Why the Fed Looks at Measures of Core Inflation

David Altig and I, and others as well, have been making the point that the Fed does not look at measures of core inflation because they measure the current cost of living for a typical consumer, they don't, the reason for focusing on core inflation is that these measures provides a better prediction of future inflation than non-core measures (e.g., see Should the Fed Focus on Core Inflation or Headline Inflation?, Which Measure of Inflation is Best for Monetary Policy?, and Trimming to the Core of the Matter, or, from David Altig, Why We Focus On Core Inflation, and Why We Focus On Core Inflation II for a few examples).

Most recently from David, he points out that the Fed believes the trimmed mean personal consumption expenditures measure (PCE), with the trim point set at 16%, is the best measure to use to forecast future inflation, better than CPI less food and energy for example:

Better, Not Best, by David Altig: In and of itself, the report on consumer price inflation in June wasn't bad. ... Except for the CPI less food and energy -- a measure of core inflation that I would invite you to jettison in favor of the trimmed-mean -- developments on the inflation front appear to be improving...

And, more from David:

Core Blind, by David Altig: The role of inflation forecasting was a prominent feature of BB's speech yesterday, and it is precisely in a forecast-based policy framework that the rationale for core inflation takes root. From today's Real Time Economics blog:

The Federal Reserve’s focus on core inflation has renewed attention to alternative measures that aim to separate the underlying inflation trend from month-to-month volatility, and some such measures are suggesting higher prices ahead.

Both the Cleveland Fed and the Dallas Fed produce “trimmed mean” price indexes that generally exclude the most volatile categories in a given month. ...

Michael Bryan, an economist at the Cleveland Fed, says the bank’s trimmed mean consumer price index does a better job in the short term at predicting future overall inflation than core inflation does. “It’s really reducing the noise and improving the signal,” Bryan said. “There’s almost no signal in the overall month-to-month CPI.”

In his comments yesterday the Chairman made it perfectly clear that the general idea is to get a bead on the inflation trend...

And it looks like the message is finally getting out with the Real-Time Economics blog noting this above, and John Berry of Bloomberg.com making the same point:

Fed Inflation Stand Criticized in 'False Debate', by John M. Berry, Commentary, Bloomberg.com: Federal Reserve officials have been criticized recently for shortchanging ordinary Americans by focusing on an inflation measure that excludes soaring food and energy prices.

That's a curious sort of populist complaint. It's really nothing less than a call for higher interest rates because the overall consumer price index rose at a 5 percent annual rate in the first half of this year and the core CPI -- which excludes food and energy -- climbed less than half as much.

It's also a complaint based on a significant misunderstanding of why the Fed pays so much attention to core prices.

Fed Chairman Ben S. Bernanke told the House Financial Services Committee yesterday that core inflation measures are a better predictor than current overall measures of what inflation is likely to be in the future. Since monetary policy affects the economy with a lengthy lag, the Fed has ''to look forward a year or two years,'' he said...

In a commentary sent to Citigroup's clients July 13, economist Robert V. DiClemente at Citigroup Global Markets Inc. called the issue of overall or ''headline'' inflation versus core inflation ''a false debate.''

''The Fed's commitment to price stability does not entail choosing one over the other,'' he said. ''History has demonstrated that so-called core measures are a good guide to headline inflation's ultimate path.'' ...

For a number of reasons, Fed officials think the personal consumption expenditure price index is a better measure of inflation than the CPI...

So, what David and others are telling us is that the Fed believes the best forecast of future overall inflation is the trimmed-mean PCE measure, with 16% the trim point on both sides of the mean.

One comment about this approach, and it is not a call for a return to the standard core measures of inflation that exclude food and energy as they have similar problems (i.e., what to exclude from the CPI is a question that is similar to determining how to set the trim point). The concern is that the trim point, currently at 16%, will change over time. It's already drifted some and it's not even clear whether a symmetric or asymmetric trimming technique is best for forecasting future inflation.

If the trim point does drift over time, and there's reason to believe it will, then in some sense we have simply pushed the forecasting problem to a different level, i.e. in order to have optimal forecasts, we will need to model how the trim point (points if its asymmetric) evolve over time in response to the state of the economy. For example, the optimal trim point could be very different for high and low rates of inflation and for different levels of overall inflation volatility. An interesting question for future research is determining what factors affect the trim point(s), and how responsive the trim point(s) are to changes in these factors.