Date: Fri, 3 Oct 2003 21:26:48 -0700 From: Norm Matloff To: Norm Matloff Subject: critique of the new GAO report To: age discrimination/H-1B/L-1 e-newsletter The long-awaited GAO report on H-1B was finally released yesterday, at http://www.gao.gov/new.items/d03883.pdf. To me, the most salient portion of the report concerns the question of whether the H-1Bs are paid less than Americans. As we all know, the industry lobbyists have repeatedly and vehemently denied that they pay the H-1Bs less than Americans, while people like me claim that the main reason the employers hire H-1Bs is cheap labor. So, this is clearly a key issue. Unfortunately, GAO has really dropped the ball on this key issue. The GAO considered various job categories. As always, I am interested in the computer-related jobs, and will focus on that category here. The GAO's first major finding on this issue is that (a) younger H-1Bs with no graduate degree are paid MORE than younger Americans with no graduate degree, but (b) older H-1Bs with no graduate degree are paid LESS than older Americans with no graduate degree. The largest single problem with this analysis concerns job type. As I have stated often, the computer-related H-1Bs are mostly computer programmers. However, programmers can have different job titles at different kinds of firms, such as Programmer, System Analyst and Software Engineer. They are all doing the same work, but because of historical reasons those with Programmer and System Analyst titles tend to be doing "old-fashioned" work (i.e. IBM mainframe) while those with Software Engineer titles are doing "modern" work. The old-fashioned work tends to be paid considerably less than the modern work. To see this, here is my analysis of mean wages from the 2000 Census, for the state of California. The means are calculated for U.S. citizens (GAO's definition of "American," which excludes U.S. permanent residents) who are under 30 years old (GAO's definition of "young"), and who have a Bachelor's degree but no graduate degree: System Analyst $43,839 Programmer $42,495 Software Engineer $57,309 GAO used the Current Population Survey, a monthly mini-census conducted nationwide but with a small sample. CPS, unlike the census, did not have a separate Software Engineer category in the 2002 data analyzed by GAO (one was added in January 2003). GAO simply used the two related categories CPS offers, System Analyst and Programmer. The large difference seen above shows that GAO's failure to separate the three categories badly skews GAO's analysis, as the H-1Bs tend to do "modern" work while some Americans do "old-fashioned" work. Even better, here is another comparison I did from the census data. I did the same analysis as above for Software Engineers, but this time for foreign-born people who had entered the country at most 8 years ago, this condition being a proxy for H-1B-ness. (The census data does not give visa/immigration status, but does give information on whether the person is a native or not, and in the latter case on year of entry to the U.S.) Again, this is for those under 30 years old, with a Bachelor's degree but not graduate degree. Here the difference is striking: citizen Software Engineers $57,309 "H-1B" Software Engineers $49,138 So, instead of the H-1Bs being paid $9,000 MORE than the Americans, as GAO found, we have the H-1Bs being paid $8,000 LESS than the Americans. That was for the Bachelor's level At the graduate level, the GAO said that it had sample size problems. However, the census data are plentiful, and I conducted the same analysis as above (under 30 years old, etc.) but for those having a Master's degree. Here the difference is again clear that the H-1Bs are paid less: citizen Software Engineers $66,290 "H-1B" Software Engineers $58,598 Note that my census figures here are for California. In other words, I held geography fixed--something the GAO did NOT do. This is crucial, since the H-1Bs tend to be concentrated in just a few states, i.e. California, Maryland, New Jersey, etc. as pointed out by GAO. (If I recall correctly, California alone has half of all the nation's H-1Bs.) My point is that the H-1Bs tend (with the exception of Texas) to be in high cost-of-living areas, more so than the Americans, and this alone would push up the H-1B salaries relative to the Americans. There are other similar problems, e.g. the fact that GAO's "no graduate degree" category includes those not having a Bachelor's, which is the case for a number of Americans but for very few H-1Bs (H-1Bs by law must have a Bachelor's degree, though there are exceptions). Even though I am deeply disappointed by the badly flawed approach GAO took to the statistical analysis described above, I view their second finding to be downright outrageous: Some employers said that they hired H-1B workers in part because these workers would often accept lower salaries than similarly qualified U.S. workers; however, these employers said they never paid H-1B workers less than the required wage. Note that this employer candor about underpaying H-1Bs occurred in the employer survey in the NRC report as well. (And needless to say, many employers who do underpay their H-1Bs are not going to admit it.) Yet GAO says it's fine, since it is "legal." GAO's attitude here is unconscionable. You don't have to be a rocket economist to notice that "something is wrong with this picture." Anyone can plainly see that if the H-1Bs are being paid less than the Americans but the H-1Bs are still being paid the legal prevailing wage, then SOMETHING MUST BE WRONG WITH THE DEFINITION OF PREVAILING WAGE. This is especially true given that the GAO also notes that employers told them that H-1Bs are willing to take lower salaries. The fact is that the prevailing-wage