(P/E)t-(P/E)(t-1)=b0+g1(P/E)(t-1)+err(t), when g1 <> imply a unit root exist, or opposite? thanks.
dont know but the ADF (augmented dickey fuller test) has a null hypothesis that there is a unit root so if your P-value is high then there is a unit root. An easier way, especially in time series models in to look at the beta of the independent variable and if it is close to 1 then you most likely have a unit root.
when g=0 (1-b1=0 and b1=1), then you have a unit root (in the t-test you would fail to reject the null).
For Dickey Fuller Ho: g1=0 you have a unit root Ha: g1!=0 Covariance stationary