Compositional shifts distort normal measures of wage growth: Median and mean changes in wages, normal and inverse probability (composition-adjusted) weights, 2019–2020 and 2020–2021

Mean (normal weight) Mean (inverse probability weight) Median (normal weight) Median (inverse probability weight)
2019–2020 0 7.4% 4.4% 0 6.6% 2.6%
2020–2021 0 -2.7% -1.7% 0 -3.4% -1.6%

Notes: The section “2019–2020” shows changes between the 12 months before the pandemic (March 2019–February 2020) and the initial 12 months of the pandemic (March 2020--February 2021). The section “2020–2021” refers to changes between the initial 12 months of the pandemic (March 2020--February 2021) and the most recent 12 months of the pandemic (March 2021–February 2022). “Normal weight” refers to data adjusted using the sample weight used in the Outgoing Rotation Groups sample. We employ a logit model to create an inverse probability weight, controlling for education, race/ethnicity, gender, age (through the quartic/fourth power), geographic region, citizenship, marital status, parental status, major industry and occupation, full-time status, and union status. Wages are adjusted for inflation to February 2022 dollars using the CPI-U-RS.

Source: Authors' analysis of Current Population Survey basic monthly microdata, EPI Current Population Survey Extracts, Version 1.0.27 (2022),

View the underlying data on