Risk Measurement Of World's Major Crude Oil Via Normal Inverse Gaussian Distribution,Variance Gamma Distribution And Copula Functions

Value at Risk(Va R)and Expected Shortfall(ES)backtesting and forecasting are playing a more and more important role in today's financial institutions.However,modeling the marginal distribution of asset or portfolio is always a crucial challenge for supervisors.In this thesis,we combine normal inverse Gaussian distribution(NIG)and variance gamma distribution(VG)with Monte Carlo simulation to model the marginal distribution of world's major crude oil futures,including West Texas Intermediate crude oil(WTI)futures,Brent crude oil(Brent)futures,Dubai crude oil(Dubai)futures and their spots.And then we estimate the Value at Risk(VaR)of them,in which process,in order to examine the value of forecasted risk,some backtesting techniques are used,including unconditional coverage test,conditional coverage test,duration-based test,generalized method of moments(GMM)test.What's more,the appropriate model we identify are combine with many kinds of copula functions as the bivariate models to estimate the VaR of world's major crude oil hedging portfolio,including WTI,Brent and Dubai.And the results are tested by several distinct backtesting techniques we mentioned before.Conclusion form two aspects are summarized: for one thing,the best model for estimating VaR of each crude oil hedging portfolios;for another,the most suitable model for three crude oil hedging portfolios under given backtesting techniques.Since the Basel Committee suggested that VaR should be replaced by ES in financial institutions,it's controversial that whether ES can be backtested or not.Nevertheless,through the tireless efforts of scholars,some theoretical achievements have been achieved in this regard.ES backtesting in practical applications is still in its infancy yet.In this part,we aim to select the appropriate distributions for individual assets and portfolios by comparing the results of two unconditional tests of ES backtesting.Finally,we find that NIG/NIG-Gumbel copula is a suitable model to estimate and forecast the ES of West Texas Intermediate crude oil(WTI)hedging portfolio.Modeling the bivariate distribution of Brent crude oil(Brent)by VG/NIG-Clayton copula is a good choice.And both NIG/VG-Gumbel copula and NIG/VG-Clayton copula are proper candidates for estimating the risk of Dubai sour crude oil(Dubai).