Abstract
We revisit the link between interest rates and corporate bond credit spreads by applying Rigobon's (2003) unique heteroskedasticity-based identification methodology to their interconnected dynamics through a bivariate VAR system. This different approach allows us to account for simultaneity issues and use this framework to test the various possible explanations for the credit spread - interest rate relation that have been proposed by the literature over the years. We find that credit spreads do indeed respond negatively to interest rates, a result consistent with Merton's (1974) structural model. This negative relation is robust to macroeconomic shocks, market uncertainty, business cycles, different sample periods, bond callability, and bond ratings. We also find the magnitude of the negative relation to be larger for high-yield bonds than for investment-grade bonds, and are able to rule out the option-like feature of callable bonds proposed by Duffee (
1998
) as the main driver of the negative nature of the relationship. These results have important portfolio and risk management implications.