Hedonic Price Indexes: A Comparison of Imputation, Time Dummy and 'Re-Pricing' Methods

The main approaches to measuring hedonic indexes in the academic literature are the imputation approach and the time dummy approach. This paper compares both approaches, discusses an alternative method called hedonic re-pricing, and comments on a recent contribution by Diewert et al. (2009). The aim is to explain the differences between the various hedonic approaches as well as their similarities, and to point to the implications for statistical agencies. Hedonic price indexes can be weighted or unweighted and the paper addresses the issue of choice of regression weights. For unweighted indexes it is shown that the 'full' hedonic imputation approach and the time dummy approach implicitly leave the matched part of the indexes unaffected, just like 'single' and 'double' hedonic imputation do explicitly.