Weekly hedonic house price indices: an imputation approach from a spatio-temporal model

Since the global financial crisis there is an increased demand for timely house price indices. The aim of this paper is to develop a method for computing house price indices at a weekly frequency using the hedonic imputation method. The hedonic imputation method provides a flexible way of constructing quality-adjusted house price indices using a matching sample approach. At annual frequencies the implementation of the hedonic imputation approach typically entails estimating the hedonic model period-by-period and then using the parameter estimates (i.e., characteristics shadow prices) to obtain the required imputed house prices. Once these imputed prices are available for a matched sample, standard price index formulas (e.g., Laspeyres, Fisher or T¨ornqvist) can be used to compute the overall price index. A common approach to control for location in hedonic models has been to include postcode dummies. This may not be feasible at higher frequencies as there may not be enough observations for each postcode and small ∗This project has benefited from funding from the samples might cause large variability in the shadow price parameters when estimated period-by-period. We develop a spatio-temporal model to obtain the imputed prices. A geospatial spline surface controls for location and is embedded in a state-space formulation that controls for trends and property quality. The advantage is that the model is parsimonious and shadow price parameters are connected over time while retaining the property that values are not revised as new time periods are added to the data set. We show the spatio-temporal specification leads to a modified form of the Kalman filter and a Goldberger’s adjusted form of the predictor to obtain the imputations. Using a recently developed measure of index performance and applying this hedonic geospatial spline/Kalman filter approach to data for Sydney (Australia) we show that it outperforms competing alternatives for computing house price indices at a weekly frequency. Furthermore, we show that weekly house price indices are much more sensitive than annual or quarterly indices to the choice of hedonic method. Hence the choice of hedonic method is of greater practical significance for weekly indices.