Proc VARMAX

[Golf]

Hello,

I have 3 variables :  LM1, LIP, TB3 and found that they are I(1) with nocointegrating vector.  

Hence I plan to use  the first difference of these variables : dLM1 dLIP dTB3 in VAR with optimal lag of 3.

   I use proc VARMAX for dLM1 dLIP dTB3 with  p = 3.  

     How can I forecast the original variables: LM1  LIP  TB3 (not in first difference) based on the results the above procedure?

Thank you

[Golf · Accepted solution]

I just got a solution.
proc varmax data=Example_var plots= forecasts ;
model LM1 LIP TB3/ p=3 dif(LM1(1) LIP(1) TB3(1));
output Back=0 Lead=12 out = Example_var_Forecast2;
run;