If I choose the optimal lag of VAR from lag length criteria function in Eviews. When I conduct Johansen Cointegration test, as Eviews tells that the lag in Johansen is for differened terms, so the lag that I need to specify is the optimal lag from VAR minus one.
Is this right?
There are some videos posted on Youtube saying that to perform Johansen Test, we can use the optimal lag from VAR. However, there are some saying it should be optimal lag minus one.
Which one is right?
You can use the optimal lag from VAR. Eviews reports VAR lag order criteria which include zero lag. Therefore, zero minus one can not be accepted as optimum lag. Regardless of its theoretical framework, it is not possible (technically) use optimum lag minus one.