Business confidence and forecasting of housing prices and rents in large German cities

The role of the housing market in the everyday life of society is difficult to overestimate. The housing rents and prices directly affect standard of living of virtually every person. Housing loans constitute the largest liability of households and account for a large proportion of bank lending. In Germany, the housing accounts for more than a half of wealth of private households. It is well known that speculative price bubbles on real-estate markets are likely to trigger financial crises, which can, in turn spill, over to the real economy by producing deep recessions accompanied by huge employment reductions. Since the end of 2010, after more than a decade of falling real housing prices, strong rent and especially price increases have been observed in Germany. This raised doubts and fears in German society. On the one hand, it is feared that Germany can follow the path of Spain, Ireland, and other bubble countries that ended in a severe economic crisis. On the other hand, the tenants that constitute a majority of German population are afraid of substantial rent increases that will erode their welfare. The tenants' discontent takes a form of massive protests and manifestations endangering political stability in the country. For this reason of the major issues debated in during recent elections and ongoing coalition negotiations among two leading German parties CDU/CSU and SPD is the housing policy. Therefore, it is very important to be able to predict the dynamics of home rents and prices in the nearest future. In this paper, we evaluate the forecasting ability of 115 indicators to predict the prices and rents for existing and new housing in 71 German cities with population exceeding 100,000 persons. Above all, we are interested in whether the local business confidence indicators can allow substantially improving the forecasts, given the local nature of real-estate markets. The forecast accuracy of different predictors is tested in a framework of a quasi out-of-sample forecasting. Its results are quite heterogeneous. No single indicator appears to dominate all others for all cities and market segments. However, there are several predictors that are especially useful, namely business confidence at the national level, consumer confidence, and price-to-rent ratios. Even better forecast precision can be achieved by combining individual forecasts. On average, the forecast improvements attain about 20%, measured by reduction in RMSFE, compared to autoregressive model. In separate cases, however, the magnitude of improvement is about 50%.