I’m trying to estimate a medium scale DSGE model. I use mode_compute=6 to find the mode and to estimate the parameters. Other optimizers for the mode computation don’t really seem to work, which also suggests that there are problems with the model or observation.
My problem is that almost all of the parameters have very small stderrs in the mode (something like 0.000X) and the posteriors look like spikes for most of the variables.
Even though the priors are very allowing and the acceptance rate is close to 0.33 and mh_replic=1000000, the posteriors are restricted to a very very tight area.
Are there some general tips about what to check or what should be done when the mode has very small stderrs?
My experience is that relaxing the priors doesn’t do much here.