Research On Systemic Risk Contagion Effect And Early Warning Of Chinese Financial Institutions Considering Network Public Opinion Index

The report of the 19th National Congress of the Communist Party of China emphasized the need to "improve the financial supervision system and keep the bottom line of systemic financial risks." Therefore,preventing and resolving systemic financial risks is the top priority of my country’s financial work in the new era,and it is also to ensure the stability of my country’s economy.,An important foundation and conditions for healthy development.Due to strict monitoring and control,there has not been a large-scale systemic financial crisis in our country.However,with the increasing linkage of the international financial market,the reform of the domestic financial market is gradually deepening,and financial innovation is changing with each passing day.The establishment of a theoretically reliable and In practice,an effective systemic financial risk early warning model is particularly necessary to avoid the occurrence of systemic financial risks and to ensure the sound and orderly operation of the economy and finance.It is also of great significance to the early warning and prevention of systemic financial risks in my country.This paper first defines the connotation of online public opinion and systemic financial risk,and from the perspective of behavioral finance,with the help of investors’ "limited attention","overconfidence" and other theories to analyze the influence of investor behavior and investor sentiment on online public opinion information.The impact mechanism of systemic financial risks.Secondly,from January 2015 to March 2019,more than 10 million postings from 45 institutions on Oriental Fortune Online were used as a research vehicle to create an emotional dictionary and set up a corresponding model to construct an online public opinion index.Then,based on the mixed causality test method,the non-linear Granger causality between online public opinion and systemic financial risk is investigated.Then embed the constructed network public opinion index into the systemic risk contagion effect measurement model to obtain a modified single-index asymmetric Co Va R model,and use the linear quantile LASSO algorithm and local polynomial estimation method to estimate the parameters,and build on this basis The directed network of finance conducts an empirical analysis of the contagion effects of systemic risks in Chinese financial institutions.Finally,the Attention＿LSTM deep neural network is used to construct a Chinese systemic financial risk early warning model,and the network public opinion index is used as a training set to be incorporated into the early warning model to test the early warning effect.Finally,the early warning results are combined with the LSTM neural network model,BP neural network model,SVR model and ARIMA model To compare the warning results.The research results show that:(1)Online public opinion has a nonlinear Granger causality relationship to systemic financial risks,and online public opinion is an important source of systemic financial risks.(2)Financial institution risk indicators represented by single indicator asymmetric Co Va R and online public opinion have a clear trend of synergy changes;(3)Securities and banking financial institutions are very sensitive to external risks and are extremely vulnerable to other financial institutions.It is also very easy to affect other financial institutions;non-banking institutions occupy an important position in the risk accumulation stage,and banks occupy an important position at the time of risk outbreak.Compared with non-banking financial institutions,banking institutions have a strong contagion ability.(4)The LSTM deep neural network has strong generalization ability.The Attention＿LSTM deep neural network model that is incorporated into the network public opinion index is more accurate in predicting the effect,and it is more accurate for most systemic financial risk indicators of different periods such as short-term,medium-term and long-term.The early warning effect has been significantly improved,and compared with the BP neural network model,the SVR model and the ARIMA model,the Attention＿LSTM deep neural network early warning model has higher accuracy.At the end of the article,based on the results of theoretical analysis and empirical research,starting from the two aspects of risk prevention and risk control,proposed systemic financial risk prevention and control recommendations related to online public opinion supervision.