Robust Control Rules to Shield Against Indeterminacy

We address robustness of inflation targeting rules in a New Keynesian model using two approaches. Firstly we use the Hansen-Sargent method, borrowed from the control theory literature, to design robust rules on the basis of the policymaker playing a game against malign nature. This welfare-based approach is intended to deal with worst case scenarios, but does not directly address stability robustness. Furthermore, in the case of forward-looking systems, it does not address indeterminacy robustness; thus a system may have good stability properties, but a small parameter change could lead to indeterminacy. Secondly, we address this latter issue by imposing a probability distribution on problematic parameters, and investigate both the probability of instability and the probability of indeterminacy of the robust rule. For comparison, we apply the same idea to inflation forecast based rules, which have the potential to perform well provided that there is enough interest rate smoothing and that the forecast horizon is not too far ahead