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

Challenge, Sept-Oct, 1998 by Daniel Mitchell

Unfortunately; standard economic theory offers less guidance to construction of a CPI than is often understood. Assume for the moment that the theory is an accurate representation of a single consumer at a moment in time. Assume that we have an index, which perfectly accords with theory. Even so, the index surely cannot simultaneously represent all consumers with their diverse incomes and tastes. A rise in the price of beer makes beer-drinkers worse off but has no effect on teetotalers. The average consumer is a cross between drinkers and teetotalers. If indexed social security benefits rose in response to beer prices, beer drinkers would be shortchanged. And teetotalers would receive a windfall. There is no theoretically correct way of adding up the conflicting welfare changes of these two groups in response to the adjustment. Consumer theory cannot even represent the changing tastes of a single consumer over time. It is premised on an unchanging utility function.

In short, a CPI based on the standard economic model is not a self-evident choice for all data users. Yet - imperfect though it actually is - such a theory-based CPI is certainly of interest for various purposes. Some users - certainly the members of the Boskin Commission - would like to see inflation measured that way. And if there is one fundamental axiom of economic theory, it is that there is no accounting for anyone's tastes.

Consistency

Macroeconomists and monetary policy-makers are often concerned with the relationship between inflation and the level of real economic activity. Generally, such users want to examine past relationships between inflation and activity, whether through econometric techniques or otherwise. These relationships are then used to predict the implications of current economic activity and policy. But if the measurement of inflation is constantly being changed, an acceleration or deceleration of inflation can be obscured. It becomes difficult to know if an observed acceleration or deceleration is a reflection of the actual trend in inflation or is merely a statistical artifact. Appropriate policy-making is made more difficult if measurement is based on a rubber yardstick. A consistently produced CPI facilitates the making of judgments about inflation trends.

Is the actual CPI produced using a constant methodology? In fact, it is not. During the 1990s, the BLS incorporated the following changes (among others) into the index, in part in response to the ongoing criticisms by economists of the CPI:

1991: Hedonic pricing introduced for apparel(4) 1991: Greater recognition of discount air fares 1992: Improved imputation methods for new product models 1994: Quality improvement recognized for reformulated gasoline 1995: Generic pricing recognized when drugs lose patent protection 1995: "Seasoning" procedures introduced for food to eliminate upward bias 1996: "Seasoning" extended to other products 1997: New procedures for pricing hospital services 1998: Hedonic pricing of home computers