Date: Tue, 21 Jul 2009 23:11:02 -0700 From: Norm Matloff To: Norm Matloff Subject: analysis of the J. Hunt paper To: H-1B/L-1/offshoring e-newsletter A couple of working papers on the H-1B work visa, and related visas, by Professor Jennifer Hunt of McGill University got quite a lot of press when they were released earlier this year. Here I will review the second one, since it is broader in scope, and will cover the first one later. The paper, accessible at http://www.nber.org/papers/w14920 is titled, "Which Immigrants Are Most Innovative and Entrepreneurial? Distinctions by Entry Visa." There is quite a bit of interesting material in this paper, and I must say at the outset that it is of high quality. It is one of the most careful in its approach among works on H-1B that I've seen. Here I use the word "careful" in the sense that Prof. Hunt adjusts for many different covariates, a statistical term meaning variables that may jointly impact the variable of interest. In the H-1B context, it means that a good study that, for example, compares H-1B wages to those earned by U.S. citizens and permanent residents, must account for many other variables besides citizenship, such as age, education, geographic region, and so on. See Section V.B of my University of Michigan law journal paper for details (On the Need for Reform of the H-1B Nonimmigrant Work Visa in Computer-Related Occupations, N. Matloff, University of Michigan Journal of Law Reform, Fall 2003, 36, 4, pp. 815-914, available online at http://heather.cs.ucdavis.edu/MichJLawReform.pdf). Again, in my view Hunt has done excellent work here (discussed below), but I do have some criticisms to make. As with my recent review of Bill Kerr's paper http://heather.cs.ucdavis.edu/Archive/KerrLincolnRejoinder.txt), I must take issue with Prof. Hunt's Introduction section (emphasis added): # ...economists have an as yet incomplete picture of the aggregate # BENEFITS TO NATIVES OF SKILLED IMMIGRATION. In this paper, I address # this by providing evidence not merely on skilled immigrants' private # productivity, as measured by their wage, but also on their success in # creating, disseminating and commercializing knowledge, activities with # PUBLIC BENEFITS LIKELY TO INCREASE U.S. TOTAL FACTOR PRODUCTIVITY... # ...[The subanalyses here in different visa categories make] the # results DIRECTLY INFORMATIVE TO POLICY­MAKERS, who can use them to # influence their decisions about which visa classes to expand or # shrink and which transitions to legal permanent residence to # facilitate. While this is carefully stated, including with a disclaimer I'll address below, I believe many readers will get the wrong impression, as I will explain. Hunt continues, # To the extent that the activities I study have a PUBLIC GOOD COMPONENT, # skilled immigrants might CONTRIBUTE TO NATIVE WELFARE simply BY # INCREASING THE SIZE OF THE POPULATION LIKELY TO ENGAGE IN THEM. That phrase of Prof. Hunt's, "increasing the size of the [skilled] population," is her key point--and is my own key point, in the sense that her statement is based on a false premise. She is saying that the nation benefits because the more skilled people we have, the better. But again as with the Kerr/Lincoln paper, this ignores the displacement effects. Employers hire the cheaper H-1Bs instead of Americans, causing many Americans to leave the tech field. And the H-1B program keeps wages down and makes careers short-lived, causing many young Americans to avoid going into the tech field, or to avoid pursuing graduate study in it, in the first place. (Recall from my analysis of the Kerr/Lincoln paper that there are also crucial short-term/long-term issues here.) In other words, H-1B causes an INTERNAL BRAIN DRAIN in the U.S. Thus we are actually not taking about "skilled immigrants [who] might contribute to native welfare simply by increasing the size of the population likely to engage in them." On the contrary, H-1B and L-1 (and a fast-track green card program, if it is enacted) cause a DECREASE in the size of the American component of that population Hunt is discussing. And the net number of skilled workers in the given field could either increase or decrease, rather than Hunt's implication that there is a clear increase. (A note on Hunt's word "native": As with many researchers, not to mention the press and the industry lobbyists, Hunt oversimplifies the situation by portraying natives as the only group impacted. Immigrants are impacted too, and gradually become just as displaceable by H-1Bs as natives.) To be sure, Hunt does make a disclaimer, # However, I do not undertake a full cost­benefit analysis of skilled # immigration's impact, as I do not capture all possible benefits of # skilled immigration and I ignore potential negative effects. Borjas # (2006b), for example, calculates that immigrants with PhDs depress the # wages of native PhDs. But this is much too weakly stated. As I've often mentioned, when the National Science Foundation called in 1989 for an increase in foreign students and an H-1B program to employ them after graduation, with the express goal of holding down PhD salaries, it projected that the American students would choose not to pursue doctoral study, rather than merely put up with depressed wages. In other words, the issue is displacement, not just wage reduction. Now, what about Hunt's findings? One could somewhat flippantly and simplistically, but pretty much correctly, summarize them as follows: (a) H-1B engineers with U.S. degrees make more money than bank tellers. (b) H-1B engineers make less money than American engineers. Item (a) doesn't say anything of interest. Engineers, even underpaid ones, do make more than bank tellers. Item (b) is key. Hunt correctly points out that much of the Great H-1B Debate involves the question of whether H-1Bs are paid less than Americans. She cites analyses with findings like (b) by John Miano and me (though she incorrectly cites my CIS article for this). (And there are also studies by Paul Ong at UCLA, the NRC, the GAO etc. showing it.) She also notes that the Mithas and Lucas study found that H-1Bs are paid more than Americans (see my critique of the Mithas/Lucas work at http://heather.cs.ucdavis.edu/Archive/MithasLucas.txt). Hunt's point is that IF ONE CORRECTS FOR ALL THE COVARIATES, ONE DOES FIND (b), though one also finds some interesting exceptions in certain visa subpopulations. And indeed, Hunt does account for covariates in analyzing all these criteria, and in almost all cases, the "immigrant advantage" (her wording) over the natives goes away, and in fact her findings jibe with what we critics of the H-1B program have been saying. Specifically Hunt finds, inter alia, that: (1) Controlling for field, education and age, immigrants make 6.9% less than natives. (2) Controlling for those same variables in (1), plus region and other variables, there is no statistical difference in wages between natives and immigrants who originally came to the U.S. as H-1Bs. This is not surprising, as the gap between former-H-1B immigrants and natives quickly disappears after they get green cards and are thus not exploitable. Note, though, my point about interaction effects below. (3) Controlling for the variables in (2), immigrants who originally came to the U.S. as foreign graduate students (or post-docs) make 13.2% less than natives. This is a very major point. It is in stark contrast to the current thinking in DC, fed by the industry lobbyists, that the H-1Bs hired from U.S. campuses are the "good" H-1Bs. That claim is prominent in virtually every press kit, op-ed and so on written by Compete America, the American Electronics Association, the American Immigration Lawyers Association, etc. This notion led to Congress establishing an extra 20,000-visa subcategory in the yearly H-1B cap in 2000. (4) She finds (Fig. 3, third graph) that immigrants from Europe with prior work experience who first came to the U.S. and have been in the U.S. a long time make a lot more than natives. The European facet is consistent with my CIS paper, in which I found that the only nationalities of foreign workers who have a tendency to be "the best and the brightest" are Europeans. There are of course some outstanding workers from Asia too, but the industry lobbyists' claim that the fact that most H-1Bs are Asian is due to superior educational systems in Asia does not jibe with the data. The "long-time U.S." aspect also is plausible. Prior to the implementation of H-1B in 1991, the old H-1 visa was officially for bringing in "the best and the brightest," with the formal title Aliens of Distinguished Merit and Ability. By 1991, it often wasn't working that way, but Hunt does seem to be picking up a lot of cases where it did work as intended. See my University of Michigan paper for the history behind H-1B. The prior experience variable is important too. Though Hunt's variable here was binary, for having any work experience at all, she's picking up a lot of workers that had considerable work experience before coming here. Remember, the key point about most H-1Bs is that they are young, thus cheap. They are also cheaper than young Americans, but as I've shown, age is key; the most important attraction of H-1Bs to employers is that it allows them to hire the young, cheap