How to create hierarchical prior for Bayesian estimation

[Jesse]

Hi All,
I am wondering that how can I create hierarchical prior for Bayesian estimation in Dynare?
Thank you.
Kind regards,
Jesse

[jpfeifer]

Please provide more details what you would like to do.

[Jesse]

Dear Johannes,
Thank you very much replying.
I am applying a mixture normal to one of the shocks in my DSGE model, and the mixture normal distribution requires a weighted average of a number of normal distributions, for Bayesian estimations, I need to sample sample the hyperparameter- the weights first, before I can sample shocks from the mixture normal distribution for my shock.
I have prepared a distribution for weights already, but I need to sample weights first, then i used the weights obtained from my sampling to formulate the mixture normal distribution (the parameters contain weights), then I can sample my shock from the mixture distribution.
Kind regards,
Jesse

[jpfeifer]

Why do you need to sample shocks for the estimation? Are you doing particle filtering?

[Jesse]

Dear Johannes,
Thank you for your reply and sorry to reply you late. No, I am using Kalman filter, however, one of my shocks follows a Gaussian Mixture distribution-a weighted average of normal distributions with different means and variances. I need to sample weight paramters, mean parameters, variances parameters first, then once i get these prior parameters, i can sample shocks from the Gaussian mixture distributions.
Kind regards,
Jesse

[jpfeifer]

Sorry, but I still don't get it. Where during Kalman filtering do you need to sample shocks?