Harvard economist George Borjas recently published an important paper on how the unexpected surge of 125,000 Cubans (henceforth Marielitos) to Miami in 1980 lowered the wages of native-born male Miamians with less than a high-school degree. Because at least 60 percent of the Marielitos were high school dropouts, Borjas found that the negative wage effects were concentrated on Miamians with the same level of education.  

There are excellent criticisms of Borjas’ paper that show his results hinge on the control cities he chose, his exclusion of women, the age group of the workers, whether Hispanics are included, whether high-school-or-less or no-high-school-at-all are included, and whether datasets with the larger samples are used. For the sake of argument, supposing that Borjas made the correct methodological choices on every single point above, the Mariel Boatlift still raised the wages for low-skilled U.S. workers collectively due to wage complementarities. That’s because native-born Miamians with only a high school degree (no associate degree, no education after high school) experienced significant wage increases immediately after Mariel relative to workers with the same levels of education in the control groups, or placebos, of other cities. Borjas’ supporters ignore this finding but he does not. 

In his Mariel paper, Borjas reports the wage of high school dropouts relative to high school graduates in Figure 3(C) and the wage of Miami high school graduates across an all cities permutation in Figure 4(B), but he doesn’t have a dramatic graph like this that shows what happened to the relative wages of high school graduates after Mariel. 

Another working paper by Borjas and Monras on the wage effects of refugees also found that “the rate of wage growth for high school graduates, a group whose size was only increased modestly by the Marielitos, is noticeably higher in Miami than outside Miami.”  They go on to write that, “the predominantly low-skill Marielitos … raised the wage of workers with a high school education, and this effect is both numerically and statistically significant. The cross-wage elasticity is about +0.7 [compared to -0.9 for high school dropouts].”  They do not find any employment effects for high school dropouts but they did uncover positive and statistically significant employment gains for those with a high school degree. Furthermore, Figure 7.5 on page 148 of Borjas’ new book We Wanted Workers hints at a wage increase for high school graduates immediately after Mariel.              

My intern Cole Blondin and I followed Borjas’ methods to create graphs for the wages of high school graduates before and after the Boatlift. We used the March Current Population survey (March CPS) and combined the May Current Population Survey and the May Outgoing Rotation Group (May CPS-ORG) datasets. The only differences are that we present the figures in dollars rather than logs, we did not use three-year averages to smooth the data, and we did not recreate the synthetic control. One final note, the wage effect of the Marielitos must be compared to placebos because there was only one Miami in 1980 and we can’t actually observe what would have happened to that city had the Marielitos not arrived. We used the same sets of placebo cities as Borjas.

Even under Borjas’ assumptions, native-born male Miamians with a high school degree or less saw a ­net-wage increase after the Mariel Boatlift.    

March CPS 

Just as Borjas reported, there is a significant drop in the wages for male native-born Miamians with less than a high school degree who were 25 to 59 years old after 1980 relative to workers with the same characteristics in the placebos (Figure 1).  By contrast, the wages for Miamians with only a high school degree also increased after the Mariel Boatlift (Figure 2). The wage changes in both figures are statistically significant relative to the placebos. Miamians with just a high school degree are complementary to high school dropouts. Figure 3 shows the wages for high school dropouts and high school graduates in Miami.       

Figure 1

High School Dropouts, March CPS

 

Figure 2

High School Only Graduates, March CPS

 

Figure 3

High School Dropout and High School Only Graduate Wages 1977-2002, March CPS

 

Relative wages for high school dropouts reached their nadir in 1985 and 1986 while the wages for workers with only a high school degree reached their peak in 1984. A mere 19 percent of Miami workers in the March CPS sample had less than a high school degree but 36 percent had only a high school degree which means that more low-skilled American workers experienced a wage gain than a wage decline after Mariel.  

We compare the pre- and post-Mariel Miami wages, weighted by the relative size of the native educational cohorts in the March CPS, to produce a rough estimate of Mariel’s net impact on native wages.  For instance, if there are 10 high school dropouts who each lose $1 and 20 high school only graduates who each gained $0.60, then the net wages earned by Americans increased by $2 [(20*$0.60)-(10*$1)]. In such a scenario, total wages rise for Americans who are high school dropouts and high school graduates. 