H-1Bs instead of the older, more expensive, Americans. Thus H-1Bs who have significant work experience really are brought here for their special talents, not for cheap labor, thus making Hunt's finding plausible. (5) Concerning patenting, Hunt says: # After I control for field of study, in the middle graph, and # education, in the bottom graph, both main work visa groups # and student/trainee visa holders have statistically # significantly lower patenting probabilities than natives, and # the three groups are not statistically significantly # different from each other... By "main work visa group" she means the H-1Bs who originally came to the U.S. as H-1Bs, as opposed to the H-1Bs who originally came to the U.S. as foreign students, whom she calls the "student/trainee visa holders." Again, these findings are radically different from the claims being bandied about in DC by the industry lobbyists, along the lines that the H-1Bs, particularly the ones hired from U.S. universities, are especially innovative and patent-generating. Again, this is a major point. (6) Concerning publishing: Hunt finds in her various groups that the immigrants have more research publications. This jibes with my observations. Among my own department colleagues, for instance, almost all the immigrants have longer publication records than natives of the same experience level. I believe they probably have more grant funding as well. There definitely is a tendency to be more agressive in building up a CV. Of course, one should not equate numbers with quality. (7) Concerning entrepreneurship: Hunt finds, "...controlling for education in the regression behind the bottom graph increases the marginal effect of immigrants who arrived as graduate students considerably, to a statistically significant 1.2 percentage point advantage over natives." Unfortunately, this does not adjust for region. In the computer industry, for example, access to the venture capital "mecca" of Sand Hill Rd. Palo Alto is crucial, and of course spinoffs are much more likely in regions where there are lots of companies to spin off from. And since the immigrants disproportionately settle in large, urban regions where there is access to VCs, this may be another confounding factor. On the other hand, in private e-mail Hunt has raised the question of whether the more entrepreneurial immigrants tend to settle in such regions. All this makes the entrepreneurship impact unclear. Furthermore, there is the question of the nature of the business. Many of the Indian firms are "bodyshops," i.e. rent-a-programmer companies, either hiring H-1Bs in the U.S. or shipping the work overseas. These do not add to the technological state of the U.S. The first Saxenian paper found that "36% of the Chinese-owned firms are in the business of Computer Wholesaling," meaning that they are simply assemblers of commodity PCs, with no engineering or programming work being done. So the type of business matters. I would add two further comments on methodology: First, note that there are other covariates that could be important. For example, over the years foreign graduate students have tended to choose the more "marketable" research areas, to maximize their chances of getting a U.S. job and green card after graduation. The domestic students tend to choose research areas out of interest, which may result in lower salaries later on. Accounting for this could accentuate some of the differences Hunt found. Data on this would be difficult or impossible to obtain, though. Second, my main criticism of Hunt's statistical analysis is that she seems to have done very little with interaction terms. I believe, for instance, that a lot of the immigrant/native differences she found would vary widely from one field to another, possibly even changing sign. She does briefly mention interaction terms, but apparently did not pursue it much. This is a major point. Overall, this is an excellent paper, in my opinion, again for the care the author has taken in accounting for many covariates. Subject to my concern about interaction terms, it explodes the major myths prevalent in DC these days, confirming the analyses of us critics. [Yes, I know, some readers are now saying, "No wonder he likes the paper." :-) ] Its two major flaws are again the interaction issue, and the lack of analysis of the impact of displacement of U.S. citizens and permanent residents by H-1Bs and foreign students. Granted, such an analysis is difficult, but I believe the author should have at least included a discussion of the possible impacts of the displacement issue, and put in strong disclaimers in the Abstract and Introduction. Norm