Testing Macroeconomic Policies with Sufficient Statistics

Abstract

The evaluation of macroeconomic policy decisions typically requires the formulation of a specific economic model. In this work, we present a framework to assess policy decisions with minimal assumptions on the underlying structure of the economy. Given a policy maker's loss function, we propose a statistic -the Optimal Policy Perturbation (OPP)- to test whether a policy decision is optimal, i.e., whether it minimizes the loss function. The computation of the OPP does not rely on specifying an underlying model and it can be computed from two sufficient statistics: (i) forecasts for the policy objectives conditional on the policy choice, and (ii) the causal effects of the policy instruments on the policy objectives. We illustrate the OPP by studying past US monetary policy decisions.