We pursue two primary tasks in this paper. Our first task is to provide empirical
evidence that controlling for location matters significantly in assessing the degree of
convergence in earnings over the period from 1940-2000. We use data from the Integrated
Public Use Micro Samples for the 1940 to 2000 Censuses to explore this question. We find little
evidence of black-white earnings convergence since 1950 in many cites. For instance, we find
that once we condition on education, age, and location, black males earned 35 and 37 percent of
what comparable white males made in 1950 in Chicago and New York. By 2000, however, the
earnings gap had increased to 45 and 41 percent in Chicago and New York. Our second task is
to develop a theoretical model that can help us interpret the evidence we present and provide
insights into important issues of policy with respect to earnings disparities. Not the least of these
is the conclusion that in an environment of significant migration a careful view of the social
welfare implications of the apparent “convergence” must go beyond any simple analysis of
changes in the earnings gap.