Sample selection and biases in local house value indices

Analysis of cross-sectional or intertemporal variations in house values among localities requires reliable house value indices. However, prior studies applying the hedonic method to transaction data have reported price series that may be biased. Our paper investigates a possible cause of bias and uses a censored regression technique extended to an intertemporal model to derive an unbiased index. We apply our method to 1971-1991 data from the Miami, Florida, MSA and find evidence in time varying amounts of bias. Our results suggest that prior studies using the hedonic method and samples consisting only of sold houses have understated the intertemporal variability of house prices. (C) 1998 Academic Press.