Local Polynomial Regressions versus OLS for Generating Location Value Estimates

Abstract

We estimate location values for single family houses using a standard house price and characteristics dataset and local polynomial regressions (LPR), a procedure that allows for complex interactions between the values of structural characteristics and the value of land. We also compare LPR to additive OLS models in the Denver metropolitan area with out-of-sample methods. We determine that the LPR model is more efficient than OLS at predicting location values in counties with greater densities of sales. Also, LPR outperforms OLS in 2010 for all counties in our dataset. Our findings suggest that LPR is a preferable approach in areas with greater concentrations of sales and in periods of recovery following a financial crisis.