Thanks very much for your response, Prof Pfeifer.
I hadn’t considered this issue before. I previously thought that in the IRF, all variables, such as interest rates and inflation, should be expressed as percentage deviations, with a unified measurement unit, expect for interest rates and inflation.
I also read your reply in this post. Question about understanding irfs in dynare - #3 by jolly816
Regarding vacancies, should I use the log percentage deviation? Like in your Born_Pfeifer_2020/BP2020_CES.mod
IRF_mat_percent_from_SSS_logged = (IRF_mat(1+burnin+1:1+burnin+options_.irf,:)-stochastic_steady_state);
For such tinty values, how can I accurately depict its changes?