited English proficiency. ### Variation in Labor-Market Outcomes Looking descriptively at the employment rates of new green card holders shows that among both men and women, all other categories of immigrants show lower rates of employment shortly after receiving their green cards relative to employer-sponsored principal applicants. However, several groups—siblings, diversity visa holders, female spouses, and other immigrants—saw significant increases in employment rates by the second survey wave, four to six years later. In contrast, employer-sponsored principal applicants show steadily high employment rates between the two periods. By the second survey wave, high shares of male immigrants were employed for all classes of entry except for parents—who are generally at retirement age, as were relatively high shares of female immigrants. On average, the employment rates of male immigrants rose from 76 to 86 percent over this period, but for female immigrants from 47 to 64 percent. Figures 2 and 3 illustrate these descriptive measures. In comparison, for the overall U.S. population, 69 percent of men were employed in 2003 and 68 percent on average in 2007–2009. Among women in the overall U.S. population, these shares were 56 percent in 2003 and 56 percent in 2007–2009. Therefore, male new green card holders were more likely to be employed than U.S. workers overall in both periods, and female new green card holders started with lower employment rates but surpassed U.S. women overall by the second wave. ![Figure 2.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F2.medium.gif) [Figure 2.](http://www.rsfjournal.org/content/6/3/70/F2) **Figure 2.** Employment Rates of Immigrant Men in Wave One and Two, by Class of Entry *Source*: Author’s analysis of New Immigrant Survey data, weighted for sampling design and for wave two nonresponse (Jasso et al. 2006, 2014). *Note:* USC = U.S. citizen. **p* < .05, denotes significance of difference from wave one ![Figure 3.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F3.medium.gif) [Figure 3.](http://www.rsfjournal.org/content/6/3/70/F3) **Figure 3.** Employment Rates of Immigrant Women in Wave One and Two, by Class of Entry *Source*: Authors’ analysis of New Immigrant Survey data, weighted for sampling design and for wave two nonresponse (Jasso et al. 2006, 2014). *Note:* USC = U.S. citizen. **p* < .05, denotes significance of difference from wave one Among those not employed in wave two, substantial shares were filling other roles in their households and not looking for work. Almost half (48 percent) said they were homemakers or on parental leave, and 13 percent that they were retired. Spouses of employer-sponsored principal applicants and spouses of U.S. citizens and LPRs were most likely to report being homemakers or new parents, 91 percent and 68 percent, respectively, of those not employed saying they were homemakers. Forty-nine percent of parents of U.S. citizens and 42 percent of siblings of U.S. citizens who were not employed in wave two said they were homemakers or on parental leave. Parents were most likely to report being retired. Of parents not employed in wave two, 29 percent said they were retired. To net out demographic and other factors that may explain differences in employment between immigrants in different classes of entry, I ran linear probability regression models predicting employment at wave two. Given strong differences in male and female employment rates, I ran these separately by gender. Absent any controls, in the male sample, all classes of entry showed lower employment rates than employer-sponsored immigrants, with the exception of spouses of employment-sponsored immigrants (see table 2). After controlling for factors not related to U.S. immigrant selection policies—age, years of U.S. residence, whether they adjusted status within the United States, country or region of birth, and the timing of the wave two interview—four groups still have significantly lower employment than employer-sponsored principal applicants: spouses, parents, refugees and asylees, and other immigrants (including adult children of citizens and LPRs). In contrast, spouses of employer-sponsored immigrants, siblings of U.S. citizens, and diversity immigrants show similar employment rates to employer-sponsored principal applicants. Parents have 25 percent lower employment rates than employer-sponsored principals, while the other three groups have employment rates only 7 to 11 percent lower. View this table: [Table 2.](http://www.rsfjournal.org/content/6/3/70/T2) **Table 2.