Date: Sun, 2 Sep 2007 00:25:57 -0700 From: Norm Matloff To: Norm Matloff To: H-1B/L-1/offshoring e-newsletter There have now been three reports by Vivek Wadhwa and various coauthors, concerning foreign workers and immigrants in the tech industry. (I don't mean to diminish the role of Wadhwa's distinguished coauthors, but Wadha has always been the lead author and almost always the one quoted in the press, so I will refer mainly to him, for convenience.) The latest, "Intellectual Property, the Immigration Backlog, and a Reverse BrainDrain: America's New Immigrant Entrepreneurs, Part III," was released last month. It can be downloaded at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1008366 In some cases I have praised Wadhwa for the fresh approach he takes to investigating these issues. For instance, in investigating the industry's labor shortage claims, he asked HR departments whether they have found it necessary to pay signing bonuses in order to hire engineers; a straightforward request for information would have gotten routed to the firm's PR people, with a politically calculated response. However, I have also been disappointed by a serious lack of depth and analytical rigor in much of his work, due to hastiness to produce reports. Specifically, I've been troubled by the failure of the Wadhwa and his coauthors to follow the standard route for academic research, which is submit their work to the scrutiny of anonymous peer review. This process takes a long time, and since Wadhwa and his coauthors apparently are anxious to have their work have an impact on the current congressional debate regarding foreign workers--certainly a legitimate goal--one can see why they have not sought peer review. Nevertheless, this lack severely weakens the value of their work. Also troubling is the fact that their later work does not address--or even mention--fundamental problems with their work that have been brought to their attention after the release of their previous studies. Unfortunately, the present study has only intensified my concerns. The present study presents two main findings: Finding 1: Immigrant participation in patent activity in the U.S. has increased dramatically in a very short period of time--more than tripling between 1998 and 2006. Finding 2: Due to its very long backlog in employer-sponsored green card applications, the U.S. is in danger of losing a huge number of world-class engineers and scientists, as they will tire of waiting and will simply return home (or immigrate to a third country). There are very serious flaws with both these findings. Let's start with Finding 2: As the report points out, for jobs in the tech field (this latter qualifier applies here and below), there are essentially three main tiers for employer-sponsored green cards. These tiers are known as EB-1, EB-2 and EB-3. The key point is that these tiers are ranked by level of talent, as codified in the law: * EB-1 is for "foreign nationals of "extraordinary ability" and for "outstanding professors" * EB-2 is for those who are either of "exceptional ability" or possess an "advanced degree" * EB-3 is for those with "bachelor's degrees" The major problem with the study concerning the backlog is that when it talks of the long waiting times, it fails to make clear that "the best and the brightest," i.e. EB-1, have to wait only a few months. In other words, the notion that we are losing outstanding talents due to the backlog is incorrect. We'll look at some numbers on this shortly, but first the point must be made that the green card wait is also dependent on the worker's nation of origin. The wait for those from China and India is longer than for other countries, so we'll look at them. (You can get all the data at http://travel.state.gov/visa/frvi/bulletin/bulletin_1360.html ) For example, currently (September 2007), any EB-1 who applied by January 1, 2007 can now get his/her green card right now, for both China and India. So, they waited only eight months. By contrast, for EB-3s, green cards are simply unavailable, period. A year ago, September 2006, the EB-1 category was designated Current, meaning there was no backlog at all, for both China and India. The waiting period for EB-3 was four years for China and five years for India. Two years ago, September 2005, EB-1 was also Current for China and India, but there was nothing available for EB-3s at all. The situation for the EB-2s is, not surprisingly, in between EB-1 and EB-3. Currently EB-2s from China have a wait of about a year and a half, and for Indians it's about three years. This certainly is not the dire wait of seven years that the study authors have told the press, such as in "Many Green Card Seekers Wait Seven Years or More: Researchers Say Backlog Could Spur `Reverse Brain Drain'", San Jose Mercury News, August 22, 2007. And remember EB-2 is not necessarily a "talent" category--anyone with a Master's degree qualifies, even if in an undistinguished degree at an undistinguished school. In other words, the study authors have been incorrect in claiming that we're on the verge of losing the geniuses because they've reached the end of their patience in waiting for their green cards. Yes, there are plenty of foreign workers who are indeed waiting seven years or more, and one can certainly sympathize with them, but they are the ordinary workers, not "the best and the brightest." The latter have no problem. Now, what about Finding 1, that foreign-born participation in patenting has skyrocketed in just eight years? Here's the exact statement: Foreign nationals residing in the United States were named as inventors or co-inventors in 25.6 percent of international patent applications filed from the United States in 2006. This represents an increase from 7.6 percent in 1998. These startling figures would be a red flag to most academic researchers. It is simply too large a jump in too short a time period. Unfortunately, the authors here took it at face value. As the authors and the industry lobbyists correctly point out, R&D in the tech field is typically done by workers with PhDs, or at least a Master's degree, and they typically obtain these degrees at U.S. universities. So, people with such degrees are the main source for patents. Yet there was no dramatic increase in the percentage of foreign students in these programs during the 1998-2006 time period in the study. For instance, in PhD programs in computer science in 1998, 46% of the students were foreign students; in 2006 the figure was 54% (Taulbee Study, CRA Newsletter, May 2007, www.cra.org). Those figures are large, yes (see explanation below), but the point is that the increase from 1998 to 2006 was mild, nothing like the jump from 7.6% to 25.6% in patents that the authors found during the same time period. Any academic researcher's first suspicion in such