Measuring and controlling inflation.

• Author: Cecchetti, Stephen G.

In many ways, the monetary policy of the last 15 years has been strikingly successful: inflation has fallen dramatically, from well over 10 percent per year in the late 1970s to only about 3 percent in recent years. But this new low-inflation environment brings with it a host of questions that do not arise when inflation is at moderate or high levels. For example, is inflation being measured accurately? And, how can policymakers control the future path of inflation? These questions have guided much of my research over the past several years.

Michael F. Bryan of the Federal Reserve Bank of Cleveland and I studied the measurement of the most popular and commonly used aggregate price statistic in the United States: the Consumer Price Index (CPI). The CPI is the most prominent measure of inflation, and thus has become a focal point in the Federal Reserve's inflation fight. Broadly speaking, there are two problems associated with using the CPI to measure inflation: First, it represents month-to-month price changes that may not reflect changes in long-term trends accurately. Measured changes in the CPI often contain substantial short-run, transitory noise that does not constitute long-term inflation, and can easily mislead the monetary authorities when making decisions. Second, there is a potential bias in the CPI that results from both the expenditure-based weighting scheme it uses and from the persistent errors associated with measuring certain prices. As a result, measured CPI inflation will not reflect the long-run trend in price changes accurately. Since the IRS code, many government expenditure programs, and numerous private contracts all use the CPI as the basis for their indexation, its proper measurement of long-term price movements is of the utmost importance.

Core Inflation

Core inflation is what remains if we can remove from our calculation of inflation changes in variables such as the weather, or movements in oil and clothing prices, and thus obtain a more reliable measure of the underlying trend in price movements. For example, food prices may rise during periods of poor weather to reflect a decrease in supply, thereby producing transitory increases in an aggregate inflation index. Because these price changes do not constitute longer-term inflation, the monetary authorities would not want to base their decisions on them.

There are numerous solutions to the problems posed by temporary movements in prices. One common technique for measuring core inflation excludes certain prices in the computation of the index, assuming that those prices are highly variable. (This is the "ex. food and energy" strategy, in which the existing index is reweighted by placing zero weights on some components.) But is it right to assume that food and energy prices contain so little valuable information about core inflation that they should be excluded completely from the index? And how can we adjust for price variability outside of the food and energy sectors?

As an alternative, Bryan and I examine the Median CPI:[1] that is, the midpoint of inflation...