Constructing Quality Adjusted Price Indexes: a Comparison of Hedonic and Discrete Choice Models

The Boskin report (1996) concluded that the US consumer price index (CPI) overestimated the inflation by 1.1 percentage points. This was due to several measurement errors in the CPI. One of them is called quality change bias. In this paper two methods are compared which can be used to eliminate quality change bias, namely the hedonic method and a method based on the use of discrete choice models. The underlying micro-economic fundations of the two methods are compared as well as their empirical implementation. Although the discrete choice model has not often been used in order to calculate quality adjusted price indexes it seems to be quite promising to do so.