报告简介  Abstract

Under the moderate high-dimensional setting, we consider high-dimensional time series which follows a factor model where both the temporal and cross-sectional dependence are assumed to be captured by a few common factors. As the efficient estimation of such factor models is in terms of eigen-decomposition of high dimensional sample auto-covariance matrices, we contribute to the establishment of the central limit theorem (CLT) for spiked eigenvalues of high dimensional sample auto-covariance matrices. The CLT is developed under general cases in the sense of allowing auto-lags either fixed or diverging. Additionally, as a statistical application of the developed CLT, a novel equivalence test of factor structures is proposed for two high-dimensional time series. This equivalence test provides statistical inferences for comparing the spiked structures of two high-dimensional time series and facilitates further statistical applications such as clustering multi-population high-dimensional time series. Simulations, as well as analysis of worldwide mortality data, are provided to demonstrate the feasibility of the proposed test statistics.

This is joint work with Daning Bi (ANU), Xiao Han (USTC) and Adam Nie (ANU).

嘉宾简介  About the Speaker

Dr. Yanrong Yang is a Senior Lecturer at The Australian National University (ANU). Dr. Yang's research interests include high dimensional statistical inference, large dimensional random matrix theory, large panel data analysis and high dimensional functional data analysis. She has published in leading journals including Annals of Statistics, Journal of the Royal Statistical Society: Series B, Journal of the American Statistical Association, and Journal of Econometrics.

讲座海报Poster