I doubt that you can just delete the seasonals---some of the regressions depend upon those being there.
As I mentioned a few posts ago, this program (aside from having errors like the missing definition of the seasonal) isn't doing what the paper describes. Instead of trying to follow a program in that type of shape, figure out what exactly it is that you want to do---what model are you estimating, and what are you testing?