This paper analyses how the properties of locational equilibrium models can be used to evaluate approaches for constructing price indexes for heterogeneous houses. Housing markets play a key role in locational equilibrium models. Prices for houses determine that implicit costs that households bear when locating in a given community. We evaluate a variety of price indexes all relying on hedonic models for predicting interjurisdictional housing prices. The application uses a unique panel data set of housing transactions in Southern California. The rank predictions of different models are robust with respect to the hedonic model and the composite commodity definition used in aggregation. They do not depend significantly on the spatial or temporal definitions used to define the choice set of local housing markets. Finally, housing price estimates are strongly correlated with education and environmental amenities.