协整向量误设对相关检验的影响:小样本模拟研究

The Effect of Vector Misspecification on Cointegration Tests:Small Sample Simulation

利用蒙特卡罗模拟方法,研究小样本中协整向量误设对两种t统计量(tDF和tECM)的实际检验水平和检验势的影响,并针对参数q′的两种取值分别对二者的检验势进行比较,认为向量误设会降低tDF和tECM的检验势,但如果误设程度不大,其具体表现依然优于相应的无约束情形。无论是否存在向量误设,统计量tECM都优于tDF,但其优势会随着q′的减小而逐渐丧失。

This paper discusses the effect of vector misspecification on the size and power of cointegration tests （ tDF and tECM） by Monte Carlo simulation of small sample, and draws a comparison between the powers of different cointegration test methods under the situation of q＇. Also, this paper shows that vector misspecification will reduce the power of cointegration test, but if its error is small enough, compared with the unrestricted situations, this specification can also improve the power of test to some extents. Furthermore, whether there is vector misspecification or not, the statistic tECM always dominates tDF,but the dominance will be decaying with the decrease of q＇.