David Autor has inaccurately reported that he has found evidence of a hollowing out of the distribution of jobs for men, with increased employment at the top and the bottom ends of the wage distribution and a loss of jobs in the middle. NYT columnist David Leonhardt seems to have largely bought this story as well.
Actually, as John Schmitt, my colleague at CEPR, and former colleague Heather Boushey pointed out, Autor's work shows the opposite. In the most recent business cycle, 2000-2007, there was a relative decline in the demand for all male occupations, except those at the bottom of the wage distribution. There was less of a decline for jobs near the top than for those in the middle, but it would be more than a bit of an exaggeration to call this a hollowing out of the job distribution. Autor's data is essentially showing an increased demand for less-skilled occupations pure and simple.
The story for the prior business cycle is also not quite what Autor describes. Between 1989 and 1999, there actually was a decline in relative employment for all occupations below the median except those near the very bottom (the bottom decile).
A large percentage of the workers in this bottom decile were immigrants. There has been considerable research (e.g. here and here) that suggests that immigrants don't compete directly with native born workers and instead fill a sub-class of occupations in which jobs would have gone largely unfilled in the absence of immigrant workers. Insofar as this is the case, it suggests that the growth in the lowest wage occupations was not a demand-side phenomenon, but rather a supply side story. In this view, if there had been a large influx of immigrants occupying the middle wage occupations, then we would have seen strong growth in employment in these occupations as well (albeit at much lower wages).
In the period from 1979-1989, the first business cycle analyzed by Autor, there is a decline in the relative shares of employment for all occupations below the 60th percentile. This also does not support the hollowing out story.
To summarize, in the first cycle, Autor finds increased relative demand for highly skilled occupations and decreased demand for less skilled occupations. In the second cycle he finds increased demand for highly skilled occupations and decreased demand for all but the lowest skilled occupations, which may be the result of an influx of low-paid immigrant workers. In the third period, there is a decline in the relative demand for everyone but less-skilled workers. In other words, he really doesn't show any evidence of a hollowing out of the job distribution.