Stochastic Production Frontier Models
Abstract
In this project we use Markov chain Monte Carlo (MCMC) methods in order to estimate and compare stochastic production frontier models from a Bayesian perspective. We consider a number of competing models in terms of different production functions and the distribution of the asymmetric error term. All MCMC simulations are done using the package JAGS (Just Another Gibbs Sampler), a clone of the classic BUGS package which works closely with the R package where all the statistical computations and graphics are done.
Key Words: Markov chain Monte Carlo, Gibbs sampler, JAGS, Bayesian model averaging, model comparison.
