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'.
Discussion(0)
No comments yet. Be the first to comment.