Here’s a letter to my incessant correspondent Nolan McKinney:
Mr. McKinney :
A part of Brink Lindsey’s and Samuel Hammond’s new paper  that you find “among the most worrying” – and allege that I, in my earlier response , “conveniently overlook” – is on page 94. There they write that “rising job polarization has hollowed out the availability of ‘middle skill’ occupations, creating bifurcated labor markets in which ‘high skill’ professionals live side-by-side with ‘low skill’ service sector workers just barely scraping by.”
Admittedly, the impression given by their portrayal of this labor-market development is worrying. Fortunately, this impression is inaccurate. Here’s economist Michael Strain, writing earlier this year:
Recall that David Autor, the economist, calculated that over the four-and-a-half decades since 1970, middle-skill employment fell from about 38 percent to 23 percent of total employment. But this fall in the middle was offset by a rise in employment at the top. Employment in high-skill occupations grew from about 30 percent to 46 percent. At the same time, low-skill employment did not increase. On the whole, the hollowing out of the middle is a story of middle-skill employment being replaced by high-skill employment.*
Only if the additional workers today, compared to in the past, holding high-skill (and, hence, higher-paying) jobs would prefer to hold middle-skill (and, hence, lower-paying) jobs, the “job polarization” that Messrs. Lindsey and Hammond lament is an unambiguously good development – and one not made otherwise by the authors putting “job polarization” in italics.
Donald J. Boudreaux
Professor of Economics
Martha and Nelson Getchell Chair for the Study of Free Market Capitalism at the Mercatus Center
George Mason University
Fairfax, VA 22030
* Michael R. Strain, The American Dream Is Not Dead (But Populism Could Kill It)  (West Conshohocken, PA: Templeton Press, 2020), page 72.
UPDATE: In response to the above letter, Scott Lincicome sent to me the following in an e-mail:
Even the Autor study is too pessimistic, because it tracks job categories instead of the wages of the actual people in those categories. I summarized the new research on the latter in a recent cato blog post (https://www.cato.org/blog/some-reasons-optimism-regarding-hollowing-out-americas-middle-class ):
Second, a 2019 paper  from Jennifer Hunt and Ryan Nunn provides similar conclusions when examining wages at the individual level (rather than occupation‐average wages) over time. They explain that “[w]hile occupations may provide reliable information about tasks and the nature of work at a point in time, average occupation wages are in general not appropriate proxies for individual wages, nor is the distribution of occupations by average wage very informative about the distribution of workers’ wages” (and thus wage inequality and polarization), in large part because wages can vary dramatically within occupations (even those classified as “low wage”).
Instead, they assign workers to real hourly wage “bins,” ignoring specific occupations and their average wages, and then track annual changes in the shares of workers in each bin between 1973 and 2018, again adjusting for inflation. The groups examined here are different from those used by Rose – the highest‐wage bin starts at $35.08 (equivalent to about $70,000 per year), and these individuals are not converted to 3‐person “family equivalents” – but the findings are nevertheless similar: low and middle‐wage jobs are slowly declining while higher‐wage jobs are increasing more quickly.