Inflation Bogeyman: Hibernating or exaggerated threat?

• Author: Delong, J. Bradford

In light of current macroeconomic conditions in the United States, I've found myself thinking back to September 2014. That month, the U.S. unemployment rate dropped below 6 percent, and a broad range of commentators assured us that inflation would soon be on the rise, as predicted by the Phillips curve. The corollary of this argument, of course, was that the U.S. Federal Reserve should begin rapidly normalizing monetary policy, shrinking the monetary base and raising interest rates back into a "normal" range.

Today, U.S. unemployment is 2.5 percentage points lower than it was when we were all assured that the economy had reached the "natural" rate of unemployment. When I was an assistant professor back in the 1990s, the rule of thumb was that unemployment this low would lead to a 1.3 percentage point increase in inflation per year. If this year's rate of inflation was 2 percent, next year's would be 3.3 percent. And if unemployment remained at the same general level, the inflation rate the following year would be 4.6 percent, and 5.9 percent the year after.

But the old rule of thumb no longer applies. The inflation rate in the United States will remain at about 2 percent per year for the next several years, and our monetary policy choices should reflect that fact.

To be sure, the conventional wisdom among economists back in the 1990s was justified. Between 1957 and 1988, inflation responded predictably to fluctuations in the rate of unemployment. The slope of the simplest possible Phillips curve, when accounting for adaptive expectations, was -0.54: each percentage point decline in unemployment below the estimated natural rate translated into a 0.54 percentage point increase in inflation the following year.

The estimated negative slope of the Phillips curve--that -0.54 figure--between the late 1950s and the late 1980s was drawn largely from six important observations. In 1966, 1973, and 1974, inflation rose in a context of relatively low unemployment. Then, in 1975, 1981, and 1982, inflation fell amid conditions of relatively high unemployment.

Since 1988, however, the slope of the simplest possible Phillips curve has been effectively zero, with an estimated regression coefficient of just -0.03. Even with unemployment far below what economists have presumed was the natural rate, inflation has not accelerated. Likewise, even when unemployment far exceeded what economists presumed was the natural rate, between 2009 and 2014, inflation did not...