The U-Shaped Career Pattern
This series examines what happens after immigrant undergraduates attend college in the United States: which employers they enter, where those employers sit in the firm-size distribution, and how those destinations compare with the paths of native students. Its central question is not whether education makes immigrants “assimilate” into one common labor market. It is why U.S.-educated immigrants are disproportionately found at two seemingly opposite destinations—very large firms and multinational enterprises on one side, and small firms or self-employment on the other. Understanding that split gives us a more useful way to judge opportunity, skills, institutions, and networks than a simple story of moving from outsider to insider.
The distinction matters because firm size is not merely a label on an employer. It changes the resources available to a worker, the kinds of jobs a firm can offer, the reach of its markets, and the infrastructure it can devote to hiring. A multinational enterprise is not just a large company: it is an organization with operations that cross national borders. A small business is not automatically a weak destination, and self-employment is not automatically evidence of independence or exclusion. The same outcome can arise from different combinations of skills, constraints, local opportunity, and social connections.
The people behind the comparison
The study follows a defined group rather than making a claim about immigrants in general. It identifies people aged 19 to 25 who were attending undergraduate college in the 2000 Decennial Census, then traces their paid employment through the Longitudinal Employer-Household Dynamics data and their self-employment through linked business records from 2001 to 2020. The comparison is with native citizens who were also college students at that time.
“Immigrant students” is an analytical category here. The researchers divide them into citizen immigrants, non-citizen immigrants who arrived in the United States by age 16, and non-citizen immigrants who arrived at age 17 or later. The last group often contains people whose migration and college decisions were closely connected. Citizenship status is observed in 2000 and held fixed for the analysis, even though some non-citizens may later naturalize. The paper uses “international students” more narrowly when discussing student visas; that population overlaps with non-citizen students but is not identical to it.
This scope changes the question. These students have U.S. educational credentials and exposure to American institutions, and many are fluent in English. They are therefore not a test of whether foreign credentials are recognized or whether a newly arrived worker can navigate an unfamiliar school system. Yet some may still need work authorization, may have selected particular fields of study, and may be concentrated in places where globally connected employers are plentiful. Their careers allow us to see how immigrant sorting persists even after a major part of the education-to-work transition has taken place inside the United States.
The data also impose important boundaries. The main employer analysis covers 25 states, about half the country, with detailed employment histories. It includes California and Pennsylvania but not Florida, New York, or Texas. The linked sample emphasizes people with a sufficiently visible connection to covered employment: the main criteria require at least two quarters of earnings, a minimum average quarterly earnings level, and a primary establishment accounting for at least half of annual earnings. These choices make employer placement measurable, but they do not make the records a complete account of every graduate’s life.
A current national statistic illustrates why that distinction matters. The Bureau of Labor Statistics reported that foreign-born workers made up 19.1% of the U.S. civilian labor force in 2025, with a 4.2% unemployment rate compared with 4.3% for native-born workers. In the full foreign-born population, median weekly earnings were lower than for native-born workers. Among workers with a bachelor’s degree or higher, however, the foreign-born median was slightly higher: $1,762 compared with $1,735. Those figures describe a much broader population, and the earnings measure covers full-time wage-and-salary workers while excluding the self-employed. They provide context, not a substitute for the college cohort studied here.
A career distribution with two peaks
The striking result is not that immigrants simply move into larger firms or simply remain concentrated in small businesses. Their observed careers show a U-shaped distribution. In the descriptive 2001–20 sample, immigrant students spent 18.2% of their working quarters in firms with 1 to 20 employees, compared with 15.0% for native students. At the other end, they spent 29.0% of quarters in firms with more than 10,000 employees, compared with 26.3% for natives. In the middle categories, immigrants were less represented: 19.7% versus 22.6% in firms with 101 to 1,000 employees, and 19.6% versus 22.1% in firms with 1,001 to 10,000 employees.
The pattern becomes clearer when the researchers look at other employer traits. Immigrant students spent 24.0% of observed quarters in multinational enterprises, compared with 18.9% for natives, and 9.3% in high-tech firms, compared with 4.9%. Their employment was also more concentrated in high-skill services and less concentrated in health and education. These dimensions overlap, but they are not interchangeable. A firm can be large without being multinational, a high-tech firm can be small, and a self-employed person is not being counted as an employee of a large enterprise at all.
