Content area
Full Text
Abstract
We show how a method that has been applied to commercial real estate markets can be used to produce high-frequency house price indexes for a city and for submarkets within a city. Our application of this method involves estimating a set of annual robust repeat sales regressions staggered by start date and then undertaking an annual-to-monthly (ATM) transformation with a generalized inverse estimator. Using transactions data for Louisville, Kentucky, we show that the method substantially reduces the volatility of high-frequency indexes at the city and submarket levels. We define submarkets in terms of both ZIP Codes and groups of contiguous ZIP Codes that approximate areas defined by the local multiple listing service. Focusing on ZIP Codes, we demonstrate that both volatility and the benefits from using the ATM method are related to sample size. Our method is clearly useful for constructing house price indexes for small areas with relatively scarce data.
(ProQuest: ... denotes formulae omitted.)
Reliable house price indexes are necessary for understanding the dynamics of urban housing markets. The Federal Housing Finance Agency publishes indexes for metropolitan areas in the United States; however, metropolitan areas are comprised of submarkets and house price dynamics can vary across submarkets (Bourassa, Hoesli, and Peng, 2003). Prices could rise rapidly in one area while they rise only moderately or even decline in another area. Therefore, it is useful to have reliable house price indexes at the submarket level.
The volume of transactions limits the frequency and the degree of geographical disaggregation at which an index can be produced. For example, due to scarce data, a monthly index for a city or a less frequent index for a neighborhood might be extremely volatile and, therefore, not particularly useful. Several papers have focused on the issue of index construction with thin data, either by parameterizing the historical time dimension (Schwann, 1998; McMillen and Dombrow, 2001; Francke, 2010) or by making use of the spatial or temporal correlation in real estate markets (Pace, Barry, Clapp, and Rodriguez, 1998; Clapp, 2004).
The focus of this paper is to show how a method that has been applied to commercial real estate price indexes (Bokhari and Geltner, 2012) can be useful in constructing high-frequency house price indexes for both cities and...