In this study, we examined the types of bankruptcy risk dependence structures of Japanese firms. We classified bankruptcy events observed in Japan into multiple event types based on industry type and then used a multidimensional Hawkes process, which has been gaining attention recently in the field of finance research, to model the self-exciting and/or mutually exciting properties of bankruptcy among the event types. For the estimation of the intensity processes associated with the multidimensional Hawkes process, in addition to the analysis by Embrechts and Kirchner (Quantitative Finance 17:571–595, 2016) using the integer-valued auto-regression (INAR) model, we tested a new approach using the INAR model comprising macroeconomic indicators. Subsequently, we compared and considered the estimation results of the two specifications and demonstrated that the INAR model comprising macroeconomic indicators is better than the INAR model lacking macroeconomic indicators in terms of estimating the intensity process. Further, we demonstrated the effectiveness of the model from a practical perspective, with applications to stress tests.

在这项研究中，我们研究了日本企业破产风险依赖结构的类型。我们根据行业类型将在日本观察到的破产事件分类为多种事件类型，然后使用多维霍克斯过程（该模型最近在金融研究领域受到关注）来模拟破产的自激和/或相互令人兴奋的性质。在事件类型之间。为了估算与多维霍克斯过程相关的强度过程，除了Embrechts和Kirchner（Quantitative Finance 17：571–595，2016）使用整数值自回归（INAR）模型进行的分析之外，我们还测试了使用包含宏观经济指标的INAR模型的新方法。随后，我们比较并考虑了这两个规范的估计结果，并证明了在估算强度过程方面，包含宏观经济指标的INAR模型要优于缺乏宏观经济指标的INAR模型。此外，我们从实际的角度证明了该模型的有效性，并将其应用于压力测试。