Calculating the price of everything: the CPI - consumer price index

Challenge, Sept-Oct, 1998 by Daniel Mitchell

A Menu Approach

Different users have different tastes with regard to transparency; accordance with theory, and consistency in the CPI. And while all users want the index to be accurate, there will inevitably be disagreement on what accuracy entails. The solution, therefore, is to respond as the private market does when there are different preferences for goods. In such a case, the private market typically offers a choice. Indexes should be offered to meet the preferences of different groups of users. The idea of a single, official CPI should be downplayed. Instead, users - even Congress, when it decides to index benefits or taxes - should be offered alternative CPIs. Users should make the choice of which index they will employ, not the Bureau of Labor Statistics.

What would this menu approach mean in practice? At present there are two CPIs offered, one weighted for "all urban consumers" (CPI-U) and the other for "urban wage earners and clerical workers" (CPI-W). In fact, however, there is only a slight weighting difference between the two, and they seldom give readings of inflation that diverge significantly. Under the menu proposal, CPI inflation would be presented in a matrix, something like that shown in Figure 2.

The hypothetical menu provides users with a choice of methodology, along both the substitution and quality control dimensions. On the vertical axis, users have a choice between a fixed-weight Laspeyres index - the current methodology, which has the advantage of transparency. The in-between option is a geometrically weighted index, along the lines that BLS has been publishing on an experimental basis. Such a weighting scheme arguably comes closer to the economic model of consumer choice. Finally, for those who want an even more elaborate index, a chained Fisher ideal of the type now used for the GDP deflator is offered? This type of index most closely approximates the theoretical preferences of the Boskin Commission, but it is the least transparent. Try explaining to your grandmother that her social security adjustments are based on the chained geometric mean of a Laspeyres and Paasche index!

The market-basket updates that accompany the three options are also varied. They range from the current practice of updating the basket about every ten years to updating annually. Cases can be made for frequent and infrequent adjustments. Again, user preference is paramount.

Quality control - the horizontal axis of Figure 2 - would also vary from modest to aggressive. The least aggressive approach would be that used before hedonic pricing began to make its way into the CPI. The in-between approach would be the current approach, as introduced in 1998. And the most aggressive would involve hedonic approaches to a broad range of products. Presumably, the members of the Boskin Commission would be most happy with the lower right corner of the matrix in Figure 2. Other users might pick another location. And all users, by studying the range of figures reported in the nine boxes, would have an indication of the sensitivity of the measured inflation rate to alternative approaches.