基于极值理论的投资组合风险研究  

A Study on Portfolio Risk Based on EVT

作　　者：周世新[1] 毛韵[2] 

机构地区：[1]江西中医学院经管学院,江西南昌330006 [2]江西师范大学政法学院,江西南昌330022

出　　处：《南昌大学学报（人文社会科学版）》2009年第5期91-95,共5页Journal of Nanchang University(Humanities and Social Sciences)

摘　　要：投资组合理论以正态分布为基础,以方差作为风险衡量指标存在缺陷。风险值用于给定的置信水平下,衡量在一定时期内最坏情况的损失。采用参数估计、非参数估计或者半参数估计,难以解决资产组合收益联合分布中的尖峰厚尾的问题。采用极值理论分析股票投资组合的风险值,结果显示采用极值分布的Copu la函数具有良好的特性,与均值-方差理论相比较,极值理论能够更加准确地刻画实际市场的极端波动和风险状况。The theory of portfolio, on the basis of normal distribution, has its own defects when evaluating ricks by variance. Within the confidence intervals, VaR is applied to the measurement of the worst - case lose in a certain period. However, parameter estimation, none parameter estimation and semi - parameter estimation are incapable of depicting a smooth and thin diagram in the joint distribution of the portfolio and income. Yet, extreme value theory is a sensible approach to fix this issue. The application of extreme value distribution of Copula Function can perfectly analyze the risks of the stock portfolio. Therefore, comparing with the theory of mean - variance, extreme value theo- ry can specifically and precisely illustrate the level of extreme turbulence and risks in the tangible stock market.

关 键 词：极值理论 投资组合 极值风险 VAR 后验测试 

分 类 号：F830.9[经济管理—金融学]