胖尾分布及长记忆下的动态EVT-VaR测度研究

发布时间：2019-04-10 19:32

【摘要】：针对金融收益胖尾分布特征及条件波动率长记忆性特征,运用FIGARCH对条件波动率建模、极值理论(extreme value theory,EVT)对标准收益序列的尾部建模,测度出金融市场动态极值风险,进而运用返回测试(back-testing)技术,对模型在样本内的测度准确性与样本外的推广能力进行稳健性检验.实证研究结果表明,无论是中国新兴市场,还是西方成熟发达市场,金融收益与标准收益均呈现出明显的有偏胖尾分布特征;金融收益条件波动率均展现出长记忆性特征;EVT与FIGARCH模型相结合的动态极值风险测度模型不仅在样本内表现出优越的风险测度能力,而且在样本外同样具有可靠的预测推广能力.
[Abstract]:Aiming at the characteristics of fat tail distribution and long memory of conditional volatility, this paper uses FIGARCH to model conditional volatility and extreme value theory (extreme value theory,EVT) to model the tail of standard income series, and measures the dynamic extreme risk of financial market. Then, by using the return test (back-testing) technique, the accuracy of the model's measure in the sample and the generalization ability of the model outside the sample are tested. The empirical results show that both the financial returns and the standard returns have the characteristics of fat tail distribution, and the volatility of financial earnings conditions show long memory characteristics, both in the emerging markets of China and the mature developed markets in the western countries, and the results show that the financial returns and the standard returns have the characteristics of fat tail distribution and the volatility of the financial returns. The dynamic extreme risk measurement model combined with EVT and FIGARCH model not only shows superior risk measurement ability in the sample, but also has reliable prediction and generalization ability outside the sample.
【作者单位】：成都理工大学商学院;西南交通大学经济管理学院;
【基金】：国家自然科学基金资助项目(70771097) 教育部“新世纪优秀人才支持计划”资助项目(NCET-08-0826);教育部社科研究基金青年资助项目(10YJCZH086)
【分类号】：F830.9;F224

【参考文献】