| 主讲人 |
王少平 |
简介 |
<p>This paper develops two robust t tests for testing the asset pricing anomalies and factor models when financial data exhibit serial dependence and time-varying (unconditional and conditional) heteroskedasticity. We model volatility dynamics with a flexible nonlinear time-varying function and construct a fixed-b long-run variance (LRV) estimator that differs from Newey–West and remains robust under strong autocorrelation and time-varying volatility. The resulting statistic, t_HAR, shows a standard limiting distribution. To improve finite-sample performance, we further propose a dependent wild bootstrap (DWB) version of the test, namely t_HAR^DWB, which preserves the dependence and time-varying volatilities; we establish its asymptotic validity. Extensive Monte Carlo experiments show that our procedures deliver near-nominal size and higher power relative to conventional t and Newey–West tests across a wide range of dependence and volatility designs. Our framework can also be applied in some other financial tests, such as factor redundancy test, spaning test, and etc. Applying our methods to 31 prominent cross-sectional anomalies in the Chinese stock market, we find only 9 significant, especially yearly turnover (YT), variation in RMB trading volume (VDTV), and one-month volatility (VOL), while many celebrated patterns vanish. This highlights the importance of a robust testing method for assessing the significance of anomalies. To check whether anomalies during the bubble period are same or not as that in random walk period, we conduct our previously proposed tests to detect random walk and bubble period first. Results show there are 2 random walk and 3 bubble periods. Finally, we implement our t tests for these periods respectively. We find 3 significant anomalies in the bubble period and 4 in the random-walk period. Typically, the one-month abnormal turnover (ABR) is significant during the random-walk period but not during the bubble period. Overall, anomaly “discovery” is highly sensitive to time-varying volatility and autocorrelation; the proposed t_HAR and t_HAR^DWB tests provide reliable inference and help tame the factor zoo in both full-sample and regime-specific periods.</p> |
| 主讲人简介 |
<p>王少平，华中科技大学经济学院教授，清华大学博士。国务院学位办（教育部）“全国百篇优秀博士论文” (2篇)指导教师，湖北省优秀博士论文（5篇）指导教师，湖北省优秀学士论文（20多篇）指导教师。著有国家统编教材和学术著作5部。其研究成果10多次获得省部社科优秀成果一和二和三等奖。主要研究方向为非平稳数据的单位根与协整检验、资产价格泡沫、定价因子模型的评估，致力于基于我国的数据和现实对前沿理论进行创新并应用于研究中国经济增长、通胀（缩）对宏观经济的影响及其相依性周期、中国资本市场的泡沫与风险等重要的现实经济问题，在《中国社会科学》《经济研究》、<em>Journal of Econometrics, Econometrics Journal, Economics letters</em>等国内外期刊发表论文30篇。其具有显著现实意义的代表性成果《中国资本市场的突出风险点与监管的反事实仿真，中国社会科学，2019.11》，《中国通货膨胀的相依性周期，中国社会科学 2013(05) 》，《中国CPI的宏观成分与宏观冲击，经济研究，2012(12)》，对如何识别资本市场的风险和有效监管、如何认识和评估当前的通缩风险及其如何把握适度宽松的货币政策的度，体现出针对性的参考意义；《Testing for moderate explosiveness, , Econometrics Journal（2019），Volume 22》，是这一方向的实质性的创新。</p>
<p class="MsoNormal" style="text-indent:21.0pt;mso-char-indent-count:2.0"><span lang="EN-US" style="font-family:"Times New Roman",serif;mso-fareast-font-family: 宋体;mso-bidi-theme-font:minor-bidi"><o:p></o:p></span></p>
<p> </p> |
期数 |
“邹至庄讲座”杰出学者论坛（第67期）暨厦门大学金融名家论坛2025-2026学年秋季学期第3讲（总第86讲） |
