Demand substitution across US cities: Observable similarity and home price correlation

This paper studies demand substitution in the context of US cities. Demand substitution occurs when individuals on the margin between certain city pairs affect demand patterns in the aggregate, causing certain cities to be better substitutes than others. Using a discrete model of city choice, I derive two predictions for migration flows and test them empirically using city-to-city migration data from the US Census. I show that cities which are similar on a variety of observable measures have higher levels of gross migration flows in the steady state and higher net migration flows in response to labor demand shocks. Finally, I propose pairwise correlation in metropolitan home prices as a price-based measure of substitutability and show that it contains substantial predictive power for migration flows relative to observable similarity.