An Evaluation Of The Use Of Hedonic Regressions For Basic Components Of Consumer Price Indices

The importance of adjusting for quality changes in the measurement of consumer prices, and the role hedonic regressions can play in achieving this, is well recognised. However, the use of such regressions can take different forms, including (i) adjustments by statistical offices for non‐comparable substitutions via the matched models method, (ii) direct estimates from the coefficients on dummy variables for time, and (iii) exact hedonic and superlative indices corresponding to a constant utility formulation from an economic theoretic approach. The literature on these approaches generally deals with each in isolation; the purpose of this paper is to outline and evaluate them in order to draw conclusions as to their practical suitability for the compilation of quality‐adjusted consumer prices indexes. The case is argued for a move towards the last of these approaches, which developments in electronic data retrieval (scanner data) now make feasible. The paper concludes with the results of some empirical work comparing the results of the direct method with those from the exact, superlative, approach.