Determining The Number Of Factor In Large Factor Models

This article develops an information criterion for determining the number of factors(r)in the general dynamic factor model and approximate static factor model.The core of paper are Inspired by Onatski(2009),and has two points as followed.Firstly,in the general dynamic factor,as long as series that obtained general dynamic factor structure satisfy our assumptions,its discrete Fourier transforms would admit an approximate factor structure in the sense of Chamberlain and Rothschild(1983).Then the criterion will be born in a way similarly Bai and Ng(2002)had done.Secondly,the original data,noted by Xj,that has approximate factor structure,is transformed by,which can be used to create a criterion.We provide sufficient conditions for consistency of the criterion,and the Monte Carlo experiments show that our methods perform better than Bai and Ng(2002)and Haliin(2007)with high correlation and noise in data.The major study of this article is as follows.In chapter one,we discuss the background,meaning and status of the research and describes the generalized dy-namic factor model and approximate factor model.Our information criterions of the two model aforementioned,The hypotheses and proof are developed in chapter two and three.Chapter four contains Monte Carlo experiments and comparison-s.Chapter five concludes.