The stock market rudely rallied when the May CPI came out -- even though the core CPI rose at a 2.8 percent rate over six months and 3.8 percent over three months. Prices of "nondurable less food, beverages and apparel" rose at an annual rate of 38 percent for the three months ending in May, and energy commodities at a 91 percent rate, which shows why it is foolhardy to convert three months into an annual rate.
The reason for focusing on a "core" inflation rate over several months is to avoid confusing such transitory gyrations in a few prices with a sustained trend in the overall buying power of a dollar. But there are many ways of estimating "core" inflation, and one is demonstrably more informative than the two the Fed chairman picked.
Whenever anyone suggests increases in the price of energy can predict higher non-energy prices, they imply that the whole idea of leaving energy prices out of core inflation is wrong. Bernanke worried about "any tendency for increases in energy and commodity prices to become permanently embedded in core inflation."
In reality, the ex-energy CPI has always slowed significantly during the year following major spikes in energy prices. That is partly because the CPI ex-energy already embeds such major indirect effects as the impact of fuel prices on consumer transportation costs, which rose 9.8 percent over the past year. Yet the CPI ex-energy nonetheless remains quite tepid by historical standards.
Others have expressed concern about "asset inflation" becoming embedded in core inflation. I cannot imagine how that idea ever gained credibility. The U.S. stock market boom of 1997-2000 was not followed by higher inflation. Neither was Japan's land and stock boom of the late 1980s. Strength in asset prices in 1929 was perhaps the world's worst predictor of inflation.
For guiding Fed policy, I would prefer to exclude only energy prices, not food. Energy prices fell in 1986, 1998 and 2001, and the Fed cut interest rates every time. Energy prices jumped (with little change in non-energy inflation) in 1989, 2000 and 2005, and the Fed raised interest rates. Energy takes the Fed's eyes off the ball.
A chain-weighted index, such as "the price index for personal consumption expenditures" that Bernanke mentioned, is superior to a fixed-weight index like the CPI, which overstates inflation. The fixed-weight core CPI was up 0.3 percent in May, for example, but the chain-weighted version was up only 0.1 percent.
Economists at two Federal Reserve Banks have proposed two alternative measures of core inflation -- a median CPI from Cleveland and a "trimmed mean" CPI from Dallas. The trimmed mean tosses out the biggest increases and declines each month -- 44 percent of all items in the CPI.
Todd Clark of the Kansas City Fed compared those novel price indexes with the CPI ex-energy. He found that "with a forecasting horizon of one year ... only the CPI ex-energy offers statistically significant explanatory power for future inflation."
Robert Rich and Charles Steindel of the New York Fed found "the 'standard' ex-food and energy core measures had ... a weaker connection to the trends than some of the other candidates." Compared with the median or stripped mean indexes, however, "the ex-energy measure generally maintained its better forecasting record for PCE inflation over the longer sample period."
Unless one believes the Fed should push interest rates down whenever global energy prices fall and up when energy price rise, it follows that they should focus on a price index that excludes energy. Fortuitously, a CPI or PCE deflator without energy prices also happens to be the best predictor of broader inflation trends.
Aside from direct energy costs, the U.S. inflation rate over the past six to 12 months is still as low or lower than in all but half a dozen of the past 40 years. That is no excuse to be sanguine, to be sure, but it might provide a calming antidote to recent hysterics about inflation.