我先做了unit root test
结果是variables都是在first difference 下stationary,
之后做xtwest,结果如下:
. xtwest l_ecpc l_gdppc l_cdepc, lags(1)
Calculating Westerlund ECM panel cointegration tests..........
Results for H0: no cointegration
With 21 series and 2 covariates
-----------------------------------------------+
Statistic | Value | Z-value | P-value |
-----------+-----------+-----------+-----------|
Gt | -0.993 | 1.697 | 0.955 |
Ga | -2.929 | 2.426 | 0.992 |
Pt | -3.407 | 0.667 | 0.748 |
Pa | -1.783 | 0.670 | 0.748 |
-----------------------------------------------+
有三个问题:(1)l_ecpc是重新定义的l.ecpc这样对吗?(2)为什么p-value会那么大?(3)做完cointegration test之后 是做面板模型的回归,请问应该是用first difference的variable做,还是原始variable?
非常感谢!