A Regime-Switching Nelson-Siegel Term Structure Model and Interest Rate Forecasts

Abstract

This article presents a dynamic Nelson–Siegel term structure model subject to regime shifts. To estimate the model, we introduce the reversible jump Markov chain Monte Carlo method, which allows jumps between the one-, two-, and three-regime models. The empirical results support the two-regime Nelson–Siegel term structure model. The empirical results also suggest that the regime-switching Nelson–Siegel term structure model forecasts better out-of-sample than the single-regime Nelson–Siegel model and other competing models. In addition, our economic analysis is favorable to the better forecasting performance of the regime-switching Nelson–Siegel model. Using the Diebold-Li bond yields, we find that the better forecasting performance is robust. Finally, two regimes are found to be related to business cycle conditions and monetary policy.