Monetary Policy in an Uncertain World: Probability Models and the Design of Robust Monetary Rules

The past forty years or so has seen a remarkable transformation in macro-models used by central banks, policymakers and forecasting bodies. This papers describes this transformation from reduced-form behavioural equations estimated separately, through to contemporary
micro-founded dynamic stochastic general equilibrium (DSGE) models estimated by systems methods. In particular by treating DSGE models estimated by Bayesian-Maximum-Likelihood methods I argue that they can be considered as probability models in the sense described by Sims (2007) and be used for risk-assessment and policy design. This is true for any one model, but with a range of models on oer it is possible also to design interest rate rules that are simple
and robust across the rival models and across the distribution of parameter estimates for each of these rivals as in Levine et al. (2008). After making models better in a number of important dimensions, a possible road ahead is to consider rival models as being distinguished by the model of expectations. This would avoid becoming `a prisoner of a single system’ at least with respect to expectations formation where, as I argue, there is relatively less consensus on the appropriate modelling strategy.