Measuring Systemic Risk In The Chinese Banking Sector Via A Generalized Dynamic CoVaR Model

Commercial banks are the main component of China's financial system.The primary task of maintaining China's financial stability is to maintain the stability of the commercial banking system.Due to the existence of business transactions between banks,there are many potential systemic risks.Once a bank becomes risky,its risks will spillover to other banks,causing a crisis in the banking system and even the financial system.Therefore,it is necessary to accurately measure the systemic risks of China's commercial banks and strengthen systematic risk supervision.In the research field of systemic risk measurement,AR-GARCH-CoVaR model is a common model,with easy data acquisition,flexible model construction,and can reflect the data autocorrelation and heteroskedasticity.However,when using this model for systemic risk measurement,the credibility of the measurement results can only be discussed through text.It is impossible to construct a backtesting for model's accuracy,which cannot provide a quantitative reference for the supervisor.In this paper,the generalized CoVaR method,which can apply the backtesting,is used to improve the AR-GARCH-CoVaR model.The improved new model is named as the generalized dynamic CoVaR model.The generalized dynamic CoVaR model needs to make distributional assumptions of innovations,which has a certain influence on the measurement accuracy and goodness of fit of the model.This paper takes full account of the characteristics of fattailedness,skewness,and tail asymmetry of financial return data,and tried five distribution assumptions: Gaussian Distribution,Student-T Distribution,Skewed-T Distribution,Generalized Error Distribution(GED)and Asymmetric Exponential Power Distribution(AEPD).Based on the generalized dynamic CoVaR model with AEPD distribution,using the returns of CSI 300 Bank Index and 14 listed Chinese banks as the sample,through backtestings and fitting graphs,this paper finds that the generalized dynamic CoVaR model with AEPD can accurately measure systemic risk and fit the data distribution characteristics.Then,this paper measures the systemic risk of the Chinese banking sector.The indicators such as risk of individual bank,long-term risk spillover,systemic risk,and short-term risk spillover ratio were studied.The correlation between the indicators has also been explored.The results show that the risk spillover effect of large-scale commercial banks is higher than that of joint-stock commercial banks and city commercial banks;the long-term risk spillover effects and shortterm risk spillover effects are significantly different.