Forecasting Regional House Prices - An Approach Based on Structural Modelling of Housing Markets

We propose forecasts of regional and local house prices based on estimates of a structural model that has several distinctive features. First, we model not only the endogenous relationship between housing prices and demand, but also microeconomic features of the market ñ in particular, the nature of search, bargaining and price formation. In other words, the wedge between supply and demand is identified by variation in degree-of-overpricing, over both space and time. Therefore, our model incorporates three structural relationships ñ for housing prices, demand and degree-of-overpricing. Second, we incorporate an additional feature of the microeconomic structure of housing markets, by including time on the market as a leading indicator for prices. Third, we allow for arbitrary spatial interaction between house prices in different regions as well as complete spatial heterogeneity in the description of the structural relationships. The above three features are in addition to explicit consideration of temporal variation in housing prices, which is standard in previous work. We use our model to forecast housing prices across boroughs in Greater London and counties in the North West of England. Our empirical exercise demonstrates the strength of the approach in terms of improved forecast performance.