Is Employment Polarization Informative About Wage Inequality and Is Employment Really Polarizing?

new NBER reinforces that the "hollowing out" of middle income jobs has been driven by demand for more high wage workers, not previous middle wage workers becoming low wage workers. paper also"....rule out the hypothesis that computerization and automation lie behind rising wage inequality...."

"...We reexamine whether U.S. workers are increasingly concentrated in both low and high wage jobs, a phenomenon known as employment polarization. We depart from most previous literature, which equates a job with an occupation, by instead considering jobs at the individual level. By assigning workers in the CPS to real hourly wage bins with time-invariant thresholds and tracking over time the shares of workers in each,we find a steady decline since 1973 in the share of workers earning middle wages, consistent with occupation based analysis. However, we find that both over the business cycle and the longer run, the shares of workers in the top and bottom bins move in opposite directions, inconsistent with employment polarization.....On the one hand, the displacement of workers from middle wage occupations to low wage occupations increases supply of low wage workers; on the other hand, demand for low wage services may rise with the rising numbers of the high wage workers.....Grouping workers into four time invariant bins defined only by workers' wages, we do find a steady decline in the share of workers in the middle two wage groups, but one that belies offsetting forces that vary over time and by gender, variously reflecting either upward mobility (workers moving faster from the middle to the top than from the bottom to the middle) or downward mobility (workers moving faster from the middle to the bottom than from the top to the middle), but not employment polarization....However, the shares of workers in the top and bottom groups generally move in opposite directions over the longer term as well, with the share in the top group rising markedly and the share in the bottom falling slightly. The patterns are very similar with ten wage bins rather than four...The slight downward trend in the share of workers with low wages could in principle coexist with the growing employment share of low wage occupations found in the literature: we show that the share of low wage occupation workers with low wages falls greatly with time. However, we also show that the share of employment in low wage occupations is trending up only from 2002-2012, and that the apparent earlier growth found in the literature is an artifact of unbridgeable changes in occupation codes. The late emergence of occupation based polarization is inconsistent with the explanation that computerization and automation have caused both employment polarization and rising wage inequality, unless underlying occupation based employment polarization in earlier periods was masked by human capital driven occupational upgrading of workers. We demonstrate that even absent breaks in occupation codes, occupation mean wages are unsuitable for the analysis of wage inequality: most of the increase in wage inequality is within and not between occupation; occupation mean wages do not capture the bottom of the individual wage distribution; and middle{wage workers are distributed widely across occupations. Computerization and automation may be increasing wage inequality, but the hypothesis cannot be studied through the lens of occupations and does not find support in our analysis...."

Jennifer Hunt and Ryan Nunn, "Is Employment Polarization Informative About Wage Inequality and Is Employment Really Polarizing?," National Bureau of Economic Research, July 2019, https://www.nber.org/papers/w26064