House Prices and Spatial Dependence : Towards an Integrated Procedure to Model Neighborhood.

This paper deals with applying GIS and spatial statistics to hedonic modeling. More precisely, it looks at spatial autocorrelation and trend surface analysis (TSA) as devices that can be used to improve model performances. Empirical analysis is performed on the Charlesbourg 1986-87 bungalow (one-story, single-family detached unit) residential market segment. Charlesbourg is the third largest municipality in the Quebec Urban Community (QUC), with some 71 000 in population by 1991. Main findings first suggest that TSA may either be viewed as a proxy for, or a complement to, detailed neighborhood analysis. While best results are obtained using an array of space-related attributes, TSA proves to be an adequate, cost-efficient alternative to expensive data collection. Secondly, the study emphasizes the need for a strict analytical framework to be followed if any spurious conclusions about implicit prices are to be avoided.