Sessão Temática 2

Co-Varying Default Risks

Beatriz Mendes (UFRJ)

We propose a new methodology for modeling joint default probabilities (DP) based on copulas combined with skew marginal distributions. We provide a comprehensive empirical investigation of the joint stochastic process of default risk using a 10-years data set of onthly probabilities of business failure within the next two years, along with the ratings for all the firms composing the S&P500 index. Our modeling strategy investigates the hazard rates of the issuers along time and includes: (1) the modeling of the skewness and kurtosis found in the individual issuer default probabilities; (2) the assessment of the linear and non-linear dependence structures shown by the PD levels across the six rating categories; (3) the investigation of the appropriate copula family linking the different categories PD levels.