Robust Optimal Monetary Policies in Behavioral New Keynesian DSGE Models

This paper addresses the challenge of uncertainty in monetary policy by incorporating local model uncertainty arising from heterogeneous expectations into a behavioral New Keynesian DSGE framework. Non-Bayesian control techniques are adopted to minimize a welfare loss derived from the second-order approximation of agents' utilities to derive robust optimal policies. In the context of uncertainty regarding the formation of behavioral expectations, the importance of gradualism in monetary adjustments is emphasized. Policymakers should consider the ratio of rational to boundedly rational agents and anticipate that a significant presence of boundedly rational agents, especially those with long-horizon expectations, may require more dynamic adjustments to interest rates. Accounting for multidimensional uncertainty exponentially increases the complexity of the analysis and model indeterminacy.