Forecasting prices of single family homes using GIS-defined neighborhoods

We estimate spatiotemporal models of average neighborhood single family home prices to use in predicting individual property prices. Average home-price variations are explained in terms of changes in average neighborhood house attributes, spatial attributes, and temporal economic variables. Models adopting three different definitions of neighborhoods are estimated with quarterly cross-sectional data over the period 2000–2004 from four cities in Southern California. Heteroscedasticity and autocorrelation problems are detected and adjusted for via a sequential routine. Results of these models suggest that forecasts obtained using city neighborhood average price equations may have advantage over forecasts obtained using city aggregated price equations.