** Linear Probability Models Predicting Employment in Wave Two, Men In the female sample, all classes of entry showed lower employment rates than employer-sponsored immigrants, even after adding controls for age, duration of U.S. residence, whether adjusted status within the United States, country or region of birth, and the interview year (see table 3). Controlling for educational attainment, English proficiency, and premigration work experience explains the lower employment of sibling, diversity, and other immigrants; spouses, parents, employer-sponsored spouses, and refugees and asylees still show significantly lower employment. Education is more strongly correlated with employment for women than for men. View this table: [Table 3.](http://www.rsfjournal.org/content/6/3/70/T3) **Table 3.** Linear Probability Models Predicting Employment in Wave Two, Women In sum, both male and female non-employer-sponsored immigrants show rapid growth in employment in the years following their acquisition of a green card. In fact, both male and female new green card holders show higher employment rates than U.S. men and women overall after several years in the United States. However, even after accounting for demographic differences between groups, many family-sponsored and humanitarian classes of immigrants show lower employment rates than employer-sponsored principal immigrants a few years after migration. Next, I look at self-employment, the share of employed workers who have a business and employ others—that is, those who are self-employed and have at least one other employee (see figure 4). Wave one shows relatively higher rates of business ownership among refugees and asylees—9 percent—relative to other groups. Overall, 3 percent of new LPRs were self-employed and employed others in 2003. By the second survey wave, the figure is more than double, 7 percent. At wave two, refugees and asylees still had the highest rates of business ownership, 12 percent, followed by 7 percent of spouses, employer-sponsored principal applicants, and employer-sponsored spouses. Parents, who are likely entering retirement, are the only group with lower rates of business ownership in wave two than in wave one. ![Figure 4.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F4.medium.gif) [Figure 4.](http://www.rsfjournal.org/content/6/3/70/F4) **Figure 4.** Share of Employed Workers Who Are Self-Employed and Employ Others *Source*: Author’s analysis of New Immigrant Survey data, weighted for sampling design and for wave two nonresponse (Jasso et al. 2006, 2014). *Note:* USC = U.S. citizen. **p* < .05, denotes significance of difference from wave one To explore how self-employment rates compare across groups net of other factors, I ran a linear probability model predicting self-employment (see table 4). In this model, I combine men and women because findings were similar. These models show that in the absence of any controls, refugees and asylees have significantly higher rates of employing others than employer-sponsored principal applicants. Other immigrants, a group that includes adult children of citizens and LPRs), have lower rates, though the differences are small in magnitude. After controlling for gender, age, duration of U.S. residence, status adjustment within the United States, country or region of birth, and interview year, refugees and asylees still show slightly higher rates of employing others. View this table: [Table 4.](http://www.rsfjournal.org/content/6/3/70/T4) **Table 4.** Linear Probability Models Predicting Being Both Self-Employed and Employing Others, Among Employed Workers in Wave Two For occupational skill level, I look at the “job zones” (the skill level required for the jobs worked by new green card holders), as shown in figure 5. Job zones classify occupations on a scale of one to five, where one is jobs that require little or no preparation and five is jobs that require extensive preparation (for more, see table A1). Looking descriptively at the mean job zones of employed workers in wave one and wave two shows that most classes of immigrants saw a slight improvement in the skill level of their occupation between wave one and wave two, though this change was significant only for diversity visa holders. Parents of U.S. citizens and employer-sponsored principal applicants did not see such an improvement. ![Figure 5.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F5.medium.gif) [Figure 5.](http://www.rsfjournal.org/content/6/3/70/F5) **Figure 5.** Mean Job Zones Among Employed Workers *Source*: Author’s analysis of New Immigrant Survey data, weighted for sampling design and for wave two nonresponse (Jasso et al. 2006, 2014). *Note:* USC = U.S. citizen. **p* < .05, denotes significance of difference from wave one The regression model predicting an immigrant’s wave two job zone shows that all groups have lower-skilled jobs than employer-sponsored principals (see table 5). The difference is smallest for spouses of employer-sponsored immigrants and greatest for parents and siblings of U.S. citizens and other immigrants (including adult children of LPRs and U.S. citizens). After adding the first set of controls, all groups still have significantly lower occupational skills than employer-sponsored principal applicants. Controlling for education, English proficiency, and U.S. work experience reduces the gaps in occupational skill level for most groups. After adding these controls, spouses of employer-sponsored immigrants show occupational skill levels similar to those of employer-sponsored principal applicants. View this table: [Table 5.](http://www.rsfjournal.org/content/6/3/70/T5) **Table 5.