Estimation And Application Of ES Under Conditional Heteroscedastic Difference

Managing risk is the most important step in financial regulation.To manage risk,we need to find a reasonable risk measurement tool.JP Morgan proposed the VaR in the 1990s.The VaR concept is simple and easy to understand.The outbreak of the subprime mortgage crisis in 2008 revealed the natural flaws in the risk of VaR that the inability to measure tail risk.In 2013,the Basel Committee clearly recommended that financial institutions use ES instead of VaR.Compared with the risk value,ES can describe the tail risk first,then the ES meets the secondary additivity,meet the risk diversification principle in modern portfolio theory,and finally the ES has convexity,which indicates that ES is used as a risk measurement tool to build investment.The combination can get the optimal solution.Dimitriadis and Bayer(2017)[1]constructed a joint regression model of VaR and ES based on the loss function(Fissler and Zigel,2016)[2].The model can measure or predict ES like a normal regression model,which greatly facilitates risk measurement and management,but the existing work does not consider the typical characteristics of financial data,Moreover,when using simple joint regression estimation,there is also a conditional heteroscedasticity in the residual,this article confirms this with actual data,in this paper,an ES metric model that fully considers the conditional heteroscedasticity is constructed.An iterative algorithm is designed to estimate the parameters,and the statistical properties of the parameter estimators are briefly discussed.In order to test the advantages of the new model in measuring ES,We use the latest method proposed by Johanna et al(2017)[3]to test the effect of the actual measurement of ES on the model.The results show that the model constructed in this paper can more accurately measure the specific value of ES compared with the traditional method.After obtaining a more accurate measure of expected insufficiency,this article continues to explore its application in practice.It mainly includes the following two aspects.First,the ES estimated by the new method is used to analyze the contribution factors of risk.Taking China Merchants Bank as an example,the paper selects the analysis variables from the three dimensions of the company's fundamental,market and macro levels to explore the main sources of risk.The results show that the source of the tail risk is mainly the liquidity risk from the stock exchange market and the system risk from the macro level,and in the extreme case,the higher the liquidity risk contribution,the lower the system risk contribution;Finally,based on the Piotroski-F scoring method(Piotriski,2000)[4],based on the more accurate ES,individual investors with value investment ideas are built to build asset portfolios,so that individual investors can control the risks while obtaining excess returns.In summary,this paper considers conditional heteroscedasticity for the first time when constructing the ES metric model.Compared with the traditional method,this method is more accurate for ES estimation.The model can provide reference for financial institutions to analyze risk contribution factors,which can help investors build asset portfolios,improve the ability of portfolios to deal with extreme risks,and provide reference value for financial regulators to monitor risks.