The unexpected twists in H1-B visa data
Two items in the news last week caught my attention.
Spectrum has an article about a new on-line tool that makes sense of the data the government releases about H1-B visas.
The tool is here, and it’s rather interesting to explore. (Late update: as of Monday morning 6-8-15 the tool is down. Hopefully it will be back soon.)
Netflix pays really well. I mean, really well. They have ten H1-B “senior software engineers” making $215k each, in an area where the prevailing wage for that job is $144K. Maybe you have a hankering to be a “senior performance architect?” With an H-1B they’ll knight you with $350k/yr, which is much more than twice the prevailing wage for that job in that region. For nearly all of the jobs listed, the H1-Bs make more than the usual wage — often much more.
With all of that high-priced talent, why is their streaming service so poor?
In the embedded world results are all over the place. Altera basically doesn’t use H1-Bs. Neither does Intel, which bodes well for Altera’s new place as part of that semiconductor giant. Their big competitor, Xilinx, bristles with them, but always pays more than the prevailing wage.
Freescale has a fair number of H1-B workers and pays about the same wage they’d offer a native worker. TI has hired over 2500 H1-B workers over the last 16 years and pays them fairly. (Did you know the prevailing wage for an electronics engineering technician in South Portland, Maine, is $26K? That seems ludicrous. TI pays the H1-B worker they hired for that position over twice that.)
The law requires that H1-B workers be paid at least the prevailing wage for a similar job description in the same location. My understanding is that this wage is determined by the company itself. Obviously, this is ripe for abuses. I was struck though, in the data I looked at, that immigrant engineers in this country are routinely paid well — nearly all above $80k and a very big chunk over $100k. It’s hard to be quantitative because the job descriptions are vague: Does digital design engineer imply someone with two or twenty years’ experience? That unknown can wildly skew the data.
So clearly the H1-B debate is over. Companies aren’t hiring them to replace expensive Americans.
Then there’s this piece in the New York Times:
Disney let 250 employees go from their fun-filled family-oriented park in Orlando and replaced them with H1-B workers. In a final humiliation many of those who lost their jobs had to train their replacements. The article claims the “savings” were 25 to 49%. That’s a little disingenuous when thinking about salaries since the new workers were contractors to outfits like Tata. The body shops will take a percentage, so if the company saves, say 49%, the H1-B worker is getting less than half of what the American he replaced earned.
Feeding Disney through the tool gave difficult-to-interpret results. 63 different Disney companies showed up. Looking at a couple of them that have hired relatively lots of people on H1-B visas showed that pay scales are at or above prevailing wages and generally appear reasonable.
Perhaps, though, the data is masked by the use of a contractor. Running Tata Consultancy Services Limited though the tool evoked over 9000 H1-B visas this year alone with six months to go. That’s a sizeable fraction of the 85k visas issued each year. There’s far too much data to look at, but what I did examine showed salaries just about equal to prevailing wages — or, to be more specific, to the prevailing wage listed on the visa application.
Computer programmers make $58k in Jacksonville? $75k in Sunnyvale? Smells fishy.
The following graph from the tool shows the number of visa applications versus salary for all of Tata’s “Computer Programmers” throughout the USA. The horizontal axis is salary, but is clipped. “620” means 62k/year. Unless Tata only hires new graduates with no experience there’s no way a reasonable person can believe that the prevailing wage they list is anything but bogus.
I don’t know all of the nuances of the H1-B program. But the graph shows that something is rotten in the system.
Do you have experience with the H1-B program? What’s your take on it?
Jack G. Ganssle is a lecturer and consultant on embedded development issues. He conducts seminars on embedded systems and helps companies with their embedded challenges, and works as an expert witness on embedded issues. Contact him at firstname.lastname@example.org. His website is www.ganssle.com.