We compared the post-Mariel period of 1981 to 1986 to the pre-Mariel period of 1976-1979. The total wage gain for high school graduates outweighs the total losses for high school dropouts in the “All Cities” placebo (the placebo that produced Borjas’ stunning graph). The Card and Low-Skill placebos also returned slightly net wage losses while the employment placebo returned a deeply negative result (Table 1).  The average for all the placebos from the March CPS data was -5.85.       

Table 1

Net Wage Effects, March CPS

March CPS Differences in Differences Changes (1976-1979, 1981-1986)
	Card Cities	Employment Cities	Low-Skill Cities	All Cities
No HS, Wage Changes Times Pop	-9.7804	-21.2793	-10.0078	-9.41663
HS Only, Wage Changes Times Pop	5.471363	4.892769	6.571003	10.13576
Net effect	-4.30904	-16.3865	-3.4368	0.71913

May CPS-ORG

Professor Borjas also used the May CPS-ORG data to test whether Mariel lowered the wages of similarly-skilled Miamians. The May CPS-ORG dataset returned a much smaller wage decrease for high school dropouts in Borjas’ paper. This is important by itself because the May CPS-ORG dataset is superior for two widely-reported reasons: First, the May CPS-ORG contains fewer errors because it asks respondents about their wages last week rather than last year (which is the case for the March CPS). Second, it has a larger sample size than the CPS in every year from 1979 onward.  

Regardless, we confirmed Professor Borjas’ findings that the wages of male high school dropouts in the relevant age range fell after the Mariel Boatlift, although they fell less than in the March CPS data (Figure 4). We also found that the wages for high school graduates in the May CPS-ORG dataset increased relative to the placebo cities after the Mariel Boatlift (Figure 5).

Figure 4

High School Dropouts, May CPS-ORG

 

Figure 5

High School Only Graduates, May CPS-ORG

 

The wage gains for high school graduates outweigh the losses of male high school dropouts in Miami for each placebo group—using the March CPS weight of the population by education (Table 2). The Card Cities, Low-Skill Cities, and All Cities placebos all returned very positive results.  The Employment Cities placebo was the only one that returned a slightly negative result. The average net gain for high school graduates and dropouts across the placebos is a positive +17.43. 

Table 2

Net Wage Effects, May CPS-ORG

May CPS-ORG Differences in Differences Changes (1976-1979, 1981-1986)
	Card Cities	Employment Cities	Low-Skill Cities	All Cities
No HS, Wage Changes Times Pop	-2.20452	-8.33419	-1.67613	-1.91032
HS Only, Wage Changes Times Pop	28.30903	7.618387	25.00548	22.89419
Net effect	26.10452	-0.71581	23.32935	20.98387

To further check our results, we also weighted the wage changes by the May CPS-ORG survey’s estimate of Miami’s education level. The May CPS-ORG survey found that the share of Miami’s native population without a high school education is higher and the percentage with only a high school degree is lower than in the March CPS. Using the May CPS-ORG education weights returns a reduced net positive wage impact for the Card, Low-Skill, and All Cities groups while the Employment Cities group is more negative. Using the May CPS-ORG education estimates, the average is still a positive +10.09.    

Evaluating the Complementary Effects from the March CPS and May CPS-ORG

We ran Borjas’ four different sets of placebo cities on two educational groups of male workers in two different datasets—the March CPS and the May CPS-ORG. There were eight final estimates of the net effects, four in the March CPS and four in the May CPS-ORG.  Four of the eight showed a positive net-wage impact – three in the May CPS-ORG and one in the March CPS.  Four of the eight also showed a negative net-wage impact—three in the March CPS and one in the May CPS-ORG. For each placebo, the positive net-wage gains in the May CPS-ORG results were far larger than the estimated loss in either the May CPS-ORG or the March CPS.

Conclusion

The Mariel Boatlift provides a wonderful natural experiment to test how a sudden, exogenous surge of immigrants affects wages. Professor Borjas’ examination of how Marielitos substituted for native-born Miamians with the same level of education is an important component of that story. However, most have ignored the important complementary effects of the Marielitos on the wages of workers with only a high school degree. A full understanding of Professor Borjas’ contributions to this subject requires acknowledging these complementary effects. To borrow the language used by Professor Borjas, the Marielitos’ redistributed wages from dropouts to workers with only a high school degree with a net positive effect on all low-skill workers.  

Special thanks to Cole Blondin for his excellent work on this blog post.