Simple versus optimal rules as guides to policy

This paper contributes to the policy evaluation literature by developing new strategies to study alternative policy rules. We compare optimal rules to simple rules within canonical monetary policy models. In our context, an optimal rule represents the solution to an intertemporal optimization problem in which a loss function for the policymaker and an explicit model of the macroeconomy are specified. We define a simple rule to be a summary of the intuition policymakers and economists have about how a central bank should react to aggregate disturbances. The policy rules are evaluated under minimax and minimax regret criteria. These criteria force the policymaker to guard against a worst-case scenario, but in different ways. Minimax makes the worst possible model the benchmark for the policymaker, while minimax regret confronts the policymaker with uncertainty about the true model. Our 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, we show that the assumption that the policymaker’s loss function is known can obscure policy trade-offs 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.