I actually am a subscriber to The Atlantic, and might even say it is one of my many sources of enjoyment in life. But good grief, it would be hard to find a bunch of journalists who are more disconnected from the way ordinary people live. So, for the second day in a row, I must cite an example of the shoddy work they sometimes do, and their disturbing bias. In turn, the article cites research that itself is demonstrably shoddy and biased.
Specifically, there is the November 30 article, “Not All Immigrant Labor Is Cheap Labor,” by Alexia Hernandez Campbell. Right off the bat, note that this is a funded article, essentially an infomercial. Can it be that such an august, 159-year-old publication such as The Atlantic runs infomercials? Sad but true. And the sponsor, the Emerson Collective, was founded by Steve Jobs’ widow, which might play some role here.
But that pales in comparison to the problems with the main research on which Ms. Campbell bases her piece, an NBER report by Gordon Hanson of UCSD and Matthew Slaughter of Dartmouth. This pair has produced similar research in the past, explicitly funded by the industry lobbying group CompeteAmerica. Needless to say in view of such bias, that old study was very badly flawed. But let’s take a look at this new one.
Best to start out with a laugher. As anyone who reads this blog or my other work knows, I believe there is far too much emphasis in the national discussion on H-1B on the “Infosyses,” and I contend that the “Intels” are just as culpable. Yet Hanson and Slaughter say that my 2013 EPI paper was about the Infosyses! It of course was not on that topic, and on the contrary, was an analysis of the fact that the H-1Bs who are former foreign students are generally of poorer quality than their American peers. They also misquote the findings of the 2010 paper by Kerr and Lincoln, overlook the NBER version of the 2011 Hunt paper and so on.
The authors’ current paper seems to have been put together hastily. There are typos, the bibliography is a mess, etc.
Most of their paper consists of umpteen graphs and tables showing that there are a lot of immigrants in STEM, which is hardly news. But their implication that this means that those immigrants were needed is of course illogical. The only justification they seem to offer is the only-moderate international test scores for U.S. 15-year-olds. I think most 15-year-olds could see through that argument too.
Hanson and Slaughter also seem to be making the “Johnny doesn’t want to go to grad school” argument, ignoring the fact that the H-1B program has made grad school unattractive to Americans. This effect was correctly forecast by the National Science Foundation, something the authors would know if they had actually read my EPI paper rather than misquoting it.
The authors do make a pretty good attempt to answer the question as to whether immigrants in STEM are paid less than Americans. In their regression analyses, they do adjust for a number of important covariates, but unfortunately not the most important of all, occupation. So, they are lumping together well-paid software engineers with poorly-paid life science lab workers, and given the prevalence of the immigrants in the software area, we clearly have a real statistical problem here. Another problem is that they are mixing together immigrants with green card or citizenship status with those who are H-1Bs, L-1s and so, who are exploitable. Nevertheless, they do find that the immigrants earn less. They also find that the immigrants do catch up, which the authors attribute to assimilation while I would attribute to the mobility the immigrants attain once they get their green cards.
The most salient problem in that analysis, though, is the authors’ failure (and that of many others) to account for AGE in the cheap-labor issue. As I have emphasized over the years, the biggest wage savings in hiring foreigners is that the latter are young. Young people are cheaper than older ones, and H-1B is largely about hiring young H-1Bs in lieu of older (35+) Americans.
Though the authors don’t make this connection, they do have a number of graphs showing that the immigrant STEM workers do tend to be young. And they define “young” to be age 25-34, just like I do. Maybe they read my EPI paper after all. 🙂