Specification of Policy Rules and Performance Measures in Multicountry Simulation Studies

Much recent analysis of international monetary and fiscal policy issues, such as the choice of an exchange-rate regime or the design of a policy coordination scheme, has been conducted by stochastic simulations with multicountry econometric models. In these studies, it has become standard practice to consider alternative policy rules of a particular form that calls for departures of a policy instrument, from some "baseline" reference path, that are proportional to deviations of a specified target variable from its own baseline path. The present paper argues, however, that this standard rule form is seriously defective for evaluating such issues because the implied rules (1) often fail to be operational and (2) have associated performance measures that can be misleading in important cases. An example is presented that concerns the international "assignment problem" of optimally pairing instruments with policy objectives.