Dr. Norm Matloff writes to his email list:
Recently a couple of people have called my attention to a study released by the Public Policy Institute of California (PPIC): H-1bs: How Do They Stack Up to US Born Workers?, by Magnus Lofstrom and Joseph J. Hayes. I believe it will be heavily circulated by the industry lobbyists, and widely quoted by the press.
You can download the paper here.
Here's the abstract:
Combining unique individual level H-1B data from U.S. Citizenship and Immigration Services (USCIS) and data from the 2009 American Community Survey, we analyze earnings differences between H-1B visa holders and US born workers in STEM occupations. The data indicate that H-1Bs are younger and more skilled, as measured by education, than US born workers in the same occupations. We fail to find support for the notion that H-1Bs are paid less that observationally similar US born workers; in fact, they appear to have higher earnings in some key STEM occupations, including information technology.
Most major think tanks have an ideological bent of some kind, which in PPIC's case is liberal Democratic. No problem for me, as I myself am in that camp, but in the case of skilled immigration PPIC has a rather hidden source of potential bias: PPIC is funded by the Hewlett family—as in Hewlett-Packard. Moreover, much of PPIC's endowment consists of Hewlett-Packard stock. See The Role and Impact of the Public Policy Institute of California: An Operating Foundation as Think Tank, David W. Lyon, presented to the PPIC Board of Directors at its quarterly meeting, March 4, 2004.
PPIC has underwritten various pro-industry studies by UCB's AnnaLee Saxenian. PPIC's Julian Betts has written in favor of expanding H-1B (San Diego Union Tribune, November 22, 2000), and so on.
Not surprisingly then, the paper by Lofstrom and Hayes concludes that those of us who've been doing research showing that the H-1Bs are underpaid have been wrong all these years. They're not underpaid, claim the authors, and in fact the H-1Bs may actually be paid more than the Americans.
Also not surprisingly, I will be very critical of this study in my posting here. The authors clearly have not done their homework. They have a number of their important facts wrong, often because they apparently haven't actually read the research papers they cite. Then in terms of analysis, they make the same methodological errors that are so common in this little niche of labor economics.
Here's an outline of what I will say below:
- The authors have their important facts wrong, ranging from their incorrect descriptions of previous research findings to inaccurate descriptions of the H-1B visa itself.
- The authors' statistical analyses are based on an inadequate understand of the nature of the labor markets in question.
- Though the authors cite two of my research papers several times, they miss my major point, well familiar to readers of this e-newsletter: The primary way that employers use the H-1B program as cheap labor is to hire younger, thus cheaper, H-1Bs in lieu of older (35+), thus more expensive Americans. Indeed, they mention repeatedly that the H-1Bs are quite a bit younger than the Americans, without realizing that THAT is the primary way that H-1B provides employers with cheap labor.
The details follow.
I believe the authors have misinterpreted the work of several authors, but I'll limit my comments here to what they wrote about my own research. They cited my 2003 University of Michigan Journal of Law Reform article about H-1B in general, and my 2008 article for CIS analyzing whether the foreign workers are "the best and the brightest."
The authors say,Matloff (2008) boldly states that these workers are not "the best and the brightest," a conclusion with which Hira (2007) concurs. The primary evidence of these contentions has been lower wages among H-1B workers compared to U.S. born workers.
This is the exact opposite of what I say in that 2008 paper. Instead, I say, look, for the sake of argument, let's assume the industry is CORRECT in claiming it doesn't underpay H-1Bs. Then if the industry is also right in saying that the H-1Bs tend to be the best and the brightest, then these workers should be making well above the average for their professions (adjusted for region, length of experience etc.). Then I show that that is not the case. In other words, again assuming the H-1Bs are NOT underpaid, the data show that they are not generally "the best and the brightest."
The authors then claim that my analyses don't adjust for age and education. On the contrary, these two variables play a central role in most of my work. Technically, the authors make this claim only about my 2008 paper, but since they did cite my 2003 paper, the reader gets the impression that my analyses don't take these variables into account, which is false. And my 2008 paper takes years of experience into account, similar to age, and takes the education variable into account in more complex way.
The authors also claim that my 2008 paper concerns H-1Bs at the time they first arrive in the U.S., which is also false. My paper is actually about employer-sponsored green card applications, not H-1Bs per se, and most of these applicants have been around for a few years.
