论文部分内容阅读
在投资组合的风险管理中我们经常使用Vine Copula来拟合资产收益率的联合分布,目前使用最多的vine结构是C-vine和D-vine,然而C-vine和D-vine只适合于描述两类特定的相依关系。基于此,本文将采用更一般的R-vine结构,结合最大生成树MST-PRIM算法,来刻画任意n个金融资产收益率的实际相依结构,并运用蒙特卡洛模拟方法,对投资组合的风险价值VaR展开研究。实证分析表明:与传统的Copula方法相比,R-vine Copula在描述资产间的相依结构上更具有灵活性和有效性,在投资组合的风险管理中准确性更高。
In portfolio risk management we often use Vine Copula to fit the joint distribution of return on assets. The most commonly used vine structures are C-vine and D-vine, whereas C-vine and D-vine are only suitable for describing two Class-specific dependencies. Based on this, this article will adopt a more general R-vine structure, combined with the maximum spanning tree MST-PRIM algorithm, to describe the actual dependency structure of the yield of any n financial assets, and the use of Monte Carlo simulation method to portfolio risk Value VaR study. Empirical analysis shows that compared with the traditional Copula method, R-vine Copula is more flexible and effective in describing the interdependence of assets and has higher accuracy in portfolio risk management.