Some questions regarding Bayesian estimation

[shuang liu]

Hello, I'm a new user of Dynare and I got some questions regarding Bayesian estimation. The User Guide says that we need the estimated_params command to implement the Bayesian estimation, and we are supposed to define the prior shape, mean and standard error. But how can I get the information about all of these without data?
Sincerely look forward to your reply.

[jpfeifer]

As the name prior says, it incorporates your (subjective) prior knowledge about the parameters under consideration. It must not depend on the data! It is completely up to you to set it. Bad, but common practice is to use the priors of seminal papers like Smets/Wouters (2007) or Justiniano/Primiceri/Tambalotti.
Please also see http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=7195 and http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=6282

[shuang liu]

Dear professor, thanks for your reply and I got another question. It seems that the prior mean and standard error are kind of subjective, so how can I test the robustness of the parameters?

[jpfeifer]

Of course priors are subjective. That is the point of Bayesian analysis.
What do you mean with robustness? Ideally, your data is informative and your posterior should look different from the prior. You could repeat estimation with a more diffuse prior and see whether it changes anything substantially.