Simulating housing prices with UrbanSim: predictive capacity and sensitivity analysis

Housing prices in the Lyon Urban area are simulated with the land use framework UrbanSim interacting with the transportation model MOSART. We focus on the Real Estate Price Model of the UrbanSim framework, which proposes the ordinary least square regression. In our simulation, the alternative geographically weighted regression methodology is applied. The model of housing prices is calibrated using a nine-year back-casting period. The calibrated model, applied in simulation, provides price dynamics similar to actual one in the very centre of Lyon. Farther from the city centre, where the available data on actual sales exist, simulated prices tend to be understated. Thus, mainly only the most central locations manifest realistic price dynamics. Sensitivity analysis demonstrates the model’s ability to capture changes in employment accessibility on price dynamics. Copyright Springer-Verlag 2013