This paper begins by comparing hedonic price indexes to the standard matched model index. The matched model index is formed as an average of the price changes of the goods in the sample in the base period that remain on the sampled stores' shelves in the comparison period. Since goods that do not remain on the shelves tend to be goods whose market value has fallen, the matched model index tends to select from the right tail of the distribution of price changes. The hedonic index corrects for the change in value of the observed characteristics of the goods that exit. Moreover we provide conditions which insure that the hedonic can be justified as an upper bound to the compensating variation (which is the standard rational for the matched model index when no goods exit). The paper then explains how the hedonic index can be constructed from exactly the same data that is currently used to construct the CPI. The matched model index can have lower variance than the hedonic when both are computed from this data, but in those cases we show that there is a hybrid hedonic index which has the selection correction properties of the hedonic and the variance reduction properties of the matched model index. We illustrate with a new study of price indexes for PC's. The hedonic index shows steep price declines which follow the pattern of technological change in this market. On average, the matched model indexes indicate no price fall at all and one commonly used matched model index is negatively correlated with the hedonic. We also consider alternative indexes. Of these the one that works well is a "Paasche style" hedonic that we introduce. Its advantage is that since it does not require computation of the current period's hedonic function, it is easier to use when monthly timetables need to be met.
Stored only when both papers are in the corpus — 408 citations exist upstream.