When the economy is subject to recurrent structural shifts, the monetary authority cannot credibly commit to a systematic approach to policy, since consistency between promises and actions is not easily verifiable; moreover, since agents have incomplete knowledge of the surrounding environment, they form expectations that may deviate substantially from the full-information case. The present paper studies the implications for the effectiveness of discretionary monetary policymaking of departing from the benchmark of rational expectations and assuming instead that agents learn adaptively. It focuses on two issues, namely whether imperfect knowledge generates a bias against stabilisation policies and whether the optimal monetary strategy takes the form of an inflation cap. Rules featuring an inflation cap are not only justified on theoretical grounds, but are also appealing because they seem appropriate to deal with imperfect knowledge and learning: by setting explicit bounds on inflation, they seem better suited to restrain expectations from drifting significantly away from target, thus removing one of the main sources of policy ineffectiveness. The main findings of the paper are the following. First, when agents do not possess complete knowledge on the structure of the economy and rely on an adaptive learning technology, a bias toward conservativeness arises. Second, a policy that involves a cap on inflation is helpful in reducing output and inflation variability, but it is not uniformly superior to a strategy aimed at minimising a quadratic loss function. Third, the bias against stabilisation policies and towards conservativeness does not depend on whether agents have finite or infinite memory.
Published in 2007 in: International Journal of Central Banking, v. 3, 3, pp. 47-85
