中国股票市场尾部风险与收益率预测——基于Copula与极值理论的VaR对比研究  

China's Stock Market Tail Risk and Return Prediction:A Contrastive Study of VaR Based on Copula and EVT

作　　者：陈坚[1] 

机构地区：[1]厦门大学经济学院,福建厦门361005

出　　处：《厦门大学学报（哲学社会科学版）》2014年第4期45-54,共10页Journal of Xiamen University(A Bimonthly for Studies in Arts & Social Sciences)

基　　金：国家自然科学基金项目"基于扇形偏好的一般均衡期权定价方法及其对股价跳跃的应用"(71201136)

摘　　要：目前,在险价值(VaR)被广泛应用于风险管理与投资组合分析中,在股票市场,它表示股票收益的尾部风险。相关研究发现VaR对未来股票收益具有预测作用。根据Copula方法和极值理论方法分别构建VaR,考察其对中国股票市场是否具有样本内和样本外的预测能力,实证结果发现:与传统经济变量作对比,基于极值理论的VaR具有较强的预测能力,而基于Copula方法的VaR的预测效果并不十分显著。Currently, Value-at-Risk （VaR） has been heavily used in risk management and investment analysis. On the stock market, VaR represents the tail risk of stock return distribution. Related studies show that VaR is able to predict future excess stock returns. In this paper, we construct VaR through the extreme value theory （EVT） and the Copula method, and investigate the predictability of two VaR measures. Empirical results show that, compared with economic variables, the VaR based on EVT has strong in-sample and out-of-sample predictive power. By contrast, the predictability of VaR based on the Copula method is not very significant.

关 键 词：样本外预测 极值理论 COPULA VAR 尾部风险 

分 类 号：F830[经济管理—金融学]