Price Indices for Austrian municipalities - Hedonic models based on Microlevel Data

Austrian municipalities must re-evaluate their real estate portfolios every year. The existing Austrian house price datasets (based on the Austrian land register) cannot fully fulfill these requirements due to a lack of descriptive variables. When constructing a hedonic model, it is vital to assemble a dataset as complete as possible to minimize the extent of the omitted variables problem. This paper shows how an existing micro-level dataset can be improved and extended to raise the data’s explanatory power. Then these data are used to compile different temporal hedonic models for the nine regional capitals of Austria. The results show that the right choice of method is essential for smaller cities with fewer transactions. For bigger cities, with more transaction data, the choice of hedonic model is less important (with all suggested model formulations giving similar results). Thus, it is vital to consider the data structure and number of transactions when compiling indices for small regions (cities).