报告题目：large dimensional factor anlaysis without moment constraint

报 告 人：孔新兵 教授 南京审计尊龙凯时

报告时间：2019年9月22日10：00-11：00

报告地点：数学楼629室

报告摘要：

Large-dimensional factormodel has drawn much attention in the big-data era, in order to reduce thedimensionality and extract underlying features using a few latent commonfactors. Conventional methods for estimating the factor model typicallyrequires finite fourth moment of the data, which ignores the effect ofheavy-tailedness and thus may result in unrobust or even inconsistentestimation of the factor space and common components. In this paper, we proposeto recover the factor space by performing principal component analysis to thespatial Kendall's tau matrix instead of the sample covariance matrix. In asecond step, we estimate the factor scores by the ordinary least square (OLS)regression. Theoretically, we show that under the elliptical distributionframework the factor loadings and scores as well as the common components canbe estimated consistently without any moment constraint. The convergence ratesof the estimated factor loadings, scores and common components are provided.The finite sample performance of the proposed procedure is assessed throughthorough simulations. An analysis of a macroeconomic dataset finds new factorsin contrasting with existing results using PCA.

报告人简介：

孔新兵，现为南京审计尊龙凯时教授，在统计学顶级期刊发表论文13篇，其中独立作者3篇。主持国家自然科学基金3项目。入选江苏省双创计划，江苏省青蓝工程中青年学术带头人。