In previous posts (here and here), I argued that we should care about high-skill immigration because of the possibility that high-skill immigrants import knowledge and capabilities that “rub off” on the rest of us, thereby increasing our productivity. I also showed that there is evidence for such spillovers in the “experimental” context (such as the sudden dismissal of renowned Jewish scientists by the Nazi regime), but the evidence is restricted to cases where the immigrants have exceptionally high skills, where there is close personal contact between the exceptional immigrants and the native workers, and where the number of high-skill immigrants is sufficiently small relative to the market.
Obviously, such flows of truly exceptional immigrants are rare. The political argument for high-skill immigration is instead presented in the context of something like the H-1B visa program, which allows employers to import 65,000 (mostly) high-tech workers annually. In the H-1B context, “high skills” typically mean that the immigrant has at least a college degree.
William Kerr and William Lincoln have a 2010 JOLE paper that carefully examines the non-experimental data. They look at what happens to patenting after Congress increases the number of H-1B visas, and they show that there is indeed additional patenting in those cities where the H-1B recipients were likely to settle (such as Silicon Valley). But their empirical analysis suggests that much of the additional patenting was due to the fact that the H-1B visa-holders themselves were patenting, rather than a spillover that made native workers more creative. To ensure there is no misinterpretation of the Kerr-Lincoln paper, this is how the authors themselves describe their evidence in the paper’s abstract:
Higher H-1B admissions increase immigrant science and engineering employment and patenting by inventors with Indian and Chinese names in cities and firms dependent upon the program relative to their peers. Most specifications find limited effects for native employment or patenting.
A more recent 2015 JOLE paper by Giovanni Peri, Kevin Shih, and Chad Sparber concludes that Kerr and Lincoln didn’t get it right. The Peri-Shih-Sparber paper instead argues that productivity and high-skill wages rise substantially when a city is hit by a supply shock of H-1B visa-holders. This is how their introduction summarizes a key result in their paper:
Our preferred specifications reveal that a rise in foreign STEM growth by 1 percentage point of total employment increases wage growth of college-educated natives by 7–8 percentage points.
In other words, the elasticity of the wage of college graduates with respect to total employment is around +8. (Yes, Virginia, you read that right). Let me rephrase this claim. If the H-1B program were used to increase total employment by 10 percent, the wage of college graduates would rise by 80 percent. Let me rephrase it yet again: If we could grant 15 million more H-1B visas, the wage of college graduates would almost double. Any reader gullible enough to believe this should contact me ASAP: I have many bridges to sell.
The H-1B visa-holders end up in Boston and Silicon Valley, and not in Flint or Detroit, for a very good reason. They will almost surely be hired by firms in cities where the tech sector is already shifting the technology frontier, and there will be patenting and productivity growth in those cities regardless of whether the H-1Bs show up or not. Put simply, correlation is not causation.
Kirk Doran, Alexander Gelber, and Adam Isen revisit the whole H-1B question using an experimental approach. It turns out that some of the H-1B visas are randomly distributed to firms through a lottery. Doran, Gelber, and Isen then compare the firms that won the lottery to the firms that did not. Here is what they conclude:
We compare winning and losing firms in the Fiscal Year 2006 and 2007 lotteries for H-1B visas…Winning additional H-1B visas causes at most a moderate increase in firms’ overall employment, and these H-1Bs therefore substantially crowd out firms’ employment of other workers. Additional H-1Bs generally have insignificant and at most modest effects on firms’ patenting and use of the research and experimentation tax credit. There is some evidence that additional H-1Bs lead to lower average employee earnings and higher firm profits.
In short, the experimental data does not document any spillovers, and instead suggests that natives get paid less and are crowded out of firms that won the lottery. Who would have thought that the labor demand curve was downward sloping?
Finally, as I was preparing this post earlier in the week, John Bound sent me a new paper (coauthored with Gaurav Khanna and Nicolas Morales) entitled “Understanding the Economic Impact of the H-1B Program on the U.S.” This is one of the most thoughtful studies I have ever read on the H-1B program. The authors construct a (very technical) model of the economy that addresses the impact on workers, consumers, and firms, while avoiding the conceptual problems and empirical issues that plague most of the non-experimental work. This is how Bound, Khanna, and Morales summarize their findings:
[H-1B immigrants] increased the overall welfare of US natives, and had significant distributional consequences. In the absence of immigration, wages for US computer scientists would have been 2.6% to 5.1% higher and employment in computer science for US workers would have been 6.1% to 10.8% higher in 2001. On the other hand, complements in production benefited substantially from immigration…Firms in the IT sector also earned substantially higher profits.
The H-1B program inevitably created winners and losers, and the losers were the native high-tech workers. And I should emphasize that “native” really means pre-existing–as many of the pre-existing high-tech workers were probably born abroad.
So what is my take on all this? The evidence that the H-1B program produces sizable beneficial spillovers ranges from non-existent to weak. I can’t say I’m surprised. After all, the experimental studies strongly suggest that spillovers can only be detected in a very specific context: The high-skill workers have exceptional skills; they interact closely with the native workers who will benefit from the spillovers; and their number is sufficiently small relative to the market. It is hard to believe that the annual flow of 65,000 college graduates in the H-1B visa program meets those conditions.
And in case the academic evidence is not convincing enough, here and here and here is some real-world reporting from the New York Times.