基于混合Copula和ARMA-GARCH-t模型的股票指数风险度量研究  

Research on the Risk Measurement of Stock Index Based on Mixture-Copula and ARMA-GARCH-t Models

作　　者：鲁思瑶 徐美萍[1] 

机构地区：[1]北京工商大学理学院,北京100048

出　　处：《广西师范大学学报（自然科学版）》2016年第1期93-101,共9页Journal of Guangxi Normal University:Natural Science Edition

基　　金：国家自然科学基金资助项目(61304155);北京工商大学研究生部促进人才培养综合改革项目(19005428069)

摘　　要：针对近期中国股票市场的剧烈波动给投资者带来的影响,本文选取中国沪深股市和香港股市中具有代表性的3种股票指数的日收益率数据进行建模分析。选用ARMA-GARCH(1,1)-t模型拟合其边缘收益率序列,利用Gumbel、Clayton和t-Copula的线性组合函数刻画其相关结构,再运用蒙特卡罗模拟方法计算不同置信水平下各股指的风险价值、预期损失和中位数损失,其中中位数损失是最近才提出的一种新的风险度量指标。本文的研究结果可以为有着不同风险偏好的投资者和管理者监管系统性风险提供借鉴。In view of the influence of recent Chinese stock market volatility on investors,three representative stock index＇s daily returns of Shanghai,Shenzhen and Hong Kong stock markets are selected for modeling and analysis.First,ARMA-GARCH（1,1）-t models are chosen to fit the marginal returns.Then,a linear combination function of Gumbel,Clayton and t-Copula is applied to describe the correlation structure of returns.Finally,Monte Carlo simulation method is employed to compute Value at Risk,expected shortfall and median shortfall of the three stock index under different confidence levels.Here median shortfall is a new risk measurement index proposed recently.The research provides references for investors and managers with different risk preferences to regulate the system risk.

关 键 词：混合Copula函数 ARMA-GARCH(1 1)-t模型 风险价值 预期损失 中位数损失 蒙特卡罗模拟 

分 类 号：F830.91[经济管理—金融学]