This paper develops a general framework for conducting local robustness analysis. By local robustness, we refer to the calculation of control solutions that are determined by the least favorable model within a set of possible models that are, along a certain metric, close to an initial core model. We provide Nash and Stackelberg equilibrium characterizations of the choice of control in such contexts. We then apply this abstract formulation to the analysis of how a desire for robustness influences the choice of control for discrete time control problems of the type often found in macroeconomics. This analysis is conducted using frequency domain methods and is shown to involve certain fundamental limits to the efficacy of controls in such environments. Finally, we use these methods to identify some implications for the robust design of monetary policy.
This work is joint with William A. Brock, University of Wisconsin.
