Dynamic Dependence in Corporate Credit∗

Characterizing the dependence in credit spreads and default intensities across companies is a central problem in credit risk. Existing practice typically relies on factor models or simple static Gaussian copula models. We instead use genuinely dynamic copula models which can capture univariate and multivariate non-normalities and asymmetries for large cross-sections of firms. Using weekly data for over nine years on 223 firms, we find that the dependence in default intensities and CDS spreads is highly time-varying and persistent, and it increases significantly in the financial crisis. The dependence between equity returns also increases in the financial crisis, but this increase is much less persistent. We document substantial multivariate non-normalities for CDS spreads and default intensities, but multivariate asymmetries are less important for credit than they are for equities. The increase in cross-sectional dependence during the financial crisis has substantially reduced diversification benefits from selling credit protection, and it is critical to take non-normality into account when computing these benefits. These findings have important implications for the management of portfolio credit risk, the pricing of structured credit products, and the role of credit-risky securities in diversified portfolios. JEL Classification: G12