Comparing House Prices Across Regions and Time: An Hedonic Approach

Panel hedonic comparisons can be made using the region-time-dummy method. This method is a natural extension of the well known time-dummy and region-dummy methods which have been used extensively in the hedonic literature. We show that these methods are all affected by substitution bias, which can seriously distort their results. We propose an alternative approach that is free of substitution bias which builds up panel comparisons from bilateral building blocks using the hedonic imputation method. This approach is very flexible. We consider a number of variants on this method, all of which are likely to be improvements on the unconstrained region-time-dummy method. We illustrate our findings using data for 14 regions in Sydney over a six year period. We find clear evidence of bias in the region-time-dummy results as well as in simple average measures such as the median that fail to adjust for quality change. For these reasons we favor the hedonic imputations approach.