A0692
Title: Estimation of large covariance matrices with mixed factor structure
Authors: Runyu Dai - Tohoku University (Japan) [presenting]
Yoshimasa Uematsu - Hitotsubashi University (Japan)
Yasumasa Matsuda - Tohoku University (Japan)
Abstract: The Principal Orthogonal complEment Thresholding (POET) framework is extended to estimate large covariance matrices with a "mixed" structure of observable and unobservable strong/weak factors, and this method is called the extended POET (ePOET). Especially, the weak factor structure allows the existence of much slowly divergent eigenvalues of the covariance matrix frequently observed in real data. Under some mild conditions, the uniform consistency of the proposed estimator is derived for the cases with or without observable factors. Furthermore, several simulation studies show that the ePOET achieves good finite-sample performance regardless of data with strong, weak, or mixed factors structure. Finally, empirical studies are conducted to present the practical usefulness of the ePOET.