Constructing Neighborhood Housing Price Indices Using Correlations of Neighborhood Housing Returns
Posted: 9 Feb 1998
Date Written: December 1997
The development of sub-metropolitan house price indices is of interest to homeowners, to home mortgage lenders, and to many regulators and urban policy makers. However, the evolution of sub-metropolitan indices is plagued by the sparsity of home sales transactions. Recent innovations in constructing such indices have relied on a weighting function by which to aggregate neighborhoods according to their similarity. The most sophisticated of these efforts has specified a multi-dimensional weighting function with distance and socio-economic characteristics as arguments. This study adopts a more inferential approach to creating an aggregation function. Adopting covariance of housing returns as the criterion for neighborhood similarity, the study first estimates provisional housing returns for traffic analysis zones (smaller regions than have been used in related work). It then uses the pairwise correlation between TAZ return series as a measure of neighborhood similarity. Numerous differences in TAZ characteristics are tested for their power to explain correlation of returns; these include Euclidian distances between TAZs, differences in socio-economic characteristics, and urban location indicators. These characteristics are found to have very little power to explain neighborhood similarity as represented in correlation of housing returns. Thus, the approach used here of directly inferring similarity of neighborhoods appears to be a more promising approach to constructing a neighborhood aggregation than to specify a function apriori.
JEL Classification: R0, R21
Suggested Citation: Suggested Citation