Housing Consumption and Prices in a Unified Metropolitan Market with Heterogeneous Preferences

We provide a new method for estimating hedonic functions in a metropolitan housing market for both rental rates and real estate asset prices. First, our method treats housing quality as unobserved by the econometrician. Second, it deals with the problem that implicit rental rates for owner-occupied housing are latent. Using a non-parametric matching approach, we show how to identify the rent-to-value ratio as a function of latent quality. Third, the paper provides a new estimation method for a generalization that incorporates heterogeneity in preferences. To our knowledge, this is the first paper that incorporates these in a unified treatment of metropolitan housing markets. We estimate the model with types obtained by using clustering techniques to learn categorization of households into types based on age and number of children. We obtain the robust result that the presence of children lowers the preference for housing quality and increases the welfare sensitivity to changes in income and price. The opposite happens as households age. Finally, we also estimated a joint model for the NYC and Chicago metropolitan areas to compare quality markets, and find that a compensating variation of approximately 20\% of the annual income would be required to induce the median household in Chicago to move to NYC. Finally, we study the recent housing market crisis in the U.S., and obtain findings consistent with the view that changes in both credit market conditions and investor expectations were key contributors to the run-up in housing values.