Same difference: A unifying framework for house value predictions through hedonic pricing, nearest neighbours, and comparable sales

This paper develops a comprehensive methodological framework demonstrating that hedonic pricing (HP), comparable sales and nearest neighbours are (constrained) versions of a general spatiotemporal hedonic specification. We meticulously account for existing real estate professional practice, illustrating that it complies with theoretical imperatives and performs well even under informational and computational constraints. The resulting specification of a parsimonious spatiotemporal Durbin model provides superior performance to alternatives, forgoing local submarket issue and multiple sources of bias common in other Hedonic pricing specifications. This framework provides exciting opportunities for further research not only in real estate, but in other applications of HP especially valuing environmental and social externalities.