2018/08/28 by Long Yu, Yong He, Yu, Long +3
Economics, Econometrics and Finance · #Spatial and Panel Data Analysis
paper · pdf · doi:10.48550/arxiv.1808.09107
The accurate specification of the number of factors is critical to the\nvalidity of factor models and the topic almost occupies the central position in\nfactor analysis. Plenty of estimators are available under the restrictive\ncondition that the fourth moments of the factors and idiosyncratic errors are\nbounded. In this paper we propose efficient and robust estimators for the\nfactor number via considering a more general static Elliptical Factor Model\n(EFM) framework. We innovatively propose to exploit the multivariate Kendall's\ntau matrix, which captures the correlation structure of elliptical random\nvectors. Theoretically we show that the proposed estimators are consistent\nwithout exerting any moment condition when both cross-sections N and time\ndimensions T go to infinity. Simulation study shows that the new estimators\nperform much better in heavy-tailed data setting while performing comparably\nwith the state-of-the-art methods in the light-tailed Gaussian setting. At\nlast, a real macroeconomic data example is given to illustrate its empirical\nadvantages and usefulness.\n