Prior sensitivity analysis for DSGE model

Estimation ML/Bayesian estimation
1

Dear Johannes,
I have some questions about prior sensitivity analysis.

  1. Should I conduct prior sensitivity analysis upon all parameters or could i just forcus on the parameter that I interest, for example, if I am interested in a parameter with support ranging from 0 to 1, for prior sensitivity analysis for this parameter, should I first estimate the DSGE model with a beta prior (0,1) for this parameter, then estimate the DSGE model with a uniform prior distribution ranging from (0,1) for this parameter, finally compare the two estimation, if the two estimation output do not vary much from each other, then I call this estimation robust to prior of this parameter?
  2. If I simply change prior mean but keep the same prior distribution type, then compare the two output with different prior mean, can this also be prior sensitivity analysis, not sensitive to prior mean?
    Thank you very much and look forward to hearing from you.
    Best regards,
    Jesse
2
  1. Usually one focuses on the parameters of interest. If a uniform prior works, then that is often sufficient.
  2. That will usually be informative as well. Prior-posterior plots usually show this.