Estimation results

Dear Mr pfeifer,
I got estimation results and I want to have your opinion about it please.

  1. Identification is ok
  2. I have a problem of convergence, I used estimation(datafile=data_furlanetto, mh_jscale=0.2,mh_drop=0.5, mode_compute=6,moments_varendo,conditional_variance_decomposition=[1 2 4 8],mh_replic=500000,conf_sig=.95,bayesian_irf,irf=20). How can I do to improve convergence, Is that recommanded to do mode_file = X_mh_mode, mode_compute = 9 in my case?
  3. Impulses response seem not good at all, I can’t see the upper and the lower bound. How Can I resolve this?
  4. After closing all figures, where can I find them again (Impulse response and MCMC convergence)? is that possible?

Thank you!

Clara
results estimation_my model_Fur.pdf (749 KB)

  1. Yes, try a different mode-finder after the one you just ran.
  2. This is a well-known problem with Dynare 4.4.3 on newer Matlab versions. It goes away when using the unstable version. Alternatively, see [Confidence band of Bayesian IRF (SW2007))
  3. See Pfeifer (2014): An Introduction to Graphs in Dynare at sites.google.com/site/pfeiferecon/dynare on where figures are saved