My understanding of the loglinear option is that it takes a log-linear approximation to the model during estimation. This matters because of the way the data are entered. Does this mean that the stoch_simul command produces impulse responses for the log-linear model? Or, if I use stoch_simul with the log-linear model, do I get impulse response functions from the linear model? I ask because there is no loglinear option for stoch_simul. If the loglinear option in estimation has residual effects for stoch_simul, this seems unfortunate, because I have to run estimation in order to produce the impulse responses I want. If I don’t get impulse response functions for the log-linear model, then the graphs don’t match the data I think they match. There is no documentation on how to see impulse response functions for a log-linear model from stoch_simul, so I am hoping someone can clear this up.