An autoregressive conditional duration model of credit‐risk contagion

Purpose – This paper seeks to discuss a modeling tool for explaining credit‐risk contagion in credit portfolios. Design/methodology/approach – Presents a “collective risk” model that models the credit risk of a portfolio, an approach typical of insurance mathematics. Findings – ACD models are self‐exciting point processes that offer a good representation of cascading phenomena due to bankruptcies. In other words, they model how a credit event might trigger other credit events. The model herein discussed is proposed as a robust global model of the aggregate loss of a credit portfolio; only a small number of parameters are required to estimate aggregate loss. Originality/value – Discusses a modeling tool for explaining credit‐risk contagion in credit portfolios.