A Panel Data Simultaneous Equation Model with a Dependent Categorical Variable and Selectivity.

This paper develops a Bayesian MCMC algorithm to estimate a Panel Data Simultaneous Equations model with a dependent categorical variable and selectivity. In contrast with previous Bayesian analysis of selectivity models, the algorithm does not require the observation of some regressors which do not enter into the likelihood function. This makes the algorithm applicable to studies of the labor market where there are typically missing regressors. In addition, the paper provides an scheme to sample the slope parameters using an analytical approximation of the posterior distribution as a proposal density. Estimation with a simulated dataset illustrates the performance of the algorithm.