The Price-Rent Ratio : Bubble Indicator? Stationary? An Investigation Allowing for Fractional Integration

Given the volatility of home values in recent decades, and the potential impact on the macroeconomy, the ratio of house prices to rents has been employed as an indicator of potential housing market froth. Researchers have attempted to find a stationary relationship between house prices and rents, or measured deviations from such stationarity as evidence of a housing bubble. Empirically, finding a stable long-run relationship between the two variables has proved difficult. In addition, there are theoretical reasons to believe that the price-rent ratio is a poor signal for bubbles. It may be that the ratio is very persistent-perhaps possesses “long memory”, but tests for non-stationarity have low power in the presence of such persistence. We thus examine the price-rent ratio for 23 metropolitan areas in the US. We first test for non-stationarity in all ratios, and find that in most cities, they appear to have unit roots. We next estimate the long-memory parameter for the ratio, at first without allowing for the possibility of structural breaks. Results still indicate that the ratio exhibits unit root behavior. Testing, however, reveals the presence of structural breaks in rents, prices and the price-rent ratio. We thus estimate the degree of fractional integration while allowing for breaks. The results show that the ratios in all cities lack unit roots, are mean-reverting and possibly stationary. This confirms that house prices, while persistent even relative to a fundamental determinant like rent, do revert to a long-run relationship with rent and that the price-rent ratio is not a reliable indicator of bubbles