Monetary policy and imperfect knowledge

How should stability-oriented central banks respond to monetary and economic developments when there is no consensus on how shocks and the associated policy actions are transmitted to the economy and on how the perceptions of economic agents change in response? This short essay reports on recent research findings regarding the policy implications of model uncertainty. It focuses on the robustness of simple policy rules in various models of the euro area economy that have been developed in the Eurosystem over the past five years and puts those results in the context of related research. Three findings are worth highlighting. First, optimal policy behaviour depends crucially on assumptions about how expectations are formed. While until recently only models without explicit expectations formation or with rational (or model-consistent) expectations were analysed, an increasing body of research has been investigating the implications of private sector learning. This literature generally supports the case in favour of focusing on price stability and anchoring inflation expectations. Second, most robustness exercises conclude that central banks should respond more aggressively towards undesired fluctuations of inflation when model misspecification is taken into account. This further undermines Brainard’s gradualism principle, which states that policy makers may want to act cautiously when faced with uncertainty about the effects of their policies. Third, while it is of utmost importance that policies are robust to various types of model misspecification, it has become evident that the features of robust policy rules are often dominated by the model that is least fault tolerant, i.e. the model in which small deviations from the optimal rule are most costly. This suggests that when allowing for model uncertainty, it is even more important to be very thoughtful about which models to consider. It also underlines the importance of further improving the Eurosystem’s macro-economic models. JEL Classification: E5