A0703
Title: Forecasting the yield curve:the role of time-varying decay parameters, conditional heteroscedasticity, and macro factors
Authors: Andre Portela Santos - CUNEF Universidad (Spain) [presenting]
Esther Ruiz - Universidad Carlos III de Madrid (Spain)
Joao Frois Caldeira - Universidade Federal de Santa Catarina (Brazil)
Werley Cordeiro - Universidade Federal de Santa Catarina (Brazil)
Abstract: The forecasting performance of several parametric extensions of the popular dynamic Nelson-Siegel (DNS) model is analysed for the yield curve. The focus is on the role of additional and time-varying decay parameters, conditional heteroscedasticity and macroeconomic variables. The role of several popular restrictions on the dynamics of the factors is considered. Using novel, end-of-month, continuously compounded Treasury yields on US zero-coupon bonds and frequentist estimation based on the extended Kalman filter shows that a second decay parameter does not have any role in obtaining better forecasts. Also, in concordance with the preferred habitat theory, the best forecasting model depends on the maturity. For short maturities, the best performance is obtained in a heteroscedastic model with a time-varying decay parameter. However, for long maturities, neither the time-varying decay nor the heteroscedasticity plays any role. The best fit is obtained in the basic DNS model with the shape of the yield curve depending on macroeconomic activity. Consequently, models for the yield curve should incorporate some sort of non-linearity depending on the maturity. Furthermore, assuming non-stationary factors is helpful in forecasting at long horizons.
