单位根的全距检验方法及其应用  

Full-range Inspection Method of Unit Root and Its Application

作　　者：舒晓惠 宋金奇[2] Shu Xiaohui;Song Jinqi(School of Business,Huaihua University,Huaihua Hunan 418000,China;Science and Technology College,Jiangxi Normal University,Nanchang 330027,China)

机构地区：[1]怀化学院商学院,湖南怀化418000 [2]江西师范大学科技学院,南昌330027

出　　处：《统计与决策》2020年第2期10-15,共6页Statistics & Decision

基　　金：国家社会科学基金青年项目(11CTJ003)。

摘　　要：文章系统研究了单位根检验的一种非参数全距检验统计量:RUR检验与FB-RUR检验,并且运用Monte Carlo模拟给出了两个全距检验的临界值及响应面函数,MC仿真比较了全距检验与秩检验、传统检验的优劣。结果表明,全距检验统计量的检验效果与秩检验相近,适用于非线性时间序列平稳性检验。实证表明,当全距检验与秩检验结果一致但与传统检验不一致时,往往存在明显的单调非线性特征;当全距检验与秩检验、传统检验结果不一致时,序列会出现周期性、非线性或结构突变等多重特征。将全距检验、秩检验与传统单位根检验方法相结合可以发现时间序列中的非线性特征,从而防止对非线性时间序列滥用传统单位根检验得到错误的结论。This paper systematically studies a kind of nonparametric full range test statistics for unit root test,i.e.RUR test and FB-RUR test,and uses Monte Carlo simulation to present the critical values and response surface functions of two full range tests,comparing the advantages and disadvantages of full range test,rank test and traditional test by way of MC simulation.The results show that the test effect of full range test statistics is similar to that of the rank test,which is suitable for the stability test of nonlinear time series.The empirical results show that when the results of full range test and the rank test are consistent but not with the traditional test,there will exist obvious monotone nonlinear characteristics,and that when the results of the full range test are inconsistent with those of the rank test and the traditional test,multiple features such as periodicity,nonlinearity or structural mutation will appear.Combining the full range test and rank test with the traditional unit root test finds the nonlinear characteristics in time series,thus preventing abusing the traditional unit root test for the nonlinear time series to draw wrong conclusions.

关 键 词：单位根 全距检验 非线性 RUR检验 FB-RUR检验 

分 类 号：O212[理学—概率论与数理统计]