Hedonic regression models for housing tax valuation

Different types of taxation include the market value of housing or housing returns in their tax base, making it essential to obtain accurate and up-to-date assessments of property values. However, to value residential property represents a major challenge for tax administrations due to informational constraints. In the present paper we present and discuss a simple, inexpensive, and transparent procedure for assigning market value to each dwelling in Norway, based on deriving estimates from hedonic regressions. The valuations are updated yearly to reflect changes in market value. This is a novel example of using predictions obtained from regression estimates to define full-scale housing values for tax purpose. We present and discuss two iterations of the method: the initial prediction model introduced in 2010 and a refined version that would offer substantial improvements without the need for additional data collection efforts.