a situation would be that the problem is that there was some dramatic change in the way the data was collected, i.e. that there was a different system in place in 1998 than in 2006, and thus the two data sets are incomparable. Yet, the authors did not raise this question, nor did they posit any other explanation for the seemingly anomalous finding. Given that it was one of their two key findings, they should have addressed this striking anomaly. Had they submitted their work for anonymous peer review, the anomaly would have met with strong objections from the reviewers. In summary: 1. The study's claim that we are on the verge of losing a lot of foreign geniuses due to long waits for green cards is disproved by taking a more fine-grained look at the data. 2. The study's claim that participation in patent activity by the foreign-born more than tripled in just eight years simply doesn't "pass the sanity test." It could be a data problem, but in any case, the researchers should have questioned it. Their failure to do so makes their data rather useless, especially given their apparent goal to influence public policy. A few other points: After Wadhwa's second study came out, several researchers in a public forum, including me, raised a more fundamental question: Did Wadhwa believe that, had the same jobs been occupied by natives, would the patent activity not occurred? He answered that he didn't know, and was not making such a claim. Yet Wadhwa and his coauthors did in fact make such claims to the press and in the report itself, such as this one in the Conclusions section of the present study: The United States benefits from having foreign-born innovators create their ideas in the country. Their departures would, thus, be detrimental to U.S. economic well-being. Again, this question of substitutability--if the employers had staffed these jobs with natives, would there still have been as much patent production?--is quite fundamental, and should have been raised in the report. Peer review would have required it. Finally, the reader here may wonder why such a large percentage of our PhD programs consist of foreign students. The answer is that the federal government actually planned for that, explicitly saying that the goal was to hold down PhD salaries. See Eric Weinstein, "How and Why Government, Universities, and Industry Create Domestic Labor Shortages of Scientists and High-Tech Workers," NBER, Harvard University, 1998, http://nber.nber.org/~peat/PapersFolder/Papers/SG/NSF.html Note that Weinstein was working at the time with Richard Freeman, one of the authors of the present study. Indeed, Freeman knows that the government plan worked. Wages for PhD scientists and engineers have fallen behind those of other professions. As he pointed out in "Does Globalization of the Scientific/Engineering Wo Leadership?," Richard B. Freeman, NBER Working Paper No. 11457, June 2005, Whichever indicator one examines, the evidence suggests that the job market for most scientists and engineers in the US has fallen short of the job markets in competitive high level occupations. Exhibit 3 records levels of pay and rates of change in pay from the Census of Population. It shows that scientists and engineers earn less than law and medical school graduate, and that rates of increase in earnings for science and engineering in the 1990's fell short of the rates of increase for doctors and lawyers and for persons with bachelor's degrees. Microsoft pays new PhDs in computer science about $90K, while it pays its lawyers fresh out of law school $140K. The MBA situation is similar. In other words, the government's plan to bring in a lot of foreign nationals in order to keep PhD salaries down worked. If Freeman thinks that's a bad thing, why is he advocating exacerbating the situation? Even if it were a good thing to import our scientists and engineers in the short run, it would be disastrous in the long run. The fact is that in the coming years, fewer and fewer foreign scientists and engineers will want to come to the U.S. in the first place. That's because economic conditions for tech people are declining here--as seen in the slow wage growth mentioned above--and because opportunities in the foreign workers' home countries are getting much better. I've mentioned before, for example, "A Chinese University, Elite Once More: Tsinghua U., Once Stripped of Its Science Programs, Now Competes with America for Graduate Students," The Chronicle of Higher Education, July 9, 2004. More recently, similar quotes have appeared, ironically, in articles that urge the U.S. to become more dependent on foreign workers. For instance, an Associated Press article, just this week ("Silicon Valley Executives Call for Immigration Reform for Workers," August 29, 2007), contained this passage: Lin Lee, who directs Sun [Microsystems'] government strategy in Asia, said many [Chinese] entrepreneurs want to join startups in China — where even five years ago they would have tried to immigrate. "The living standard has really gone up. You can have any food you like. The apartments are really nice," said Lee, who moved to China after her husband was transferred to Beijing. "Things are very, very different now — not just the living standard but the opportunity entrepreneurs see. They almost see another wave of opportunity that they don't see [in the U.S.]." And then there was this one in Forbes on March 5, 2007 ("Silicon Valley's Immigration Problem," Elizabeth Corcoran). Quoting Rosen Sharma, a graduate of the Indian Institute of Technology (IIT, India's top set of engineering universities) who later became a CEO in the U.S., the article reports: Indian immigrant Rosen Sharma opted for the U.S. in 1993 and has done extraordinarily well here. But if he were just coming out of college these days, he says, he would pick India. The business opportunities are better, he says, and quality of life issues are at least as good... In 1993, he says, after graduating with flying colors from the Indian Institute of Technology in Delhi, Sharma headed straight for the U.S. So did most of his classmates. Of the 40 people in Sharma's graduating class at IIT Delhi, he says, all but three came to the U.S... But Sharma, who is president of the IIT Delhi Alumni Association, says the next generation of Indian engineers are unlikely to feel the way he does: Last year, only 10 of the 45 IIT graduates who went through the same program Sharma did decided to pursue jobs in the U.S., he says. In other words, the policy advocated by Wadha and his coathors would not only worsen the wage suppression which makes tech unappealing to our own best and brightest, but it would count on a labor source which is in the process of drying up. The proposed policy would be harmful not only from a labor point of view but also even a corporate one. Norm