Response to the American Immigration Council on the Value of Foreign Degrees

By Jason Richwine on March 12, 2019

The Center recently published my report (and hosted an event) showing that immigrants with foreign degrees are less skilled than U.S. degree holders. The figure below summarizes the results.

Graph: Percentile Scores by Test-Takers with College or Advanced Degrees

For example, among those with at least a college degree, native-born Americans score at the 74th percentile on the PIAAC test of literacy skills, while U.S.-educated immigrants score at the 66th percentile, and foreign-educated immigrants score at just the 42nd percentile. The numeracy and computer operations PIAAC tests show a similar pattern. "Policy-makers should therefore be cautious in treating foreign degrees as evidence of 'high-skill' immigration," I wrote.

Remarkably, Walter Ewing of the American Immigration Council (an arm of the American Immigration Lawyers Association) calls this conclusion "far beyond" what the data can support. He makes two arguments. First, he says the PIAAC covers only basic skills rather than the specialized knowledge possessed by foreign-educated immigrants. Second, he claims that the high demand for H-1B workers proves their skills are genuine.

The first argument is curious. If the PIAAC covers only basic skills, shouldn't the test be a cinch for immigrants with advanced skills? When a person who purports to be, say, an engineer cannot demonstrate basic numeracy, common sense suggests the person is exaggerating his or her qualifications. In any case, although the PIAAC does not measure specialized knowledge, it is far from easy. As the National Center for Education Statistics puts it, the PIAAC covers "a broad range of abilities, from simple reading to complex problem solving skills". Less than 1 percent of Americans managed to score at the highest of the six skill levels on the literacy and numeracy tests. Furthermore, PIAAC scores are strong and robust predictors of income. Even after accounting for educational attainment, "soft" skills, and demographic factors, PIAAC scores retain significant predictive value. Therefore, the PIAAC scores of foreign-educated immigrants are clearly relevant to their economic outlook here in the United States.

As for Ewing's second argument, which cites the desirability of H-1B workers, my report cannot distinguish H-1B workers from other types of "high-skill" immigrants. The report presents PIAAC scores for immigrants with college degrees regardless of whether they arrived on H visas, L visas, O visas, spousal visas, or any other pathway. How the skills of immigrants vary across those different admissions categories is an interesting question, but not one that can be answered with the PIAAC data.

That said, outside evidence suggests that the skills of H-1B workers are not especially impressive. In fact, critics suspect that H-1B is desirable to employers not because it allows them to hire Einsteins, but because they can use the program to replace Americans with cheaper workers — all under the guise of "high-skill" immigration. Ewing claims that H-1B workers earn more money than U.S. natives in similar positions, but because that claim requires combining income data from different (and possibly incompatible) sources, it should be viewed skeptically. Certainly it is not consistent with the most notorious uses of H-1B, such as Disney requiring its U.S. workers to train their foreign replacements.

Part of the problem with quantifying the wage and employment impacts of H-1B is that there are so many different variables to disentangle. To get around that problem, the best study I've seen takes advantage of the lottery that ensues when the number of H-1B applications exceeds the number of available visas. The study's authors were able to isolate the impact of H-1B workers by comparing companies that applied for an H-1B visa and won the lottery with companies that applied and lost. I summarized the results in a piece for National Review:

If H-1B workers are truly exceptional talents for whom there are few American substitutes, then we would expect to see that lottery-winning companies increase their employment relative to the lottery losers by roughly the number of H-1B workers they receive. (In fact, if the more grandiose claims about immigrant productivity are true, then we would expect the lottery-winning companies to generate even more jobs beyond the ones that go to the H-1Bs.) Instead, both lottery-winning and lottery-losing firms ended up with employment levels that were statistically indistinguishable. Presumably, the losing companies just went ahead and hired someone who was not an H-1B.

Although the evidence is less robust, it also appears that lottery-winning companies lowered median wages and increased their profits relative to the lottery losers. So the lottery data are more consistent with the view that firms use the H-1B program to hire cheaper substitute workers from abroad, not to bring in the next Einstein.

Therefore, the disappointing performance by "high-skill" immigrants in general on the PIAAC tests might apply to the performance of H-1B workers specifically. However, there is no way to know for sure without testing a sufficiently large sample of immigrants whom we know to be H-1B workers. In fact, why not do this as a matter of policy? If H-1B brings in as many brilliant minds as Ewing implies, let's all agree to require that evidence, in the form of standardized test scores, from every future H-1B applicant. I look forward to the American Immigration Council joining me in this proposal.