The loglikelihood at the posterior mean
Posted: Wed Oct 31, 2012 3:50 pmDear all
We can extact the mcmc draws of the parameters from the subfolder metropolis , and compute the mean of the posterior draws. And in this file, there exists a
variable "logpro2", I think it is the loglikelihoods at diffrent draws of parameters, am i right?
Another question is how to compute the loglikelihood at the mean of the parameters? I think mean may diffrent from each of the mcmc draws.
Now I am trying to estimated the model in An and Frank (2007), I estimated his model by the simulated data first, then subsititute the posterior mean of parameters into
the model and solve it to state space form by Dynare. Then I use Kalman Filter to compute the loglikelihood by my own code.
Am I right? Is there some order in dynare can do this?
Best Regards
Zeng Tao
We can extact the mcmc draws of the parameters from the subfolder metropolis , and compute the mean of the posterior draws. And in this file, there exists a
variable "logpro2", I think it is the loglikelihoods at diffrent draws of parameters, am i right?
Another question is how to compute the loglikelihood at the mean of the parameters? I think mean may diffrent from each of the mcmc draws.
Now I am trying to estimated the model in An and Frank (2007), I estimated his model by the simulated data first, then subsititute the posterior mean of parameters into
the model and solve it to state space form by Dynare. Then I use Kalman Filter to compute the loglikelihood by my own code.
Am I right? Is there some order in dynare can do this?
Best Regards
Zeng Tao