acjlhong 发表于 2008-8-29 07:54 
如果误差修正项的系数不能显著为负,说明所谓的“协整”关系并不可靠;如果显著为负,就可以分析短期和长期效应了。
e.g. if I have following outcome ( coeffecients are significant at 5%) , could I say in the long run AP Granger cause attendance, meaning that causality runs interactively through the error correction term from AP to ATT and Y ??
Thanks for your reply
Error Correction: D(ATT) D(AP) D(Y)
CointEq1
-0.143398 0.000752 -2.66E-05
(0.07293) (0.00024) (1.1E-05)
[-1.96625] [ 3.11695] [-2.52247]