I've been using VAR's for quite some time but have not found literature on obtaining an IRF for a exogenous shock to the VAR. An SVAR would seem to come closest to resembling exogenous shocks to the system by structuring the VAR such that the "exogenous" variables only shock the dependent response variable. However, the feed back between the endogenous and exogenous variable is still calculate in the process. Any tips?