We establish that the Phillips curve is persistence-dependent: inflation responds differently to persistent versus moderately persistent (or versus transient) fluctuations in the unemployment gap. Previous work fails to model this dependence, so it finds numerous “inflation puzzles”—such as missing inflation/disinflation—noted in the literature. Our model specification eliminates these puzzles; for example, the Phillips curve has not weakened, and inflation is not “stubbornly low” at present. The model’s coefficients are stable, and it provides accurate conditional recursive forecasts through the Great Recession. The persistence-dependent relationship we uncover is interpretable as being business-cycle-phase-dependent and is thus consistent with existing theory.
The origins of the Great Inflation, a central 20th century U.S. macroeconomic event, remain contested. Prominent explanations are poor forecasts or deficient activity gap estimates. An alternative view: the FOMC was unwilling to fight inflation, perhaps due to political pressures. Our findings, based on a novel approach, support the latter view. New econometric tools allow us to credibly identify the particular activity gap, if any, in use. Persistence-dependent unemployment (gap) responses in the 1970s were essentially the same pre- and post-Volcker. Conversely, FOMC behavior vis-à-vis inflation—also persistence-dependent—changed markedly starting with Volcker, consistent with (though not proving) the political pressures view.
We use recently developed econometric tools to demonstrate that the Phillips curve unemployment rate–inflation rate relationship varies in an economically meaningful way across three phases of the business cycle. The first (“bust phase”) relationship is the one highlighted by Stock and Watson (2010): A sharp reduction in inflation occurs as the unemployment rate is rising rapidly. The second (“recovery phase”) relationship occurs as the unemployment rate subsequently begins to fall; during this phase, inflation is unrelated to any conventional unemployment gap. The final (“overheating phase”) relationship begins once the unemployment rate drops below its natural rate. We validate our findings in a forecasting exercise and find statistically significant episodic forecast improvement. Our analysis allows us to provide a unified explanation of many prominent findings in the literature.
The historical analysis of FOMC behavior using estimated simple policy rules requires the specification of either an estimated natural rate of unemployment or an output gap. But in the 1970s, neither output gap nor natural rate estimates appear to guide FOMC deliberations. This paper uses the data to identify the particular implicit unemployment rate gap (if any) that is consistent with FOMC behavior. While its ability appears to have improved over time, our results indicate that, both before the Volcker period and through the Bernanke period, the FOMC distinguished persistent movements in the unemployment rate from other movements; implicitly such movements were treated as an intermediate target, one that departs substantially from conventional estimates of the natural rate. We further investigate historical FOMC responses to inflation fluctuations. In this regard, FOMC behavior changed in the Volcker-Greenspan-Bernanke period: its response to the inflation rate became much stronger, and it focused more intensely on very persistent movements in this variable. Our results shed light on the “Great Inflation” experience of the 1970s, and are consistent with the view that political pressures effectively limited the FOMC response to the buildup of inflation. They also suggest new directions for DSGE modeling.
We estimate a monetary policy rule for the US allowing for possible frequency dependence (the central bank can respond differently to persistent and transitory innovations) in the unemployment and inflation rates.