Using priors for shock standard deviations from other papers

Estimation ML/Bayesian estimation
1

Dear Dynare community,

I am estimating a DSGE model in Dynare and I have a question about setting priors for the standard deviations of shocks.

If another paper reports priors (distribution and hyperparameters) for the shock standard deviations, can I directly use those in my model? Or should I be cautious about possible differences, for example:

  • how the shocks are defined in each model,
  • whether the data are expressed in percent vs. log deviations,
  • annualized vs. quarterly growth rates,
  • or differences in observables and detrending.

In other words, is it acceptable to copy priors from another paper, or is it better to treat them only as a guide and specify wider priors in my model?

Thank you for your advice.

2

You should always proceed with caution. You can only rely on other people’s priors if they apply to the same concept. Time aggregation and whether something is in percent is of course important.

The following thread may also be relevant: Choosing an appropriate standard deviation for prior

1 Like
3

Thanks a lot Professor!