组合信用风险测度的藤copula方法  

Vine copula approach to measure portfolio credit risk

作　　者：唐振鹏[1] 黄友珀[1] 

机构地区：[1]福州大学管理学院,福建福州350108

出　　处：《系统工程学报》2013年第4期488-496,共9页Journal of Systems Engineering

基　　金：国家社会科学基金资助项目(07BJY164);国家自然科学基金资助项目(71171056)

摘　　要：研究组合信用风险测度问题,用藤copula描述违约相依结构,提出一种测度组合信用风险的藤copula方法.实证结果表明:常用多元copula方法往往低估或高估风险值,而藤copula方法在各种常用的多元copula相依结构假定下的VaR和ES估计值与实际风险值很接近,VaR都通过了回测检验,ES回测检验指标也表明藤copula方法估计的ES更准确.因此,相对于常用多元copula,藤copula方法更具灵活性,能提高组合信用风险测度的准确性.This paper focuses on the measurement of portfolio credit risk.Using vine copula to capture default dependency,we propose a vine approach to measure portfolio credit risk.Empirical results show that common copula methods often underestimate or overestimate value at risk,but value at risk and expected shortfalls derived from the vine copula approach are very close to the real values under different assumptions of dependency structure described by common copulas.Value at risk passes the corresponding back testing,and the back testing indicator of expected shortfalls also suggests that the vine copula approach can estimate expected shortfall more accurately.The evidence above can prove the superiority of the vine copula approach relative to common multivariate copulas.

关 键 词：组合信用风险测度 藤copula 相依结构 在险价值 回测检验 

分 类 号：F830[经济管理—金融学]