Just to be sure: to calculate the PSRF as show in the beginning oh the Section 3 of Brooks and Gelman (1998), all I need to do is to divide the blue and the red line shown in the MCMC univariate diagnostics?
Conceptually, the answer is yes. But not that Dynare does not plot the standard deviations but interpercentile ranges. That is different from the original paper.
Thank you. Is the interpercentile range referring to the interval chart? And the m2 chart is simply Equation 3.1 from Brooks and Gelman (1998), correct?
If I understood correctly, the implementation for the PSRF using Equation 3.3 is attached.
mcmc_diagnostics.m (25.7 KB)
Yes, exactly. I had forgotten that they also use the non-parametric version.