基于Lévy测度的动态操作风险度量  

Dynamical models of operational risk based on Lévy measure

作　　者：徐驰[1] 汪冬华[1] XU Chi;WANG Donghua(School of Business,East China University of Science and Technology,Shanghai 200237,China)

机构地区：[1]华东理工大学商学院

出　　处：《系统工程理论与实践》2018年第9期2177-2187,共11页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(71171083,71771087);上海市教育委员会科研创新项目(14ZS058);上海市浦江人才计划(15PJC021)~~

摘　　要：依据我国商业银行1994-2012年操作风险历史数据,本文引入Lévy测度描述操作风险损失所具有的非连续跳跃行为,利用稀疏序列法产生动态操作风险损失过程,采用了同时考虑损失频率相关性和强度相关性的Lévy Copula模型,给出了具有时变参数和时变相关性结构的动态操作风险度量模型和数值实验技术,计算了不同置信水平上的VaR与CVaR.实证结果表明：Lévy Copula模型能在减少模型设定风险的同时,较好地描述风险单元间的相关性结构;Lévy Copula模型相比于传统Copula模型对相关性结构刻画更为细致,能降低风险资本,并且通过稳健性检验;动态Lévy Copula模型能捕捉到风险的变化趋势,减少由于时变参数导致的风险资本估计偏差.Based on historical data of operational risk in Chinese commercial banks between 1994 and 2012,this paper introduces Lévy measure to describe discontinuous jumping behavior of operational losses, uses thinning method to simulate the dynamical process of operational risk losses and adopts Lévy Copula model considering frequency dependence and severity dependence simultaneously. In this paper, a dynamical operational risk model with time-varying parameters and time-varying correlation structure is given as well as the corresponding numerical experimental technology to calculate VaR and CVaR in different confidence levels. The empirical result shows that Lévy Copula model of which setting risk is decreased shows priority in describing the dependence structures between risk cells. In comparison with traditional Copula model,Lévy Copula model depicts dependence structure more explicitly and intensively lowering total risk capital and passes model robustness test as well. Furthermore, the dynamical Copula model displays trends of risk and reduces the estimation bias of risk capital deriving from time-varying parameters.

关 键 词：Lévy测度 Lévy COPULA 共同冲击 动态模型 

分 类 号：F832.2[经济管理—金融学]