摘要
本文利用连接函数(Copula)解决整合风险管理中不同类型风险的联合分布建模问题,提出了基于连接函数的整合风险度量Copula-VaR及其蒙特卡洛模拟算法;以深圳发展银行和上海浦东发展银行为研究对象,将Copula-VaR与N-VaR和Add-VaR这两种业界常用的近似整合风险度量方法进行了实证比较分析,发现:与Copula-VaR相比,N-VaR和Add-VaR存在高估风险的倾向,而其主要原因则是由于N-VaR和Add-VaR对信用收益率与市场收益率之间的相关结构进行了不符合实际的假设。
This article uses Copula to solve the problem of joint distribution modeling of variant risks in integrated risk management, puts forward a method of integrated risk measurement based on Copula： Copula-VaR and its Monte Carlo simulation algorithm. Using Shenzhen Development Bank and Shanghai Pudong Development Bank as research objects, this article compares Copula-VaR with approximate integrated risk measurements N-VaR and Add-VaR empirically, finds that N-VaR and Add-VaR tend to overestimate risk and the main reason is that they do reasonless assumption about correlation structure between credit return and market return.
出处
《统计研究》
CSSCI
北大核心
2008年第11期72-80,共9页
Statistical Research
基金
国家自然科学基金项目“基于期望收益率时变性和背景风险的战略资产配置理论研究”(70801046)资助
