Stoch_simul after the estimation

Sorry if this is a stupid question from a stupid beginner. I get the same decision rules both if I stoch_simul after I estimate the model by ML (I’m setting mh_replic=0 in the estimation) and if I do it after I estimate it by Metropolis (set mh_replic>>0). Shouldn’t I expect to get linearizations around different steady states (i.e., doesn’t the estimation command set the parameters at the mean of the posterior if it is doing it by Metropolis?). Many thanx, Fabio