Common breaks in means for panel data under short-range dependence. Issue 2 (17th January 2021)

• Record Type: Journal Article
• Authors: Xi, Daiqing; Pang, Tianxiao
• Is Part Of: Communications in statistics. Volume 50:Issue 2(2021)
• Page Start: 486
• Page End: 505
• Publication Date: 2021-01-17
• Subjects: Common break -- fractionally integrated error -- least squares estimator -- limiting distribution -- panel data -- short-range dependence; 62F10 -- 62F12; Mathematical statistics -- Periodicals; Mathematics; Statistics; 519.2
• DOI: 10.1080/03610926.2019.1637000
• Languages: English
• ISSNs: 0361-0926

Abstract

Abstract: In this paper, we revisit the model studied in Bai (2010 ), in which a common break in means for panel data with linear-process errors was studied. The errors admit the form of x it = ∑ j = 0 ∞ a ij ε i, t − j for series i . Bai (2010 ) conducted the change-point analysis under the situation where ∑ j = 1 ∞ j | a ij | < ∞ for all i . In this paper, we suppose ∑ j = 1 ∞ j | a ij | = ∞ and ∑ j = 1 ∞ | a ij | < ∞ for all i, that is, the errors are short-range dependence. Then the least squares estimator of the common break is examined. In comparison to Bai (2010 ), the theoretical results in this paper reveal that: (1) Short-range dependence does not have the impact on the consistency of the estimator of the common break under the case of fixed magnitude of breaks. (2) It also does not have the impact on the convergence rate while it contaminates the limiting distribution of the estimator of the common break under the case of shrinking magnitude of breaks. Monte Carlo simulations are conducted to examine the finite-sample performance of the estimator. Our theoretical findings are supported by the Monte Carlo simulations.