** OLS Regression Models Predicting Job Zone (1–5) Among Employed Workers in Wave Two Another way to look at occupational skill levels is whether individual immigrants experienced occupational mobility, finding more-skilled jobs by 2007–2009 than directly after receiving their green card. Figure 6 shows the mean change in job zones for immigrants, by class of entry, between the first and second NIS survey, focusing now on just those who were employed in both wave one and wave two. Several groups—spouses, diversity visa holders, refugees and asylees, and other immigrants (including adult children of LPRs and U.S. citizens)—show significant improvement in the skill level of their jobs from wave one to wave two. ![Figure 6.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F6.medium.gif) [Figure 6.](http://www.rsfjournal.org/content/6/3/70/F6) **Figure 6.** Mean Change in Job Zone Between Wave One and Wave Two, Among Those Employed in Both Waves *Source*: Author’s analysis of New Immigrant Survey data, weighted for sampling design and for wave two nonresponse (Jasso et al. 2006, 2014). *Note:* USC = U.S. citizen. **p* < .05, denotes significance of difference from zero change in job zone An OLS regression model of the change in job zones between wave one and wave two, among workers employed in both waves reveals that spouses, diversity visa holders, and refugees and asylees all experienced greater occupational mobility than employer-sponsored principal applicants (see table 6). No groups show significantly less occupational mobility than employer-sponsored principals. After controlling for age, gender, duration of residence, adjustment of status in the United States, country or region of birth, and year of interview, this higher mobility persists for diversity visa holders and for refugees and asylees. View this table: [Table 6.](http://www.rsfjournal.org/content/6/3/70/T6) **Table 6.** OLS Regression Models Predicting Change in Job Zone (1–5) Between Wave One and Wave Two, Among Workers Employed in Both Waves However, some of this job mobility may represent immigrants regaining the occupational standing that they enjoyed in their home countries—that is, it may show recovery from brain waste. Considering only those who had jobs in their home countries before migration, and who were working at waves one and two, figure 7 presents trajectories in immigrants’ occupational standing over time. Employment-sponsored principal applicants, who were selected for a visa based on their match with a U.S. employer, have better occupational standing in their first U.S. job than in their last job outside the country. No other group saw occupational progression, however, and several groups—siblings, diversity visa recipients, refugees and asylees, and other immigrants (including adult children of citizens and LPRs)—saw occupational downgrading. This change is likely the result of a combination of limited English skills, lack of recognition of foreign professional credentials, and limited familiarity with how to navigate U.S. labor markets. For siblings, diversity visa holders, and other immigrants, their occupational standing at wave two of the survey was still lower than their last job outside the country. Refugees were able to recover to their premigration occupational standing. Parents saw occupational downgrading over time in the United States. ![Figure 7.](http://www.rsfjournal.org/https://www.rsfjournal.org/content/rsfjss/6/3/70/F7.medium.gif) [Figure 7.](http://www.rsfjournal.org/content/6/3/70/F7) **Figure 7.** Mean Job Zone Before Migration, in First U.S. Job, at Wave One, and at Wave Two *Source*: Author’s analysis of New Immigrant Survey data, weighted for sampling design and for wave two nonresponse (Jasso et al. 2006, 2014). *Note:* USC = U.S. citizen. **p* < .05, denotes significance of difference from pre-migration job zone ## CONCLUSION The data show that on obtaining green cards for permanent residence, employer-sponsored immigrants show the highest level of human capital and employment, and the highest-skilled occupations, relative to family-sponsored immigrants or those entering under a diversity visa or humanitarian channels. But other classes of immigrants—family-sponsored, diversity, and refugee and asylee—are not poorly educated on the whole. In 2003, most groups of new permanent immigrants had higher rates of college completion than the U.S. population overall. The biggest exception is parents