As part of reviewer's requests/recommendations, I am trying to compare forecast accuracy of the (almost) same model using fixed effects, random effects, and mixed effects. My data is panel data. I see that forecast support xtreg but does not support xtmixed. Has anyone succeeded using Stata to forecast based on xtmixed model outputs?

If anyone from Stata is reading, can anyone briefly explain, under the hood, why forecast does not support xtmixed models? Is this some feature that can be added/updated in future Stata version? If for now forecast cannot support xtmixed models, what is the "work around" solution using Stata to forecast based on xtmixed model outputs?

Thanks for taking your time reading and/or answering in advance!