“好”“坏”方差与股票市场收益率预测

Good Volatility,Bad Volatility,and Stock Return Predictability

风险与收益的关系是金融资产定价研究领域中亘古不变的话题。基于指数日内高频数据和Barndorff-Nielsen等提出的已实现方差分解为“好”的方差和“坏”的方差框架,研究了方差非对称衡量指标——符号跳跃变差(SJV)对中国股票市场收益率的预测能力。实证研究显示,无论是样本内还是样本外,符号跳跃变差对深圳综指未来一个月超额收益率均有显著的预测能力,并且预测回归系数显著为负;即使控制了其他预测变量,效果仍然显著。此外,为了进一步说明符号跳跃变差稳健的预测能力,还考察了上证指数和中证500指数,结果显示符号跳跃变差对其都有预测能力,尤其在中证500指数上表现更佳。

Using the framework of variance decomposition provided by Barndorff-Nielsen et al.(2010),we analyze the predictive power of sign jump variation(SJV),a new asymmetric measure,for the excess return of the Chinese stock market.The empirical results show that SJV has a significant ability to predict the next month excess return in the Chinese stock market with a negative coefficient,both in-sample and out-of-sample.After controlling for other predictors,this conclusion still holds.Moreover,similar results exist in the Shanghai Composite Index and the CSI500,especially the latter.