Research On Extreme Value Model And Its Application In Risk Measurement

With the increasing internationalization of financial market in Various countries,the fluctuation of domestic and foreign financial markets is intensified,which leads to the possibility of extreme loss events,resulting in great financial pressure and high bankruptcy cost risk.It is very important to accurately measure extreme risk in financial market risk management.At present,most of the researches on extreme risk focus on the distribution form and volatility of return on financial assets.How to build a risk measurement model that can accurately measure extreme Va R is the purpose of this paper.Because the traditional EGARCH model can not well describe the extreme value of the tail of the distribution in extreme events,this paper proposes a new EGARCH-CF model based on this model and the Cornish-Fisher expansion method,and uses the Cornish-Fisher method to asymptotically expand the residual distribution of EGARCH model.The new model can reflect both the volatility clustering and asymmetry of the rate of return.EGARCH-CF model and traditional EGARCH model are applied to measure the extreme risk of stock return of China high speed railway.The results of measurement and Kupiec back test show that EGARCH-CF model can be used to measure extreme risk,and the extreme risk measured by EGARCH-CF model is more accurate than EGARCH model.The distribution function of return on financial assets often has the characteristics of peak and thick tail,which can not be described by normal distribution and t distribution.In this context,based on the extreme value theory and the Gamma mixed generalized error(GGED)distribution,this paper constructs the POT-GGED model by approximating the excess distribution in the POT model with the GGED distribution.POT-GGED model not only has the advantages of flexible application of GGED distribution,but also has the advantages of POT model which can describe the extreme risk at the tail of return distribution.In the empirical analysis,POT-GGED model and the existing POT model are used to measure the extreme Va R value of Shanghai 380 index.The empirical results and Kupiec back test show that POT-GGED model can measure the extreme risk of Shanghai380 index return.To a certain extent,POT-GGED model is more accurate than POT model in measuring extreme risk.The two new models proposed in this paper provide reference for measuring extreme risk in financial market.