An accessible implementation of interest rate models with Markov-switching

We examine the performance of interest rate models with regime-switching feature through a straightforward implementation. In particular, three short-rate models, the Vasicek, CIR and Black–Karasinski models, are extended to capture the switching of economic regimes using a finite-state Markov chain in discrete time. The Markov chain modulates the parameters of the model. We illustrate numerically that the resulting extended models are capable of reproducing various shapes of the yield curve. A quasi-maximum likelihood method based on James and Webber (2000) is employed to estimate the parameters of the regime-switching models. We demonstrate the implementation using actual financial datasets of Canadian yield rates. The numerical results show that under some model validation metrics, the two-state regime-switching models are more flexible, have better forecasting performance and provide better fit than the models without the regime-switching characteristic.