报 告 人：刘志 副教授

报告题目：Limiting theorem of realized covariance when zero returns are present

报告时间：2023年12月18日（周一下午4：00 )

报告地点：bat365中文官方网站学术报告厅（静远楼1506室）

主办单位：bat365中文官方网站、数学研究院、科学技术研究院

报告人简介： 

       刘志，澳门大学科技学院副教授，主要研究方向为统计学及其交叉方向。研究兴趣包括随机过程统计，金融统计，机器学习在生物信息和医学数据方面的应用。其工作的主要贡献为发展了连续时间模型中波动率在不同场景下的估计理论，以及建立了澳门本地慢性肾病病人肾小球滤过率的预测模型。论文广泛发表于统计学及其交叉学科优秀期刊，如统计学期刊Annals of Statistics, Journal of American Statistical Association, Journal of Business and Economic Statistics, 计量经济学期刊Journal of Econometrics, Quantitative Economics, Econometric Theory, Econometrics Journal, 金融学期刊Journal of Banking and Finance, Finance and Stochastics, 生物信息学期刊Bioinformatics, 以及机器学习会议AAAI , 等. 目前主要关注的问题为复杂神经网络的泛化误差和估计误差上界，和慢性肾病的影响因子探测。

报告摘要：

       Considering the presence of bivariate price staleness, we study the problem of estimating the covariation of two semimartingales. We propose a consistent estimator and establish a unified limiting theory for the realized covariation under the presence of price staleness, which includes the existing results as special cases. Our results demonstrate that the presence of idiosyncratic price staleness can induce a bias in the realized covariation but the systematic price staleness does not affect the limit of the realized covariation. Moreover, the bias induced by price staleness makes the realized covariation closer to zero than that without the presence of price staleness, hence this explains the well-known Epps effect appropriately. We conduct Monte Carlo studies to assess the finite sample performance of the proposed theory, and some empirical applications to real high-frequency data are considered to illustrate our theory. This is a joint work with Haibin Zhu.