摘要
在操作风险领域,损失样本的不一致问题,使以损失分布法度量的操作风险价值存在不确定性。假定损失强度为Weibull分布,损失频率为Poisson分布,并以K.Bocker等人得出的操作风险价值的解析解为理论基础,进一步研究操作风险价值误差的产生机理,得到在特征参数置信度为1-β时,操作风险价值的置信区间。并通过对误差传递系数与特征参数η、λ与m弹性的研究表明:参数m的误差是操作风险价值误差的主要来源,从而也是影响操作风险价值置信区间长度主要敏感因子。
Unconformity of loss data makes to the uncertainty of the operational VaR in LDA to rise. When the Weibull severity distribution and the Poisson frequency distribution are premised, the mechanism of producing error is researched in the operational VaR On the basis of the operational VaR's closed-form approximation which is solved by K. Bocker and oth- ers. And a confidence interval is built in a confidence level 1 -- βof the characteristic parameters. After the coefficient of error propagation and the elasticity of the parameters η,λ and m are researched, the solutions show that the error of the parameter m is the dominating source of the error of the operational VaR. And the error of the parameter m is also the great cause of influencing the length of the Confidence Interval.
出处
《系统工程》
CSCD
北大核心
2007年第10期33-39,共7页
Systems Engineering
关键词
操作风险
操作风险价值置信区间及敏感性
误差传递
弹性
Operational Risks Confidence Interval and Sensitivity of the Operational VaR~ Error Propagation~ Elasticity