I’m modeling a simple NK model and trying to get simulated fake data with various shocks at random.
To do so, I’m using stoch_simul(order=1,irf=0,hp_filter=1600,simul_replic=1,periods=300)
but I cannot understand the result.
I thought that by decreasing the stderr of each shock in a shocks block, the variance of the replicated data (eg. y, i, pi) tends to be smaller.
However, even though I changed the stderr of each shock in linear_nk_mod, the simulated replicated data (fake y1, fake i1, fake pi1 in the code below) did not change at all.
Is my understanding wrong?
Or, how can I get different replicated data with various volatility?
You need at least simul_replic=2 for get_simul_replications to work with the current run. My guess is you are always loading an old _simul.mat-file still present on the disk.
Thank you for quick reply.
By changing simul_replic from 1 to 2, I managed to get the expected result.
Thank you very much!
By the way, I wonder why simul_replic = 1 cannot work well. In my understanding, simul_replic determines the number of simulation, and I thought even when simul_replic = 1, the simulation can run well (though it generate only one future path.)
What is wrong?
When you set simul_replic=1, the simulation is stored in oo_.endo_simul. Only if there is more than 1 simulation Dynare will save the additional simulation in a binary file on the disk. See the manual.
I think you have to do it manually. datafile requires .csv or .mat or some other form as mentioned in manual. Maybe oo_.endo_simul is not in these forms.