Discovering the urban structure using a spatial hedonic house price model
This paper employs a spatial hedonic house price model that is capable of incorporating temporal dynamics within the model specification in addition to the location, the neighborhood and structural variations of house prices within the Ede municipality in the Netherlands. The dataset includes transaction prices and other information regarding houses sold within the municipality. The impact of distance from the town center on values of housing units is examined and the variants of the monocentric and polycentric models are estimated. Additional tests employed to search if there is spatial autocorrelation in the residual series suggest use of the spatial autoregressive error model (SEM) as appropriate. The estimation results are then used to shed light on the urban structure of the municipality. The spatial model suggests that the polycentric structure reflected in the non-spatial model may be an outcome of spatial dependence. The spatial model also indicates that the Bennekom sub-center is preferred to the Ede town which was purposely planned as the formal urban center. The cluster of neighborhood in which the housing units are located is the most important location variable compared to district and distances from the town centers. House prices are also sensitive to trend and the cyclical variation of income.