In formulating monetary policy, central banks must cope with substantial economic uncertainty. Economic uncertainty can arise from different sources: the state of the economy, the nature of economic relationships, and the magnitude and persistence of ongoing shocks. Robust control theory instructs decision makers to investigate the fragility of decision rules by conducting worst-case analyses. In this paper we show how state space methods and structural-form solution methods can be applied to robust control problems, thereby making it easier to analyze complex models. We illustrate the state space solution methods by applying them to an empirical New Keynesian business cycle model of the genre widely used to study monetary policy under rational expectations. A key finding from this exercise is that the strategically designed specification errors will tend to distort the Phillips curve in an effort to make inflation more persistent, and hence harder and more costly to stabilize. The optimal response to these distortions is for the central bank to become more activist in its response to shocks.
