Your thoughts on a paper, please?

A few days ago, Marginal Revolution's Tyler Cowen linked to a recent paper by UC Davis researcher Giovanni Peri, "The Effect of Immigration on Productivity: Evidence from US States". This paper is the latest entry into the debate on whether or not immigration depresses wages.

Below is the abstract.

Using the large variation in the inflow of immigrants across US states we analyze the impact of immigration on state employment, average hours worked, physical capital accumulation and, most importantly, total factor productivity and its skill bias. We use the location of a state relative to the Mexican border and to the main ports of entry, as well as the existence of communities of immigrants before 1960, as instruments.

We find no evidence that immigrants crowded-out employment and hours worked by natives. At the same time we find robust evidence that they increased total factor productivity, on the one hand, while they decreased capital intensity and the skill-bias of production technologies, on the other. These results are robust to controlling for several other determinants of productivity that may vary with geography such as R&D spending, computer adoption, international competition in the form of exports and sector composition. Our results suggest that immigrants promoted efficient task specialization, thus increasing TFP and, at the same time, promoted the adoption of unskilled-biased technology as the theory of directed technologial change would predict. Combining these effects, an increase in employment in a US state of 1% due to immigrants produced an increase in income per worker of 0.5% in that state.

Below is the conclusion, which as conclusions ought to do restates things nicely.

We present three main findings, two of which are quite new in this literature. First, we confirm that immigrants do not crowd-out employment of (or hours worked by) natives but simply add to total employment. Second, we find that they increase total factor productivity significantly and, third, that such efficiency gains are unskilled-biased—larger, that is, for less educated workers. We check that these effects are robust to including several control variables (such as R&D spending, technological adoption, sector composition or openness to international trade) and that they are not explained by productivity convergence across states or driven only by a few states or particular decades. We conjecture that at least part of the positive productivity effects are due to an efficient specialization of immigrants and natives in manual-intensive and communication-intensive tasks, respectively (in which each group has a comparative advantage), resulting in an overall efficiency gain. Preliminary empirical evidence supports this claim. In conclusion, we also check that these findings are in line with the analysis of the wage effect of immigrants on less educated natives, which is close to 0, and on highly educated natives, which is positive.

Thoughts? It holds together well enough for me, but I'd like to stay away from the last commenter's unthinking cheeriness. Does the paper make sense to you? Is its thesis unique to the United States?