Data generation of DSGE model

[aiq]

I am trying to replicate data as shown kindly by Stephanie in this file
https://github.com/DynareTeam/dynare/bl ... lambda.mod

This is the replication of the standard model, and she shows us how to generate artificial data which is then used for estimation.

The command she uses is "stoch_simul(periods=500)" and "datatomfile('datarabanal_hybrid',[])"
This command generates series for 500 periods for the variables.

My question is this: If I want to generate say 5 sets of data for 200 periods, and discard 5,000 or so for each data set, how do I go about doing so in this model?

[jpfeifer]

You cannot use this command. The best option is

Code: Select all

stoch_simul(order=1,periods=5200,simul_replic=100);
[sim_array]=get_simul_replications(M_,options_);


using the get_simul_replications from my homepage. In sim_array you will find all simulated values. You would have to drop the initial 5000 simulation points.

P.S.: Stéphane is a he.

[aiq]

Thanks a lot for your assistance.

i did what you said, and I got the series I wanted. Just one thing, wouldn't it be more efficient to repeat this process multiple times?

Also a general question is how can one extrapolate artificial data from a model, to match the data, because for example, data generated from the RBC model does not match the predictions of the real world data at all! A good model would be one that matches, then?

Sorry, I am doing an MA and this stuff is not very clear to me.

[jpfeifer]

What do you mean with "more efficient"? The data is generated from indepent simulations.

You mean you want to do moment matching?

[aiq]

yes, thanks a lot, I have figured out the issue.