SAS Proc Autoreg

[xiaoling]

When I use proc autoreg (AR1), should I do transformation to make all the dependent(Y) and independent variables(Xs) stationary? After I do the stationary transform, fit the model with autoregression, why i can not get any significant variables? How to solve these problems? Thank you for the help!

[PGStats · Accepted solution]

Residuals from the regression should be stationnary, not inputs.

[xiaoling]

Even the dependent variable Y desn't need stationary transform for AR1?