Recent CPI Revisions

10/01/2000
Summary of working paper 7759
Featured in print Digest

Primarily because of the reforms instituted since the Boskin report, the current upward bias of the CPI is in the range of 0.65 percent, down from the 1.1 percent that the report estimated existed for the 1995-6 period.

In 1996 the Boskin Commission released a report analyzing the performance of the Consumer Price Index (CPI) and making recommendations to remove various biases that had crept into the index over the years. Among the most important observations of the Boskin Commission was that the CPI overstated inflation by 1.1 percent per year in 1995-6.

The media devoted an enormous amount of attention to the Commission's findings, since the CPI is a bedrock economic indicator. Biases in the CPI contaminate estimates of productivity growth, median income, and wages, and alter growth rates of government spending programs that are indexed to inflation. Biases also give misleading information to monetary policymakers and make comparative economic performance between nations more difficult to assess.

In The Boskin Commission Report and Its Aftermath (NBER Working Paper No. 7759), NBER Research Associate Robert Gordon reviews the comments and criticisms of the Boskin report, provides responses to the more important criticisms, and offers his own observations on the current status of the CPI and of price measurement research. He confines his analysis entirely to the technical issues involved.

Primarily because of the reforms instituted since the Boskin report, the current upward bias of the CPI is in the range of 0.65 percent, down from the 1.1 percent that the report estimated existed for the 1995-6 period. Gordon discusses several biases inherent in the CPI calculation and the Bureau of Labor Statistics' (BLS) efforts to eliminate them. Among them is outlet substitution bias, which refers to the BLS's practice of ignoring differences in prices for identical items across sales outlets. Superstores and discount chains are increasingly capturing market share, with prices markedly lower than those available at traditional retail outlets. Gordon estimates that the shift in market share toward superstores alone, not reflected in BLS price-gathering practices, adds 0.1 percent of upward bias to the CPI. To date, the BLS has not addressed this bias.

However, the BLS has implemented revised procedures to eliminate or moderate upward bias attributable to substitution bias, quality change, and new product introductions. These include new indexes for television sets and personal computers as well as an improved methodology for measuring medical care prices.

Gordon also points out that much of the initial criticism of the Boskin Commission report was over the political ramifications of its recommendations, namely those that would affect tax and Social Security benefit indexation, rather than any technical flaws in its work. He further points out that a fortunate side-effect of the report is the rapid pace of change at the BLS and the new attention to research issues raised by measurement bias.

-- Lester A. Picker