Discretion Versus Commitment with Information Acquisition

This paper revisits the classic discretion versus commitment debate when a policymaker issues signals and agents choose to ignore them or not. Besides responding with discretion to agents’ actions or committing to a plan ex ante, a policymaker can also choose an adaptation policy, where she commits after agents make learning decisions but before agents take actions. Two mechanisms are at work in the choice of monetary policy. First, the policymaker could give agents incentives to learn if she commits before agents make learning decisions. Second, a discretionary policy can react to the aggregate shock flexibly, track agents’ actions closer, and partly control the effect of noisy information. However, this flexibility results in an inflation bias and an information bias. Overall, commitment is always the most efficient policy and adaptation may be worse than discretion under certain circumstances, especially when the policy is less important to agents than their idiosyncratic conditions. JEL Classifications: D82, D83, E52, E58