Separating causes of spatial autocorrelation: an empirical analysis of the price formation process

Abstract

Researchers deal with spatial dependence in urban transaction data due to econometrical reasons – to maintain model reliability. Few research is focused on the economic causes of the phenomenon. Researchers are unable to distinguish between econometrical causes (omitted variables) and economic ones (e.g. price formation). This work separates the price formation process from other causes by reweighting the spatial weight matrix according to the similarity of characteristics between observational pairs. This unique pattern captures only the spatial dependence effect that pertains to the price formation process. Based on more than 33,000 transactions in the heterogeneous residential area of Hong Kong, we find that spatial dependence structure is to a large degree linked to similarities in hedonic characteristics. This confirms the price formation – spatial dependence conjecture.