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

Challenge, Sept-Oct, 1998 by Daniel Mitchell

The author proposes a range of alternative CPIs. One estimate cannot satisfy all needs.

A cynic, Oscar Wilde observed a long time ago, "knows the price of everything and the value of nothing." More recently, many economists have become cynics about our official index of the price of (virtually) everything, the Consumer Price Index (CPI). The Boskin Commission in 1996 summarized most of the complaints about the CPI.(1) The commission argued that the CPI, as the Bureau of Labor Statistics (BLS) produces it, is beset by inflation overstatement. It identified two faults in the CPI: insufficient adjustment for product quality and lack of recognition of consumer substitution. The Boskin Commission also criticized the BLS for other technical deficiencies in the index. In the background of this debate loomed the fact that, by law, increases in the CPI cause social security payments and other government benefits to rise. And, because of indexation of income tax brackets, federal revenues increase with any rise in the index.

By no means do all economists agree with the criticisms of the CPI.(2) Nor are most familiar with the details of its construction. However, all economists and policy-makers should understand that continued debate about the index undermines public confidence in official economic data. At various points, both Republicans and Democrats in Congress have threatened the BLS with extinction if it did not change its methodology. Republican Speaker of the House Newt Gingrich once proposed to "zero...out" the BLS if the agency did not alter the CPI to his liking. And so-called Blue Dog Democrats, led by Representative Charles W. Stenholm of Texas, once proposed mandating methodological changes.(3) If the Bureau of Labor Statistics makes changes in the CPI that move in the direction suggested by the Boskin Commission, the agency might be accused of kowtowing to political pressure. On the other hand, if the BLS does not make changes, the old complaints will continue and the agency will be seen as unresponsive.

I am proposing a resolution to the CPI controversy that is not related to the technical issues entailed in measuring quality or accounting for substitution, one that is also applicable to other official data series. I am suggesting that the CPI "problem" has been framed by an approach to official data that can only lead to continued conflict: the assumption that there is but one theoretically correct CPI. Even if perfection cannot be fully achieved in practice, proponents of this erroneous approach believe that statistical policy should at least aim at approximating a particular idealized index. My menu approach, in contrast, is based on the view that there are many plausible and reasonable versions of the CPI. I argue that a variety of alternatives should be offered to meet the preferences of CPI users.

Taylorism or Choice?

In the late nineteenth century; management guru Frederick W. Taylor argued that there was "one right way" for factory production tasks to be accomplished. By using "scientific" methods, that one right way could be both determined and imposed on factory workers. Nowadays, of course, the rigidity of Taylorism has fallen from grace in management circles. What is peculiar is that Taylorism should continue to be applied by knowledgeable economists to the quest for a single, perfect price index.

Economics, after all, is often defined as the science of choice. And Taylorism implies that there is no choice. Suppose that, for example, I were to argue that the public would be best served by having government and academic experts design an ideal automobile for consumers. Suppose that I further proposed thereafter that only one ideal model should be available. Most economists would recoil in horror at such a proposition. They would point to the varied preferences of automobile consumers. Some consumers prefer full-size cars; others want compacts. Some like sporty convertibles; others prefer utilitarian sedans. Why should a group of experts, however well-meaning or technically proficient, impose their will on others?

Taylorism and Official Data

If the one-right-way approach is not particularly appealing for automobiles, then consider the way major economic data series, including the CPI, are typically produced. Essentially, government experts - sometimes advised or critiqued by official task forces (such as the Boskin Commission) - design a methodology of data collection, assembly, and presentation. The end product is then presented to the public as the "official" Consumer Price Index (or Producer Price Index or gross domestic product [GDP] or other series). This approach to official statistics is simply "data-Taylorism." Why is Taylorism still the underlying methodology of official data?

There are several justifications for the current system of CPI production that might be cited. First, government has been perceived as a "neutral" source of data. A price index produced by, say, the American Association of Retired Persons might be suspect; perhaps such an index would be deliberately designed to exaggerate inflation and pump up social security benefits. Second, the government has the authority to compel or induce cooperation by data respondents. Absent such authority, failure to obtain cooperation could produce undesirable biases in data collection. Third, official statistics are public goods. Particularly for general indexes such as the CPI, it is not clear that there would be a sufficient private market to make collection and sale profitable. Even if the CPI were sold commercially, it would be difficult to prevent knowledge of it from being disseminated without payment.