But how strong is the anchor? There are at least three types of inflation expectations: those priced into financial markets; those that appear in surveys of households and businesses; and those of professional forecasters. Market expectations have been spooking hawks. The ten-year bond yield has risen to about 1.7%, up from 0.5% in early August. However, the inflation expectations incorporated in these yields remain broadly consistent with the Fed's target. The bigger problem is tail risk. William Marshall of Goldman Sachs, a bank, calculates that the implied inflation risk premium—in effect, the price of insuring against very high inflation—has risen. The market-implied probability of average consumer-price inflation exceeding 3% per year for the next five years is over 30%, according to the Minneapolis Fed. That does not imply 1970s-style inflation, but would be uncomfortable for the Fed.

The evidence suggests that survey expectations are more important than market prices. Households' inflation expectations have not budged much, though consumers, like investors, have become less certain about the future. The danger is that the public is poorly informed, and its expectations are therefore fickle. Even firms do not seem to pay much attention to inflation nowadays. When Olivier Coibion of the University of Texas and three co-authors surveyed top executives in April 2018, 55% said that they did not know what inflation would be over the next year. When they do have a view, both firms and households chronically overestimate price rises. Consumers seem unduly swayed by the price of petrol. The authors concluded that the public's expectations looked "anything but anchored".

Professional forecasters can give Mr Powell most comfort. They are nearly unanimous and unwavering in believing what the Fed says about the long term. Yet their historical record as an early warning signal is not encouraging. As the economy overheated in the late 1960s prognosticators were behind the curve, according to the Livingston survey, the best available record of their views.

Part of the explanation is that forecasting inflation is hard. Even with today's vastly improved methods, after two years the consensus inflation forecast is on average off by 0.4 percentage points in one direction or another, calculates Goldman Sachs. Someone who forecasts that a central bank's target will lose credibility before it happens can look unhinged. Even Mr Summers—who does not suffer from excessive humility—couches his predictions in probabilities which make it nearly impossible for him to be proved wrong.

Joseph Gagnon of the Peterson Institute, a think-tank, says the Fed should promise "dramatically" higher interest rates if inflation rises and does not fall back. Saying this too soon would knock confidence. Arguably, however, the Fed is undermining the implicit understanding that it will tackle overheating by emphasising its duty to ensure a thriving jobs market that reduces inequality. That makes it harder to imagine the central bank crushing inflation by engineering a recession, as happened in the 1980s. Should enough people doubt its hypothetical resolve, the door to persistently higherinflation—or to a painful credibility test—would be ajar.

译文由可可原创，仅供学习交流使用，未经许可请勿转载。