摘要
本文基于关联网络视角,提出度量行业系统性风险贡献的新方法——"留一法"(leave-oneout,LOO),将条件Granger因果检验方法与LOO相结合,评估行业的系统重要性程度。研究结果表明,虽然在传统无条件Granger因果检验方法和LOO方法下金融行业的系统重要性排名普遍靠后,但是各行业在两种研究方法中对应的系统重要性排名表现出很大的差异性;在全样本期内各行业之间普遍存在着较强的联动性,各行业的系统重要性排名会因不同极端事件冲击而发生明显变化。
From the perspective of the correlation network, this paper proposes a new method measuring the systemic risk contribution of industry-"leave-one-out"(LOO), which combines the conditional Granger causality test method with LOO to evaluate the industry’s degree of systemic importance. The research results show that, although the systemic importance ranking of financial industry is generally low by both the traditional unconditional Granger causality test method and the LOO method, the systemic importance ranking of each industry by the two methods shows great difference;in the entire sample period, generally there is a strong linkage among various industries, and the systemic importance ranking of each industry will change significantly due to the impact of different extreme event.
作者
卜林
刘凯迪
BU Lin;LIU Kai-di
出处
《金融论坛》
CSSCI
北大核心
2021年第8期60-69,80,共11页
Finance Forum
基金
国家社科基金一般项目“混业经营下中国交叉性金融风险的识别、度量及预防研究”(20BJY240)。