Some years ago I knew a researcher whose nonprofit organization employer had gotten a grant to study how HR works in Silicon Valley. It was arranged that the researcher would act as a “fly on the wall” in the HR office at a major firm, Cisco if I recall correctly. Unfortunately, her employer folded, and the study never took place.
For you researchers and policy makers out there, let me tell you–you would have been profoundly shocked at the results had the project gone through. It would be clear in a very concrete way why the tech industry’s claims to be unable to find qualified workers is a joke. Well, I’ll present to you here a weak but I believe powerful substitute for that ill-fated Cisco study。
Over the weekend the Baltimore Sun ran a piece on the claimed tech labor shortage, and the claimed need to remedy it by expanding the H-1B work visa program. It ran true to the usual formula: Give examples of local employers who turn to the H-1B program out of desperation when they can’t find qualified U.S. citizens or permanent residents; quote “experts” stating that the current H-1B cap is woefully inadequate; and present a couple of token quotes from some naysayers. Not much new so far, but let’s take a closer look at the article’s poster firm, Merkle, Inc., which “builds more than 1,000 complex computer models each year for marketing clients who want to know more about their customers’ behaviors and preferences.”
Merkle is an excellent example for a change of pace from my usual analyses, in that it’s not a software company. Instead, it specializes in data science and analytics, fancy new terms referring really to old statistical methodology applied to huge data sets (“Big Data”). This is a currently hot area, and for instance I would estimate that over half the startup tech firms in San Francisco’s trendy South of Market (SOMA) area are in this business.
Those who are well familiar with my research know what telltale signs to look for first in a company–a hiring focus on the young. The First Central Employer Motivation for hiring H-1B visa workers is to cut costs by employing young H-1Bs in lieu of older (35+) Americans, as the latter cost more in both wages and benefits. (Hiring young H-1Bs is cheaper than hiring young Americans too, but the age aspect is where the big savings are to be made.)
Sure enough, Merkle’s Careers Web page has openings for Junior Analyst, Junior Data Scientist and so on, and one of the positions as a “Senior” Analyst requires only two years of experience! And the firm can’t even give the usual software firm’s excuse that it hires young “because the technology is changing so fast” (a red herring anyway)–because it isn’t. Most of the methodology is well over 20 years old. There are a couple of new tools for implementing them, and one or two new-ish methods, but the pace of change is certainly much slower than in the software world.
Now go a level deeper, by reading Merkle employee comments on Glassdoor. No, not a scientific sample, but telling nonetheless, especially since most of the commenters say they would recommend the company to a friend; these are not malcontents. Yet there is a consistent theme in the comments that the firm has an alarmingly high turnover, e.g.:
“…they struggle to retain talent”; “The turn-over rate is way too high, and most people live[sic] for more money”, “The turnover rate is quite high”, “Pathetic pay”, “the great talent mentioned above gets tired and leaves
-I’d estimate at least 80% of the employees have been there less than 2 years. Think about that turnover for a minute with almost 2000 employees”; “no promotion, no real bump in salary”; “Poor attrition rate, due to burn-out or dissatisfaction, both really common with most exits”; “Incessant burn out pace- good people leave for lack of any sense of accomplishment or gratitude”; “Lots of turnover – job burnout – aggressive delivery timelines – long hours. Projects often were disrupted because someone left the company”; “Burn n churn”
That ties right in with the Second Central Employer Motivation for hiring H-1Bs: A desire for immobile labor. Unfortunately, this aspect is almost never mentioned in debates about the foreign tech worker programs, and indeed is not widely known even by researchers. But to many employers, especially the newer ones, it is of the utmost importance. Clearly, Merkle, with its horrendous turnover problem, would find the “solution” of hiring immobile foreign workers irresistible.
Immigration lawyers openly pitch this to employers. By sponsoring a foreign worker for a green card, the employer in essence renders the worker immobile, trapped there for years. I cited several public statements by lawyers in my University of Michigan Journal of Law Reform article, and more recently have been quoting David Swain’s document. (Swain, it should be noted, writes that he was the architect of the immigration policy for Texas Instruments, one of the most active firms promoting expansion of the H-1B visa.)
