The Form and Function of Price Indexes: A Historical Accounting

At least when speaking theoretically, most economists today are likely to define a price index for measuring the cost of living in explicitly utilitarian terms. That is, they would define it as the ratio of expenditure functions required to hold utility constant in two different scenarios. Indeed, this perspective is typically referred to as the "economic approach" to index numbers, and the formula as the "true cost-of-living index" at the relevant level of utility. Likewise, in practice, economists have tended to recommend that official price indexes try to approximate this theoretical definition as much as possible. The 1961 Stigler commission (Stigler et al. 1961) on the U.S. Consumer Price Index (CPI) was the first to recommend this standard, and the more recent Boskin commission (Boskin et al. 1996) made it their "overarching recommendation" for reforming the CPI. In response, after years of hesitation, the U.S. Bureau of Labor Statistics (BLS) now accepts this standard in evaluating how best to construct the CPI (U.S. Bureau of Labor Statistics 1997).¹ Most contemporary academic research on price indexes, both within and outside the BLS, works from this premise.²

However, there continues to be some skepticism about making this the objective of official price indexes such as the CPI. Most notably, after the BLS accepted the goals of the cost-of-living framework, the most recent external review of the U.S. CPI (Schultze and Mackie 2002) reversed the suggestions of the Boskin commission, refusing to endorse the cost-of-living concept.³ Some objections coming from the committee and from others are quite broad, speaking to the coherency of welfare economics generally—or even much of neoclassical economics.⁴ However, the committee's primary objection is that in fact the price index has many competing objectives. One set of objectives relates to the cost of living, including the indexing of wages and benefits. These objectives are consistent with the utilitarian or "economic" approach. But another set relates to a measure of inflation for the purposes of macroeconomic forecasting and monetary policy. For these objectives, it is not as clear that constructs from utility theory are relevant. The Schultze committee refers to these two index concepts as the cost-of-living index (COLI) and the cost-of-goods index (COGI), respectively, or "the cost of living" versus "the cost of things." Since both objectives are important, the Schultze committee refused to endorse the cost-of-living framework for the CPI.

While the Schultze committee is quite clear about these different goals—and the way that they would each inform specific problems in constructing the index—the problem of reconciling them remains. That is, if it is not possible to have a separate index for each purpose, what is the single best index for competing purposes? A reading of the history of index numbers in the United States shows that this is not a new conundrum. This article considers a similar period in the history of price indexes, from approximately 1915 to 1950, a time when something like the "cost-of-things" approach was the norm and the utilitarian approach was new.⁵ The article begins with what I call the "physical approach" to price indexes, exemplified by Irving Fisher and Wesley Clair Mitchell—two early giants in the field. It then goes on to introduce the "economic" or indifference-defined index. It focuses on two debates in this time period and suggests that the disagreements (and confusion) were grounded in different conceptions about what price indexes measure. The first was a well-known debate between Fisher and Mitchell—about whether there was a single "ideal" index that could address all contexts. The second occurred during World War II, when another review panel, chaired by Mitchell, addressed questions about the index raised by severe quality deterioration during wartime. In telling this story, a major theme will be...