Immigrants cluster in a small number of cities. A third live in three metropolitan areas (New York, Los Angeles and Chicago). In the past, the stereotypical study exploited this clustering by correlating wages and immigration across cities. A negative correlation, indicating that wages are lower in cities penetrated by immigrants, would suggest that immigrants reduce the wage of competing workers. In fact, the estimated correlations bunched around zero, creating the impression that immigrants had little impact.
This inference is not correct for two reasons. First, immigrants are not randomly distributed across cities. If, as seems sensible, high-wage areas attract immigrants, there would be a spurious positive correlation between immigration and wages. This positive correlation could easily swamp any negative wage effect that immigrants might have had.
Natives also respond to immigration. Employers in Michigan, for example, see that Southern California cities flooded by low-skill immigrants pay lower wages to laborers. The employers will want to relocate to those cities to increase their profits. The flow of jobs to the immigrant-hit areas cushions the adverse effect of immigration on the Southern California wage, while slightly worsening conditions in Michigan. Similarly, laborers living in California, who now face stiffer competition, might leave the state to search for better opportunities. These flows of jobs and workers diffuse the impact of immigration across the national economy and tend to equalize economic conditions across localities.
From this, Borjas posits that immigration is better measured on the national level. That's sensible enough for wages, although there are lots of analyses to be done on the local level to show that immigration harms the well-being of the native working and underclasses (e.g. home ownership rates, affordability, school test scores, etc.
Sticking with the data at the national level and looking at how immigrant jobs effect natives with similar skill-levels, Borjas finds that from 1980 to 2000 immigration lowered the wages of high school drop-outs 5% over the long-term (its effects were more detrimental in the short-run). For other groups, immigrants had a marginal effect on wages (people who had completed high school but not college benefited slightly; natives with college degrees suffered slightly). He concludes that immigration has resulted in a wealth transfer from labor to capital owners, with a question mark on the total economic utility realized in the US.
It makes sense that low-end earners suffer more from immigration than high-end earners, at least on the service side of the economy. Globalization means that Microsoft, Pfizer, or KPMG seek out professionals wherever they reside. If KPMG needs to hire ten accountants to run numbers for managerial accounting consultation, it can pull them from anywhere in the world. If the five chosen who do not happen to be American natives are living in the US or in India at the time of hire is not going to significantly effect the compensation offered to the five natives who are picked. Having the five Indians in the US likely actually boosts the pay for natives because the Indians become more costly to hire than if they were in Bangalore.
Menial service jobs, however, are much more geographically contingent. Dishwashers, maids, and yard workers have to be in close proximity to the entity they are doing the work for. If the non-native laborers are in Mexico City, they can't cut lawns in Los Angeles. Angelinos desiring lawncare will have to look locally. Thus the wages of native yard workers will be directly effected by the location of non-native yard workers.
In other words, higher-end work can increasingly be done irrespective of the worker's physical location. American-born programmers (or researchers, writers, etc) compete with Indian-born programmers whether the Indians live in the US or in India. Low-end service work depends more on where the workers are. American-born yard workers (or burger flippers, janitors, construction workers, etc) only compete with Mexican-born yard workers if the Mexicans are in the US.
This is why I am less concerned by an increase in highly-skilled immigrants than by low-skilled immigrants. The US standard of living is a competitive advantage America can leverage to attract brains. Better to have a Japanese-born researcher (who brings few social problems like crime, disease, school deterioration, etc) become a US citizen to create wealth inside the country than have a multinational pay him in Japan. Or at least the first scenario is preferable to bringing in unskilled labor that is going to suddenly compete with unskilled natives and depress wages, consume services that natives pay for, and bring other pathologies. Whether or not the first scenario is preferable to no immigration at all is a tough one. But encouraging the second situation certainly makes no sense to me.