Analysis Of Text-based Business Relevance And Systemic Risk Prediction In Commercial Banks

Recently,Silicon Valley Bank and Signature Bank in the United States have successively closed down,spreading market concerns and triggering a banking crisis.The incident once again reflects that with the development of digital technology,the speed of information dissemination has accelerated,and the structure of financial products has become increasingly complex,greatly increasing the correlation between financial institutions and the market,as well as between financial institutions.This correlation is the fundamental reason for the accelerated spread of systemic risks within the financial system.The issue of risk spillovers caused by the correlation between financial institutions has aroused high attention from regulatory authorities and academia in various countries.China’s regulatory level has initially established a correlation evaluation method for commercial banks based on the indicator method.Taking correlation as an important consideration dimension for evaluating systemically important banks is an important measure to prevent systemic risks.However,looking at the current measurement methods for the correlation between commercial banks in the academic and practical fields,it is found that the current methods are mainly based on institutional financial data or financial open market data.However,issues such as the insufficient effectiveness of China’s financial market and the low availability of some financial data may affect the feasibility and accuracy of the current methods,and determining a measurement method suitable for China’s actual situation is particularly important for subsequent effective supervision and risk prevention.Therefore,exploring new methods for measuring the correlation degree of commercial banks has practical significance in promoting the improvement of the monitoring mechanism for the correlation degree between commercial banks and optimizing important systematic bank evaluation methods in China.Based on the meaning of correlation between commercial banks,this article focuses on depicting the correlation generated by the homogenization of commercial banks’ asset and liability businesses.Using text data from commercial banks’ annual reports,and introducing text similarity analysis techniques,this paper constructs a correlation index for China’s commercial banks’ asset and liability businesses through two text similarity calculation methods: cosine similarity and jaccard similarity.Through the Granger causality test,it is found that the constructed correlation index is forward-looking compared to the evaluation ability of the correlation index based on financial data.Through a single threshold regression model,it is found that business correlation has a threshold effect on systemic risk,and values exceeding the threshold value will have a significant positive impact on systemic risk.According to the business relevance index of commercial banks constructed in this article,it is found that the business relevance of China’s commercial banking system is increasingly frequent,especially the asset business connection,and the connection between small and medium-sized corporate banks such as urban commercial banks and other banks is becoming increasingly close.After 2016,the relevance of the sample banking business has successively exceeded the threshold,and the upward trend is significant.The indirect risk exposure it may generate needs to attract the attention of regulatory authorities.Finally,this article proposes policy recommendations to improve the relevance evaluation mechanism of China’s commercial banks,strengthen risk prevention management of commercial banks,and strengthen the use of relevance indicators.The relevance index constructed in this article is incorporated into text data,broadening the data perspective of commercial bank business relevance measurement and systemically important bank identification research.Due to the high suitability and availability of data,it also provides new ideas for promoting to the measurement of relevance of non bank financial institutions and the evaluation of systemically important non bank financial institutions.However,there are still shortcomings in this article,mainly including insufficient analysis of the quality of text data to be improved,and the direction and extent of risk spillovers from a single institution to the banking system.