(Update: Microsoft responds.)
One issue that’s come up regarding an earlier post about declarations by Bill Gates on what Microsoft pays H-1B visa holders is whether the analysis purporting to catch Gates in a fib actually compare apples to oranges: In this case green-card data vs. H1-B conclusions.
There is no mixing of fruit here, says Norm Matloff, a professor of computer science at the University of California, Davis. And Matloff has plenty more to say about Microsoft’s H-1B practices, as well as those of the tech industry as a whole. Here’s an excerpt from his e-mail to me:
In almost all cases, when an employer sponsors a foreign worker for a green card, the worker is already working for the employer as an H-1B. The H-1B gives the employer the services of the worker while the green card application goes through, which takes several years. So, almost all employer-sponsored green card applicants are H-1Bs. (The converse is largely true for big companies like Microsoft.) The post in Daily Kos was thus not an apples-and-oranges comparison.
What is significant about the post is that it illustrates the fact that the big firms abuse H-1B just as much as the small ones, quite contrary to the claims often made that the only abusers are the Indian job shops. I have a similar analysis for Intel in my archived e-newsletters.
WashTech has done an excellent job of exposing Microsoft’s lies on H-1B and offshoring. See my post reporting on the WashTech investigations here and here.
Those lies (a term I do NOT use lightly) are just the tip of the iceberg. The tech industry and the American Immigration Lawyers Association hire the best PR people in the business, and everything they say is incorrect or highly misleading. Yes, everything. That includes their claim that we have a shortage of engineers and programmers, a claim shown false by a Businessweek article, plus my own followup. See my e-newsletter postings here and here.
The earlier post has been picked up over at Slashdot, and, as might be expected, Gates isn’t faring well in the comments section. No word yet from Microsoft PR or David Broder.
And in a later update, Ron Hira, an assistant professor of public policy (on leave) at the Rochester Institute of Technology, explains how/why he culled the data used in Oak’s post:
The source of the data is the U.S. Department of Labor’s Employment & Training Administration. They publish the greencard (“Permanent”) case data. You can download it here. I used the FY04 data because that was all that was available last week when I ran the analysis. I noticed, while writing this e-mail, that they have the FY06 data currently available.
One could use the H-1B LCA database, available here.
The reason I used the greencard data is that the LCA H-1B data is sometimes criticized because they are applications rather than actual people, and the wages listed are not necessarily the actual wages paid. In the case of the greencards, the actual wage must be specified, and as Norm Matloff has pointed out it is a conservative (in favor of Microsoft) way to calculate the wages being paid.
So, I don’t think the objection from the DailyKos contributor has a whole lot of substance.
Plus, keep in mind that if you run the data with the LCA H-1B data, then only about 11% of the workers are paid $100K or more. So Gates is still misrepresenting how his company is using the H-1B program, which was the whole point of the post in the first place.
Even later update … OK, now Matloff and Hira are engaging in one of those “academic discussions” that nearly drove me to quit college. I’m getting out of the way. Here’s Matloff:
Ron Hira wrote: “The reason I used the greencard data is that the LCA H-1B data is sometimes criticized because they are applications rather than actual people, and the wages listed are not necessarily the actual wages paid. In the case of the greencards, the actual wage must be specified, and as Norm Matloff has pointed out it is a conservative (in favor of Microsoft) way to calculate the wages being paid.”
Matloff’s reply: This is not quite correct. Actually, it won’t matter, for the reason I’ll give in the next paragraph, but still it should be corrected. The H-1B form actually does require the employer to specify the actual wage, in the column labeled Rate of Pay.
However, as Ron says, an LCA is merely an application for the employer to hire some foreign worker, possibly in the future, and need not correspond to an actual person. Nevertheless, in most cases the application is for an actual person, and more importantly, those values in the Rate of Pay column do track quite well with the USCIS’ own data. (As Ron pointed out, the USCIS refuses to release individual records corresponding to the LCAs, but they do release the macro data.) John Miano found that the LCAs indicated that 85% of the computer-related H-1Bs were paid less than the overall median for their occupations.
My analysis, using the published aggregate USCIS data for 2001, indicated that the figure was around 90%. So, speaking as a statistician, the LCA data is reliable. Moreover, the wages filled out by the employers in the Prevailing Wage column of the LCAs state what the employers consider prevailing wage, which is quite telling, as they are far below the market medians.
Please note that the employers are generally NOT breaking the law here.The problem is NOT one of enforcement of the law. Instead, the problem is that the law is riddled with gaping loopholes, which the employers make aggressive use of of (just like they do with the loopholes in the tax code). And it must be kept in mind that the employers and the AILA are the ones who got those loopholes into the law in the first place.
Norm
