《保险研究》20250802-《全球保险业系统性风险演化与关键驱动因素研究——基于复杂网络与可解释机器学习的融合分析》（徐昕、周圣杰、高杰英、张胜）

［摘   要］在全球经济不确定性加剧的背景下，金融体系的系统性风险问题受到广泛关注。本文基于LASSO-VAR模型，对全球80家保险机构的风险溢出效应展开分析，并构建了风险溢出网络。在此基础上，结合PageRank中心度与可解释机器学习方法，识别出风险溢出网络中的关键节点和系统性风险的关键驱动因素。结果表明：重大事件冲击下，保险业系统性风险呈显著上升趋势，美国、荷兰和英国的保险公司主要为风险净输出方，中国大陆、法国和德国的保险公司主要为净接受方；美国、瑞士和百慕大的保险公司是风险传导的关键节点；在地缘政治冲突叠加美联储加息时期寿险公司在风险网络中居中心地位，而在其他时期非寿险公司主导风险扩散；市场因素与货币政策对系统性风险的长期贡献突出，反映出资本流动与市场传染机制的主导作用。基于上述发现，本文提出完善宏观审慎监管框架、提升国际风险传染应对能力等对策，以防范和化解保险业系统性风险。

［关键词］全球保险机构；系统性风险；风险传染；LASSO-VAR

［基金项目］本研究获国家社会科学基金一般项目“基于大数据及可解释集成学习的信用风险模型构建与应用研究”（22BTJ016）资助。

［作者简介］］徐昕，首都经济贸易大学金融学院教授、博士生导师，研究方向：金融风险、机器学习与精算；周圣杰，首都经济贸易大学金融学院博士研究生；高杰英，首都经济贸易大学金融学院教授、博士生导师，研究方向：风险管理、商业银行；张胜（通讯作者），玉林师范学院商学院副研究员，研究方向：风险管理、保险机构。

Evolution of Systemic Risk in the Global Insurance Industry and Key Driving Factors：An Integrated Analysis Based on Complex Networks and Explainable Machine Learning

XU Xin，ZHOU Sheng-jie，GAO Jie-ying，ZHANG Sheng

Abstract：Under the backdrop of intensifying global economic uncertainties，the systemic risk issues of the insurance industry as a crucial pillar of the financial system have attracted widespread attention. This study employed the LASSO-VAR model to analyze risk spillover effects among 80 global insurance institutions and constructed a risk spillover network. Building upon this framework，we integrated PageRank centrality with explainable machine learning methods to identify critical nodes in the risk spillover network and key drivers of systemic risk. The findings reveal three main conclusions. First，systemic risk in the insurance industry rise significantly under major event shocks，with U.S.，Dutch，and British insurers primarily acting as net risk transmitters，while Chinese Mainland，French，and German insurers serving mainly as net risk receivers. Second，insurers from the United States，Switzerland，and Bermuda emerge as pivotal nodes in risk transmission. Life insurance companies occupy central positions in the risk network during periods of geopolitical conflicts combined with Federal Reserve rate hikes，whereas non-life insurers dominate risk diffusion in other periods. Third，market factors and monetary policy demonstrate prominent long-term contributions to systemic risk，reflecting the dominant role of capital flows and market contagion mechanisms. Based on these findings，the study proposed countermeasures including improving macroprudential regulatory frameworks and enhancing international risk contagion response capabilities to prevent and mitigate systemic risks in the insurance industry.

Key words：international insurance institutions;systemic risk;risk transmission;LASSO-VAR

>