One big problem with empirical economics is that some disputes never get resolved.
If two physicists disagree about the mass of the electron, they can do an essentially infinite number of experiments to see who’s right, and eventually the undeniable truth will emerge.
In the social sciences, though, most evidence can’t be produced at will — it comes to us at random, from out in the big wide world. That means that when two economists differ about something like the labor market impact of immigration, sometimes the only way to settle the issue is to argue about who’s empirical methods are superior.
That sort of squabble has been going on for about two years now, between a bunch of economists who think immigration isn’t very harmful to native U.S. workers, and one very famous economist who thinks the negative effects are large.
The dissenter is no slouch; it’s George Borjas, a professor at Harvard University’s Kennedy School of Government (and a Cuban immigrant himself), who has been described by major publications as “America’s leading immigration economist” and “the pre-eminent scholar in his field.”
Of all economists who study the immigration issue, Borjas is perhaps the only one capable of holding his own against the combined might of the others.
The argument is over a 1990 paper by celebrated labor economist David Card. Card studied the 1980 Mariel boatlift, in which Fidel Castro suddenly decided to let about 125,000 Cubans emigrate to the U.S.
Most of them ended up in Miami. Card, making the assumption that the whim of a dictator like Castro is unlikely to be affected by labor market conditions in Miami, realized that this presented a perfect “natural experiment” with which to study how immigration affects native-born workers.
Surprisingly, Card found that the surge of immigrants had no effect on wages or employment levels for low-skilled Miamians.
That result, and many subsequent studies reaching the same conclusion, changed the way most economists think about the costs and benefits of immigration.
Borjas, an inveterate immigration skeptic, disagrees. In 2015 he wrote a paper claiming that if you look carefully enough, you can see a big harmful impact of the Mariel immigration wave on the wages of a certain group of native-born Miamian high-school dropouts.
Borjas’ paper immediately drew fire from other immigration researchers. Card sniffed that Borjas was “flailing around.”
Giovanni Peri and Vasil Yasenov of the University of California-Davis noted that by limiting his sample to male high-school dropouts between the ages of 25 and 59, and excluding non-Cuban Hispanics, Borjas had limited himself to a sample of just 17 to 25 observations.
That’s probably too small of a sample to draw strong statistical conclusions, since a small amount of random noise can easily create spurious effects.
Peri and Yasenov found that even slight changes to the set of workers being measured eliminates the negative effect that Borjas claimed to find. A 2017 follow-up paper by Peri and Yasenov, using more sophisticated statistical methods, concluded that Borjas’ result was due to measurement error.
Other studies also found the opposite of what Borjas claimed.
Recently, Michael Clemens of the Center for Global Development and Jennifer Hunt of Rutgers University found an even bigger problem with Borjas’ study.
Clemens and Hunt noted that in 1980, the same year as the Mariel boatlift, the U.S. Census Bureau changed its methods for counting black men with low levels of education.
The workers that Borjas finds were hurt by the Mariel immigration include these black men.
But because these workers generally have lower wages than those the Census had counted before, Borjas’ finding of a wage drop among this group, the authors claim, was almost certainly a result of the change in measurement.
Borjas dismissed the contrary findings, suggesting they were motivated by the political bias of the Silicon Valley philanthropists who contribute to the think tank where Clemens works.
He also argues that the low-wage workers whom he claims were hurt by the Mariel immigration weren’t just black workers, and that low-skilled white workers might have moved out of Miami to avoid having to compete with the new Cuban arrivals.
This acrimonious debate is unlikely to be resolved soon. But stepping back, a larger pattern becomes clear.
A number of studies of refugee waves have found little or no impact on the employment of native-born workers in the cities where the refugees go.
These small or zero effects hold whether the refugees go to rich countries, poor countries, or simply to neighboring countries.
In 2005, Card found little or no harm from Mexican immigration to the U.S. Some studies even find that immigration raises native-born wages, by prompting locals to go back to school and improve their skills.
So overall, the weight of evidence is solidly against Borjas on the immigration question. Borjas has written many papers showing harmful impacts of immigration, but an overwhelming majority of researchers in the field have found the opposite.
Borjas might still be right. But despite his fame, high position and voluminous output, it’s getting harder to side with him against his many opponents.
The ongoing dispute highlights the unfortunate fact that in empirical social science, long-running arguments rarely get definitively resolved — so don’t expect Borjas or his critics to ever surrender. But the weight and volume of evidence must count for something, and the emerging consensus is that immigration just doesn’t hurt native-born workers very much.
Noah Smith is a Bloomberg View columnist. He was an assistant professor of finance at Stony Brook University, and he blogs at Noahpinion.