Essays on imperfect information, learning, and regime shifts: Implications for monetary policy

The first two chapters of the dissertation focus on the imperfect information structure that exists between monetary authorities and the public, and on how the formation of expectations affects policy choices. In the first chapter I present a modified version of Barro and Gordon's (1983b) model of time-consistent monetary policy, incorporating incomplete information regarding the preference of the policymaker and learning behavior on the part of the public. I show that lack of commitment, which is inherent in discretionary policy, is necessary to achieve high inflation persistence and high unemployment rates during a disinflationary process. As a consequence, disinflation costs are shown to be highly dependent on the degree of credibility and transparency of monetary policy.; In the second chapter I present a model of dynamically inconsistent monetary policy where an exchange rate target is implemented to combat excess inflation. The model highlights three important determinants of credibility and stability of an exchange rate target regime: (1) competing government objectives, (2) market distortions and the design of monetary institutions, and (3) verifiability. It is shown that large distortions in the economy coupled with bad design of monetary institutions reduce the credibility of the exchange rate target regime and push the long-run exchange rate away from parity. It is also shown that, under the assumption of perfect information, greater flexibility in terms of tending to competing government objectives improves the credibility of the regime. However, when there is uncertainty regarding the true commitment of the monetary authorities, then greater flexibility reduces agents' ability to verify the true nature of the exchange rate target and imposes short-term welfare costs which may destabilize the regime.; The third chapter (co-authored with Alberto Chaia) examines the forward discount puzzle and attempts to determine whether it is due to irrationally formed expectations or a time-varying risk premium. We find strong evidence against the hypothesis that agents have complete information and form their expectations rationally. The evidence in favor of a time-varying risk premium is considerably weaker. We then estimate Engel and Hamilton's (1990) regime-switching model to determine whether the ex-post irrationality could be explained by incomplete information and learning. However, a comparison of the survey data with the derived exchange rate forecasts from the regime-switching model does not provide such evidence indicating that a more sophisticated model is needed.