I am writing a mod file in which I declare the equations in log-levels, ie, exp(x) = …, where x is the log of my variable (say X).
When I use the stoch_simul command, I know Dynare computes a linear approximation around the steady-state of my model, so it is actually a log-linear approximation, given the way I declared my variables.
But I also want to use information of the model’s Jacobian which is stored in the modname_dynamic function. My question is, are these the jacobian associated with the variables in log-levels or in log deviation from the steady-state? Or does this make no difference at all?