Content area
Full Text
The purpose of this study is to develop a regime-switching extension of the dynamic Nelson-Siegel term structure model and apply it to Japanese corporate bond spread data on an individual firm basis for the period April 1997 through December 2011. The results indicate that the estimated regime probability is closely linked to business and market sentiment. The regime-switching model is extended by adopting a time-varying transition probability matrix driven by leading macroeconomic indicators. The overall fit is improved by incorporating a time-varying transition probability matrix. Our results imply the importance of incorporating regime shifts into modeling the term structure of credit spreads.
(ProQuest: ... denotes formulae omitted.)
INTRODUCTION
Recent studies on the term structure of credit spreads explain credit spread dynamism using an affine term structure model or a Nelson-Siegel model (Nelson-Siegel (1987)). These models are consistent with principal component analysis in which the first, second, and third factors explain over 90% of credit spread variation.
However, the question of whether the current term structure model explains 100% of the variation of the yield curve considering the non-normality or discontinuity of credit spreads remains unanswered.
In the research on the term structure of government bond yields, the models incorporating the jump process, or regime shifts, have considered the non-normality or discontinuity of government bond yield. The development of a term structure with regime shifts was earliest introduced by Hamilton (1989), Garcia and Perron (1996), and Gray (1996). The authors developed and estimated time series models to capture the dynamism of short-term interest rates. Landen (2000) solved the closed-form solution of bond price with two regime shifts using the Gaussian model, while Bansal and Zou (2002) studied a two-factor Cox-Ingersoll-Ross model with mean- reverting parameter shifts and shifts in both volatility and the market price of risk. Both studies insist that the regime-switching model has superior goodness of fit to the three-factor affine term structure model, and regime is closely linked with the business cycle and monetary policy.
However, as Litterman and Schenkman (1991) note, over 90% of the yield curve variation is explained by three principle components (level, slope, and curvature) of the term structure. One or twofactor models require modification even if they admit regime shifts.
To overcome this difficulty, Dai and Singleton...