Estimating Of VaR Based On Dynamic Extreme Risk Management Model

In the financial market, the financial risk management becomes more and moreimportant, because of aggravation of fluctuation of interest rate and exchange rate,and occurrence of extreme financial event. VaR and CVaR become the most widelyrecognized risk measurement tools, and have been widely used in the study offinancial risk management. Traditionally, the calculation method of VaR generallyassume that financial income obey which type of distributions, and on this basis,wecan get risk value at a given confidence level. However, the randomness of thisassumption usually leads to underestimated risk by this traditional VaR method, andalso greatly increases the model risk when we calculate VaR value.Therefore, extreme value theory (EVT) got more and more attention, and it is akind of the method based on estimation of statistical parameters. In recent years, it hasbeen widely applied in the measurement of financial extreme risk. It overcomes thedefect of traditional VaR method, which assumes the overall distribution function inadvance, and only takes sample data as the research object, establishes model ofextreme value distribution which is a kind of tail distribution. If given a specific tailprobability, we can calculate the excess loss value of extreme market fluctuations,namely risk value (VaR). Hence, extreme value theory has the ability over the sampleestimation, solve the "thick tail" problem of financial assets return data, can get moreaccurate estimation value of market risk.The distribution of financial assets return data, not only has characteristics ofgathering fluctuation and fat tail, but also its heteroscedasticity of fluctuationcondition has certain effect to the risk value (VaR). GARCH family model reflects thechanging process of volatility, so it is widely used in analysis of financial time series.In this paper, we introduced volatility into the calculation formula of riskmeasurement based on the extremum theory, which combine extreme value theoryand heteroscedastic model, construct dynamic extreme risk management model,estimate tail quantile of return distribution conditions and depict the characteristics offat tail and stochastic volatility of the loss, and get the dynamic VaR value.In this paper, firstly introduce the research background of the estimation methodand risk value theory based on VaR model at home and abroad, then summarize thesimple method of extreme value theory, mainly discusses the BMM model and POTmodel. Finally, we build a dynamic extreme risk management model, which the extremum theory is introduced into the GARCH family model. This dynamic model isapplied in empirical test of the public bank logarithmic return series, we can see thatthe traditional static model can’t measure the losses during the period of violentfluctuation, and introduction of dynamic extreme risk management model still hashigh accuracy. First of all select the series of return rate by GARCH model, andobtained pareto distribution by relating series of standard residual and EVT, at lastcalculate risk value using extreme value theory.