of U.S. citizens, who represent the lower educational attainment of past generations. Immigrants coming through different classes of entry also represent different mixes of ages, gender, and years of U.S. residence, which must be accounted for when generalizing about their outcomes in the United States. Although most new LPRs have high rates of employment relative to the overall U.S. workforce, particularly several years after gaining their green cards, family, diversity, and humanitarian immigrants do not have the same levels of occupational success as employer-sponsored immigrants. New LPR men have employment rates higher than those of U.S. men overall, and even higher levels of employment several years later. New LPR women start with lower employment than U.S. women overall but see strong increases after several years, with most groups exceeding U.S. women’s employment overall. All the same, most classes of immigrants still have significantly lower employment rates at wave two than employer-sponsored entrants—the exceptions being male siblings of U.S. citizens, male spouses of employer-sponsored immigrants, and male diversity visa holders. Four groups—spouses, employer-sponsored principal applicants, diversity visa holders, and other immigrants (including adult children of citizens and LPRs)—showed significant increases in rates of self-employment and employing others. Relative to employer-sponsored immigrants, refugees and asylees had significantly higher self-employment rates at wave two. Finally, looking at occupational skill levels shows that employer-sponsored principal applicants have the highest-skilled jobs at wave two, significantly higher than those of other LPRs. But refugees and asylees and diversity visa holders see the fastest growth in the skill level of their jobs from wave one to wave two. Several categories of family-sponsored immigrants, as well as diversity visa holders and humanitarian migrants, experience substantial occupational downgrading in their first U.S. job. Although some groups are able to recover over time, others, such as parents, siblings, diversity visa holders, and other immigrants (including adult children of LPRs and U.S. citizens), do not, at least not during their first four to six years in the country. ### Limitations This work has two main limitations in application to current policymaking. First, the 2003 cohort of new LPRs does not necessarily represent today’s immigration streams. Although immigrants in 2003 entered under the same set of rules and policies as today, and the top sending countries to the United States have remained the same, global political and economic changes have meant that some immigration streams, namely diversity visa applicants and refugees and asylees, come from different places than they did in 2003. Further, changes in sending countries and changes in the costs and incentives for migration may lead to different types of immigrants from within sending countries. However, the NIS represent the latest and best data available for studying differences in outcomes by class of entry, and so shed at some much-needed light on an otherwise murky topic. Second, this study is subject to various forms of selection bias. The wave two NIS response rate was low, and the nonresponse weights may not have been able to fully account for differences in response rates by those with different labor-market outcomes. Even more, self-employment and occupational skill levels are only observed for those working, leading to upward bias in both outcomes. In particular, longitudinal models showing within-person change in occupational skill level between wave one and wave two represent only the selected sample of survey respondents who were employed in both survey waves. Without an ideal instrument that predicts employment but is not correlated with other labor-market outcomes, it is difficult to overcome this selection. Nevertheless, these findings, interpreted carefully, present the most up-to-date look available at how labor-market trajectories vary among groups of new legal immigrants. ### Policy Implications If the country’s goal is to bring in only immigrants with greatest labor-market success, these findings suggest that employer-sponsored migrants do have the greatest success in the first few years after obtaining permanent resident status, showing the highest employment rates and the highest-skilled jobs. On the other hand, nearly all LPRs have higher employment rates than U.S. workers. And refugees and asylees show highest rates of self-employment and employing others, relative to other LPRs. LPRs may be strongly contributing to the country’s economy, even if through lower-skilled work. Decisions about immigration policies should consider how immigrants across the skill spectrum affect U.S. economic growth, and which immigrants best complement U.S. workers. That the occupational