Doing no harm.

Photographer: Alan Oxley/Getty Images

An Immigrant Won't Steal Your Raise

Noah Smith is a Bloomberg View columnist. He was an assistant professor of finance at Stony Brook University, and he blogs at Noahpinion.
Read More.
a | A

Immigration is one of the most contentious topics in American politics. It’s also a case study in how empirical economics is coming to dominate public policy debates. Instead of theories about how immigration should affect labor markets, people are turning to the evidence. And the most powerful evidence we have comes in the form of quasi-experimental studies. 

The most important and widely cited such study is a 1990 paper by economist David Card of the University of  California-Berkeley. Card studied the impact of the Mariel boatlift, in which Fidel Castro suddenly sent thousands of immigrants to the U.S. in 1980. Most of those immigrants stayed in the Miami area. Standard Econ 101 theory says that a big increase in labor supply should reduce wages for local workers, especially for those who are in direct competition with immigrants. Since most of the Cubans who came in 1980 had little education, we would expect low-skilled Miamians to suffer the biggest wage drops from the labor shock. We might also expect the Mariel boatlift to raise unemployment levels, especially for less-educated Miamians.
QuickTake Skilled Immigrants 

But Card found something startling: the negative impact on native Miamians was negligible. Neither wages nor employment fell by a measurable amount. Card’s result was the first of a series of papers showing that immigration doesn’t do much -- if any -- harm to native-born workers. 

In order to guage the impact of the Mariel boatlift, however, Card had to measure how well Miami did relative to other cities. The early '80s were a bad time for the U.S. economy and labor markets in general, since that was when Paul Volcker, the Federal Reserve chairman at the time, was raising interest rates in order to combat inflation. Card chose four comparison metropolitan areas -- Atlanta, Houston, Los Angeles, and Tampa-St. Petersburg, Florida -- based on similar demographics and similar growth rates during the late 1970s and early 1980s. 

But in 2015, George Borjas of Harvard University’s Kennedy School came out with a shocking claim -- the celebrated Card result, he declared, was completely wrong.  Borjas chose a different set of comparison metro areas -- Anaheim, San Jose and Anaheim in California, and Rochester and Nassau-Suffolk counties in New York -- that had employment growth trends similar to Miami's before 1980. He also focused on a very specific subset of low-skilled Miami workers. Unlike Card, Borjas found that the Mariel boatlift immigration surge had a big negative effect on native wages for this vulnerable subgroup. 

Now, in relatively short order, Borjas’ startling claim has been effectively debunked. Giovanni Peri and Vasil Yasenov, in a new National Bureau of Economic Research working paper, use a much more robust comparison method than either Card or Borjas. This approach is called synthetic controls, and involves using data on real cities to construct imaginary cities that were extremely similar to Miami in 1980. Peri and Yasenov find that Card’s result, not Borjas’, was based on the more robust set of comparisons. 

Even more damning, Peri and Yasenov find that Borjas only got the result that he did by choosing a very narrow, specific set of Miami workers. Borjas ignores young workers and non-Cuban Hispanics -- two groups of workers who should have been among the most affected by competition from the Mariel immigrants. When these workers are added back in, the negative impact that Borjas finds disappears. 

But it gets worse. Borjas was so careful in choosing his arbitrary comparison group that his sample of Miami workers was extremely tiny -- only 17 to 25 workers in total. That is way too small of a sample size to draw reliable conclusions. Peri and Yasenov find that when the sample is expanded from this tiny group, the supposed negative effect of immigration vanishes. 

All of this leaves Borjas’ result looking very fishy. He would have had to have searched hard to find the one small group of workers who seemed to suffer from the Mariel influx. Borjas could well have been subject to heavy confirmation bias -- he might have been so fundamentally certain that immigration was bad for native workers that he searched and searched until he found one group that seemed to confirm his pre-existing beliefs. In science terms, that is called data mining;  it's a big no-no. 

In debates about immigration, the anti-immigrant side inevitably cites Borjas. He has gained fame and notoriety for being the most prestigious economist who thinks that immigration is a disaster for native workers. All of Borjas’ papers seem to arrive at this same conclusion. Participants in immigration debates really should stop citing Borjas’ research so much.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

To contact the author of this story:
Noah Smith at nsmith150@bloomberg.net

To contact the editor responsible for this story:
James Greiff at jgreiff@bloomberg.net