You’d think the question of whether there is an effect of immigration on native employment opportunities would be an open and shut case. Carefully comparing nations, states, cities and occupations using cross-sectional and quasi-experimental (here, here and here) data has shown there to be no correlation. Nations, states, cities and occupations with higher proportions of immigrants, all else equal, do not have lower native wages.
Houston was one of of the cities Card compared to Miami to see if immigrants had an effect on wages after the Mariel boatlift. At any given moment, there are quite a few investors in Houston that are willing to invest their money somewhere besides Houston. Miami perhaps. If those investors catch wind that several tens of thousands of low-skill workers are going to flood the Miami labor force, they may decide to invest in factories in Miami to take advantage of this new labor supply. Or they may invest in service businesses there to service a new batch of customers. Perhaps some less efficient businesses in Miami, who would have otherwise gone out of business, get new injections of capital. In any case, because capital is mobile, it flows to Miami in short order. What would the impact of this capital mobility be on the Houston labor market and so what would its impact be on what Card saw?
Machines make workers more productive. If you take machines away, workers become less productive. By our normal assumption that productivity equals wages, this means that when capital leaves Houston and goes to Miami, wages in Houston go down. Assume, too, that through the mechanisms of the simple supply and demand story, the Mariels decrease native wages in Miami. So natives have two effects on natives. They directly reduce wages in Miami and they indirectly, through induced capital flows, reduce wages in Houston.
Now Card comes along and compares wage changes between Houston and Miami. He simply subtracts the change in Houston wages before and after the boatlift from the change in Miami wages. He gets a zero and concludes the Mariels had no effect. But that is wrong. He is wrong because he assumes there is no indirect effect of immigrants on native wages.
Capital moves from nation to nation, state to state, city to city and even occupation to occupation relatively easily. Goods move relatively easily too across political borders. Trade, then, is another channel for immigrants to have an indirect effect on native wages because changing trade patterns change the patterns of derived demand for labor. Textile companies in Miami, for example, may find that their production process is cheaper. This reallocates resources towards textile companies in Miami and, because there’s now less demand for Houston produced textiles, away from textile companies in Houston. Textile workers in Houston see a reduction in their wages. The same effect is at play for workers in general (relative to capital) and so trade can cause wages in Houston to decline. Trade is another source of indirect costs of immigration for native workers.
Finally, labor is mobile too, at least within a country’s borders. Like capital mobility, labor movement — geographic or otherwise — can result in equalization of wages across regions and even occupations. Natives in Miami, on the realization of the horde of immigrants thundering upon them from Cuba, may have just lost hope, packed up and moved to other cities. Houston perhaps. Then the increase in labor supply induced by the new Cuban immigrants would seep throughout the country, rising wherever forlorn natives chose to resettle. At the national level, then, analysts would see an effect of immigrants but at the local level no effect could be detected.
And all of the studies we’ve looked at have assumed no indirect effects of immigration. The zero effects found by those studies, then, become negative when indirect effects are added in. Borjas, in his 2003 paper, suggest the only way to account for indirect effects is by looking at relatively closed markets; to look at the nation as a whole.
So its not a slam dunk case. Economics intervenes. People respond to incentives.