Dear all,
I have been wondering for a long time about the following. When I simulate my model over 200000 periods in order to generate observations to calculate the sample moments, it takes ages in Dynare 4, about 2 hours. By contrast, in Dynare 3, it takes 4 minutes. I am solving the model using the second order approximation to the policy function.
Did any of you encounter the same thing?
Best,
Ansgar
Hi Ansgar,
The difference between dynare v3 and v4 is that the last version simulates (by default) fifty times the same model with 200000 periods… I guess this is not what you want (but this may be useful in some cases). The problem was corrected under dynare 3 but not under version 4. The problem is that we use the same variable (options_.replic equal to 50 by default) for different purposes… We will correct this bug as soon as possible.
Meanwhile, if you are not interested in IRFs, a workaround is to add the following options in the stoch_simul : replic = 1. If you are also interested in IRFs with a second order approximation you have to call again stoch_simul with options (say) irf=40 and replic=50.
Best,
Stéphane.
[quote=“Ansgar”]Dear all,
I have been wondering for a long time about the following. When I simulate my model over 200000 periods in order to generate observations to calculate the sample moments, it takes ages in Dynare 4, about 2 hours. By contrast, in Dynare 3, it takes 4 minutes. I am solving the model using the second order approximation to the policy function.
Did any of you encounter the same thing?
Best,
Ansgar[/quote]
Hey Stephane,
many thanks for your reply! I have used the replic option and it now works in acceptable time!
Best,
Ansgar