Is There a Stationary Home Price-Rent Relationship in US Housing Markets?

Abstract

Asset pricing models imply there should be a stationary relationship between house prices and rent. Several previous studies for the U.S. examine this hypothesis by testing for stationarity in the ratio of home values to rents, and fail to reject nonstationarity. However, this method implicitly tests for both a unit root and a particular cointegrating relationship. In addition, most previous studies employ an index for home prices in such tests, which of course entails issues such as representativeness of the housing transactions included and coverage of a particular housing market. In this paper, we employ the actual median dollar price of houses sold in four regional U.S. housing markets, and divide by the rent of primary residence index-, essentially “deflating” house prices by rents. We do, as previous studies have, test for unit roots in the ratio of prices to rents, but we also test for cointegration between the two to avoid imposing a particular cointegrating relationship. The results indicate clear stationarity in the price/rent relationship for three of the four regions, with findings for the West region mixed. In addition, there are positive linear trends for the ratio of house prices to rents, so while prices and rents are together (trend) stationary, the relationship has changed over time, similar to the long-term increase in the price/earnings ratio for equities. Finally, for error correction, we find that rents typically respond to disequilibrium, whereas home prices usually do not, which is contrary to some previous findings that have used different data or examined markets subject to rent control.

JEL Codes: R30, R31, G12.