Modelling Belgian Housing Prices: a cross-sectional and time series analysis

This paper presents the results of two econometric analyses using existing models applied to and elaborated for Belgian housing data. First, a cross-sectional analysis is done based on mean housing transaction prices for 43 Belgian districts. The results show that introducing a dummy variable distinguishing the French and Dutch speaking region of Belgium significantly improves the model. This finding is consistent with the dual character of the Belgian economy. Due to many reasons, but mainly because of a more severe land planning regime in the Flanders region, mean level of housing prices are quite different in both parts of Belgium. Second, a time series analysis is done to explain the development of average housing transaction prices in Belgium. Although the cross-section analysis indicated substantial differences in the level of housing prices for both parts of the country, the development of housing prices appears to be rather equal. Hence, a model explaining housing prices in a spatially aggregated way for the whole country is developed. For this purpose a pseudo ECM model based on Malpezzi is used. A related model developed for the Netherlands was also tested. However, the Malpezzi model appears to be more adequate for the Belgian situation. Finally, some additional insights about affordability of housing are provided.