Big Picture Can Distort Immigration Policy Research

By David North on June 9, 2010

As the amnesty/legalization debate heats up, there will be many a research report on the subject that suggests the impact of immigration on the rest of us is pretty bland.

One type of research that produces these seemingly soothing results was on display on June 7 at a seminar sponsored by the Migration Policy Institute in Washington. Giovanni Peri, a professor at UC-Davis, used regression analysis to examine census data in a report entitled "The Impact of Immigrants in Recession and Economic Expansion," He concluded:

Immigration unambiguously improves employment, productivity, and income, but this involves adjustments. These adjustments are more difficult during downturns, suggesting the United States would benefit most from immigration that adjusts to economic conditions.

My principal reaction to such research is to compare it to George Bush flying over New Orleans in Air Force One at, say, 20,000 feet for his first look at Katrina. If you are distant enough from the reality, it does not look very troubling.

Similarly, Peri's analysis is based on national data and national averages. The problem is that an unemployed, or badly paid, resident worker, who is hurt by immigration is not an average; he or she is an individual who lives and works in a particular place. The economic impact of immigration may be bland in Maine and devastating in Houston, but the problems in Houston are totally lost in the national over-view.

Further, because he was dealing with census data (and social scientists choose the data sets they will use) Peri could not parse out the differential impact of legal or illegal aliens, or those on temporary visas or permanent ones. When they are all blended together, the impacts are considerably less visible than when you check on what illegal aliens do to the agricultural labor market in a particular place, or what H-1B beneficiaries do to wages and working conditions in a particular software shop.

The problems with the impacts of immigration on the labor market, particularly the problems caused by migrants with limited bargaining powers (such as temporary workers and illegals) can best be examined by looking at what happens when a bunch of them congregate in a particular labor market, and distort it.

And the resident workers who do lose out are not the powerful who make laws or write research papers; they are people that you do not encounter in college reunions or suburban cocktail parties. Much of the time they do not know that international migration may be hurting them, and even more of the time they are silent and invisible to the establishment.

As a further, (probably unconscious) contribution to the blanding of the research, Peri uses regression analysis which causes economists to sound like the Latin-speaking clergy of old; using an elite language, thus making it hard for non-initiates to question the speaker. He is, of course, not the only professor to use this scholarly approach.

In one of his tables using regression analysis there are 64 cells, with the numbers in them ranging from -0.57 to +1.50; not the stuff of either drama or clarity for the layman, but probably of interest to the cognoscenti.

Peri's talk did have one saving grace, however. He raised the issue of the very real subsidy to corporations flowing out of the H-1B nonimmigrant worker program. He suggested one way to reduce the subsidy and secure some economic data – a twofer for an economist – was to distribute these visas through a public auction process. The government would get the auction proceeds and might use the data to decide how many H-1B visas to issue during the next round.

I like the idea, but surely Microsoft and Goldman Sachs will not.

Peri also made the useful point (i.e., I agree with him) that, while levels of nonimmigrant worker visas and the numbers of illegal aliens in the nation vary sharply with economic activity, the state of the economy has no impact whatsoever on the number of green cards issued in a given year, at least in recent decades.