Bayesian Analysis of Econometrics Systems with Discrete Variables and Inequality Constraints

In econometric models, sign or inequality constraints on parameters arise in a wide variety of applications. In this paper, we incorporate linear and nonlinear inequality constraints into systems of equations with both discrete and continuous variables serving as the dependent variables. The posterior sample is generated using Gibbs and Metropolis algorithms with data augmentation. One illustrative example analyzes the choice between fixed and adjustable mortgage rates with inequality restrictions on the financial and borrower characteristics. A second application focuses on certification and performance of auditors.