Abstract
In absence of a model-based measure of occupational skill-intensity, the literature on wage inequality cannot consistently track technological progress on occupational level -- a key ingredient of recent theories of labor market polarization. In this paper, I use the March CPS data from 1983 to 2002 to estimate such a measure corresponding to occupation-specific relative productivities of college and high-school educated.
With imperfect substitution across skill types, the measurement of relative productivities requires estimation of substitution elasticities, and I propose a simple strategy to obtain these. The resulting measure is used to shed light on the modified skill-biased technological change hypothesis proposed by Autor et al. (2006).