Abstract: Financial security is an important guarantee for the stable operation of a country’s macroeconomy. Faced with an increasingly complex global financial system, it is particularly important to accurately perceive and measure systemic risks and financial security status across countries. This paper builds a multi-quantile intelligent risk perception model and uses financial market index, commodity prices, and national macroeconomic state variables of G20 countries from October 2004 to September 2023 to measure systemic financial risks, risk sources, and security levels of different countries under different risk states. The results show that the spillover of financial risks faced by countries mainly originate from global stock markets, especially neighboring countries, while commodities and macroeconomic factors generally play a risk-mitigating role. China’s systemic financial risk and security level are relatively less affected by shocks from the global financial system but are more sensitive to its own economic conditions. In addition, systemic risks in developing countries are generally higher than those in developed countries, and their financial security levels—calculated under the unified consideration of external shocks and macroeconomic characteristics—are also generally lower. The intelligent risk perception model expands the research perspective on systemic risk and provides regulatory authorities with a practical tool for monitoring systemic risks and financial security levels under various risk conditions.

Key words: financial systemic security, systemic risk, multi-quantile intelligence risk perception, network vulnerability

摘要： 金融安全是一国宏观经济稳定运行的重要保障。面对日益复杂的全球金融系统，准确识别和测度各国的系统性风险与金融安全水平具有重要意义。本文构建了多分位风险智能感知模型,利用2004年10月至2023年9月G20国家的金融市场指数、大宗商品价格以及各国宏观状态变量数据，系统测度了不同风险状态下各国金融系统的系统性风险、风险来源及其安全水平。研究结果显示：各国面临的金融风险溢出主要来源于全球金融市场，尤其是来自相邻国家的溢出效应更为显著；而大宗商品和宏观经济则普遍为各国提供了风险缓释作用。相较而言，中国金融系统性风险与安全水平受全球金融系统冲击相对较小，但对自身经济状态的变化更为敏感。此外，发展中国家的系统性风险整体高于发达国家，在综合考虑外部冲击与宏观特征后，其测算出的金融安全水平也普遍低于发达国家。该风险智能感知模型拓展了系统性风险研究的视角，为监管部门提供了一种能够监测多种风险状态下系统性风险与金融安全水平的实用工具。

关键词: 金融系统安全, 系统性风险, 多分位风险智能感知, 网络脆弱性