Dynare forums

Dear all,
I ran a model with 3 chains and I obtained the following acceptance rates:
Chain 1: 6.8844%
Chain 2: 17.4192%
Chain 3: 28.5928%

Which one should I choose in order to play with mh_scale ?

I mean, if the acceptance rate of the last chain is comprised between 0.33 and 0.25, is it sufficient to conclude that we obtained satisfying acceptance rates ?
Or should all the 3 chain's acceptance rates be comprised between 0.33 and 0.25 in order to conclude that we obtained satisfying acceptance rates ?

Thank you for you answer,
Jonathan

When your are sampling from the ergodic distribution of the MCMC, the chains should have exactly the same properties. The hugely differing acceptance rates suggest that at least one chain has not properly converged yet. Try a trace_plot to find out which one(s) are affected. If it is only chain one, the scale factor is OK. If it is the other two, you should change it.