Date: Mon, 24 May 2010 00:12:04 -0700 From: Norm Matloff To: Norm Matloff Subject: CIO article on Mithas/Lucas study To: H-1B/L-1/offshoring e-newsletter (Note: See a further update at http://heather.cs.ucdavis.edu/Archive/MithasLucas4.txt--NM) I reviewed the Mithas/Lucas study, discussed in the enclosed CIO Magazine article, about a week ago, archived in http://heather.cs.ucdavis.edu/Archive/MithasLucasPublished.txt I have some further remarks to make, and the enclosed article brings up some a new issue to be addressed. The study is both (a) fatally flawed, due to its use of an invalid data set, and (b) disturbingly lacking in evenhandedness. Here is a summary of my previous posting, with some updated points: 1. Mithas and Lucas' data set is irrelevant to the issue of H-1B wages in the IT field. The data consists of a reader survey for a magazine whose audience is largely NOT in the IT field. The magazine is mainly aimed at managers and nontechnical people. Its self-description states, "InformationWeek.com is the industry-leading source of news... serving business technology executives..." Its job postings section currently lists the following positions: % D. E. Shaw Research seeking Chief of Staff in New York, NY. % Switch and Data seeking Sr Product Marketing Manager in Tampa, % FL. % Kadrmas, Lee and Jackson seeking Network Architect in % Bismarck, ND. % Osram Sylvania seeking Benefits Specialist in % Danvers, MA. % AccuWeather seeking Business Development Manager % in Atlanta, GA. Only one of these jobs is in IT. This is NOT the type of magazine read by engineers at Google or Apple. That qualitative point can be seen in the data. The respondents in the ML data have much lengthier job experience than government data show for computer-related H-1Bs. Moreover, wages in the ML data are much higher than the official data show for H-1Bs in the computer field, as Prof. Hira points out in the enclosed article. In short, the data set relied upon by Mithas and Lucas is simply irrelevant to the issue of H-1Bs in the computer area. 2. Mithas and Lucas AGREE that H-1Bs are underpaid, relative to Americans of comparable background and skills. Let me repeat: Mithas and Lucas AGREE that H-1Bs are underpaid, relative to Americans of comparable background and skills. That means they agree with me, with Ron Hira, with John Miano, with Rob Sanchez, with Kim Berry, and for that matter, with Vivek Wadhwa, who had admitted that he himself underpaid H-1Bs when he was a tech CEO. As I have been saying for years, you can tell that H-1Bs are underpaid relative to comparable Americans with a simple "thought experiment," i.e. without looking at any data. H-1Bs have very limited mobility, and thus cannot swing as good a deal as Americans. The latter can move to another employer if they get a better offer elsewhere; this is very difficult for an H-1Bs and essentially out of the question if the H-1B is being sponsored for a green card. The 2001 National Research Council study, commissioned by Congress, also discussed this point in detail. Well, Mithas and Lucas AGREE with this. Their paper points out: Possession of a green card provides GREATER BARGAINING POWER and job security for an IT professional compared to someone with a[n H-1B] work visa because (1) employers typically hold work visas, which makes it difficult for an IT professional to easily change his or her employer and (2) work visas are of a limited duration... In other words, although there are many aspects of H-1B which can be discussed, there is one aspect in which everyone agrees, INCLUDING Mithas and Lucas: H-1Bs are on average underpaid relative to comparable Americans. 3. The Mithas/Lucas study is disturbingly lacking in evenhandedness. Again, I went into great detail on this in my previous posting, but let me cite two examples, the second of which is illustrated in the enclosed article. First, Mithas and Lucas dismiss the studies they cite that indicate that H-1Bs are underpaid (though, remember, M/L actually agree with this), in a rather nitpicky manner, and omit many of the most important studies on H-1Bs as cheap labor altogether--yet they give a free ride to, making no criticism at all of, the studies that are viewed as pro-industry regarding H-1B. For instance, M/L complain that the studies showing underpayment of H-1Bs often do not report statistical significance values. Well, this is silly in that with the large samples used in most studies, including M/L's, almost anything will be statistically significant (this does NOT mean important). But much more to the point, Mithas and Lucas do not complain that the