Output when running osr

Dear all,

I would like to ask for a clarification, please: when I run the command osr, it automatically generated a) graphs with IRFs (with respect to all the shocks I have activated) for the endogeneous variable I mention after osr (i.e. the variable whose variance enters the loss function). However, b) the IRFs that I extract from oo_.irfs look a bit different. Can it be that the IRFs at aforementioned point a) are with respect to unitary shocks, while the ones I get at b) are with respect to standard deviation shock (as I have specified in the “shocks” block in the mod file)? Otherwise, what are the graphs I mentioned at point a) above?

Moreover, I noticed that, for the endogeneous variable (slet’s call it “b”) whose variance enters the loss function, for each active shock, say epsiX, two new variables are generated: 1) b_epsiX and 2) b_obs_epsiX. Further, I noticed that the variable plotted in the automatic graph generated by osr is the same as b_epsiX.

In other words, summing up, what do they represent the graphs automatically generated by running osr?

Thank you very much and best regards,


  1. I would need to see the code to investigate the source of the difference.
  2. Dynare by default saves IRFs also in the workspace by appending _shock_name to the name of the variable. They should be identical to the content of oo_.irfs .

Dear Johannes,

thanks a lot for your answer! Can I send the codes to your email, please? I could not find how to attach the codes here.

Thanks a lot,


You can simply drag and drop the file into the reply window. Or send a private message.