Search Now

Recommendations

Sunday, December 24, 2006

United States: Inflation Uncertainty


Richard Berner | New York

Inflation appears to have peaked in September, and inflation risks seem to have moderated, as both inflation expectations and growth have cooled over the past few months. For example, year-over-year “core” inflation measured by the CPI has declined by 0.3% in the past two months to 2.6% in November, and measured by the Fed’s preferred gauge, the personal consumption price index (PCEPI), it probably declined to 2.2%. Surprising softness in a broad range of categories — motor vehicles, air fares, communication and apparel — yielded a flat core rate in November.

Adding to the good inflation news, longer-term inflation expectations calculated by the University of Michigan’s consumer canvass remain below their summer peaks, although they edged up to 3.1% in early December. And distant-forward (5-year, 5-year) inflation compensation has moved down by 25 basis points from summer levels. Together with slower growth, those factors prompted us to reduce slightly our baseline inflation forecast for 2007 to 2.4%, and to predict a steady monetary policy until late next year. Is that inflation forecast now too high?

It could be, but before jumping to that conclusion, it’s worth remembering that there’s still considerable uncertainty over inflation measurement and key inflation determinants, and thus about the outlook. That uncertainty will probably dominate the inflation outlook and thus the monetary policy debate in 2007. Some officials legitimately take comfort from today’s well-anchored inflation expectations. But as I see it, neither policymakers nor investors should take them for granted; today’s well-behaved readings could change and don’t guarantee that inflation will recede. The commitment of monetary policy and possible policy action to assure that outcome is the missing link. Thus the Fed’s policy bias may be slow to change.

There is, to start, uncertainty over the “right” measure of underlying or core inflation. The two popular measures of core inflation both moved up over 2006, but the core CPI accelerated by 60 bp but core inflation measured by the PCEPI rose by only 0.1%. The main culprit for the divergence: Shelter, which has twice the weight in the core CPI as in the PCEPI, took off with increased demand for apartments and a sympathetic response in the so-called owners’ equivalent rent category. As these and other factors fade, these two metrics are converging. In the three months ended in November, the core CPI rose at just a 1.6% rate, while the core PCEPI probably decelerated to 1.8%.

Nonetheless, these data may exaggerate the inflation downshift. We’re suspicious that some of November’s price softness may exaggerate reality or may not last. In particular, motor vehicle discounting may ebb with inventories of new cars and trucks back to desired levels. Moreover, while airlines may have passed on lower fuel costs to fares in recent months, load factors are high and anecdotal reports point to a recent rebound in fares. And the unusual weakness in the communications category this month largely reflected a sharp drop in the price of internet access services — perhaps tied to recent price slashing by a major provider.

What’s more, there’s much less certainty over how to measure key inflation determinants and the model that links them to inflation. What are those determinants? The workhorse “markup over cost” inflation model has proven increasingly less reliable, courtesy perhaps of good monetary policy, globalization, and changes in firm pricing behavior. Indeed my own analysis suggests that firms now price “to market,” setting prices based on conditions of demand and supply in global product markets.

Both models do include three key elements, however: A measure of inflation expectations, a gauge of slack in the economy, and factors that “pass through” to underlying inflation, like changes in energy or import prices. But it appears that the slack-inflation relationship has loosened over the past several years, and that the pass-through has also diminished. This flattening of the so-called “Phillips curve” means that as slack dwindles, inflation may not rise as much today as it did in the past. But it also means that the cost of bringing inflation down may have risen.

Or has it? The price to market model may help explain this phenomenon, as companies absorb costs, including currency swings more readily into margins. But lower and more stable inflation expectations may also have shifted the relationship rather than altered its slope, so that empirical analysis must consider all these factors. Indeed, recent studies show that inflation expectations may exert a “gravitational pull” on inflation so long as a credible monetary policy provides a “nominal anchor” for them (see Brian Sack and Joel Prakken, “Inflation Modeling,” Macroeconomic Advisers, December 13, 2006). The pricing dynamics of such models are consistent with my price-to-market hypothesis.

Operationally, however, our inability to measure economic slack and inflation expectations with any precision also adds to inflation uncertainty. Measures of slack in the economy, like the output gap, are unobserved, and the unemployment rate only measures slack in labor markets, not in product markets. Some fear that potential growth has recently shrunk by as much as 1 percentage point to 2½%. In my view, it has declined, but to about 3%. In any case, that issue is a key source of today’s inflation uncertainty among policymakers and investors alike.

Likewise, Fed Vice-Chairman Kohn recently noted that “the reliability and usefulness of the existing data [on inflation expectations] are less than we might like.” And “inflation compensation measures are ‘contaminated’ both by an inflation risk premium and by differences in liquidity between the markets for nominal and indexed Treasury securities…and give only a sense of where inflation is expected to go, not why it is going there.”

That statement highlights a risk in using market-based measures of inflation compensation as independent evidence on inflation expectations: Fed policies affect market prices, so breakeven inflation reflects the Fed’s own views. Thus, Vincent Reinhart, FOMC secretary, opined in 2003 “to rely exclusively on market prices to inform policy decisions is like looking in a mirror” (“Making Monetary Policy in an Uncertain World,” August 28, 2003).

But there is also a positive element to such market-based measures: They serve as barometers of the Fed’s commitment to keeping inflation both low and stable. Fed officials can look to such measures as one barometer of their commitment to assure the right outcome. But they are not the only such measures. Richmond Fed President Lacker worries that three years of inflation running above the Fed’s presumed comfort zone will allow inflation expectations to drift higher. In that context, the Fed’s tightening policy bias serves as a commitment to cap inflation, and a contingent signal for action if needed.

Given inflation uncertainty, inflation risks seem evenly balanced around our baseline outlook: A stumbling economy could reduce inflation faster than we think likely, while stronger growth that reduced product and labor-market slack would boost it. In the spirit of the holiday season, however, it’s worth noting that one admittedly uncertain metric puts inflation well above the Fed’s presumed “comfort zone:” PNC’s Christmas Price Index. According to the 22nd annual survey, the cost of the gifts in “The Twelve Days of Christmas” is $18,920 in 2006, a 3.1 percent increase over last year. Even so, I’m most certain that the Fed’s tolerance for higher inflation than today’s is limited.