studies they cite in support of their findings, e.g. NFAP and Wadhwa, don't report statistical significance values either. This is a double standard, pure and simple. Just as troubling is the fact that in their statements to the press, M/L highlight the numbers from their coarser analysis, which sound more impressive, while not mentioning that their more refined analysis resulted in much milder, less impressive statistics. For instance, though their initial analysis found that H-1Bs get a 6.8% salary premium over Americans, their more refined analysis that controlled for location and job title produced a number only about a third as high, 2.6%. Yet in their comments to the press, both in the article enclosed below and in interviews with other publications, the authors cite the 6.8% figure, not the 2.6%. And this is in spite of their stressing, correctly, that as many variables as possible must be controlled for. Now let's turn to another issue brought up in the enclosed article. There is the opinion, expressed by several of those quoted in the article, that there are "two kinds" of H-1B employers, a Good Kind and a Bad Kind, the latter hiring H-1Bs as cheap labor and the former doing so to acquire genuine talent and skills unavailable in the U.S. This is currently a popular view in some government and academic circles, and is implicitly behind the proposals to give fast-track green cards to foreign workers with U.S. graduate degrees. Yet it is totally at odds with reality. The big mainstream firms like Intel and the Bank of America are just as culpable as the Indian bodyshops such as Tata Consultancy Services. (Mithas, by the way, worked for Tata for 10 years, though apparently not for the TCS branch of the firm.) Intel and the BofA tend to hire people with Master's degrees while TCS does not, but both are using the H-1B program for cheap labor, as I've explained before. It's interesting, for instance, that the enclosed article says, Jacob Funk Kirkegaard, a research fellow at the Peterson Institute for International Economics, examined the Department of Labor's database of labor condition applications (LCAs) and surmised that many large U.S. corporations and educational institutions frequently offer their H-1B recipients salaries substantially above the prevailing U.S. wage. However, he says, "offshoring/outsourcing type IT services providers" aggressively pursue all legally available paths to cut labor costs, including paying foreign workers only the legally mandated 95% of the prevailing wage. Well, guess what--Intel does the same thing! When that 95% rule was in effect (it was removed a few years ago), Intel took advantage of it too. You can see it in the government data. Intel had a lot of prevailing wage figures with an overabundance of nonzero digits, say $59,850, which when divided by 0.95 became much rounder, in this case $63,000. In the infamous Cohen & Grigsby immigration law firm video series, which the C&G law firm put on YouTube to their later regret, you see many examples of LARGE, MAINSTREAM FIRMS taking advantage of that 95% rule. (Note: The prevailing wage policies apply both to H-1B and green cards, though C&G was referring mainly to the latter.) Of course that 5% saving is just peanuts, and the video explained the real savings are much larger (see below), but again my point is that Kirkegaard's implying that the large mainstream corporations did not take advantage of that 95% rule is demonstrably false. Again, the major point is that in video 12 the law firm talked about how to get a prevailing wage that can be, in the Cohen & Grigsby attorney Jennifer Pack's own words, "$10,000 to $15,000" below the market. That is a key point--the legal prevailing wage laws and regulations are riddled with huge loopholes, so the legal prevailing wage is NOT the market wage. C&G explains this in their video And C&G is a LARGE, PROMINENT MAINSTREAM law firm with LARGE, PROMINENT MAINSTREAM clients. It is NOT true that the big mainstream firms are the Good Kind, while the Indian ones (talk about scapegoating!) are the Bad Kind. See http://heather.cs.ucdavis.edu/Archive/CohenAndGrigsbyPrevailingWage.txt for details. The article follows below. Norm http://www.computerworld.com/s/article/print/9177045/H_1B_visa_holders_earn_more_than_U.S._born_IT_pros_study_says_?taxonomyName=Careers&taxonomyId=10 H-1B visa holders earn more than U.S.