Financial Risk Management Based On Vine Copula-quantile Regressions

With the influence of economic globalization,financial scientific,information technology innovation and so on,the global financial market is undergoing great changes.How to achieve effective management of financial risk,such as effective portfolio selection and financial market crises prevention,has become an important issue faced by venture capitalists and financial institutions.To the best of our knowledge,the effective management of financial risk depends on the joint distribution information between financial assets or variables,which needs to be obtained through joint distribution modeling.However,it is difficult to accurately estimate the joint distribution function for two reasons: model selection and dimension disaster.Fortunately,copulas provides a solution for this problem.It can transform the joint distribution modeling into two independent processes: marginal distribution modeling and dependence structure modeling.To address these issues,we develop a vine copula-quantile regression model through combining quantile regression and copula.In the proposed model,quantile regression is used to fit the marginal distribution and copula is used to investigate the dependence structure.We then apply the novel model to financial risk management.In the portfolio,we apply it to conduct portfolio selection in multiple financial assets.First,we establish a vine copula-quantile regression model,which extends the Zhu's method from binary to multivariate case.Second,we propose a portfolio selection scheme to optimize the generalized Omega ratio,which includes simulating the variation of financial returns and presenting a algorithm for threshold acceptance.Third,the efficacy of the novel method is illustrated through empirical studies on practical portfolios of three kinds of commodities and five stocks respectively.In the financial risk contagion,the "multiple-to-one" risk spillover effects among financial markets have been depicted.This procedure first build a vine copula-CAViaR model to estimate joint distribution of multiple conditions,which includes using CAViaR to fit marginal distribution and using a vine copula method to describe nonlinear correlation relationship.It then derive risk measure method of CoVaR based on joint distribution of multiple conditions.Third,its efficacy is illustrated through empirical studies on stock markets among China,United States and Japan in terms of CoVaR measure.To sum up,the contributions of this dissertation are two-fold.First,the proposed vine copula-quantile regression model provides an effective way to solve problems in joint distribution of multiple conditions modeling.Second,the empirical results of portfolio selection and risk spillover effects are promising and can provide desicision support for those practioners and regulators in financial market.