Comparing Theories Of Endogenous Protection: Bayesian Comparison Of Tobit Models Using Gibbs Sampling Output

Bayesian inference and model comparisons are easily performed quite accurately using Gibbs sampling, even if (1) the likelihood is analytically intractable and (2) nonstandard prior probability density functions (pdfs) are required. In this study Bayesian model comparisons are performed among five competing theories of endogenous protection. Tariff and nontariff barrier data from 1983 between the United States and five OECD partner countries--Japan, France, Germany, Italy, and the United Kingdom--are used in the analysis. Posterior odds based on two priors show special-interest models to be more likely than other models in determining U.S. protection. © 1998 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology