Hedonic Prices, Economic Growth, and Spatial Dependence

This thesis consists of three papers on econometric modeling of spatial dependence. The awareness of interactions between actors is fundamental for understanding property markets as well as the growth of regions. In both cases, neighbors and neighboring markets may stimulate or hamper growth of values. From a modeling point of view, these interdependencies calls for spatial econometric models. In the thesis we introduce such methods in the analysis of regional property markets as well as in a comparative regional growth analysis. In the first paper, we estimate hedonic prices in the market for co-operative flats in the city of Umeå, Sweden, during 1998 and 1999. Structural, neighborhood, and accessibility characteristics are used as attributes in the hedonic price function. Important attributes were the rent, floor space, age, and population density. Two attractive nodes, although with different characteristics, were found. Thus there are signs supporting the view that Umeå has developed into a multi-nodal structure for property values. SAR-GM estimation was used due to signs of spatial error dependence. In the second paper, hedonic prices for single-family homes in two Swedish counties are estimated for two years. Parameter estimates are compared and changes in space and time analyzed. Spatial lag dependence is found to influence the results. Hence, four independent variables are lagged with a spatial weights matrix. Additional spatial error dependence is treated by SAR-GM estimation. Structural, neighborhood, and accessibility characteristics are used as attributes. The regional price pattern and its changes over time, is illustrated and identified with GIS maps. Proximity to the two county capitals, as well as the other municipality centers, influence property prices positively. This is also noticable over time, where values have risen for homes located near major population centers and those which have water provided by the municipality. Values are in addition largely a function of the quality of each home. The third paper examines the provincial pattern of growth in China during the period 1985-2000, testing the hypothesis that provinces with similar growth rates are more spatially clustered than would be expected by chance. The provincial economic growth is explained by the distribution of industrial enterprises, foreign direct investment, infrastructure, and governmental preferential policies. The neoclassical hypothesis of convergence is also tested. Indications of unconditional convergence does occur during the periods 1985-2000 and 1985-1990. In addition, conditional convergence is found during the sub-period 1990-1995. Evidence of spatial dependence between adjacent provinces has also been established, and in the econometric part, solved by a spatial lag, or alternatively a spatial error term, in the growth equation.