Simulation moments

[kyri82]

Hello everyone,
Why when I compute the moments of the simulated data myself do not exactly match those provided by Dynare? I mean, I can always produce the mean by the command:
mean(oo_.endo_simul(1,:))
but the result does not match the first moment of the first simuated endogenous variable.

Thanks!

Kyriacos

[jpfeifer]

That is not the correct way. Dynare drops by default 100 points at the beginning of the simulation (see manual) plus the number of points required to initialize the autoregressive state space representation. Hence, it should be:

Code: Select all

mean(oo_.endo_simul(1,options_.drop+M_.maximum_lag+1:end))

[kyri82]

jpfeir,
thanks for your response! So just to make it clear to me: even though in the matrix oo_.endo_simul(:,:) I can find data for as many periods as I have asked for in the stoch_simul(periods = INTEGER), dynare computes moments by excluding the first 100+sthng data?
Thanks again!

Kyriacos

[jpfeifer]

Yes, you are right. By setting options.drop=x you can change the number of points to be dropped. Note that this is also documented in the manual.

[kyri82]

Yes I know. I just thought that the data points are dropped all together. Thanks again!
Kyriacos