面板数据下序贯模型变点的渐近检验法

THE ASYMPTOTIC METHOD OF SEQUENTIAL CHANGE-POINT IN PANEL DATA

本文研究了面板数据模型下变点是否存在的检验问题.对于面板序贯模型下的变点,利用构造的检验统计量及其极限分布的方法,获得了关于变点的一种渐近检验法.随后进行Monte-Carlo数值模拟,在模拟中对检验方法的经验检验水平、检验功效以及变点后的停时三个判断标准进行了考察.结果显示无论在理论还是数值模拟上,我们提出的渐近检验法均表现优良.推广了现有文献的关于单序列变点的序贯检验方法.

In this paper, we study the statistical inference of change-point for panel data.For the change-point in a sequential panel data model, using the test statistic and the limiting distribution, we obtain an asymptotic method of change-point. Then we make a Monte-Carlo simulation, in the simulation, we investigate the effectiveness of our test method from three aspects： empirical test size, test power and average run length. Both the theory results and the Monte-Carlo simulation show that the asymptotic method we obtained is very excellent,in addition, the existing sequential method about the change-point in univariate sequence is generalized.