Putting time into space: the temporal coherence of spatial applications in the housing market

Relationships between past events, future expectations and present decisions, typically examined through a temporal prism within applied economics, have been lately moving to the spatial dimension through spatial econometrics. However, violations of the "arrow of time", and thus causality, have been identified in spatial econometric techniques applied to spatio-temporal data consisting of observations each at a specific location and distinct moment in time. A comprehensive review classifies for the first time several redresses to this issue in a currently fragmented literature. This paper puts back the temporal dimension into spatial Hedonic Pricing models through a unique specification of a spatio-temporal model that successfully isolates three distinct effects. First, past sale prices affecting current prices, which exemplifies the "sales comparison" approach. Second, a contemporaneous peer effect that is occurring within the narrow time frame of interaction between market participants prior to a sale. Third, the signals of sellers' expectations are captured in the asking prices of other houses not yet sold, while the buyer is active in the market. This affects the final sale price, due to strategic behavior and anchoring. This is the first instance that effects other than the own asking price can be handled in such models. In "boom" market conditions, this mechanism introduces the expectation of increasing prices to the hedonic price function, potentially contributing to housing market "bubble" propagation.