基于AST分布和HGARCH模型的金融资产收益率波动非对称性刻画与VaR预测  

Characterization of the Volatility Asymmetry of Financial Asset Returns Based on AST Distribution and HGARCH Model and VaR Forecasting

作　　者：王杰 杨爱军[1] 林金官 WANG Jie;YANG Ai-jun;LIN Jin-guan(College of Economics and Management,Nanjing Forestry University,Nanjing 210037,China;School of Statistics and Mathematics,Nanjing Audit University,Nanjing 211815,China)

机构地区：[1]南京林业大学经济管理学院,江苏南京210037 [2]南京审计大学统计与数学学院,江苏南京211815

出　　处：《数理统计与管理》2020年第6期1121-1140,共20页Journal of Applied Statistics and Management

基　　金：教育部人文社会科学基金(18YJC910001);国家自然科学基金(11971235);江苏省高校哲学社会科学基金项目(2018SJA0130);江苏省青蓝工程项目(2020)。

摘　　要：选择合适波动模型和概率分布成为影响VaR预测可靠性的重要因素,本文首次结合HGARCH模型和AST分布来对金融资产收益率进行建模,并将所构建模型用于VaR预测研究中。我们重点比较研究不同头寸和不同风险水平下HGARCH模型及其子类共20种GARCH族模型的VaR预测效果,并系统性研究HGARCH模型中两个波动非对称参数在波动非对称性刻画和VaR预测中的作用。研究结果表明,在样本内,AST分布下的非对称GARCH族模型具有更好的波动拟合效果、分布拟合效果和VaR预测效果;HGARCH模型的两个波动非对称参数虽然在理论上是互补品的关系,但实际建模效果类似,相互之间更接近替代品的关系。在样本外,AST分布下的非对称GARCH族模型在波动拟合、分布拟合和VaR预测方面的优越性有所下降;两个波动非对称参数的实际效果也近似为替代品,其中放缩参数相比位移参数更具优势。Selecting the appropriate volatility model and probability distribution becomes an important factor affecting the reliability of VaR prediction,this paper combines the HGARCH model and AST distribution for the first time to model the return of financial assets,and use the constructed model for VaR prediction research.We focus on the performance of VaR prediction of HGARCH model and its subclasses in total 20 GARCH model under different positions and different risk levels,and systematically study the role of two volatility asymmetrical parameters in the HGARCH model in asymmetry characterization and VaR prediction.The empirical results show that,in the sample,the asymmetric GARCH model under AST distribution has better volatility fitting effect,distribution fitting effect and VaR prediction effect;the two volatility asymmetry parameters of the HGARCH model are theoretically complementary,but the actual modeling effects are similar,and the relationship between them is closer to each other.Out the sample,the superiority of the asymmetric GARCH model under AST distribution in volatility fitting,distribution fitting and VaR prediction has declined;the actual effect of the two volatility asymmetric parameters is also approximated as a substitute,in which the scaling parameters have advantages over the shifting parameters.

关 键 词：AST分布 HGARCH族模型 波动非对称性 VaR预测 

分 类 号：O212[理学—概率论与数理统计]