Credit risk identification with Hawkes processes: Theory and evidence

This study utilizes the Hawkes process as an alternative to the Poisson distribution assumption in the jump-diffusion KMV (JD-KMV) model, thereby enhancing the assumption of the expected asset jump size and deriving the Hawkes jump-diffusion KMV (HJD-KMV) model. Subsequently, a regression analysis is conducted on the default distance calculated by the model using bond spreads as a proxy variable for credit risk. The findings reveal that the enhanced jump frequency assumption in the HJD-KMV model enriches its representation of asset information, leading to an improved ability in identifying credit risk. In terms of heterogeneity research, we find that the enhancement of the jump frequency assumption consistently grants the HJD-KMV model superior capacity in identifying credit risk. Moreover, relaxing rigid payment structures proves beneficial to the model's ability in identifying credit risk, while the implementation of financial "deleveraging" policies and the occurrence of epidemics tend to diminish the model's effectiveness in credit risk identification.