When Did Inflation Persistence Change?
Policymakers and academics have noticed that the inflation process in the United States and other countries has changed markedly. Two formerly characteristic features of the process have been deviating from their historical norms. First, inflation persistence - the degree to which current inflation depends on past inflation - appears to have declined. Second, the relationship between current inflation and the output gap has also fallen. (The output gap is the percent by which actual output deviates from its potential.)
The timing of this decline suggests that something else may be going on. Before 2000, every percentage point in the previous year's inflation was associated with almost a 1 percentage point increase in current inflation. Six quarters later, that number had fallen to 0.4. This roughly coincides with the period of time in which the decline in inflation that had been occurring more or less steadily since the early 1990s had abated and leveled off.
To the extent that the steady decline in inflation until 2000 reflected a lowering of the Fed's implicit long-run inflation target, the timing of the change in inflation persistence may be mismeasured. A sustained decrease in long-term inflation would artificially increase measured inflation persistence since it would be picking up the persistence in the declining trend of long-term inflation. Thus, the actual decline in inflation persistence may have occurred much earlier. Survey data also suggest that over this period of time, professional forecasters were expecting inflation over the next 10 years to fall.
To correct for this effect we need some measure of long-run inflation. Unfortunately, the Fed's implicit long-term inflation target is not directly observable. We address this problem by smoothing the data. By smoothing the data, we are left with a measure of the underlying trend in inflation. This gives us a reasonable measure of long-term inflation.
By filtering out the high frequency (eg., quarterly and annual movements) we have a relationship that best captures whether inflation persistence would be declining in a period where long-term inflation is constant, as appears to be true during the current period. We also have a better measure of how the output gap affects inflation in such an environment. While the current decline in inflation persistence is historically unusual, the decline in the gap-inflation trade-off does not seem unusual. This coefficient has declined but the decline is modest and its current value is not low by historical standards. The impact of the output gap on inflation is currently (and is typically) very small.
Comparing our estimates of inflation persistence and the inflation-gap relationship for both the raw inflation data ("constant long-run inflation") and where the monetary authority's implicit long-term inflation target changes over time ("variable long-term inflation"), we see some interesting differences. The decline in inflation persistence is more pronounced and has been pushed back to around 1990.
Note that since these are 10-year rolling windows, any possible change that may have led to the decline in inflation persistence could conceivably have occurred anywhere between 1980 and 1990. Two obvious suspects, both of which occurred in the early 1980s, come to mind: the sharp decline in output variability (the so-called "Great Moderation") and the change in the central bank's operating procedure. Since 1983 the operating procedure has de-emphasized monetary targets and reacted much more aggressively to control inflation than it did in earlier periods.