Spatial Effects in Hedonic Price Estimation: A Case Study in the City of Toronto

Recognition of the limitations of traditional hedonic models to account for spatial effects has led in recent years to the development and use of spatial econometric and statistical techniques in real estate applications. It seems appropriate, as the number of applications grows, to evaluate the relative ability of some newer approaches in terms of producing accurate spatial predictions. This article compares a selection of techniques to assess their performance. The focus is on moving window approaches that can be conceptualized as sliding neighborhoods (i.e. soft market segmentations), and that can incorporate spatial dependency effects. Comparison of moving windows regression (MWR), geographically weighted regression (GWR), and moving windows Kriging (MWK) sheds light into the relevance of different spatial effects. Results using Toronto as a case study indicate that market segmentation may be more important than spatial dependencies. The findings suggest practical guidelines with regards to the use of the models investigated.