Parameters´ Instability, Model Uncertainty and Optimal Monetary Policy

Observed policy rates are smooth. Why should central banks smooth interest rates? We investigate if model uncertainty and parameters instability are a valid reason. We do so by implementing a novel ´´thick recursive modelling´´ approach within the framework of small structural macroeconomic models. At each point in time we estimate all models generated by the combinations of a base-set of $k$ observable regressors. Our econometric procedure delivers 2$^{k}$ models for aggregate demand and supply at any point in time. We compute optimal monetary policies for each of these specifications and then take their average as our benchmark optimal monetary policy. We then compare observed policy rates with those generated by the traditional ´´thin modelling´´ approach to optimal monetary policy and to our proposed ´´thick modelling´´ approach. Our results confirm the difficulty of recovering the deep parameters describing the preferences of the monetary policy makers from their observed behaviour. However, they also show that thick recursive modelling can, at least partially, explain the observed interest rate smoothness.