Credit Spread Modeling with Regime-Switching Techniques

Abstract

Tests on Moody9s AAA and BAA corporate bond yield data consider the determinants of the excess yield earned on corporate debt over U.S. Treasuries. Cointegration estimation techniques reveal significant long- and short-run relationships. Models that allow for distinct regimes over time generate a better fit to the data, and there are some interesting differences in the key determinants of credit spreads across different regimes. Regimes are assumed to be determined by a first-order Markov process or a self-extracting threshold autoregressive process.