摘要
杠杆效应(leverage-effect)描述了风险资产收益冲击对波动的非对称影响,是金融市场数据三大特征之一。大多数文献将杠杆效应作为常数,本文在EGARCH模型的框架内对杠杆效应参数进行时变化处理,以捕捉杠杆效应的时变特征。为克服得分驱动模型对冲击反应迟缓的缺陷,本文采用加速广义得分驱动(aGAS)设定杠杆效应的时变动态模型。蒙特卡洛模拟结果显示,时变杠杆效应EGARCH模型具有更好的样本内拟合效果和更强的样本外预测能力,以Qlike为损失函数的模型置信集(MCS)检验对三种EGARCH模型的选择结果表明,时变杠杆效应EGARCH模型的波动预测能力总是在最优模型集合内。对上证指数和沪深300的实证分析表明,在波动拟合和预测方面,采用aGAS去时变化EGARCH模型比采用GAS更能灵活地捕捉市场变化,具有更好的效果。为进一步验证杠杆效应时变化的价值,本文用时变杠杆效应滤波值改进技术交易规则,结果表明改进后的技术交易规则在年化收益率、年化波动率、最大回撤和夏普比方面都优于原来的技术交易规则,显示出时变杠杆效应具有边际信息增量。
The leverage effect that describes the asymmetric reaction of volatility to risk asset return shocks is one of three main features in financial data.This paper handles with time-varying leverage effect parameter in the frame of EGARCH model to seize the time-varying characteristic of leverage effect while most literatures regarded leverage effect as the constant.In order to overcome the defect of the slow response of the score-driven model to shock,this paper adopts the accelerating GAS model to set the time-varying dynamic model of leverage effect.The Monte Carlo simulation results show that the EGARCH model with time-varying leverage effect has better in-sample fitting power and stronger out-of-sample prediction capabilities.The results of the Model Confidence Set test with Qlike loss function for selecting the three EGARCH models demonstrate the volatility prediction ability of the EGARCH model with accelerating GAS is always within the best model set.Empirical analysis on the SSE Composite Index and CSI 300 index shows that the aGAS-based time-varying EGARCH model exhibits superior flexibility in capturing market dynamics and enhanced performance in volatility fitting and forecasting compared to the GAS-based one.To further validate the value of time-varying leverage effect,this paper uses the filtered time-varying leverage effect to improve the technical trading rules.The results reveal that the improved technical trading rules perform better in terms of annualized return,annualized volatility,maximum drawdown and Sharpe ratio than original counterpart,displaying the time-varying leverage effect has marginal information increment.
作者
杨亮亮
沈根祥
YANG Liangliang;SHEN Genxiang
出处
《中央财经大学学报》
北大核心
2025年第7期80-94,共15页
Journal of Central University of Finance & Economics
作者简介
通讯作者:杨亮亮,男,1993年10月生,上海财经大学经济学院博士研究生,研究方向为金融计量经济学,联系方式为liangliang@stu.sufe.edu.cn;沈根祥,男,1964年4月生,上海财经大学经济学院教授,博士生导师,研究方向为金融计量经济学和资本市场数量分析。
