Real Estate Sub-markets

Abstract

The idea that location is important in real estate economics is not new. However, the spatial dimension of real estate data is not always taken into account in traditional real estate models. Spatial econometrics is a tool that could remedy this problem. The objective of this paper is to review spatial models and apply them to typical hedonic real estate data. By doing so we gain insight into the size of bias that can occur in parameters if we do not take spatial effects into account. The conclusion is that spatial autocorrelation is present, least-square estimates may be biased and inefficient and spatial hedonic models do explain more of the price variation. However, the choice of spatial structure does affect the interpretation of parameters for variables with which it is correlated, i.e. it is a sort of multicollinearity problem. Hence, uncritical use of spatial econometrics may cause problems in the interpretation of individual parameters.