Do Nonlinear Monetary Policy Rules Outperform the Taylor Rule in Forecasting Exchange Rates?

This study compares out-of-sample exchange rate predictability between nonlinear monetary policy rules and the linear rule. The nonlinear rules are derived under the circumstances of central banks’ asymmetric preferences or a nonlinear Phillips curve. The results show that the predictability of the five interest rate rule based models is related to the monetary policies operated by domestic and foreign central banks. For inflation targeting countries, the Taylor rule provides the best performance and its significant predictability occurs especially at the one-month horizon. For non-inflation targeting countries, on the other hand, monetary policy rule based models have significant longer-term predictability. In addition, the monetary policy model including interaction terms, for non-inflation targeting countries, has better predictive power than the Taylor rule.