Autor's 2019 Richard T. Ely Lecture addresses the paradox that non-college workers' real wages fell over four decades despite rising educational attainment in the overall workforce (which under standard gross-complementarity assumptions should have raised non-college wages). The paper's central contribution is demonstrating that occupational polarization — the hollowing out of middle-skill jobs in favor of high- and low-skill work — has been disproportionately urban. Dense cities once offered non-college workers distinctively skilled production and clerical jobs; automation and trade have eliminated those jobs preferentially in cities, collapsing the urban occupational skill gradient for non-college workers and, consequently, eroding their urban wage premium.
"Not only has technology change been transformational, it has been broadly deskilling—by which I mean that it has narrowed the set of jobs in which non-college workers perform specialized work that historically (and currently, as I show below) commanded higher pay levels."
"Distinct from earlier decades, non-college workers now perform essentially the same work in both urban and non-urban labor markets: custodial work, food services, protective services, recreation, and health services; transportation services; and laborer occupations."
"The skill escalator that non-college workers once ascended as they entered urban labor markets has lost altitude."
The geographic angle is the paper's genuine contribution — documenting that the urban middle-skill gradient existed, was steep, and has completely disappeared is new and important. The DFL exercise is elegant in structure: you can account for essentially all of the fall in non-college wages once you account for both occupational shift and the fact that those occupations used to concentrate in high-wage cities. The limitation is the partial-equilibrium framing — "an occupation is not a labor market," as Autor himself acknowledges, so the exercise likely understates the general-equilibrium wage effects. The Baum-Snow/Freedman/Pavan (2018) independent finding of the declining non-college urban wage premium validates the empirical result. The "wealth work / frontier jobs / last mile" taxonomy introduced in Autor and Salomons (2019) is illustrative but speculative as to what it implies for future wage trends. The paper does not address disability insurance directly, but the mechanisms it documents — displacement of non-college workers from stable, middle-skill jobs into low-skill work or unemployment — are precisely the supply-side shocks that generate disability insurance (DI) applications (see China Trade Shock and Conditional DI Applicants).