Page 1 of 1
convergence problem
Posted:
Tue Sep 25, 2012 9:01 amby killergabin
Hi,
I've been trying to estimate a RBC model with Dynare but my model don't converge and f(x) stay around 0.26. Being a new dynare' user I think i made some mistakes in my .mod file. I hope i could find some useful advices here.
Thanks!
(.mod file attached)
Re: convergence problem
Posted:
Tue Sep 25, 2012 10:06 amby kyri82
Have you tried to change the initial conditions?
Re: convergence problem
Posted:
Tue Sep 25, 2012 10:42 amby killergabin
Yes, too many times actually and these values are the one giving me the best values for f(x); may be the problem is somehere else! I don't know
Re: convergence problem
Posted:
Tue Sep 25, 2012 10:49 amby killergabin
Dear Kyriacos if you have something in mind...
Re: convergence problem
Posted:
Tue Sep 25, 2012 11:43 amby kyri82
I am not sure, what can I tell you... I do not do much estimations my self. I calibrate my models but I use fsolve - myself, not via Dynare - to estimate the steady-state values. Make sure that the problem is stationary and well defined. Make sure that there are no unit roots. Some parameter values are sometimes associated to unit roots. Make sure that this is not your case. My other guess would be to try different initial conditions but you already did that...
Good luck!
K.
Re: convergence problem
Posted:
Tue Sep 25, 2012 12:02 pmby killergabin
ok i'll try that! thank you
Re: convergence problem
Posted:
Tue Sep 25, 2012 2:05 pmby jpfeifer
Put
- Code: Select all
resid(1);
before steady and trace down the residuals. Either your steady state computation is wrong or the entered model equations. Currently there are large residuals suggesting that did a mistake somewhere.
Re: convergence problem
Posted:
Tue Sep 25, 2012 2:20 pmby killergabin
ok thanks.
Re: convergence problem
Posted:
Mon Jan 21, 2013 6:55 amby killergabin
Hello,
I finally find some time to try to run my model again. I still have some issues there! Can someone tell me what's wrong?