law is a total sham, so riddled with loopholes as to be useless. The GAO researcher had been told this quite clearly in discussions with labor groups. She could have verified it by interviewing some immigration lawyers, and in any case at the very least should have raised the question. But no, instead, she says it's fine to pay the H-1Bs less than Americans, because it's complying with the (loophole-riddled) law. Ironically, this goes to the issue of job title I mentioned above, as one of the favorite loopholes used by immigration lawyers is to exploit that difference in job titles. Once again, look at the notorious Bank of America case, analyzed at the Programmers Guild Web page, "How to Underpay H-1B Workers," http://www.programmersguild.org/archives/howtounderpay.htm The analysis there points out that one of the ways that bank H-1Bs were underpaid was to call them System Analysts, which for reasons I explained above carries a much lower prevailing wage even though the work they are doing could be called something else with a higher wage. What the H-1Bs did falls within the very broad generic definition of System Analyst, so it is FULLY LEGAL to underpay them by choosing that title, one of the many loopholes which the employers exploit. Moreover, GAO was fully aware of these issues. It had had a lengthy conference call with the Programmers Guild and had had their literature, and there were prominent references to that Guild Web analysis in both the phone call and the literature. The GAO then dismisses any underpayment of H-1Bs as an enforcement issue, another favorite industry line. Using this spin, the industry gets Congress off the hook, insuring that Congress won't make needed reform of the law. All Congress has to do, the claim goes, is give the Dept. of Labor more funding and inspection powers, thus sidestepping the real issue of loopholes in prevailing wage. And sure enough, that (expanding the enforcement role of DOL) is what GAO recommends at the end of this report. Again, that recommendation completely misses the mark, as enforcement is not the real issue. What is really disappointing about this is that their own quote of the employers, which I will display here again for the reader's convenience Some employers said that they hired H-1B workers in part because these workers would often accept lower salaries than similarly qualified U.S. workers; however, these employers said they never paid H-1B workers less than the required wage. shows that the problem is the loopholes, not enforcement. It is LEGAL to pay H-1Bs less than Americans. Again, this is something the GAO was told in its Programmers Guild conference call. Bottom line: The GAO tells only the industry's side of the story here, not labor's. And that is a consistent theme. The employers' comments are stated repeatedly throughout the report, with only a few cases in which labor's points are mentioned. In most cases, the employer' comments are cited without citing any countering arguments from labor whatsoever. For example, the report says that the employers stated that they hire H-1Bs because American programmers and engineers are lacking in certain specific skill sets, but the report does not mention that the evidence labor cites to show that the skills issue is just a phony pretext (e.g. the major firms which lay off Americans and force them to train their H-1B/L-1 replacements, showing that it is the foreign workers, not the Americans, who lack the skills). Another example is that the report cites the employers as claiming that the 2000 legislation removed the H-1Bs' de facto indentured servant status, but the report does not say that labor showed that the H-1Bs still have that problem due to green card sponsorship. GAO had a responsibility to present both sides, and it didn't do so. Even a quick glance at the report shows that the report is something like 95% employer views, 5% labor views. Again, GAO's recommendations for reform include NOTHING about prevailing wage. Instead, their recommendations consist of better tracking of the H-1Bs, in order to gauge the impact of the H-1B program on the American workforce, and they make such a big thing out of this that they make it the title of the report. Certainly better tracking would be nice, but plugging the gaping loopholes, which quite plainly are at work (again, as any immigration lawyer would have confirmed), is what is really needed. They also recommend expanded enforcement, which as I said is not the solution. The other major section of the GAO report concerns how many H-1Bs have graduate degrees. This too is a favorite theme of the industry lobbyists, with the point being that employers supposedly hire H-1Bs because they have graduate degrees, i.e. advanced training in their field of work. And once again, the GAO analysis is very misleading. The report says that 38% of the computer-related H-1Bs have a graduate degree. (It is not shown in the report, but that breaks down to about 2% PhD, 36% Master's.) As I've explained in the past, this is correct but misleading, because at the Master's level, just like the case at the Bachelor's level, many computer-related H-1Bs come from fields other than computer science. It is very common for a foreign national to get a Master's degree in, say, economics or biology at a U.S. university, taking a few undergraduate computer courses, and then be hired by an American employer. The point is that such a person does not have Master's-level training in computer science, as implied by the industry lobbyists. Based on educational statistics data (see my updated congressional testimony), among computer-related H-1Bs, only 0.6% have a PhD in the field and only 7.6% have a Master's in the field, totaling about 8% overall. Moreover, I don't