This thesis analyze the conduct of monetary policy in the presence of uncertainty. By adopting the framework proposed by Hansen and Sargent (2003), we analyze the behaviors of monetary authorities and private agents when faced with various sources of uncertainty, as well as their consequences in terms of macroeconomic performances. Our work shows that, in order to guard against the possibly catastrophic results of the worst-case scenario, central bankers have to react in an active manner, by manipulating the interest rate. However, in an open economy, the magnitude of this adjustment decreases with the degree of openness. Also, greater transparency of the central bank's objectives, by reducing preference uncertainty, will attenuate the variations of macroeconomic variables that follow the consideration of possible erroneous specifications. It is thus advisable to reveal informations about the central bankers' preferences, including their own estimates of the degree of model uncertainty.

The Federal Reserve Bank of San Francisco presents the full text of an article entitled "Is the Fed Too Thin? Monetary Policy in an Uncertain World," by Glenn D. Rudebusch. Estimates of the Taylor rule suggest that monetary policy in the United States can be characterized as having reacted in a moderate fashion to output and inflation gaps. The article discusses an attempt to match the historical policy rule with an optimal policy rule by incorporating uncertainty into the derivation of the optimal rule and by examining plausible variations in the policymaker's model and preferences.

Abstract: My dissertation grapples with the issues of model uncertainty in macroeconomics, and analyzes its consequences for monetary policy. It consists of three essays. In the first essay (Chapter 1), "Monetary Policy under Misspecified Expectations", I examine policy choices for the central bank that faces uncertainty about the process of expectation formation by economic agents. The economy contains both "rule-of-thumb" agents who base their expectations on recent observations and agents who have rational expectations. The central bank is uncertain about the fraction of the rule-of-thumb agents. This uncertainty concern enables me to partially rationalize the over cautious policy stance of the Fed: empirically observed policy in the past two decades involves much weaker responses than optimal policies derived from various micro-founded models. It is well understood that when the economy is more forward-looking, the central bank displays more aggressive responses to inflation and output. But the uncertainty-averse central bank evaluates policies by the performance in the worst case. In my economy this has a high fraction of agents that are backward-looking. The best policy the central bank chooses thus involves moderate responses. That is to say, this minimax policy moves closer toward actual less responsive policy. In the second essay (Chapter 2), "Phillips Curve Uncertainty and Monetary Policy", I investigate the effect of model uncertainty on policy choices employing a more general approach, which nests the minimax and Bayesian approaches as limiting cases. The central bank is uncertain about whether the economy has a sticky price Phillips curve or a sticky information Phillips curve. I argue that how the central bank chooses a policy depends both on its perception of uncertainty environment and on its attitude towards uncertainty. I find that as the central bank either becomes more uncertainty-averse or considers sticky information more plausible, the response to inflation decreases and to output increases. The third essay (Chapter 3) is entitled "Optimal Simple Rules in RE Models with Risk Sensitive Preferences". This paper provides a useful method to solve optimal simple rules under risk sensitive preference in macro models with forward looking behavior. An application to a new Keynesian model with lagged dynamics is offered and risk sensitive preference is found to amplify policy responses.