Hierarchy and spatial autocorrelation effects in hedonic models

Abstract

Models of hedonic housing prices have to deal with spatial autocorrelation in order to avoid bias and inconsistency in the estimates of the coefficients. Multilevel models have been presented as a way to adequately consider the effects of neighborhood amenities that operate at different spatial hierarchies. In this paper this feature is tested by specifying a three-level model for a data set in the center of Madrid, which includes spatially lagged explanatory variables. The results show that hierarchical models are not able to fully grasp spatial autocorrelation so that a greater effort should be made ​​in the development of appropriate spatial multilevel models.