基于AM-BPNN模型的系统性金融风险评估及预警

China’s Systematic Financial Risk Assessment and Early Warning Based on AM-BPNN Model

文章结合主成分分析法构建了基于AM-BPNN算法的系统性金融风险预警模型,并对中国的系统性风险状态进行评估和预警。主成分分析结果显示,中国的系统性风险状态主要与货币银行风险、宏观经济贸易风险、外部冲击风险及财政风险等密切相关,在此基础上可以将1994—2018年中国历年的系统性风险状态划分为四类。进一步基于优化后的BP神经网络算法构建了有关中国系统性金融风险预警的AM-BPNN模型,并对2019年中国的系统性金融风险状态进行仿真预警,结果显示,2019年中国的系统性金融风险状态基本安全,而且风险主要来自货币银行系统积累的风险及金融开放所带来的外部冲击风险。

This paper combines with principal component analysis(PCA)to construct an early warning model for systemic financial risk based on AM-BPNN algorithm,and evaluates and warns the status of systematic risk in China.The results of PCA show that China’s systematic risk status is closely related to monetary bank risk,macro-economic trading risk,external shock risk and financial risk.On this basis,China’s systematic risk status from 1994 to 2018 could be classified into 4 categories.Furthermore,based on the optimized BPNN algorithm,the paper constructs an AM-BPNN model for China’s systematic financial risk early warning,and also conducts simulated early warning on China’s systematic financial risk status in 2019.The results show that China’s systematic financial risk status in 2019 is basically safe,and the financial risk mainly comes from the accumulated risk of the monetary banking system and the external shock risk brought by financial liberalization.