Hi professor, I came across some questions when estimating the parameters.

When the mode_check_plot reports some red dots, may I just ignore the those red dots? Does that mean that the posterior distribution or smoothed decomposition has some errors when those red dots shows up?

The HPD intervals of bayesian irf is almost zero. Is that feasible? How can I solve this problem?