Is it possible to make an unanticipated shock in a simulation of a deterministic model in for instance period 2? I.e. that the economy is hit by an unanticipated shock at time t=2 but such that the path for the shock is completely known afterward? I’m not sure whether this break with the “determinism”
It depends what you want. In a rational expectations framework, once a shock has occured, it is perfectly known. Hence, what you describe pretty much describes a standard IRF from stoch_simul. So without further context, I don’t understand why you need to conduct this in a determinstic model. Do you need the full non-linearity?