A Hybrid Dynamic Risk Measure Approach Based On SV-SKST-EVT Model

Financial risk management is a major subject in the field of financial asset management.Despite the fact whether you are a large financial institution or a general fund manager,and even individual investors,everyone need to have enough knowledge of the potential risks that their positions is facing,that is,we need more accurate means of risk measurement.Especially in recent years,extreme financial events occur more frequently,and every time these abnormal fluctuations will bring great losses to the national wealth.Such as the 1997 financial crisis in Southeast Asia,the US subprime mortgage crisis in 2008 let those countries suffer from different degree of economic depression in the next few years and hurt badly.A recent case is China's stock market crash in 2015,leading to fierce fluctuations in the stock market,the Shanghai Composite Index once raise up to 5,000 from somewhere about 2000,and then suffered a cliff-style fall back to 3000 points just in a year.Countless investors and institutions suffer a huge loss in the stock market,which is due to the lack of a good risk measurement approach.In order to cope with the losses caused by the abnormal fluctuations of financial assets,financial risk management approach is particularly important.Through good financial risk management approach,we can benefit from risk diversification and quantify the potential risks,in this case we will be able to respond quickly when a risk event occurs.As an international measurement risk approach,VaR is widely used in various types of financial instruments' risk measurement.Thus how to measure Va R more accurately and build a more precise model for risk measurement is an important issue of financial risk management.Basically,there are two types of VaR estimation methods: parametric method and non-parametric method.In general,the nonparametric estimation method can easily deal with “fat tail” while it has no idea with heteroskedasticity.In contrast,the parameter estimation method can effectively reflect the dynamic evolution of the time series,that is,heteroskedasticity.But it will act poorly when the assumed distribution differs from the real distribution.For the reasons above,this paper uses a VaR estimation method that combining the advantages of the two methods.First,we try to find a model that can accurately describe the distribution characteristics of financial assets.And here we refer to a SV model with generalized hyperbolic partial t distribution in order to reflect the fat tail,skewness and other characteristics of the financial asset better.And a nonparametric method,that is,extreme value theory,is used to model the tail of a standard residual sequence.The advantage of the extreme value theory is that only the tail distribution of the sample needs to be considered,and the prediction ability of the EVT is stronger.Also the MCMC algorithm is used when estimating the parameters.