CVaR的鞍点解析式及其在CreditRisk+框架下的应用  

The Saddlepoint Analytic of CVaR and Its Application in CreditRisk+ Model

作　　者：林清泉[1] 张建龙[1] 

机构地区：[1]中国人民大学财政金融学院,北京100872

出　　处：《系统工程》2008年第2期25-30,共6页Systems Engineering

摘　　要：CVaR是满足一致性的风险度量指标,它测度了超出VaR部分的条件期望。本文在Daniels(1987)基础上,独立导出CVaR的鞍点解析式。利用真实CVaR值已知的伽玛分布和贝塔分布做检验,结果表明CVaR鞍点解析式是CVaR的稳健近似。此外,本文还探索了该方法在风险管理中的应用,所推出的解析式可应用于CreditRisk+框架下损失分布CVaR的计算。CVaR is a coherent risk measure and quantifies the conditional mean beyond VaR. We independently derive the analytic of CVaR using saddlepoint approximation technique referred to Daniels（1987）. Gamma distribution and Beta distribution, whose CVaRs are known, are tested, and the results show that the analytic of CVaR based on saddlepoint can accurately and stably approximate the real CVaRs. Otherwise, new applications are explored in risk management. Under the framework of CreditRisk＋ issued by CSFB, the analytic we derived can be applied to the computation of CVaR fvr loss distribution of portfolio.

关 键 词：鞍点近似 一致性风险度量 CVAR 期望短缺 

分 类 号：F830[经济管理—金融学]