But even better is Swain’s newest, pitching specifically to employers who wish to recruit foreign students (F-1 visa category). Imagine how appealing a retention-challenged company like Merkle would find the following statements, taken from that new document:
“Most college graduates leave in less than two years. F-1 students who
want permanent residence must stay seven to twelve years…In most cases
the employee is required to wait under an extensive quota system which
can be anywhere from five to ten years, depending on the position and
the employee’s place of birth. It is this step in the process which
allows the employer to maximize the retention advantages of hiring
international students…[Concerning the cost of green card
sponsorship:] Since the legal system requires the international employee
to keep the same employment for seven to twelve years, the ‘cost’ [the
lawyer says $8,000] of that employee should be viewed in the greater
context of the value that employee brings to the company…minor
compared to the overall compensation.”
(Swain adds that, of course, the employer has no obligation to keep the worker for “seven to twelve years.” )
Again, this issue is not widely known, but it is one of the two central employer motivators for hiring H-1Bs. It is de facto indentured servitude, highly abusive, all perfectly legal, and has a clear adverse impact on Americans: If an American and a foreign student of equal quality apply for a job, the above considerations would give the foreign student a major advantage over the American.
Now what about the type of H-1B abuse discussed more often, the payment of lower wages to an H-1B than to an equally-qualified American? To those who don’t know the nature of the jobs, the PERM data for Merkle would appear to indicate that in some cases the firm is paying its foreign workers quite generously, well above the legally-required prevailing wage. For instance, here is one of Merkle’s PERM records from 2011:
A-11187-90670,09/27/11 15:25:25,Certified,PERM,MERKLE INC.,7001 COLUMBIA GATEWAY DRIVE,,COLUMBIA,MD,21046,5416,Management, Scientific, and Technical Consulting Services,Other Economic Sector,15-2041.00,Statisticians,Statisticians,Level I,OES,56243,yr,85000,,yr,Columbia,MD,CHINA,H-1B
The prevailing wage is $56,243 while the offered wage is $85,000. (Actually, the offered wage may be less than this, due to a loophole I won’t go into here.)
I’ve explained before why the legal prevailing wage is typically substantially below the market value the foreign worker would have if he/she were a free agent. But in the case of data analytics jobs like those at Merkle, the inadequacy of prevailing wage is especially acute. Here’s why:
The government job category often used in Merkle’s PERM records is 15-2041.00, Statisticians. As mentioned above, this category is technically right, as the workers are using statistical methodology. But those in data analytics are perceived as knowing far more, such as use of the cloud programming language Hadoop. (And Merkle’s job ads do ask for experience in Hadoop and similar tools.) So, the going rate for data science is $90,000 on the low end, and goes much higher than that. See for instance the O’Reilly survey.
In other words, the 15-2041.00, Statisticians wage category is not the right one for Merkle’s foreign workers at all. The prevailing wage figures Merkle is using are for, say, a Bachelor’s graduate with a Statistics degree who is doing small-data statistical work at, say, a small bio lab. So no, the evidence does not show that Merkle is paying its foreign workers well, and indeed it would appear the opposite is the case, as many of the Merkle PERM records show it is paying only the prevailing wage.
Finally, what about the other industry claim, to hire H-1Bs because they are “the best and the brightest”? Sounds like Merkle would be hiring foreign students from the likes of MIT, Stanford and UC Berkeley, right? Well, a search for Merkle in LinkedIn doesn’t confirm that at all. Of the ones from China that I found, they had attended universities at home such as Tsinghua, Zhejiang, Huadong, Hong Kong Baptist College and the Shanghai University of International Business and Economics. Of these, Tsinghua is first-rate, Huadong and Zhejiang are middling, and the rest are considered weak. The Chinese students had then attended American schools such as Penn, San Francisco State, BYU, Sonoma State, Bentley University, University of Chicago, George Mason, Johns Hopkins, University of Louisville, University of Delaware and so on. So, there is a wide range of selectivity, and with my usual disclaimer that quality of institution does not necessary mean quality of its graduates, we see there is no general pattern here of Merkle hiring “geniuses” from China. The results are similar for India.
Wrapping up: The above information suggests that Merkle, notorious for retention problems, must hire like crazy to fill the jobs vacated by the high turnover, rather than indicating a “labor shortage,” as Merkle claims. Moreover it is likely, I believe from my experience, that Merkle views the indentured nature of foreign workers to be the solution to its retention problems, as opposed to hiring Americans who can and will leave after experiencing the company’s onerous working conditions.
The statistical evidence, notably flat wage trends, shows the claim of a tech labor shortage to be false. But I believe a look at the qualitative side, as we’ve had here for Merkle, yields further understanding of the invalidity of the shortage claims.