A Direct Monte Carlo Approach for Bayesian Analysis of the Simultaneous Equation Model

Computationally efficient methods for Bayesian analysis of simultaneous equation models (SEMs) are described and applied that involve use of a direct Monte Carlo (DMC) numerical algorithm to implement Bayesian estimation and prediction procedures employing diffuse prior densities for parameters and normal likelihood functions. This DMC approach is shown to be very useful for the computation of posterior densities, moments, intervals and other quantities for parameters and future values of the endogenous variables. The results of Monte Carlo experiments and analyses of two models using empirical data show that the new approach performs rather well. JEL classification: (C11; C31