The construction of housing price indexes raises many conceptual and practical problems, mainly because houses are heterogenous assets whose prices can only be observed when they are sold. Hedonic pricing methodologies are a prominent approach in this area, which allow not only to compile house price data controlling for quality differences but also the assessment of the value of a particular property in absence of market transactions.

The computation of an hedonic index involves two main choices: (i) which type of price index to use; and (ii) which hedonic method to employ to decompose the (unadjusted) price index into a quality index (the part of the price change that is explained by changes in the characteristics of the dwellings) and a quality-adjusted price index. Regarding the first choice, we analyse Dutot and Jevons price indexes. Concerning the second issue, the most popular techniques, imputation price, time dummy variable, re-pricing, and characteristics price index methods, are examined. It is shown that the type of price index selected has a decisive influence on the choice of the form in which the dependent variable appears in the hedonic regression. This implies that the hedonic functions employed to construct Dutot and Jevons indexes must include as dependent variable the price and the log of the price, respectively. Moreover, a detailed analysis on the alternative hedonic methods shows that, under particular circumstances, the last three methods are encompassed by the imputation method. This integrated approach allows a detailed analysis and comparison of the assumptions and specification requirements of the four methods.

Keywords: Hedonic price indexes; Specification analysis; Hedoninc methods

Biography: Assistant Professor: Universidade de Evora

Researcher: CEFAGE-UE (http://www.cefage.uevora.pt/)

Consultant: Coordination and Support for the Owner-Occupied Housing Statistics Pilot Project