Several other bloggers already have covered this important paper, but there remain underexplored details. Overall the main result is that trade has had more of a negative impact on employment than we used to think. I won’t attempt a summary, but here are a few further results of note:
1. In the Providence, Rhode Island area the trade exposure to China for 2000-2007 went up by $3,490 per worker. For New Orleans the same increase was only $490 per worker.
2. Technology gains and mechanization in a region do not predict employment declines, but they do predict polarization of wage returns. (I do think that automation will create problems for labor markets, but I think that issue is more about our future. It also was true, for a while, in our more distant past, as outlined by David Ricardo.)
3. The negative employment effects of technology on manufacturing jobs peaked in the 1980s, and since have declined. The negative employment effects of technology on service sector jobs have been rising. On net the effect on employment across all sectors has stayed roughly constant over the last few decades.
4. Women and older workers are those most likely to lose their jobs because of technology.
5. The employment effects of exposure of a region to Chinese imports are significant. A good deal of this effect works through the labor force participation rate rather than through measured unemployment per se. This by the way is one indication that the labor force participation rate does contain relevant information about the health of the labor market.
6. The authors classify jobs into the categories of abstract, routine, and manual, and suggest that routine jobs are most vulnerable to automation. Maybe, but I would not take this for granted. Better software in a car can forestall mechanical problems, and thus replace the manual labor of the automobile mechanic, even if we cannot imagine how a robot could itself do the car repair work.