Abstract
A median-based quantile estimator is less prone to the upward bias caused by unobserved renovations and missing “quality” variables when estimating appreciation rates using the repeat sales approach. As expected, the quantile estimator indicates lower appreciation rates for a sample of repeat sales of single-family homes in Chicago for 1993-2002. The quantile estimator is much less sensitive than the standard repeat sales estimator to the omission of observations for which building permits have been issued during this time. Thus, the quantile approach appears to be less prone to sample selection and missing-variable bias than the standard estimator.