Dynamic Probit Model with Gibbs Sampling

[TomDoan]

This estimates using Bayesian methods a probit model with the latent index variable assumed to follow an AR(1) process. This uses the typical Gibbs sampling technique of drawing the continuous latent variable given the observable 0-1's, but, because of the dynamic process for the latent variable, you can't just draw those independently. Instead, the YSTAR are done one-at-a-time given the values of the YSTAR at all other data points.

The same idea, applied to a VAR (where the latent variable is one of the variables in the VAR), is at http://www.estima.com/forum/viewtopic.php?f=8&t=1191.

gibbsprobitdynamic.rpf

Program file

(2.15 KiB) Downloaded 75 times