A State Space Model For Berlin House Prices

Abstract

How risky are investments in residential real estate? To answer this question, information is needed about the behavior of house prices. The hedonic methodology has become a standard approach for modelling the prices of heterogeneous assets. Although intuitively appealing, it is often criticized that this approach has no sound theoretical background. We have developed a model that partly circumvents this criticism. Based on an approximation for the present value, our model delivers a state space form for the determination of house prices. Thus, we can incorporate in an economically meaningful way other economic variables like the inflation rate, mortgage rates and returns of other assets. Under some restrictive conditions, our model reduces to the standard hedonic approach. We use the EM algorithm with a final scoring step to estimate our model with monthly data of single-family house sales from the four South-West districts of Berlin for the years 1982:7 to 1999:12.