Census data show that low-skill workers in the United States are increasingly employed
in non-traded jobs. As a consequence, their employment opportunities and their earnings
are increasingly dependent on physical proximity to high-wage earners that are the
customers for these services. We present a model for a local economy that captures this
idea and delivers testable implications that can help in separating our consumption-driven
story from the effects of production complementarities and human capital externalities.
We find evidence consistent with the predictions of the model and we explore how this
approach can improve our understanding of the recent observed divergence in inequality
trends at the top and the bottom of the U.S. wage distribution.