Estimation Errors of Measuring Housing Quality from Tax Records

Abstract

Hedonic measurement of housing quality is normally based on a wide range of data including information on structural, neighborhood, and geographic attributes of the housing stock. Readily accessible tax assessment database across many U.S. communities typically contain limited data on housing structural attributes. It is always very expensive to augment such variables with additional data on housing neighborhood and location. Consequently, hedonic estimation of quality based on these limited data perennially suffers specification bias. Because few hedonic variables are capable of proxying the statistical influence of omitted ones, there is really no need for more expensive data that some analysts insist are necessary for proper use of hedonic index to make housing quality judgments. This makes tax records a valuable source of information for hedonic estimation.