Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/124310 
Year of Publication: 
2014
Series/Report no.: 
54th Congress of the European Regional Science Association: "Regional development & globalisation: Best practices", 26-29 August 2014, St. Petersburg, Russia
Publisher: 
European Regional Science Association (ERSA), Louvain-la-Neuve
Abstract: 
"Spatial thinking" is increasingly popular in housing market studies and spatial dependence across properties has been widely investigated in the intra-city housing market. The contribution of this paper is to study the spatial dependence and spillover effect of house prices from an interurban perspective, referring to the spatial interaction across local housing markets. The extensive literature study concludes that following behavior, migration and equity transfer and spatial arbitrage of capital are the main behavioral reasons for interurban spatial interaction. Using a cross-sectional data set in eastern China, our empirical results from both parametric and nonparametric approaches provide strong evidence of spatial interaction in the interurban housing market. The parametric results suggest that the spatial lag model (SAR) is the best model specification to describe the interurban house price process, indicating an endogenous interaction pattern. Ignoring such interaction effect in the house price model will produce biased coefficients estimators and misleading interpretation. In SAR model, Spillover effects of explanatory variables caused by spatial interaction are calculated by partial derivative interpretation approach and are demonstrated to have the magnitude as much as half of their direct effects. Moreover, the comparison between different spatial weighted matrices reveals that the spatial interaction depends not only on distances, but also on the economic situation of each jurisdiction. Meanwhile, nonparametric approach draws a flexible relationship between spatial dependence and geographical distances. Using spline correlogram, we find monotonically declined spatial autocorrelation of house prices and explanatory variables within larger distances, whereas the significant spatial autocorrelation of OLS residuals can only be observed at short distance (60 Km). The spillover effect, being obtained from spatial covariance decomposition, is highly significant and declines within the radius of 250 Km. All the nonparametric results imply that though the house price determinants can satisfyingly account for the interurban house prices, the importance of spillover effect cannot be neglected within certain distances. That is the neighbor's housing market situation is quite useful in predicting the house price of a particular city. This study provides a good insight into explaining why the house prices in some cities always run above the level indicated by fundamentals, and highlights the importance of cooperation between local governments in making the housing policy.
Subjects: 
Spatial autocorrelation
spillover effect
interurban housing market
spatial econometrics
nonparametric estimation
China
JEL: 
R31
C21
Document Type: 
Conference Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.