Bayesian versus robust control approach towards parameter uncertainty in monetary policymaking: how close are the outcomes? Some illustrating evidence from the EMU economies

This paper tries to assess the proximity of the macroeconomic outcomes which could arise from a monetary policymaking process based upon either a robust control or a Bayesian (à la Brainard) approach towards parameter uncertainty. We use a small, structural, backward-looking, aggregate model of the EMU economies as the basis for this empirical exercise. After deriving the optimal feedback rules which correspond to the two approaches that we consider in this study, we assess their relative performances with respect to the behavior of the output gap and the in.ation rate volatilities and compare with the no-uncertainty benchmark case. We are particularly interested in the output-in.ation variability trade-o¤ which is usually associated with the implementation of the optimal monetary policy rule in the literature and in the distortions that the presence of parameter uncertainty and its taking into account via the robust control approach or the Bayesian method may induce to this trade-o¤. The results show that the performances of the rules are not too divergent but they appear to be highly contingent upon the preference parameters in the model, ie the relative weight that the monetary authorities attach to output variability (w.r.t. in.ation variability) in the loss function and the robustness aversion of the policymaker which is associated to the robust control approach. In particular, non-standard shapes of the output-in.ation variability trade-o¤ obtain in the robust control case what may be due to the way the misspeci-.cations associated with the worst case scenario feedback into the structural equations of the model. When the rules are considered within the nominal model, the volatility outcomes appear to be closer to each other.