Bayesian Full Information Analysis of Simultaneous Equation Models Using Integration by Monte Carlo
by Luc Bauwens
I. The Statistical Model -- 1.1 Notation -- 1.2 Interpretation -- 1.3 Likelihood function -- II. Bayesian Inference: The Extended Natural-Conjugate Approach -- II.1 Two reformulations of the likelihood function -- II.2 The extended natural-conjugate prior density -- II.3 Posterior densities -- II.4 Predictive moments -- II.5 Numerical integration by importance sampling -- III. Selection of Importance Functions -- III.1 General criteria -- III.2 The AI (?) approach -- III.3 The AI(?) approach -- IV. Report and Discussion of Experiments -- IV.1 Report -- IV.2 Conclusions -- V. Extensions -- V.I Prior density -- V.2 Nonlinear Models -- Conclusion -- Appendix A: Density Functions: Definitions, Properties And Algorithms For Generating Random Drawings -- A.I The matricvariate normal (MN) distribution -- A.II The inverted-Wishart (iW) distribution -- A.III The multivariate Student distribution -- A.IV The 2-0 poly-t distribution -- A.V The m-1 (0 < m ? 2) poly-t distribution -- Appendix B: The Technicalities of Chapter III -- B.I Definition of the parameters of (3.3) and (3.6) -- B.II Computation of the posterior mode of ? -- B.III Computation of (3.15) -- Appendix C: Plots of Posterior Marginal Densities And of Importance Functions -- Appendix D: The Computer Program -- Footnotes -- References.
