Deal All,

BVAR-DSGE are now implemented in DYNARE (I still have to document it in the official release). Here is an example that shows how a BVAR-DSGE may be estimated with DYNARE as in Del Negro & Schorfheide (2004): …

This archive contains a data file and a commented *.mod file.

Best, St

Hi Stephane
trying the BVAR-DSGE program, Matlab do not find BVARDSGE1_mode


If you uncomment the “estimation” command at line 99 (and comment out the same command in lines 104-108), that will compute the mode and store it in oo_.posterior_mode. You can then retrieve the modal values of the parameters and construct your own mode_file (see the documentation for the estimation command).




It’s even easier – after running the first estimation, the BVARDSGE1_mode file is created automagically. So just run the model again with the other estimation command.


Hi P.Summers
Thanks a lot

Hi St

This is fixed. It was my mistake preparing for the server migration.



Dear Stephane

After running the BVAR-DSGE code with my model, dynare generates the following error messages after provding the prior plots. Is this related to my chioce of the number of first observation? or my observation sample size is too small: nobs = 36. Also I have five observables for the VAR model (serven shocks for the DSGE)

**??? Index exceeds matrix dimensions.

Error in ==> d:\dynare_v3.065\matlab\VarSampleMoments.m
On line 41 ==> Y = data(FirstObservation:LastObservation,:);

Error in ==> d:\dynare_v3.065\matlab\dynare_estimation.m
On line 235 ==> evalin(‘base’,’[mYY,mXY,mYX,mXX,Ydata,Xdata] = ’ …

Error in ==> D:\Matlab7.0\work\dsgevarhag.m
On line 302 ==> dynare_estimation(var_list_);

Error in ==> d:\dynare_v3.065\matlab\dynare.m
On line 26 ==> evalin(‘base’,fname) ;**

Thanks very much in advance

Hi, I really need that
But I can’t download it
Can you update the link?

Thank you in advance

Sorry, but the link is dead…
Is someone could update it?
Thank u in advance


Hi, the example can be found on our wiki here.
Best, Stéphane.

Following on this, is there any way to obtain larger graphs from the “bayesian_irf” option? - they are very nice, but very small!