How are moments of simulated variables computed

[beanbag]

Hello

How does dynare compute the moments of the simulated variables ?
For your info: I'm running "stoch_simul(periods = 1000,IRF = 200,nograph)" on a fully calibrated model, no estimation or other commands whatsoever.

if I store the IRF values of variable "Consumption" in matrix "C" and I give in the matlab command "std(C)" I get a value that is a lot smaller than the value displayed in the dynare output under "MOMENTS OF SIMULATED VARIABLES ST. DEV"

What is the difference in computation ?
I have got a gut feeling that maybe some observations might be dropped once a certain treshold of proximity towards steady state is breached, that would explain the smaller values of my own calculations. I could be completely wrong though.

Any insights are sincerely appreciated.



Kind regards

[MichelJuillard]

IRF s are the response to a single shock at the start of the simulation.

The variance and standard deviation reported for theoretical of simulated moments take into account repeated shocks to each of the exogenous variables.

Michel

[beanbag]

Thank you for your response. That makes sense.