Bad luck, bad policy, and learning? A Markov-switching approach to understanding postwar U.S. macroeconomic dynamics

This paper evaluates the role of monetary policy in understanding the U.S. post-war macroeconomic fluctuations. We construct a medium-size DSGE model in which the central bank—through a real-time learning process—can make mistakes in estimating key elasticities between macroeconomic aggregates and the central banks objectives may change over time. The model also features time-varying volatilities of non-policy shocks. We find that misperceptions of the persistence of inflation and the slope of the Phillips curve could affect optimal monetary policy decision making. Therefore, their accurate prediction is key for achieving successful monetary policy outcomes. In addition, we argue that the monetary policy shift geared toward inflation stabilization in the late 1970s changed the composition of output making the marginal efficiency of investment shock—that proxies for the functioning of the financial sector—the most important contributor to the business cycle.