Spatial Variations in Amenity Values: New Evidence from Beijing, China

Using parks as an example, this paper explores the robustness and sources of spatial variation in the estimated amenity values using an extended geographically weighted regression (GWR) technique. This analysis, illustrated with estimates using geo-coded data from Beijing's residential land market, has three important implications. First, it provides a powerful estimation strategy to evaluate how sensitive GWR parameters are to unobserved amenities and complementarities between amenities. Second, it compares the spatial variation patterns for the marginal prices of proximity to parks, estimated using a range of GWR model specifications. The answers generated using the GWR approach still reveal a significant underlying problem of omitted variables. Finally, it highlights the importance of conceptualizing amenity values not just in terms of their structural characteristics but how those characteristics interact with or are conditioned by local social, economic, and other contextual characteristics.