Run the random walk Metropolis algorithm with a normal or a Student-t proposal density, the slice sampler, the fixed blocking RWM sampler with a normal or a Student-t proposal density, or the random blocking RWM sampler with a normal or a Student-t proposal density to sample from the posterior distribution of the DSGE model parameters when the marginal likelihood of the DSGE-VAR is used. The number of posterior draws, the number of burn-in draws, the number of parallel sampling chains, etc., are determined in the posterior sampling frame on the Options tab.
The parameterization of the proposal distribution for the random walk Metropolis is also determined from your settings in the posterior sampling frame. In addition, the user can choose which posterior mode results to make use of for the proposal density.
The posterior sampler is computed a user selected value of the λ hyperparameter of the DSGE-VAR. YADA supports three choices:
1. | The posterior mode of the DSGE model; |
2. | The marginal posterior mode of the DSGE model parameters based on the DSGE-VAR for the same value of the λ hyperparameter; and |
3. | The joint posterior mode of the DSGE model parameters based on the DSGE-VAR for the same value of the λ hyperparameter. |
To have access to a specific choice, the estimation must already have been performed. Hence, if only the posterior mode estimates of the DSGE model exist, then this is the estimator that YADA will use when parameterizing the posterior sampler.
Additional Information
• | A detailed description about posterior sampling of the DSGE model parameters in the DSGE-VAR model can be found in Section 15.5 of the YADA Manual. |
Page url: http://www.texlips.net/yada/help/index.html?dsge_posterior_sampling2.htm