How to make immigration assessments fairer

Immigrants to the U.S. with job offers often apply for work authorization. But immigrants from Latin America are less likely to have those requests granted than are immigrants from other regions, according to a new study conducted by scholars at MIT and Brown University — a study that also suggests a potential remedy for this problem, by finding that this regional disparity does not exist when officials examine cases in greater detail.

“This could have significant implications for both understanding inequality [in immigration], and also how to think about testing solutions that will mitigate some of these demographic biases,” says Emilio J. Castilla, an associate professor at the MIT Sloan School of Management, and co-author of a new paper detailing the findings.

The study shows that over a recent period of more than three years, the U.S. federal government approved about 91 percent of labor certification applications from Asian immigrants and 90 percent from Canadian immigrants, but just 67 percent of applications from Latin American immigrants. The study controls for variations in the offered salaries and job titles; characteristics of the firms making the offers; and the visa histories of the immigrants in question.

Ben A. Rissing PhD ’13, who is the corresponding author of the paper, conducted the work with Castilla. Rissing began collecting data for the study while at MIT; he is currently a visiting assistant professor at Brown.

Intriguingly, the researchers also found that the pattern of a higher rate of rejection for Latin American applicants is not present for instances in which the U.S. government conducted an “audit” of the labor certification application — that is, when agents evaluated the application using detailed supporting documentation. In those cases (about 13 percent of the total) there is no evident pattern of regional inconsistency in the work authorization decisions.

“We find that when the government audits applications and collects detailed supporting documentation for some of these individuals, it becomes a very fair process,” Rissing says. “We don’t see differences in that regard based on individuals’ countries of citizenship.”

A potential solution?

The paper, “House of Green Cards: Statistical or Preference-Based Inequality in the Employment of Foreign Nationals,” appears in the latest issue of the American Sociological Review.

To conduct the study, Castilla and Rissing examined roughly 198,000 labor-certification requests with the U.S. Department of Labor, filed from June 2008 through September 2011. The requests are one key initial stage in the granting of most employment-based green cards. Those cases involved immigrants from 190 countries and about 68,000 U.S. firms offering jobs to the immigrants. All cases are reviewed by federal officials working in the same location.

Immigrants from Asia represented about 66 percent of the roughly 198,000 applicants, and Latin America about 15 percent of all applicants, followed by Europe (8 percent), Canada (5 percent), the Middle East (3 percent), Africa (2 percent), and Oceania (less than 1 percent).

A majority of the applications involved cases where the immigrant had lined up a job in one of four industries: information technology, advanced manufacturing, education, and finance.

However, the pattern of geographic inequality extended to other occupations as well: The study finds that immigrants from Asia seeking jobs in the U.S. as cooks and construction workers are, respectively, 42 percent and 81 percent more likely to have their applications approved than immigrants from Latin America seeking jobs in those areas.

Under a 1990 federal law, Castilla notes, “No evaluation criteria should pertain to the country of origin” of immigrants. However, that demographic information is available on each application, although government officials are at least instructed not to use country-of-origin data while reviewing cases.

In the paper, Castilla and Rissing acknowledge they cannot rule out “unobserved heterogeneity in immigrants’ qualifications or application characteristics,” which might help explain the uneven distribution of application acceptances. The authors also emphasize that they would welcome further empirical studies of the subject.

The researchers believe a couple of potential remedies could lead to an application system that yields greater geographic parity.

“One potential solution to the kind of outcomes that we observe … would be if all of these applications were to be audited and thus evaluated with detailed employment information,” Rissing says. However, both he and Castilla acknowledge that this would likely add additional costs and administrative burdens for the federal government.

Alternately, the researchers note, simply masking the country-of-origin data during the review of each application might create a relatively inexpensive fix, addressing the current situation.

This article is published in collaboration with MIT News. Publication does not imply endorsement of views by the World Economic Forum.

To keep up with Forum:Agenda subscribe to our weekly newsletter.

Author: Peter Dizikes writes for the MIT News Office.