-born IT pros, study says Stephanie Overby May 20, 2010 (CIO) One of the biggest complaints about the federal government's H-1B and L-1 visa programs is that they could be used by corporations to hire skilled workers born outside the U.S. at wages lower than the U.S. market rate. Indeed, anti-H-1B visa activists say the program depresses American IT workers' salaries and robs them of jobs. But new research from the University of Maryland seems to contradict anti-H-1B visa activists' claims about the immigration program's impact on U.S. wages. The research suggests that foreign-born IT professionals with temporary skilled worker visas actually earn more than their American counterparts, not less. Hank Lucas, professor of information systems at the University of Maryland's Robert H. Smith School of Business, and assistant professor Sunil Mithas examined the effect of immigration policies on IT salaries using data from online salary surveys conducted from 2000 to 2005 by InformationWeek and management consultancy Hewitt Associates. After adjusting for educational qualifications, work experience, and other individual characteristics, Lucas and Mithas found that IT professionals without U.S. citizenship earned 8.9% more than U.S. citizens. Tech workers on temporary visas, such as the H-1B and L-1, were paid 6.8% more than those with U.S. citizenship, and green card holders took home 12.9% more than their American-born counterparts, according to Lucas' and Mithas' research, published this month by the Institute for Operations Research and the Management Sciences. The professors say restrictive visa policies resulted in even higher salary premiums. In years when applications exceeded the annual caps for H-1B visas, salaries for all non-U.S. citizen IT workers--that is, visa or green card recipients--rose relative to the salaries of American-born IT professionals, say Lucas and Mithas. Mithas says the study was driven by the lack of compelling data around claims that foreign-born IT professionals are taking away jobs from American workers. "Much of the immigration debate in this country ignores skill levels," says Lucas, adding that the influx of non-U.S. citizens has a much different impact on job availability and wages for unskilled labor than it does for skilled workers. U.S.-born citizens and foreign workers can potentially benefit from an influx of skilled workers, Lucas says. B salary survey ignites controversy The Lucas-Mithas research deviates from the findings of other studies investigating the effect of temporary visa programs on the salaries of U.S. IT professionals. According to Lucas and Mithas, H-1B visa holders earned an average of $75,358 from 2000 to 2003, compared with the average U.S. citizen's salary of $66,836. (The InformationWeek survey did not ask about visa status in 2004 and 2005). But according to the U.S. Citizenship and Immigration Service (USCIS), the median salary for H-1B visa holders in computing professions during the 2000 to 2003 period was just over $50,000. "It [seems] strange to me that the authors would depend on sampled data when we have the whole census of new H-1B recipients' salaries reported [by] the USCIS, at least in aggregate terms," says Ron Hira, associate professor of public policy at the Rochester Institute of Technology. "For computing occupations, those data show low wages relative to Bureau of Labor Statistics wages for Americans. The median salary for new H-1Bs is comparable to the entry-level wages for freshly minted bachelors in computer science, as reported by the National Association of Colleges & Employers. So half the new H-1Bs are being paid at- or below entry-level wages." Lucas and Mithas say the USCIS and BLS numbers aren't granular enough to make meaningful comparisons. "You don't get a good sense of who these people are, what is their educational background, how long have they been in IT, what industry are they working in," Mithas says. "If you don't have data at an individual level, you don't know if you're comparing apples to apples." Hira suggests there may be a self-selection bias at play when using a sample population. The data Lucas and Mithas used comes from 50,000 IT professionals, including 809 temporary visa holders, who opted to participate in an online salary survey. The researchers say the overall sample and sample of non-U.S. citizen foreign-born IT professionals in their study is reasonably representative of the U.S. population. While those numbers may line up, it's unlikely that H-1B or L-1 grantees who depend on their employers for their visas and who earn lower than average wages would participate in such a survey, says Hira. "The [Lucas-Mithas] report may be able to control for some additional factors that affect wages, but there is no doubting the USCIS characteristics data," says Hira. "It is a census, not