What is the core for?

Suddenly I have so much to say about the core CPI, in response to the latest attacks by Barry Ritholtz and Daniel Gross. I've already said most of it in comments to a post by Brad DeLong (also note kharris' insightful comments) and one to the original Barry Ritholtz post. I may reproduce some of them in future posts here, but my last (thus far) comment (from the DeLong post) is probably what needs to be said first:

There are really 3 separate questions here:

(1) What is the best measure of retrospective changes in purchasing power?

(2) What is the best indicator of the general trend in prices (with respect to what can expected in the immediate future)?

(3) What is the best target for monetary policy?

For the first question, obviously the full index is better than the core, and nobody denies that.

For the second question, there is room for debate, but if the only choices are the core and the full index (to measure inflation for a specific period of a year or less), I would still choose the core. (If you let me smooth the inflation rate with, say, an exponential smoother, I might prefer the full index.)

For the third question, I think there is little room for reasonable debate: the core is better. When food and energy prices go up relative to other prices, the optimal policy rule would accommodate those increases so as to allow other prices to remain stable. The increase in the general inflation rate does little harm; the alternative of deflation in the non-core component would do considerable harm.

Messrs. Ritholtz and Gross, and their supporters in this commentary, are finessing the issue by not distinguishing among these three purposes.

The answer to the third question only works if people know roughly what to expect in advance. If people expect the Fed to control the overall inflation rate but the Fed only attempts to control the core, the outcome will not be good when the two start to diverge. People who attack the core index are contributing to the likelihood of such a bad outcome, as well as to the likelihood of the other bad outcome in which the Fed actually does control the full inflation rate even in situations where it shouldn't.