Some perils of policy rule regression: the Taylor rule revisited. IDEI working paper

This paper analyzes the potential misidentification involved by the estimations of policy rules trough single equations. We perform this for the Taylor rule regression in the context of two model economies, which are only different by the monetary rules they present. We show that one can easily find reduced forms that resemble the Taylor rule, even though other policy were in place. Also, we illustrate that in the case of misidentification, limited–information estimation techniques (as GMM) not necessarily deliver consistent estimators for the reduced form parameters. Thus, ruling out policy misspecification requires performing several regres-sions with important changes in the instrument set. In this sense, the Taylor rule given in Clarida, et al. (2000) fails this robustness exercise, showing that the misidentification of the rule could be important. Notably, weak evidence is found in terms of robust consistency favoring the monetary policy inertia hypothesis