Hey! Again I’m trying to replicate a DSGE model. I believe this is an interesting way to learn how to use Dynare and also to better understand the use of these models in the economy.
This time it looks like I entered the equations correctly, but I don’t know if the model is correct. The output graphs present curves similar to the author’s, however, the values are different. I don’t know if the shock values are different or again there is an error.
This led me to the question: how good are DSGE models? And how transparent are the researchers who use these models? It seems that it is a very arduous task to replicate the results of some article.
Anyway, I would like to know if there are any errors in my code or in the author’s article? Are my shocks poorly sized? In the article the author does not inform the value of the shocks, I assume it is 1 standard deviation.
Paper: A small-scale DSGE-VAR model for the Romanian economy
Model.mod (4.0 KB)
pop2016.pdf (1.1 MB)
Replicating estimated models is tough and often close to impossible without access to the original codes and particularly the data. As you figured out the hard way, information in papers is often not complete. Dynare by default uses 1 standard deviation shocks and those are clearly Dynare figures in the paper. But without information on what the standard deviation was, you will not be able to perfectly replicate the paper.
Apparently the model was inserted correctly in Dynare, although, I don’t know sigma I think everything is ok. However, I get this message:
“There are 4 eigenvalue(s) larger than 1 in modulus
for 4 forward-looking variable(s)”.
I think some problem with the time of the salary equation. Could you help me understand and solve this problem? I read some things here in the Forum, however, I didn’t understand very well.
I would also like to apologize for posting questions often, but it helps me to better understand the DSGE tool and models. I believe that what I ask may also be of interest to beginners.
Model.mod (13.3 KB)
pop2016.pdf (1.1 MB)
Nominal variables have a unit root in the New Keynesian model. It seems you are considering the nominal wage. You may want to consider wage inflation instead. That being said, having the unit root in the model is typically no reason to worry.