Housing Depreciation Revisited: Hedonic Price Modeling Versus Assessor Estimates

Age and condition driven rates of structural depreciation for single-family housing based on hedonic price modeling representing the perceptions of home buyers/sellers are compared to tax assessor depreciation estimates for 47,000 homes in Sarpy County Nebraska. A hedonic price model with age specified as linear generated depreciation rates 11% below assessor rates with differences ranging from 43% lower to 13% higher across four classes of home values. A quadratic-age specification generated depreciation 39% above assessor rates with a range of 15% to 162% higher. A third model, with both quadratic-age and age-condition interaction variables, generated depreciation 27% higher than assessor rates with a range of 8% to 128%. If the goal of hedonic-based housing depreciation modeling is to converge with assessor-derived depreciation estimates based on widely used proprietary cost estimation software and data, then a linear model specification with respect to home age is recommended. Regardless of functional forms chosen, quantile regression where depreciation is estimated across different classes of home values is recommended for all types of hedonic depreciation models.