报告题目: Prediction for High-dimensional Cointegrated Series
主 讲 人: 张荣茂(浙江大学)
报告时间:2021年6月3日(周四)15:30-16:30
报告地点:腾讯会议 ID: 475 882 619
点击链接入会:https://meeting.tencent.com/s/RPAQLu4xbqrM
报告摘要:
Cointegration inference is often built on the correct specification of vector autoregressive regression (VAR) model for the short-run effects. When the dimension of time series is high, using VAR model will lead to many problems. First, the specification for the dynamic structure could be wrong. Second, it would be difficult to estimate the dynamic structure. To avoid these problems, we propose a factor error correction model, which use a dynamic factor model to capture the short-run dynamic effect. Based on eigenanalysis analysis and Yule-Walker equation, the loading matrix, latent factors and their dynamic structure are estimated. It is shown that the proposed method increases the predictability over a pure factor model and the classic error correction model. Asymptotic properties of the estimators are established for both fixed and slowly diverging dimension cases. Illustration with both simulated and real data sets is also reported.
主讲人简介:
张荣茂,浙江大学数学科学学院教授、数据科学中心与经济学院兼职教授。2004年在浙江大学获得博士学位,2004年7月-2006年6月在北京大学从事博士后研究,2006年至今在浙江大学工作,多次访问香港科大、香港中文大学和伦敦政治经济学院。主要从事非平稳时间序列和高维空间数据的理论与应用研究,已发表SSCI/SCI论文40多篇,发表的杂志包括Ann. Statist.,J. Amer. Assoc. Statist.,J. Econometrics等。主持过国家自然科学基金、浙江省杰出青年基金、省重点和其他省部级基金项目多项,现任浙江大学统计所所长,浙江省现场统计研究会副理事长和J. Korean Statist. Soc.的Associate Editor。
主办单位:山东大学金融研究院,山东大学数学学院
邀 请 人:王汉超
欢迎各位老师同学积极参加!