Sunday, August 05, 2007

Does "Core" Inflation Predict All-Items Inflation?

I've criticized the concept of "core" inflation and the habit of completely focusing on that and only that before, a criticism which has spread to an increasing number of people. Now some people try to defend the concept by asserting that core inflation supposedly predict future all-items inflation. Yet after having studied the historical data available here I found many troubles with this view. I here use the PCE deflator rather than the CPI, not because I somehow think this index is best (I don't), but because that is what the Fed thinks is best so for the sake of the argument I use it.

First of all, particularly in recent years all-items inflation (I use the word "all-items" instead of "headline" as "core" inflation now is what typically dominates the headlines) have typically been higher than core inflation, so core inflation systematically under-predicts all-items inflation. During the latest 7-year period, all-items PCE deflator has increased at an average annual rate of 2.35% while core PCE deflator increased 1.93%.

As Peter Schiff commented on Ben Bernanke's assertion that energy prices would soon moderate:

"Finally, Bernanke dismissed concerns about the wisdom of favoring core inflation over headline by asserting that oil prices will soon moderate. Considering that oil prices rose another 2% during his two-day testimony, and that he and his predecessor have consistently underestimated oil prices for years, what now makes his crystal ball any clearer?"

Another and even more serious problem is that there is in fact no tendency at all for movements in core inflation to somehow predict future movements in all-items inflation. During 2001 for example, core inflation accelerated from 1.5% to 2.2% (with most of the increase coming during the first half of the year), yet all-items inflation fell during 2002. During 2002 and 2003 core inflation decelerated sharply, yet all-items inflation accelerated during 2004. Between 2004 and the first half of 2006 core and all-items inflation both reached and stayed at above average levels. In mid-2006 (the second and third quarters) "core inflation" reached post-1999 highs, yet all-items inflation fell back sharply during the fourth quarter of 2006 and first quarter of 2007. Only in the second quarter of 2007 did all-items inflation again accelerate, but only after core inflation had started to moderate significantly. So to the extent there is any correlation between current core inflation and future all-items inflation, this correlation is negative.

By contrast, there does seem to be some positive correlation between current movements in all-items inflation and future core inflation. The acceleration in all-items inflation in early 2000 was followed by acceleration in core inflation in 2001. The deceleration of all-items inflation during particularly the first half of 2002 was followed by a sharp deceleration of core inflation during 2002 and 2003. During 2004, we as previously mentioned saw all-items and core inflation accelerate significantly at about the same time, but this significant acceleration was preceded by a smaller acceleration all-items inflation while core inflation remained low. Then during the second half of 2006 we saw a significant deceleration of all-items inflation followed by a significant deceleration in core inflation during the first half of 2007.

Anyone who understands sound economic theory shouldn't be surprised by these results. After all, if you have a central bank which inflates the money supply, which prices would you expect to increase first? Relatively flexible prices such as food and energy whose commodities are traded on financial markets, or relatively sticky prices such as most items in the "core" index? Well, you don't have a genius to realize it is going to be the flexible prices which will be increased first and that the sticky prices will be affected only later. Which means that the movements of the flexible prices should have a relatively good predictive power for the movements of the sticky prices. On the other hand, there is no reasons to expect the movements of the sticky prices to predict the movements of the flexible prices. Indeed, to the extent inflation comes in waves, you should expect the sticky prices to at least sometimes start to rise just before the flexible prices starts to fall.

There are two noteworthy implications of this: first of all, again it has been shown that there are no legitime reasons for the Fed to focus on "core inflation". And the only reason for the rest of us to focus on it is because the Fed focuses on it and because of this core inflation provides a hint of their interest rate decisions.

Secondly, if historical patterns persist, we should see core inflation accelerating soon reflecting the sharp acceleration in all-items inflation during the first half of 2007. That will make it difficult for the Fed to lower interest rates to counter the recessionary effects of the subprime mess.

2 Comments:

Anonymous Anonymous said...

I can not be short on this item.
As a matter of fact, headline CPI and core CPI are twins, i.e. there is a trade-off for the mid-term evolution of these variables .
http://ideas.repec.org/p/pra/mprapa/6900.html

8:26 PM  
Anonymous Anonymous said...

I have recently been doing some research on "core inflation" and have learned that CI was created by Fed Chairman Arthur Burns at the behest of Presdent Rchard Nixon.

Kevn Phillps, a Nixon aide, tells the story http://www.mindfully.org/Reform/2008/Pollyanna-Creep-Economy1may08.htm:

Richard Nixon, besides continuing the unified budget, developed his own taste for statistical improvement. He proposed albeit unsuccessfully—that the Labor Department, which prepared both seasonally adjusted and non-adjusted unemployment numbers, should just publish whichever number was lower. In a more consequential move, he asked his second Federal Reserve chairman, Arthur Burns, to develop
what became an ultimately famous division between "core" inflation and headline inflation. It the Consumer Price Index was calculated by tracking a bundle of prices, so-called core inflation would simply exclude, because of "volatility," categories that happened to he troublesome: at that time, food and energy. Core inflation could he spotlighted when the headline number was embarrassing, as it was in 1973 and 1974.

12:36 PM  

Post a Comment

<< Home