consider a graduate degree important to begin with. See http://heather.cs.ucdavis.edu/Archive/ProposedMSPhDExemption.txt Finally, there is the question as to why the GAO took so long to release this report, which was released only yesterday, even though it is dated September 10 and all but one other of the reports issued in the last two days were dated September 30 and October 1. One person who had checked with GAO a few weeks ago had been told that the report was actually ready in August, but was being "reviewed." So why the delay? Well, a partial answer may be in this passage of the report, in which GAO addresses Rep. Mark Udall, who had requested GAO to undertake this study: As arranged with your office, unless you publicly announce its contents earlier, we plan no further distribution of this report until 30 days from its issue date. At that time, we will send copies of this report to the Secretary of Homeland Security, the Secretary of Labor, appropriate congressional committees, and other interested parties. In addition, the report will be available at no charge on GAO's Web site at http://www.gao.gov. That certainly shows that there was some political manipulation of this report, at least in the timing of the release but very possibly also in terms of its content. I make the latter statement both because of the clear imbalance in the report--there is no denying that employers are quoted FAR more than labor throughout the report, something like 95%/5%--but also because of very disturbing comments I heard from one H-1B critic who talked to the supervisor of the researcher responsible for the report. I regard the researcher herself to be competent and fair, but what I heard second-hand about the supervisor would, if true, suggest that political pressure was placed on the GAO. The pressure need not be explicit. It could be simply something like, "Make sure your analysis is on data, not just anecdotes." It is yet another favorite industry lobbyist line to dismiss anything said by critics of H-1Bs as "anecdotal," in spite of the critics supplying lots of data addressing various aspects of the H-1B debate. In the GAO report, GAO does indeed characterize one of the points made by labor as "anecdotal," though none of the anecdotal statements made by industry is characterized that way. Since some of the issues are indeed hard to quantify, notably impact of H-1Bs on American workers, the best way to avoid the report having an outcome unfavorable to industry is to insist that everything be quantified on both sides. Certainly it is clear that many in Congress, especially the Democrats--who are supposed to at least look pro-labor, even if they are just as tied to industry as the Republicans--were hoping that the GAO report would come out exactly like it did, saying "We need more data," letting the politicians off the hook. I've heard that a number of politicians have said (and have myself heard one of them say it), "We need to wait for the GAO report" before doing anything on H-1B/L-1. An inconclusive report--one whose main recommendation is to collect more data--is exactly the vehicle they want for stalling. Another of the report's recommendations, giving DOL more power or funds for enforcement action, is even more pernicious, since as I stated above, the problem lies in the law itself, not enforcement; by saying it's enforcement issue, the GAO is enabling Congress to sidestep the central issue, which is the gaping loopholes in the law. No wonder Silicon Valley congressperson Anna Eshoo, in meeting with the Programmers Guild a few weeks ago, was so dismissive of the NRC report of 2000, commissioned by Congress. She insisted it was "too old" (what, the employers starting paying the H-1Bs better since then????), and that she couldn't do a thing until the GAO report came out. Now, of course, it appears that the report will enable her to not do a thing, if that was her goal. I should mention my own involvement here. The GAO researcher contacted me before the study began, in August 2002, referring to my "expertise in this area" (her phrasing), and asking me what kind of data might be analyzed. I gave her an answer, and she said she would call me back the next week for more advice. She never did call back. I did get to talk to her many months later, but only because I was invited by labor representatives to participate in the conference call mentioned above. I felt at the time, and still feel now, that she was sincerely interested in what everyone had to say in the conference call. After that, I offered to give her a draft of a long academic article I had written (due to come out soon now), and she replied that she would like to see it. But she ignored everything that was in it, as can be seen from what I said above. I suppose that sometime after she called me an "expert," some industry lobbyist (or third party who had talked to industry lobbyists) convinced her that I am instead some sort of wide-eyed radical. If that is what did happen, it's a shame, because I could have pointed out analysis which would have made for a much more meaningful report, e.g. the data I've shown above. In any case, again the point is not that she didn't *support* my/labor's side in the debate with employers. The point is that she didn't even *present* labor's side in most aspects. By word count or any other measure, the employers' voices were given 95% coverage and labor was given only 5%. That undeniable fact comes from a simple reading of the report. GAO is highly respected in Congress (even if their political constraints mean that they don't always heed GAO's advice), so this report is especially insidious. Sadly, the report comes nowhere near living up to GAO's slogan, "Accountability, Integrity, Reliability." Norm