skill levels of family, diversity, and humanitarian migrants remain lower than those of employer-sponsored migrants, even when controlling for differences in education and English proficiency, also suggests that many LPRs may be facing durable barriers to their occupational success, such as difficulties gaining recognition of their foreign degrees or credentials, ongoing lack of savvy about how to navigate U.S. job markets, or discrimination. Research showing high rates of brain waste among highly educated immigrants suggests that immigrant workers might benefit from greater assistance in finding employment commensurate with their skills and education or in getting their foreign credentials recognized. Such efforts may help the country derive the greatest benefit from immigrant workers already here. On the other hand, it could be that the education and skills that family-sponsored, diversity, and humanitarian migrants bring are not a good match with U.S. labor-market needs, and that employer sponsorship is key to identifying immigrants likely to have high labor-market success. Finally, policymakers should consider broader objectives of immigration policy. Most immigrants operate as parts of family units, not just as individuals. Family unity has long been a goal of U.S. immigration policy. But family unity may also help support immigrants’ labor-market contributions. As shown by high rates of spouses, parents, and siblings reporting that they do not work because they are homemakers, family members may migrate to the United States to fill vital childcare and housekeeping duties, and thus support the high employment rates of other immigrants. A survey of sponsors of family-sponsored immigrants in Canada revealed that 40 percent of sponsors of spouses or partners and 48 percent of sponsors of parents or grandparents said having their relative in Canada helped them work more hours (Citizenship and Immigration Canada 2014). Similarly, the ability to bring their parents to live with them in the United States may allow some LPRs and naturalized citizens to continue living and contributing to the United States, rather than moving back to their countries of origin to care for aging parents (Treas and Mazumdar 2004). Finally, U.S. refugee and asylum policies were established to support human rights and foreign-policy goals, rather than in consideration of such entrants’ economic impacts on the country. ## Appendix View this table: [Table A1.](http://www.rsfjournal.org/content/6/3/70/T7) **Table A1.** Job Zone Classification Scheme ## FOOTNOTES * 1. I attempted to also control for prior legal status in the United States for those who were adjusting within the United States to green cards. However, the measures of prior status have substantial incomplete information, limiting their utility. After imputing missing information, and adding a variable for prior legal status into the regression models, the results did not change, so I removed this control variable from the models. * 2. Given that the largest shares of employment-based immigrants are from India, Europe, Canada, and Oceania, the Philippines, and “other Asia”—many of which are places where English is commonly taught in school, it is a bit surprising that the LEP share is as high as it is for employment-based immigrants. But 31 percent of employment-based immigrants from India, 28 percent from Europe, Canada, and Oceania, 43 percent from the Philippines, and 69 percent from “other Asia” report being LEP, along with higher shares from Mexico, other Latin American countries, and China. * © 2020 Russell Sage Foundation. Gelatt, Julia. 2020. “Do Employer-Sponsored Immigrants Fare Better in Labor Markets Than Family-Sponsored Immigrants?” *RSF: The Russell Sage Foundation Journal of the Social Sciences* 6(3): 70–93. DOI: 10.7758/RSF.2020.6.3.04. The author thanks participants at the Russell Sage Foundation conference The Legal Landscape of U.S. Immigration in the Twenty-First Century, Sarah Bohn, and three anonymous reviewers for comments on earlier versions of this article. Direct correspondence to: Julia Gelatt at jgelatt{at}migrationpolicy.org, Migration Policy Institute, 1400 16th St., NW, Suite 300, Washington, D.C. 20036, United States. Open Access Policy: *RSF: The Russell Sage Foundation Journal of the Social Sciences* is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. ## REFERENCES 1. Batalova, Jeanne, and Michael Fix. 2008. *Uneven Progress: The Employment Pathways of Skilled Immigrants in the United States*. With Peter A. Creticos. Washington, D.C.: Migration Policy Institute. Accessed August 7, 2020. [https://www.migrationpolicy.org/research/uneven-progress-employment-pathways-skilled-immigrants-united-states](https://www.migrationpolicy.org/research/uneven-progress-employment-pathways-skilled-immigrants-united-states). 