To me the most outrageous lapse here is that the authors, allegedly having read my work, do not know my central point, which as noted above, is that the primary way that employers use the H-1B program as cheap labor is to hire younger, thus cheaper, H-1Bs in lieu of older (35+), thus more expensive Americans.
As many of you will recall, I distinguish between Type I salary savings—hiring H-1Bs at lower wages than comparable Americans would get—and Type II salary savings, hiring young H-1Bs instead of more established Americans. In my work I've shown that while employers can save money via Type I, the Type II savings are much larger. As I've often said more succinctly, "H-1B is about age."
The authors, in pointing that the H-1Bs are younger—about 10 years, they say for IT—and in recognizing, correctly, that younger means cheaper, are unwittingly making my point for me: Employers save a bundle by hiring the H-1Bs, because they're younger and thus cheaper. That's H-1B in a nutshell.
The authors state, "For an H-1B visa to be approved, the US Department of Labor (DOL) must certify that the intended foreign worker does not displace...US workers," again, completely false except for the minuscule H-1B-dependent subcategory.
The authors fall into the same traps so many others have fallen into. They break their analyses down into occupation groups, which is good, but don't understand the specialized labor markets. One of their job categories is IT.
My frequently-expressed complaints about economists—as usual, my apologies to the economists on this list—is that even though someone is a great number cruncher, their crunching can be meaningless (or worse, misleading) if they don't know WHICH numbers to crunch and what the numbers MEAN. In studies of H-1B (whether they support H-1B or are critical of it) in the IT area—by far the largest H-1B occupation—this problem is quite acute.
The problem is that IT is much too broad a term, including many jobs that are not the types taken by H-1Bs. Such jobs are generally of the technician type, which pay less. That results in a built-in downward bias for the American salaries relative to those of the H-1Bs; in other words, the H-1Bs misleadingly look like they're doing well compared to the Americans, misleadingly so because they're working in different jobs from the Americans. This is the same central problem with the 2003 Zavodny study that I've discussed before.
It is for this reason that in my H-1B analyses, I stick to the job market I know well, that is for programmers, software engineers and electrical engineers. So for instance I do NOT analyze H-1B mathematicians, as I don't know that market, which has huge salary disparities due to the fact that many mathematicians now work as "quants" on Wall Street. The authors of this study presumably don't know any of these market sectors sufficiently well to avoid the pitfalls.
Region is important, as wages are much higher in some regions than authors, and the H-1Bs are disproportionately in the high-wage areas. In my case, I've always addressed that issue, either by limiting to Silicon Valley, or by using the green card data (which is adjusted by region), or by using a region variable in my regressions.
The authors understand this point, and lament the fact that their primary data does not include information on region. They try to remedy that by doing a small supplementary analysis using the ACS data, but do not supply the details. They do say that this analysis narrows the gaps they found previously between H-1B and native wages, to about 2-4%. But without knowing exactly what they did, I can't comment.
What also always surprises me about economists is that they often fail to understand how basic economic principles show that the H-1Bs actually MUST be underpaid on average:
a. After they're hired, the H-1Bs have limited mobility in the labor market, and if they're being sponsored for a green card, they have essentially NO mobility. If you can't move around in the labor market, then you can't have employers bid against each other for your services. In other words, you can't swing as good a deal for yourself as you would if you were a citizen or permanent resident—so the H-1Bs are on average underpaid.
b. The H-1Bs are getting very large nonmonetary compensation, either in the form of a green card, or if they wish to go home, in the form of a U.S. work experience. They are thus willing to work for less (have a lower "reservation wage," in economics terminology). It's just like you and I might take a lower wage if the job had a shorter commute or it offered onsite child care, say.
c. Advanced degrees, special technological skill sets and the like command a salary premium on the open market. The employers claim they hire the H-1Bs for these special qualities. Yet the government prevailing wage does NOT take these into account. Therefore the law allows employers to underpay H-1Bs relative to their worth. Furthermore, the data show that most of the foreign workers make either the prevailing wage or just a bit higher, so most actually ARE underpaid.
None of these considerations shows up in the PPIC study, which for this and the above reasons is simply not a useful contribution to the H-1B debate.