According to literature, the linearization in Dynare(order=1 in stoch_simul) should have the same result with 1-order approximation in Dynare++. However, when I simulated a simple RBC model in DYNARE and DYNARE++, I got huge differences between variances(or standard variances) of variables in DYNARE and those in DYNARE++. Can anybody give me a reasonable explanation for this phenomena? Thanks for your help!
Please provide the mod-files for both.
I have attached both DYNARE code and DYNARE++ code here. Thanks for your reply.
RBCDYNARE.mod (582 Bytes)
RBCDYNARE++.mod (491 Bytes)
You are using different orders, the HP-filter in one case, and you are using a variance of 0.01 in one case and of 0.01^2 in the other one.
I have corrected what you mentioned. The variances now are close to each other. By the way, how to implement HP-FILTER in dynare++? By using the dynare_simul to get the simulated data and then applies HP-FILTER to the simulated data?
Yes. Dynare++ does not support the HP-filter. You need to apply it manually to the simulated series.
Thanks, got it!