Segmentation spatiale et choix de la forme fonctionnelle en modélisation hédonique

This paper investigates the possibility of accounting for market heterogeneity within the hedonic price equation using a nested model. Using a predefined definition of submarkets, different specifications of the hedonic price equation are estimated for Quebec City, Canada. If the options considered are at least better than the classical approach given the spatial drift of hedonic prices of the residential amenities, nothing suggests that a disaggregate approach is necessarily the better modelling option. We show how the log-linear model can be improved by integrating location dummy variables capturing submarket fixed effects. Although such an approach does not solve all the problems associated with the classic hedonic pricing model, it largely lessens their detrimental impact while insuring the stability of regression coefficients related to physical amenities of properties.