Essays on incomplete information, model uncertainty, and macroeconomic policy

In this dissertation I study the role of incomplete information and model uncertainty in the design of macroeconomic policies.;In Chapter 1, I study the information contained in the equilibrium aggregate price level of an economy where firms make output price decisions faced with incomplete information about economy-wide disturbances. This chapter makes three contributions to the literature on monetary business cycle and incomplete information. First, it proposes a set of techniques in the frequency domain that allow for the explicit derivation of individual heterogeneous expectations while preserving the tractability of the fixed point condition typical of rational expectations. Second, it develops and solves a stylized model where aggregate price plays a key informational role of the type advocated by Hayek. Finally, it presents an application to monetary policy.;Chapter 2 characterizes the frequency domain properties of linear systems with feedback control rules. The goal of the analysis is to derive restrictions on how feedback rules restrict the frequency by frequency fluctuations that underlie a time series of state variables. Existing results in the control theory literature are expanded to account for discrete time bivariate systems with rational expectations. The basic methods presented in this chapter provide ways to understand how fluctuations at different frequencies are subject to tradeoffs via the choice of a feedback rule.;Chapter 3 contributes to the policy evaluation literature by developing new strategies to study alternative policy rules. I compare optimal rules to simple rules within canonical monetary policy models. The policy rules are evaluated under minimax and minimax regret criteria. These criteria force the policymaker to guard against a worst-case scenario, but minimax regret confronts the policymaker with uncertainty about the true model. The results indicate that the case for a model-specific optimal rule can break down when uncertainty exists about which of several models is true. Further, it is shown that the assumption that the policymaker's loss function is known can obscure policy tradeoffs that exist in the short, medium and long run. Thus, policy evaluation is more difficult once it is recognized that model and preference uncertainty can interact.