Navigation:  Menu Commands > View Menu >

 

Posterior Sampling Summary

 

Print this Topic Previous pageReturn to chapter overviewNext page

View summary information about the posterior draws for up to 10 parallel chains. The results are divided into the following categories:

Model information: The data includes the files used by the estimation routine.
Sample and data information: The data contains the selected sample, the length of the training sample, the estimation sample, names of variables, etc.
Posterior draws information: Presents the number of posterior draws, the number of burn-in draws, the acceptance ratio of the random-walk Metropolis algorithm, the choice of inverse Hessian estimator, the scale factors for the proposal density and the initial value.
Marginal likelihood estimation: If you have chosen to estimate the marginal likelihood in the DSGE posterior sampling frame on the Settings tab, then the full sample estimates a presented here.
Parameters: The mean, median, and mode of the marginal posterior distribution of the original parameters are presented along with related parameter data. Specifically, the standard deviation based on assumed independent draws and autocorrelated draws (Newey and West, 1987, corrected), relative numerical efficiency (RNE), inefficiency factor (IF), numerical standard error of the posterior mean, as well as prior distribution information. Finally, quantile values are presented based on the percentiles specified in the data construction file. An example is provided in Table 2.

 

Additional Information

A more detailed description about the Newey and West (1987) numerical standard errors also be found in Section 8 of the YADA Manual.
A more detailed description of how to setup the data construction file for quantile selection is given in Section 18.5 of the YADA Manual.
RNE is equal to the ratio between the variance under independent draws and the variance under autocorrelated draws.
IF is equal to the inverse of RNE. This means that

IF = 1 + 2Σk=1 ρk,

where ρk is the autocorrelation at lag k. and the summation is taken until period K.

 

 


Page url: http://www.texlips.net/yada/help/index.html?posterior_sampling_summary.htm