Estimating serial cross-correlation in real estate returns

Appraisal smoothing understates the true volatility of real estate returns, and consequently affects asset allocation decisions. The high level of smoothing observed in commercial property index returns can be shown to be largely influenced by the effect of serial cross-correlation. This paper examines this phenomenon by analysing the time series properties of a sample of individual property returns using the generalized autoregressive conditional heteroskedasticity (GARCH) model. The autoregressive conditional heteroskedasticity (ARCH) model, proposed by Engle (1982) and Bollerslev (1986), is used to characterize time series data that exhibits clustering in the residuals. The time-varying conditional variance in ARCH models is often used as a measure of intertemporal risk. We use this property to analyse the cross-correlation coefficients of a sample of properties. Our results suggest that there is positive skewness in the serial cross-correlations. This is consistent with a process that would generate high serial correlation at the index level. We demonstrate that serial cross-correlation can be used as a proxy of the proportion of sticky values in an index. The results indicate that serial cross-correlation is time varying and positively skewed. Copyright © 2001 John Wiley & Sons, Ltd.