Intra-urban earnings differences: spatial mismatch or selective migration?

We pose two models for earnings differences between residential locations in an urban area. The spatial mismatch model is based on barriers to labor mobility and suggests that observed earnings differences are real. The selective migration model suggests that observed earnings differences reflect unobserved differences in individual abilities. Using a sample of white men with a high school degree or less, we estimate a model of location choice and location-specific earnings for three portions of the Allegheny County (Pittsburgh). Our results indicate that the earnings decline in the old manufacturing portion of the county reflects selective emigration rather than continuing excess labor supply