Coincident with the latest news on consumer inflation (it was tame in June), the Dallas Fed today rolled out its new and improved measure of core inflation based on the price index for personal consumption expenditures.  The motivation for this new measure, which is similar in spirit to the Cleveland Fed's median CPI statistic, is nicely articulated in a background working paper by Dallas economist Jim Dolmas:

Measures of inflation that exclude food and energy prices are probably the most well-known estimators of core inflation.  In fact, the "excluding food & energy" measures (XFE for short) are often spoken of as if they are synonymous with core inflation.  Properly speaking, though, they represent just one of many potential core measures...

... in any given month, excluding only food and energy items still leaves many volatile items in the price index... excluding all food and energy items may throw out some useful information.

What the Dallas folks have proposed is a "trimmed mean" estimator. Here, in a nutshell, is why:

In a study focusing on the CPI and the PPI, Michael Bryan, Stephen Cecchetti and Rodney Wiggins make a statistical case for the use of trimmed means as a method for estimating core inflation.That case relies on the fact that trimmed means can be more efficient estimators of a distribution’s location, as compared to the sample mean, when the distribution is characterized by heavy tails. Intuitively, samples drawn from a heavy-tailed distribution will contain relatively large numbers of extreme values. As a result, the sample means, which are sensitive to outliers, will have a high variance. Trimming some fraction of those extreme observations produces a less volatile estimator, a fact which Bryan, Cecchetti and Wiggins illustrate…

(You can regularly find a picture of the trimmed mean calculated from the Consumer Price Index in the Cleveland Fed's Economic Trends publication.)

Here's the bad news from the trimmed-mean PCE calculations:

Compared to the story told by the usual "excluding food & energy" measure, the trimmed mean PCE tells us that the lows reached in 2003 weren't quite as low and that the highs reached in mid-2004 were really a bit higher. On a 12-month basis, the new measure suggests that core PCE inflation is currently about 1/2 a percentage point higher than what is being indicated by the "excluding food and energy" inflation rate.

That difference, based on the monthly inflation rate, actually got bigger in June.  The better news is that twelve-month trend in both the trimmed and ex-food-&-energy core measures is down.  Here's the table, from the Dallas website:

12-month PCE inflation

Jan.
05

Feb.
05

Mar.
05

Apr.
05

May
05		 June
05
PCE
2.7
2.6
2.7
2.9
2.5
2.2
PCE excluding food & energy
2.2
2.2
2.1
2.0
2.0
1.9
Trimmed mean PCE
2.4
2.3
2.3
2.2
2.2
2.1

This does contrast with the 12-month change in the median CPI, which has mainly held steady for most of the year (at 2.3 to 2.4 percent).

That difference is worth monitoring, but in any event kudos to the Dallas Fed for adding another useful statistic to the toolkit.