The purpose of Discussion Paper No. 1492 is to estimate the value of a time-varying NAIRU for the US over the period 1956-1996. The model required to explain the time-series behaviour of the inflation process is based on three factors: Inertia of the inflation rate, demand and supply conditions. Using a smoothness criterion to limit jumps in the inflation rate, Robert Gordon finds that the NAIRU stays remarkably close to 6% over the entire time interval, ranging between 5.3% and 6.5%. The author rejects recent findings of wider ranges and resulting claims of its uselessness for the conduct of policy. Since the inflation rate for the GDP deflator was roughly constant from 4/94 to 1/96, and, at the same time, the actual unemployment rate was 5.6%, Gordon concludes that NAIRU must also have been very close to 5.6% during that time period.

On the basis of his findings, he suggests that excess demand during certain periods in the past was even higher than earlier research had implied. Thus, monetary policy must have been even more openly expansionary during these periods than previously thought. From '94 to '96, though, according to his estimates, monetary policy was almost exactly on target. Gordon thinks that his study makes two valuable contributions to the conduct of monetary policy: First, it quantifies in a systematic way the intuitive feeling that the US NAIRU must have fallen in the 1990s, because as of early 1996 inflation had not accelerated as it did in 1988-90. Second, it highlights the differences in the NAIRU series implied by alternative inflation measures, and thereby forces the Fed to clarify which inflation concept it is trying to stabilize.

The Time-Varying NAIRU and its Implications for Economic Policy
Robert J Gordon