Hi Tommaso,
On your guide (chapter 5 I think) you mention that dynare plots a green dashed line at the posterior mode. In a few cases, I found that the green line does not coincide with the peak of the simulated posterior. In `Practicing Dynare', for example, there are a few graphs where this happens.
Is this an indication that Metropolis-Hastings has found a `better' mode and that one should re-run the estimation from different starting values?
Thanks,
Mariano
Dynare forums
Forums for asking questions, posting comments and uploading examples related to Dynare
Green dashed lines
Forum rules
This forum is closed. You can read the posts but cannot write. We have migrated the forum to a new location (https://forum.dynare.org) where you will have to reset your password.
This forum is closed. You can read the posts but cannot write. We have migrated the forum to a new location (https://forum.dynare.org) where you will have to reset your password.
2 posts
• Page 1 of 1
Green dashed lines
by mkulish » Wed Sep 26, 2007 10:53 am
- mkulish
- Posts: 10
- Joined: Sat Aug 11, 2007 7:23 am
Re: Green dashed lines
by StephaneAdjemian » Wed Sep 26, 2007 4:24 pm
Hi Mariano,
The difference between the green dashed line and the top of the marginal posterior density may either indicate that the posterior mode previously found with the optimization routine is not the global mode or that the Markov chain is not long enough (in this case, the discrepancy should narrow if you increase the number of simulations).
If your convergence diagnostics (of the mcmc) are okay, the first possibility is the more likely. If you really needs to present the posterior mode in your tables you may run again the optimization using the top of the posterior marginal densities as an initial guess.
Best,
Stéphane.
The difference between the green dashed line and the top of the marginal posterior density may either indicate that the posterior mode previously found with the optimization routine is not the global mode or that the Markov chain is not long enough (in this case, the discrepancy should narrow if you increase the number of simulations).
If your convergence diagnostics (of the mcmc) are okay, the first possibility is the more likely. If you really needs to present the posterior mode in your tables you may run again the optimization using the top of the posterior marginal densities as an initial guess.
Best,
Stéphane.
- StephaneAdjemian
- Posts: 429
- Joined: Wed Jan 05, 2005 4:24 pm
- Location: Paris, France.
2 posts
• Page 1 of 1
Who is online
Users browsing this forum: No registered users and 1 guest