Unit Roots and Structural Change: An Application to US House-Price Indices

This paper addresses two issues. First, we employ unit-root tests that allow for two endogenous breaks as suggested by Lumdaine and Papell (1997) and, more recently, Lee and Strazicich (2003) to investigate the integration properties of the returns on the S&P/Case-Shiller Home Price Indices. The findings of the tests that assume structural stability provide no evidence against the unit-root hypothesis in all returns series. Conversely, the Lumdaine-Papell and Lee-Strazicich tests indicate that significant structural breaks exist in the US housing market. Only the Lee-Strazicich test, however, which incorporates structural changes under the null hypothesis, finds that the returns to houses exhibit trend stationarity with structural breaks, in most cases, rather than a random walk. Second, we apply these tests to analyze what UK researchers call the "ripple effect" in the British housing markets. Following Meen (1999), we investigate the stationarity of the metropolitan house-price ratios. The findings of the Lumsdaine-Papell test provide no evidence against the unit-root hypothesis in all house-price ratio series. Conversely, the Lee-Strazicich test finds broken-trend stationarity of the metropolitan house-price ratios for Boston, Miami, and New York. This provides limited evidence that some ripple effects do indeed exists in the US housing market.