Model-Based Ex Post Evaluation of Monetary Policy

We present a model-based methodology to conduct an ex post evaluation of monetary policy decisions, by testing whether alternative policy decisions could have brought a Pareto improvement in terms of inflation and output volatilities. This involves simulations of counterfactual scenarios under alternative monetary policy shocks, and computation - for each such simulation - of the root mean square (RMS) of the inflation and output gaps during and following the evaluated year. It is then possible to compare the actual RMS with simulation-based frontiers, with each frontier reflecting different constraints on interest rate volatility, which can be viewed as a third objective variable. The actual RMS is also compared with the counterfactual RMS that would have been obtained under the case of no policy shocks. Such comparisons enable testing whether monetary policy shocks were “ex post efficient.” The methodology is implemented in an evaluation of Bank of Israel policy decisions during the years 2001–11. The implementation shows several distinct sets of years: years in which actual RMSs were close to the efficient frontiers and years in which they were distant from them; years in which monetary policy shocks led to an absolute improvement in economic outcomes (by reducing the RMSs of both inflation and output gaps) or an absolute deterioration; and years in which policymakers faced a trade-off between all three objective variables or between a subset of the variables. For most evaluated years, the results seem qualitatively robust, considering the uncertainty in the historical shocks extracted, as well as in alternative definitions for the output gap.