The adjusted career estimates preserve the same shape. Depending on the immigrant group, students spent about 2.4 to 4.5 percentage points fewer working quarters in firms with 101 to 10,000 employees and 3.4 to 6.0 percentage points more quarters in multinational enterprises. The shift toward the smallest firms was strongest among non-citizens who arrived at age 17 or later. That pattern is consistent with a later-arriving student facing a different mix of visa constraints, networks, field choices, and entrepreneurial opportunities than someone who arrived as a child or had already become a citizen.
The middle is therefore analytically important. If the result were only that immigrants were more likely to work for large employers, the explanation could focus on technical skills, urban labor markets, or sponsorship. If it were only that they were more likely to be self-employed, the focus could be ethnic networks or barriers in mainstream firms. The relative absence from the middle suggests that different mechanisms are operating at different scales. It also warns against using a single career ladder as the implicit standard for everyone.
This does not mean every immigrant student experiences a two-track career, or that either endpoint is inherently better. Large firms may offer scale, training, and higher pay but can also impose bureaucratic constraints. Small firms and self-employment can provide autonomy and access to a community market but can also carry volatility and limited institutional support. The study observes where people work; it does not directly observe which destination they preferred, what alternatives they had, or whether a move between firm sizes later improved their options.
The central puzzle can be represented as four channels converging on the same employment pattern:
Four channels, one interpretive problem
The first channel is geography. Immigrant college students were more concentrated than natives in major metropolitan areas such as Los Angeles, New York City, and San Francisco, and less likely to study outside a metropolitan area. That matters because students often search for their first jobs near where they study, while large firms and multinational enterprises also cluster in dense, globally connected labor markets. A student’s location can therefore shape the menu of employers before the student makes an individual application decision.
The second channel is degree specialization. Immigrant students were more likely to study engineering and computer science and less likely to study fields such as education, fine arts, psychology, and communications. The study finds that this degree-field alignment is the strongest predictor of differences in placement across immigrant groups, particularly in larger firms and MNEs. Technical and quantitative capabilities are portable across places and are often demanded by organizations with R&D, high-tech, or cross-border operations. The relationship is much weaker in the small-firm split, which is why degrees alone cannot explain the U-shaped result.
The third channel belongs to employers rather than students. A graduate may have relevant skills but still need a firm that can support work authorization. Sponsorship requires administrative capacity, and uncertainty in programs such as the H-1B system can make scale and experience valuable. The paper’s proxy for firm capability is positive but modest once geography, degree specialization, and other factors are included. It helps explain why some matches are feasible; it does not explain most of the cross-group pattern, and the data do not show each student’s legal pathway or visa history.
The fourth channel is ethnic self-employment specialization. Group-specific entrepreneurial histories and co-ethnic networks can provide information, workers, customers, trust, or knowledge of a market. In the study, this mechanism has its strongest explanatory role among firms with 20 or fewer employees, where geography and degree specialization do not show a comparable role. That finding does not tell us whether any particular student entered a small firm by choice, because of family expectations, through network advantages, after encountering discrimination, or because visa and labor-market barriers narrowed the alternatives. Agency and constraint can coexist.
The earnings evidence sharpens the interpretive problem. Immigrant students who remain observable in U.S. employment records earn more on average than native peers in the study’s linked sample. During 2016–20, average annualized earnings in real 2000 dollars were $65,270 for immigrants and $55,620 for natives. The paper reports an adjusted earnings advantage of about 11% for citizen immigrants and higher self-employment earnings for immigrants in the relevant comparison. Those results are consistent with productive matching: the two endpoints may reflect different ways of using technical skills, local opportunity, and entrepreneurial networks.
But “consistent with” is not “proves.” People who leave the United States, never appear in covered employment, work outside the data, remain out of the labor force, or cannot be linked reliably are less visible in the long-run outcome. The study’s earnings results therefore describe immigrants who remain observable in U.S. work, not all people who once attended an American college. Nor do higher average earnings erase the possibility of unequal access, blocked choices, or discrimination within particular pathways.
The useful starting point for the rest of this series is thus a conditional one. U.S. education does not place immigrant students into a single mainstream destination. It interacts with where they study, what capabilities they acquire, which employers can make hiring feasible, and which communities can support entrepreneurship. The next question is how the first of those channels—place—turns a college location into a structured set of career possibilities.