摘要
假设风险满足一定条件,利用Copula和蒙特卡洛模拟方法对边际分布的选择和边际分布间的尾部相关关系两个方面进行分析,并得出结论:第一,在一定条件下,边际分布的选择并不影响对总风险的估计;第二,相关系数模型可能因为边际分布间的尾部相关关系而低估总风险.在此基础上,提出了尾部相关系数的概念,得出了修正的相关系数模型.
Copula is not an efficient tool for risk aggregation.Correlation coefficient model is an important method in the area of risk aggregation.But it is criticized for ignoring the marginal distributions and not capturing the tail correlation between risks.Based on some assumptions,this paper does research on these two subjects with copula and Monte Carlo.It concludes:first,under certain conditions,the choice of marginal distributions does not affect the estimation of the total risk;second,correlation coefficient model may underestimate the total risk if the marginal distributions are tail correlated.At last,this paper defines the concept of tail correlation coefficient and modifies the risk aggregation model.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2009年第12期73-79,共7页
Systems Engineering-Theory & Practice
基金
教育部人文社会科学项目(07JA790055)
关键词
全面风险管理(ERM)
风险整合
相关系数模型
边际分布
尾部相关性
enterprise risk management
risk aggregation
correlation coefficient model
marginal distribution
tail correlation