a sample." Lucas admits that selection bias could be a factor in any survey, but he remains confident in his data. "In situations like this, there's always the possibility for the sample not to be truly random," he says. "But I feel more comfortable with this survey with 50,000 respondents than I would if we did a random sample of a couple hundred IT professionals on our own." corporations pay more for H-1Bs Lucas and Mithas say their research proves that corporations use foreign-born IT professionals as a complement to, not as a cheaper substitute for, their American workforce. But the data does not provide any explanations for why employers would pay non-citizen IT workers more. Lucas and Mithas have their own theories. For one, they think companies recruit foreign IT professionals for skills or expertise that they can't get from American workers, whether it's a stronger work ethic, multi-cultural experience, or willingness to travel. "We were searching for an explanation, and it wasn't education or anything we could measure. So it had to be something intangible, like how aggressive you are or how much of a risk taker," says Lucas. "I'd have to say it's motivation. You have to be motivated to break out of the rut you're in, get out of the city you were born in, go to another country and work in IT." Hira is not buying it. "There's no doubt in my mind that, in general, H-1B workers are underpaid. That's why the offshore outsourcing majors rely almost exclusively on H-1Bs rather than hiring Americans," he says. "Plus they are beholden to their employer, making it more difficult for them to protest against poorer [wages and] working conditions." Hira notes that it's possible the compensation among IT workers on temporary visas falls into two camps: "lots of low wage workers and a good portion of high wage workers." B. Lindsay Lowell, director of policy studies for Georgetown University's Institute for the Study of International Migration, has been analyzing the salaries of H-1B visa holders in science and technology using data from the National Science Foundation's National Survey of College Graduates. He has found that H-1Bs are paid lower average wages during the first three years of their permitted stay, but once they reapply for another three years or change employers, they may earn more than comparable U.S. citizens. "The NSCG is an odd survey itself, but it's the largest random survey of graduates in the United States," says Lowell. "The reason we think there's an increase in wages is due to a change in bargaining power on the part of the H-1B. Of course, the first three years still reflects a savings to the employer." Jacob Funk Kirkegaard, a research fellow at the Peterson Institute for International Economics, examined the Department of Labor's database of labor condition applications (LCAs) and surmised that many large U.S. corporations and educational institutions frequently offer their H-1B recipients salaries substantially above the prevailing U.S. wage. However, he says, "offshoring/outsourcing type IT services providers" aggressively pursue all legally available paths to cut labor costs, including paying foreign workers only the legally mandated 95% of the prevailing wage. "That suggests another point I've long argued, and which is supported in my and others' research," says Lowell, "which is that the H-1B labor market is softly segmented with different types of employers who pay less than the mainstream to the H-1Bs in their employ." Lucas and Mithas consider the LCA data unreliable because it does not provide actual salary data, and many approved LCAs don't result in the granting of an actual visa. "There are an awful lot of anecdotes of this or that company bringing in foreign workers and paying them less than the going wage, but you can't take that and generalize it to thousands of people," says Lucas. "We'd love to get inside of Infosys and Accenture and get detailed individual information, but privacy concerns enter into this, and they wouldn't want to give us the information anyway. They're afraid, no matter which way the research comes out, they'll be criticized." e research into H-1B wages needed There are as many as 700,000 temporary high-skill foreign professionals in the U.S. on visas today (approximately 500,000 in the H-1B category alone), and 60% of them work in IT, according to some estimates. More detailed data from USCIS or some other source would go a long way toward settling the debate over whether these visa programs depress the wages of IT professionals. "This may be a good thing for the DHS or GAO to do," says Hira.