A Tale of Two Studies

Academics and professional economists have critiqued many well known academic papers on immigration in the last year. The first was by Alan de Brauw and Joseph R.D. Russell and it replicates and expands a famous 2003 paper by Harvard University economist George Borjas entitled “The Labor Demand Curve is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market.” 

Borjas famously found that from 1960-2000 there was a  wage elasticity of -0.38, meaning that a 10 percent increase in the size of the labor force due to immigration in a particular skill-cell lowered the average weekly wages in that cell by 3.8 percent relative to workers in other skill-cells.  Borjas’ paper is an impressive piece of scholarship and has been the lynchpin of arguments to close the border in order to protect wages.  Many economists disagree with Borjas

De Brauw and Russell had three findings.  Their first finding was that the wage elasticity dropped to -0.22 when they extended Borjas’ study to 2010. That is an important finding by itself – if the Borjas model was correct then why would the impact of immigrants on wages decrease as more of them entered the labor force between 2000 and 2010? 

Their second set of findings is that small changes in variable definitions turned some of Borjas’ ideas into statistically insignificant results. While not definitive, that suggests that the conclusions in his paper are not reliable.     

That leads to De Brauw and Russell’s third set of findings. They looked at the relationship between annualized male and female wages in the skill-cells when women entered the workforce in significant numbers. The correlation turned out to be positive­, which means men and women with the same skill level are complementary.  Thus, they argued that Borjas’ model is misspecified as it assumed immigrants and natives in the same skill-cells are more substitutable than they really are. If this finding is true, it would call into question the assumptions Borjas’ built in to his model, namely that immigration and natives are substitutable rather than complementary.

I’m still eagerly awaiting Borjas’ response to De Brauw and Russell’s paper. The critique of Borjas’ paper was serious because it replicated his work, extended it another decade, and found the results didn’t hold up. Many academics have already contested Borjas’ claims in numerous ways as I document here and here but this challenge cuts deep.

The second important working paper of the last year is by George Borjas and it critiques an important paper by David Card called “The Impact of the Mariel Boatlift on the Miami Labor Market.”  Card’s paper studied the infamous 1980 Mariel boatlift that resulted in 125,000 Cubans arriving in Miami in a very short period of time – the type of exogenous shock that economists love to test as they have for Portugal, France, and Israel.

The Mariel flow increased Miami’s workforce by 7 percent in 3 months– 28 percent in a single year if it had continued at the same pace.  The Mariel boatlift was an extreme event that can help economists understand the wage effects of immigration and the elasticities in the labor market. But it’s also important to point out that in recent years the annual flow of immigrants to the United States is equal to about 0.3 percent of the U.S. population – two orders of magnitude smaller than what happened to Miami.

Card compared Miami wages prior to the boatlift and afterward with a set of cities that did not experience such a shock, concluding that Mariel lowered the relative wages of lesser-skilled Miami workers slightly before they rebounded entirely around 1984. The finding that a massive influx in lower-skilled workers did not affect wages has been an understandably important contribution to the literature.  

Borjas’ critique of the Card paper changes all of the treatment cities and control groups. Fair enough, maybe Card selected the wrong cities to begin with and Borjas selected four comparison cities that would better approximate Miami’s labor market without the Mariel boatlift. By changing the placebo, Borjas found that wages for less-skilled Miamians dropped by 10 to 30 percentage points due to the Mariel boatlift but fully recovered by the late 1980s. If Borjas had selected a better group of comparison cities than Card did, then this would be a very interesting finding indeed, but the relatively quick recovery in wages is still remarkable.      

Interestingly, using Borjas’ way of measuring the economic costs and benefits of immigration, this larger negative wage effect could actually increase the economy-wide benefits from immigration. The size of the wage decrease for some workers increases the economic surplus for other workers and capital owners – meaning that Borjas’ earlier estimates of the economy-wide benefits could be several times too small.

Neither Borjas nor Card have responded to the critiques of their classic papers – yet. Regardless of how this debate turns out, this is a great time to watch several excellent scholars and academics debate this important topic. Stay tuned.