The EGARCH-GPD Model And Its Application Of Risk Measurement In Stock Market

VaR,developed from the early 90 s of 20 th century,is a tool in the spotlight in the area of financial risk management.Basel Accords also consider it as a very important assessment of today's international banking's operation and management.Various methods and models of the estimation of VaR were presented by the scholars during in the study of risk management to describe the volatility and distribution of financial profit.However,because of the complicated and changeable market it is difficult to fit the characteristics of financial data in a single model or method.Therefore,how to construct a model which can accurately quantify VaR is the purpose of this paper.This paper first introduces the related knowledge of financial risk management,summarizes the research status of VaR estimation at home and abroad,and makes a detailed introduction and comparative analysis of the definition and calculation methods of VaR estimation.In view of the calculation of VaR estimation,there are two key issues: first,the description of the volatility of financial assets,and the second is the distribution fitting of sample data.In this paper,the comparative analysis of GARCH models,SV model and other volatility models found that the EGARCH model can effectively capture the volatility of financial assets,such as the time variability,volatility clustering and asymmetry.Combined with the actual situation that extreme events in financial markets occurs frequently in recent years,the extreme value theory which can effectively describe the tail characteristics of financial assets is used to fit the residual sequence of EGARCH model.This paper combines with the advantages of EGARCH model and extreme value theory,construct a combination EGARCH-GPD model,the combined model not only has the advantages of the both two models,but also overcome the shortcomings of themselves: residual sequence of EGARCH model eliminate the serial correlation of sample,reached the distribution requirements that the sample must be independent and identically distributed in the statistical inference of the extreme value theory,the application of extreme value theory to overcome the problems that traditional EGARCH model cannot deal with the "sharp peak and fat tail" of the sample and the influence of extreme events.After the detailed introduction of the specific operation process of the combined model,VaR calculation of the combined model is derived from the positive homogeneity and shift invariance of VaR.Finally,CSI 300 index,the Shanghai Composite Index and Shenzhen Component Index,three representative of the Chinese stock market stock indices,are taken as samples,the application of combined model and EGARCH model based on normal distribution were compared and analyzed,the VaR back testing results show that: the dynamic VaR estimation of combined model effectively overcome the shortcomings that the traditional EGARCH model underestimate the risk,strengthen the prediction and management of risk,has stronger applicability and higher accuracy.The research on the GARCH-GPD model provides a new VaR estimation model for the financial risk measurement,enriches the theoretical methods to prevent and resist financial risk of the investors and risk management institutions.