Unusually long estimation speed

Estimation
1

I’m running RBC_Est.mod (796 Bytes)along w/ simuldataRBC.m (140.9 KB) . I’m trying estimate the parameters using mode_compute=6 (since all mode_compute= 1 to 4 ends up with error: Error using chol Matrix must be positive definite.). However, I’ve noticed MCMC estimation is taking hours and hours to complete.

My system config;

  1. Windows 10 (64 bit)
  2. Intel 1.5GHz (2 cores) processor
  3. 4GB RAM
  4. Dynare 4.6.3
  5. Octave 5.2.0.

My questions are;

  1. Is this pace of estimation unusual for a small model like this?
  2. If I run the same files on Matlab would I experience a notable increase in performance?
  3. If I reduce the number of parameters to estimate would the process be faster?
2

There are various issues here:

  1. Regarding speed, your machine is slow and you are using Octave. That is a very bad combination. Matlab would be a lot faster.
  2. You are not providing an analytical steady state. That slows things down, too.
  3. Use mode_compute=5 to speed things up.
  4. Running identification will show you that your model is mostly not identified given your observables. For that reason, estimation will not work.
3

Thanks professor Pfeifer for the prompt reply.