2. Batalova, Jeanne, and Michael Fix. 2017. “New Brain Gain: Rising Human Capital Among Recent Immigrants to the United States.” Washington, D.C.: Migration Policy Institute. Accessed August 7, 2020. [https://www.migrationpolicy.org/research/new-brain-gain-rising-human-capital-among-recent-immigrants-united-states](https://www.migrationpolicy.org/research/new-brain-gain-rising-human-capital-among-recent-immigrants-united-states). 3. Batalova, Jeanne, Michael Fix, and James D. Bachmeier. 2016. “Untapped Talent: The Costs of Brain Waste Among Highly Skilled Immigrants in the United States.” Washington, D.C.: Migration Policy Institute. 4. Citizenship and Immigration Canada. 2014. *Evaluation of the Family Reunification Program*. Ottawa: Government of Canada. Accessed May 13, 2020. [https://www.canada.ca/content/dam/ircc/migration/ircc/english/pdf/pub/e4-2013-frp.pdf](https://www.canada.ca/content/dam/ircc/migration/ircc/english/pdf/pub/e4-2013-frp.pdf). 5. Duleep, Harriet Orcutt, and Mark C. Regets. 1996. “Admission Criteria and Immigrant Earnings Profiles.” International Migration Review 30(2): 571–90. DOI: 10.2307/2547394. [CrossRef](http://www.rsfjournal.org/lookup/external-ref?access_num=10.2307/2547394&link_type=DOI) 6. Jasso, Guillermina, Douglas S. Massey, Mark R. Rosenzweig, and James P. Smith. 2006. “The New Immigrant Survey 2003 Round 1 (NIS-2003-1) Public Release Data.” March 2006. Accessed May 20, 2020. Funded by NIH HD33843, NSF, USCIS, ASPE & Pew. [http://nis.princeton.edu](http://nis.princeton.edu). 7. Jasso, Guillermina, Douglas S. Massey, Mark R. Rosenzweig, and James P. Smith. 2014. “The New Immigrant Survey 2003 Round 2 (NIS-2003-2) Public Release Data.” Accessed May 20, 2020. Funded by NIH HD33843, NSF, USCIS, ASPE & Pew. [http://nis.princeton.edu](http://nis.princeton.edu). 8. Jasso, Guillermina, and Mark R. Rosenzweig. 1995. “Do Immigrants Screened for Skills Do Better Than Family Reunification Immigrants?” International Migration Review 29(1): 85–111. DOI: 10.2307/2546998. [CrossRef](http://www.rsfjournal.org/lookup/external-ref?access_num=10.2307/2546998&link_type=DOI) 9. Massey, Douglas S. 2011. “The New Immigrant Survey and Research on American Stratification.” Social Science Research 40(5): 1287–91. DOI: 10.1016/j.ssresearch.2011.06.001. [CrossRef](http://www.rsfjournal.org/lookup/external-ref?access_num=10.1016/j.ssresearch.2011.06.001&link_type=DOI) 10. Massey, Douglas S., Guillermina Jasso, and Monica Espinoza. 2017. “Weighting for Nonresponse on Round Two of the New Immigrant Survey.” Princeton, N.J.: Princeton University. 11. National Academies of Sciences, Engineering, and Medicine. 2017. *The Economic and Fiscal Consequences of Immigration*. Washington, D.C.: National Academies Press. DOI: 10.17226/23550. [CrossRef](http://www.rsfjournal.org/lookup/external-ref?access_num=10.17226/23550&link_type=DOI) 12. Office of Immigration Statistics (OIS). 2004. *Yearbook of Immigration Statistics 2003*. Washington: U.S. Department of Homeland Security. 13. Office of Immigration Statistics (OIS). 2019. *Yearbook of Immigration Statistics 2018*. Washington: U.S. Department of Homeland Security. Accessed May 20, 2020. [https://www.dhs.gov/immigration-statistics/yearbook/2018](https://www.dhs.gov/immigration-statistics/yearbook/2018). 14. Schwabish, Jonathan A. 2009. “Identifying Rates of Emigration in the United States Using Administrative Earnings Records.” *CBO* working paper no 2009-01. Washington, D.C.: Congressional Budget Office. Accessed May 20, 2020. [https://www.cbo.gov/sites/default/files/111th-congress-2009-2010/workingpaper/2009-01\_1.pdf](https://www.cbo.gov/sites/default/files/111th-congress-2009-2010/workingpaper/2009-01_1.pdf). 15. Stoops, Nicole. 2004. “Educational Attainment in the United States: 2003.” Suitland, Md.: U.S. Census Bureau. [https://www.census.gov/content/dam/Census/library/publications/2004/demo/p20-550.pdf](https://www.census.gov/content/dam/Census/library/publications/2004/demo/p20-550.pdf). 16. Treas, Judith, and Shampa Mazumdar. 2004. “Kinkeeping and Caregiving: Contributions of Older People in Immigrant Families.” Journal of Comparative Family Studies 35(1): 105–22. 17. U.S. Department of Labor, Employment and Training Administration. 2020. “Job Zone Reference.” O-NET Resource Center. Accessed May 20, 2020, [https://www.onetcenter.org/dictionary/21.3/excel/job\_zone\_reference.html](https://www.onetcenter.org/dictionary/21.3/excel/job_zone_reference.html). 18. Vincenza Desiderio, Maria, and Kate Hooper. 2016. “The Canadian Expression of Interest System: A Model to Manage Skilled Migration to the European Union?” Washington, D.C.: Migration Policy Institute Europe. Accessed May 13, 2020. [https://www.migrationpolicy.org/research/canadian-expression-interest-system-model-manage-skilled-migration-european-union](https://www.migrationpolicy.org/research/canadian-expression-interest-system-model-manage-skilled-migration-european-union).