Working Paper: NBER ID: w31387
Authors: Kyle Butts; Taylor Jaworski; Carl Kitchens
Abstract: We estimate the urban wage premium in the United States from 1940 to 2010. Drawing on recent advances in the literature on selection on unobservables, we show how to control for heterogeneity in the characteristics of individuals that choose to live in cities to address endogenous sorting. Estimates from naive comparisons of individuals living in urban versus non-urban areas substantially overstate the urban wage premium. We find that the premium is highest in the middle of the twentieth century (about 12 percent in 1940 and 1950) relative to the early in twenty-first century (declining to a few percent by 2010). Overall, the urban wage premium is decreasing and sorting explains a larger fraction of the difference in urban versus non-urban earnings across our sample period.
Keywords: urban wage premium; agglomeration; sorting; economic history
JEL Codes: N92; R0
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| endogenous sorting (D00) | overstate urban wage premium (J39) |
| urban wage premium in 1940 and 1950 (J39) | urban wage premium in 2010 (J39) |
| sorting explains difference in urban versus non-urban earnings (R23) | urban wage premium (J31) |
| agglomeration effects (R11) | urban wage premium (J31) |
| urban wage premium (J31) | national wage compression patterns (J31) |