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

This paper employs linear and nonlinear unit-root tests to investigate: a) the price dynamics of the home price indices included in the S&P/Case-Shiller Composite10 index, and b) the validity of the “ripple effect,” following the approach outlined in Meen (1999). In general, the findings lack uniformity and depend upon the assumptions imposed by the testing procedures. The tests that assume structural stability and linear adjustment fail to provide evidence in favor of stationarity in the price dynamics of all series. Conversely, the nonlinear test of Kapetanios, Shin and Snell (2003) provides evidence that the price dynamics of Los Angeles and San Francisco follows a nonlinear stationary process. The Lumsdaine and Papell (1997) and Lee and Strazicich (2003) tests indicate that significant structural breaks exist in all series. However, whereas the Lumsdaine-Papell test finds evidence of broken-trend stationarity only in the price dynamics of Las Vegas, the Lee-Strazicich test finds that the price dynamics of Miami and San Diego also exhibit broken-trend stationarity. The tests of the “ripple effect” also display conflicting evidence. The tests that assume structural stability and linear adjustment provide partial evidence in favor of the “ripple effect” in the case of Chicago and Denver, while the nonlinear test finds that the “ripple effect” is present in Boston, Denver, Miami, New York, San Diego, and San Francisco. The Lumsdaine-Papell test provides no evidence in favor of the “ripple effect”. Conversely, the Lee-Strazicich test finds broken-trend stationarity in the case of Boston, Miami, and New York