An Empirical Examination of Term Structure Models with Regime Shifts

We examine several continuous-time term structure models in which the short rate is subject both to continuous changes and to discrete shifts. Several regime-switching term structure models are developed, with regime-dependence in various combinations of their drift and diffusion parameters. We examine their predictive power. Our empirical analysis suggests that it is important to attempt to specify the switching model correctly: badly parameterized switching models may not be an improvement (in terms of pricing) over models which do not allow for regime switching, even when there are clear breaks in the data.