Fractional Persistence and Common Persistence in the Conditional Variance of US Bond Yields

This paper examines persistence and common persistence in US monthly zero coupon bond yields by utilising fractionally integrated GARCH and fractionally integrated stochastic volatility models. Both conventional maximum likelihood and indirect estimation procedures are considered for these models. Simulation results are then reported on the effectiveness of the alternative estimators before empirical results are presented. The concept of cointegation in conditional variance is examined by estimating the extent of co-persistence in the bond yield spread series. The empirical results suggest that the conditional variances of the yields and spreads are best characterised as fractionally integrated, or persistent, processes.