Need help with VAR


I am planning to run VAR to gather insight about the feasibility of inflation targeted monetary policy. I have been reading up on the three types of VAR and their usage, as I don't have prior academic knowledge. I have a few basic questions to start with:

1) If my variables in question are not stationary, can I take the first difference and proceed? (Of course if the first difference makes it stationary). I also read that if there is co integration among variables, then its better to use VECM. However, the first-difference stationary variables are still applicable for a reduced VAR right?

2) While determining the lag length, should I use the varsoc command with the log first-differenced stationary variables or just the log variables?

3) Lastly, I have a set of variables that are exogenous so can I incorporate these in a reduced VAR or do I need to use SVAR? Kindly elaborate with commands in either cases.

I will really appreciate a quick response as I have to meet a deadline.

Thank you in advance!VAR