Intra-industry credit contagion: Evidence from the credit default swap market and the stock market.

摘要

Credit contagion has attracted considerable research interest in recent years, spawned by episodes of correlated default, the burgeoning use of synthetic credit derivatives, and the need to manage credit risk within the context of a portfolio. While new generation credit risk models proposed contagion mechanisms and factored in the impact of contagion, systematic empirical evidence in support of credit contagion is lacking.; Applying the event study methodology to a comprehensive Credit Default Swap (CDS) dataset, I investigate intra-industry credit contagion by examining the impact of credit events on industry default risk, captured in the CDS Market. I observe heterogeneity in industry CDS spread reactions conditional on the type of credit events. On average, jump events are associated with a large widening in CDS spread of industry rivals, suggesting strong credit contagion. Contagion effects also dominate for Chapter 11 bankruptcies but with a smaller magnitude. In contrast, industry rivals CDS spread narrows in response to Chapter 7 bankruptcies, implying competitive effects. Overall, contagion effects are stronger for less anticipated credit events, whereas competitive effects are stronger for more severe credit events. This pattern is justified by several models.; Conditional on type of credit events, I further investigate drivers of contagion and competition effects within a unified framework incorporating macroeconomic, industry and firm-specific factors. Particularly I identify two drivers of contagion effects undocumented in previous studies, namely, the size of firms filing for bankruptcies and the credit qualities of rival firms, which help explain the level and time variations in default correlations.; Furthermore, this paper uncovers evidence of pure contagion beyond macroeconomic and industry common factors. Finally, this study for the first time provides evidence that credit contagion is captured in the CDS market in an earlier, cleaner and stronger way than in the stock market. The empirical findings can be used to improve specification of default correlation in portfolio credit risk models, and have rich implications for credit risk portfolio management.