Application of Neural Network with Autocorrelation in Long-Term Forecasting of Systemic Financial Risk

Abstract

Carrying out early warning of systemic financial risk is a prerequisite for timely adjustment of monetary policy and macroprudential policy to effectively prevent and resolve systemic financial risks. This paper constructs a systemic financial risk monitoring and early warning system for China’s banking industry based on isolated forest anomaly detection and neural network with autocorrelation mechanism and uses low-frequency data with high credibility to effectively identify the ten factors that have the greatest impact on systemic financial risk in China’s banking industry, improving the prospective and accuracy of risk early warning. The conclusions can help regulators to adjust their policies prospectively to curb the rise of systemic financial risks.