Long run credit risk diversification: empirical decomposition of corporate bond spreads

Following the reduced-form models of Duffee (1999) and Jarrow, Lando and Yu (2003), this study investigates the risk diversification issue of corporate bond portfolios. Considering especially long run market behavior, our empirical decomposition of corporate bond yield spreads indicates that the idiosyncratic component serves as a good vehicle for risk diversification. Moreover, the diosyncratic spread provides significant inferences about observed conditional corporate bond default rate, while full spread does not. Applying an affine model from Duffie and Singleton (1999), we find that the idiosyncratic credit spreads do not respond empirically to Treasury yields, unlike what is suggested in the structural model of Longstaff and Schwartz (1995) and literatures that follow. Systematic credit spreads are however positively related to Treasury yields in the long-run, but negatively so in the short run, suggesting the validity of both the tax and the option hypotheses. A long-run and optimal decomposition scheme yields an idiosyncratic credit spread measure at a median of 60 b.p. for the Baa index and is specifically compatible with Duffee’s model. It is insensitive to interest rate in the short-run, but would rise slightly with a positive shock in the long run at a rate of one to a hundred. Our findings in the study contribute to the risk practice of bond portfolio diversification.