Abstract

What explains the significant dispersion in rent-price ratios units across and within US cities? Conventional urban models attribute this heterogeneity to differences in rent growth expectations. Using a Campbell-Shiller decomposition I find that since 1980 rent growth only accounts for 10-30% of cross-sectional dispersion, leaving 70-90% to be explained by differential returns. To rationalize persistent differences in returns I build a theoretical framework that embeds an asset pricing model within an urban setting. In the model, landlords invest only in properties that are nearby, an assumption supported by the data. Geographic segmentation among landlords and inelastic housing supply cause spatial dispersion in rental returns in response to local heterogeneity in wealth levels and inequality, as well as rent volatility. Model-implied comparative statics hold in the data both within and across cities: returns fall and prices rise with average wealth and the share of investors, and vice-versa with the share of renters. Calibration of the model suggests that equalizing investor wealth and risk across space would reduce rent-price dispersion by 30% and 70%, respectively. My analysis suggests that the geography of wealth and rent growth volatility are essential in understanding spatial differences in housing prices.