报告题目：Projected Estimation for Large-dimensional Matrix Factor Models

主 讲 人：虞龙

报告时间：2020年11月11日 10:00-11:00

报告地点：腾讯会议 会议 ID：807 555 196

点击链接入会： https://meeting.tencent.com/s/4YU4evGuRXNk

报告摘要：

In this article, we propose a projection estimation method for large-dimensional matrix factor models. We construct a projection direction such that the strength of common factors remains while the effects of idiosyncratic errors are weakened, resulting in a nearly noise free factor model. Theoretically, we prove that the projected method achieves faster convergence rates under mild conditions. Asymptotic distributions of the projected estimators are also studied. A novel iterative procedure is also proposed to specify the factor numbers. Thorough numerical studies verify the empirical advantages of the projected method. We apply the projected method to analyze two real examples on finance and macroeconomics and the resulting factor loadings show patterns cross rows and columns coinciding with financial, economic or geographical interpretation.

主讲人介绍：

虞龙，2020年6月于复旦大学统计学系获得博士学位，现于新加坡国立大学从事博士后研究工作。主要研究兴趣为高维因子模型的统计推断，厚尾分布下的稳健统计和随机矩阵理论。研究内容于Journal of Business & Economic Statistics, Journal of Multivariate Analysis期刊发表。

邀请人：何勇

欢迎各位老师同学积极参加！