Research on China's Energy Financial Risk Early Warning and Internal Risk Control Spillover Characteristics

Abstract

It is of great significance to scientifically and effectively assess the spillover effects of internal and external risks in China's energy finance and build an accurate risk early warning system to steadily promote the realization of the " carbon peaking and carbon neutrality " goal. Multiple internal and external market data are selected, and the spillover index model based on quantile vector autoregression is used to capture the internal and external risk spillover characteristics under different market conditions and visualize them through complex networks. In addition, the early warning index of nonlinear Granger causal test is incorporated into the Attention-CNN-LSTM model to construct a risk early warning system. The empirical results show that: (1) There are significant risk spillover effects both inside and outside China's energy financial market under different market conditions, and the risk spillover index under extreme market conditions is greater than that based on conditional mean and conditional median. (2) The internal crude oil and fuel oil markets and the external energy and stock markets occupy an important position in the overall risk spillover system. (3) Comparing and analyzing the prediction effects of six different models, the MAE and RMSE of the Attention-CNN-LSTM model were 0.7686 and 0.9077, respectively, which were optimized by 12.9% and 21.4% respectively compared with the second-best performing CNN-LSTM model; Moreover, after adding the early warning indicators, the prediction effect of the Attention-CNN-LSTM model is improved by 19.8% and 31.9% respectively in MAE and RMSE compared with the original model, so it is more suitable for constructing China's energy financial risk early warning system.