Second‐​Guessing the Fed

The Fed's critics make it sound as though the past three months have revealed an enormous spike in "core" inflation (excluding direct energy expenses and, unimportantly, food). On Aug. 2, a Wall Street Journal report said, "The price index for personal consumption expenditures excluding food and energy (PCE)... rose 2.4 percent in June compared with a year earlier, matching the fastest annual rate since 1995." Comparing the second quarter to the first, that same core PCE index was said to have increased "at a 2.9 percent annual rate, the fastest pace in more than a decade."

The Fed statement remarked that "readings on core inflation have been elevated in recent months," yet predicted that "inflation pressures seem likely to moderate" as a result of past monetary tightening. Contrary to Fed critics, the phrase "baked in the cake" refers to lags between actions and effects, not only to excessively easy policy in 2003.

In reality, the core PCE index rose at the same 0.2 percent pace in April, May and June, leaving only March "elevated" at 0.3 percent. Monthly increases also averaged 0.2 percent during the first quarter and during the last quarter of 2005.

Martin Feldstein's recent Wall Street Journal article worried that "the doubling of the price of oil is being reflected in transportation costs and in the costs of goods that use petrochemical inputs." But all of those indirect energy expenses are included in the core PCE deflator, which is nonetheless still rising at a fairly routine 0.2 percent monthly pace. Prices of oil and natural gas cannot rise forever, and when they stop rising, that 0.2 percent monthly rate (2.4 percent a year) is all that will be left, minus today's elevated costs for transportation and petrochemicals.

Why are these unchanged 0.2 percent monthly increases now being described as the fastest since 1995 when they have not changed since last August? This mass hallucination results from statistical pitfalls involving comparing price indexes from year to year and multiplying quarter-to-quarter changes by four to fabricate an "annual rate."

The reason the year-to-year core PCE deflator appeared to increase from 2 percent to 2.4 percent over the past three months had nothing to do with faster inflation this year. It had to do with unusually slow inflation last summer -- just 0.1 percent from June to August. That pulled the year-to-year figure down from 2.3 percent in November 2004 to as low as 2 percent for a while, and it made last summer a tough act to beat on any year-to-year comparison. If the increases stick to the 0.2 percent monthly routine, the year-to-year percentage changes will look better this fall because core inflation was back to normal last fall. These year-to-year gains have fluctuated up and down between 2.0 and 2.4 percent for 27 months.

What about expressing quarter-to-quarter changes at an annual rate? On that basis, core inflation looked like 1.6 percent in the third quarter of 2005, yet I recall no front-page stories about how inflation had suddenly returned to the "deflation" levels of 2003. The first quarter of 2006 included one low 0.1 percent increase in February, so the quarter-to-quarter rate appeared to drop to 2 percent, from 2.5 percent in the fourth quarter. That created an artificially low base for the second quarter, which appeared to be 0.713 percent higher than the first -- the so-called 2.9 percent annual rate. Yet nobody dared to claim the annualized figures of 1.6 percent or 2 percent were proof of a new inflation trend, so why believe that only when the number looks higher?

Monthly increases in the core PCE index averaged 0.2 percent in all of the past three quarters, briefly touching 0.3 percent only in October 2005 and March 2006. The market-based core PCE index, which is less reliant on estimates, is up 2 percent over the past year. The similar chain-weighted CPI (which is not seasonally adjusted) increased by 0.1 percent in May and June. There is no evidence of "elevated" inflation in recent months from any chain-weighted index that excludes direct energy expenses.

But suppose we really are suffering the fastest core inflation "since 1995." What did the Fed do about this supposedly horrific inflation back then? It reduced the Fed funds rate from 5.6 percent in December 1995 to 5.3 percent in December 1996 and 4.7 percent by December 1998.

Double standards aside, those who now condemn the Bernanke Fed for failing to raise interest rates ad infinitum would have to condemn such lowering of rates by the Greenspan Fed in 1996-1999 as wildly inflationary. Yet the consumer price index excluding energy fell from 3 percent in 1995 to 2 percent in 1999.

That same CPI ex-energy was up 2.6 percent over the past year -- not the lowest it has ever been, but close. That 2.6 percent is lower than in 1995 or 2001, for example, and the total CPI including energy rose just 0.2 percent in June. Yet the Wall Street Journal editorial nonetheless bemoans "the inflationary pressures... now showing up with a vengeance in the consumer price... indices."

That editorial rebukes the Fed for drifting "back to the era of the Phillips Curve," which blamed inflation on "wage push" resulting from low unemployment. Yet the same editorial enthusiastically embraced the Phillips Curve, fretting that "unit labor costs are now 3.2 percent higher than a year ago; that's the fastest increase since 2000, when monetary policy was considerably tighter than it is now."

The increase in unit labor costs was indeed faster in 2000, which soon proved to be a better omen of a dangerous squeeze on profit margins than of inflation. The Fed's policy was unquestionably tighter in 2000. But please do not forget what happened to stock prices and industrial production in late 2000 and 2001 before emulating the Fed's policy in 2000. And recall that the resulting recession, in turn, provoked the Fed to overcompensate, belatedly, with a 1 percent interest rate.

The Phillips Curve notion that increases in unit labor costs predict or cause higher inflation was debunked in studies from the Federal Reserve Banks of Richmond (Yash Mehra), Dallas (Kenneth Emery and Chich-Ping Chang) and Cleveland. The latter paper, by Gregory Hess and Mark Schweitzer, found "little systematic evidence that... unit labor costs are helpful for predicting inflation."

A March 2005 study for the Bureau of Labor Statistics by Anirvan Banerji found that "upturns in unit labor cost growth actually lag upturns in CPI inflation 80 percent of the time."

The Fed is right this time, for a change. I regret to say that their toughest critics, including many of my oldest and best friends, are wrong.