经济变量的协同波动是学术界、宏观政策制定者长期关注的重要问题,本文将时间序列数据的共同周期检验方法扩展至面板数据,提出非平稳面板数据的非线性共同周期检验方法。本文根据数据特征,将共同周期划分为强降秩结构共同周期和弱降秩结构共同周期,分别在强降秩结构数据和弱降秩结构数据中提出共同周期检验统计量,并提出区分强降秩结构数据和弱降秩结构数据的统计量。研究结果表明,本文检验统计量的极限分布都是卡方分布,并且各检验统计量都表现出良好的有限样本性质,因此,本文提出的非平稳面板数据的非线性共同周期检验方法具有较高的实用性。
The co-movements of economic variables are important problems for economist and the government. This paper expands the time series common cycle testing method, and proposes new nonlinear common cycles testing method in nonstationary panel data models. Based on the characteristic of common cycle data, this paper divides the common cycle into strong form reduced rank structure and weak form reduced rank structure. This paper puts forward statistics to test strong form reduced rank structure common cycle and weak form reduced rank structure common cycle, also proposes statistic to distinguish between strong form reduced rank structure and weak form reduced rank structure. The results show that these testing statistics are all follow chi-square distribution and have good finite sample properties. Therefore, the nonlinear common cycles testing methods developed by this paper can be conveniently used in empirical studies.