The Performance Of Forecast-Based Monetary Policy Rules Under Model Uncertainty

  • 1 January 2000
    • preprint
    • Published in RePEc
Abstract
In recent years, a number of researchers have advocated monetary policy rules for setting the short-term nominal interest rate rules in response to forecasts of inflation, rather than recent outcomes of a limited set of macroeconomic variables such as the well-known Taylor rule. Furthermore, such inflation-forecast-based rules are considered to describe the policy strategies of several inflation- targeting central banks fairly well. While outcome-based rules express the interest rate as an explicit function of available information, forecast-based rules are equilibrium relations which require a forecasting model in order to generate an interest rate prescription. In this paper, we compare the performance of outcome- and forecast-based rules in four different macro-econometric models of the U.S. economy---the Fuhrer-Moore model, the MSR model of Orphanides and Wieland, Taylor's Multi-Country Model and the FRB/US staff model---and investigate directly how robust such rules are to model uncertainty. We start by looking at a set of rules taken from the literature and then turn to investigate the characteristics of forecast-based rules that are optimized in one of our four models more systematically.We find that forecast-based rules yield at best only small benefits in stabilizing inflation, output and interest rates relative to optimized outcome-based rules, which respond to inflation, the output gap and the lagged interest rate. While forecast-based rules have been recommended in the literature, because they can account for policy transmission lags and embody much information on the state of the economy, these potential advantages are quantitatively unimportant in our models. However, we find that even if output stabilization only receives a small weight in the policy objective, constraining the policy rule to respond to the inflation forecast alone and not also directly to output causes a significant deterioration in performance. As to the choice of forecast horizon,

This publication has 0 references indexed in Scilit: