Default Correlation Dynamics with Business Cycle and Credit Quality Changes

Abstract

The distinction between structural and reduced form credit risk models is now well known. Structural models arise from more compelling theoretical principles, but reduced form models are easier to implement in practice. As derivative instruments tied to portfolios of credit–sensitive securities, such as collateralized debt obligations (CDOs) and basket credit derivatives, become more common, the need to understand and model default correlations along with default probabilities also becomes important. But structural models have had a hard time producing the amount of correlation needed to match market prices for these products, partly because default intensities are typically modeled as being constant, or at most, functions of largely firm–specific variables. In this article, Kim and Kim introduce a tractable structural model in which default intensities also depend on broad macroeconomic variables that are assumed to follow correlated diffusions. This joint dependence on macro factors can induce strong default correlations, and can also help to explain certain results from the literature, like the fact that credit risk seems to be more highly correlated among high–grade firms than among low–grade firms, even though the latter have higher individual probabilities of default. The model can also produce a rich dynamic in credit risk exposure.