Disaggregated property price appreciation: The mixed repeat sales model

Abstract

Real estate is, by its nature, a highly heterogeneous asset class with most investors owning a single property rather than a diversified portfolio. As a result their returns are likely to reflect not only the overall dynamics of the asset class but also the way these interact with the unique features of the property. Yet the way that property characteristics influence price dynamics is not well measured or understood. We propose a Mixed Repeat Sales (MRS) model which allows for the nature of the asset to influence price dynamics. This is illustrated empirically using a large data set for the state of New South Wales in Australia (of which Sydney city is the capital) from 2000-2014. The MRS approach enables us to estimate returns at a disaggregated level based upon property characteristics. We find significant dispersion in appreciation rates across locations and that returns were higher for separate houses compared with both apartments and attached houses. However, the size of the dwelling, reflected in the number of bedrooms and bathrooms, and the land area play a lesser role in price developments.