Ridge Estimators in House Valuation Models

Statistical models are being applied in the valuation of different types of real estate. This article discusses Ridge estimators for the valuation of single family residences in the presence of multicollinearity (ill-conditioned data). We distinguish between the prediction case, where the purpose is to estimate the selling price, and the explanation (or control) case, where the purpose is to infer the response in selling price from changes in regressor values, in particular property characteristics. The economic inferences from the latter ease are particularly interesting.