只要是采用时间序列的动态模型就一定要考虑数据的平稳性，做相关检验是必要的，Eviews6就可以做。但是由于面板数据的复杂性，协整关系的确定和模型系数的估计目前似乎还没有现成的软件，有些模型的性质也是未定的，很难见到运用于实证的文献。

the asymptotic properties in dynamic panel data model is based only on cross-section dimension, N.   the time dimension T is assumed to be fixed. but  the asymptotic theory of panel cointegration model  is based on both N and T. So the two models have differences in many ways.  basically, if N>>T, it seems reasonable to use dynamic panel data.  However, if N and T both large,  you need to consider the stationarity of the time series, this is one of most important topics in high dimension data.

面板协整 是非平稳数据 楼主概念不清晰