Skip to main content

Rethinking NAIRU

Top of the Ninth

June 1, 2000


The U.S. economy, which has performed admirably since the early 1980s, has been especially impressive over the past five years. While the record is remarkable along a number of dimensions, one of the most striking has been the size and consistency of gains in employment. Another has been continued progress in reducing inflation. The employment increases have surpassed growth of the labor force, so the unemployment rate dropped below 6 percent in 1994, below 5 percent in 1997 and below 4 percent recently. And today inflation, however measured, is less than 3 percent per year, far below the rates of the early 1980s.

Low unemployment accompanied by low inflation is not what many commentators and economists had expected, however. Instead, they had expected declines in unemployment—and concomitant tightening of labor market conditions—to lead to pressures on wage costs and inflation. The argument, in its straightforward guise, is that as the labor market tightens, upward pressure on compensation increases, and this pressure is reflected, sooner or later, in rising prices. A more sophisticated version of the story invokes the concept of the nonaccelerating inflation rate of unemployment (NAIRU) and posits that, when measured unemployment falls below NAIRU, compensation and inflation accelerate.

Although estimates vary, NAIRU is usually put at somewhere between 5 percent and 6 percent for the U.S. economy. The economy, thus, has been decidedly below NAIRU for some time, and yet there has been little discernible evidence of the anticipated wage-price acceleration. Explanations for this unexpectedly favorable performance have tended to emphasize factors that have characterized the past several years but that may not persist: acceleration of improvement in labor productivity, strong international value of the dollar, declining (for a time) energy prices, flat (for a time) health care costs and so on. The argument is that erstwhile inflationary pressures have been offset by the good luck experienced with these other variables.

This explanation, which relies on fortuitous special circumstances, could be correct but, as usual in economics, it appears that the situation is more complicated. In particular, recent analysis by Andrew Atkeson of our staff strongly suggests that there has not been a significant statistical relation between unemployment and inflation since the mid-1980s. The important implication of this research, then, is that NAIRU has failed to help predict inflation for quite some time.

Empirical work is rarely definitive in economics, so it is premature to discard the NAIRU concept or the general notion that labor market conditions may be a precursor of inflation. But it would seem that the burden of proof has shifted to those who continue to assert NAIRU's value in understanding and predicting inflation.

There are, perhaps, at least two lessons in all of this. First, an effort to understand short-run inflation pressures is likely to require a
broad-based review of economic conditions going well beyond labor market data. And secondly, it is well to recall that there is an impressive body of research indicating that in the long run inflation is a monetary phenomenon, and thus it is the responsibility of Federal Reserve policy to maintain a low-inflation economy.