MCMC in econometrics

This paper presents the methodology of Markov chain Monte Carlos (MCMC) to statistical inference in Ecometrics. MCMC theory is reviewed and some relevant pratical aspects associated with convergence of the chain are discussed. The most common forms of MCMC using Gibbs sampling and the Metropolis-Hasting algorithm are described. The methods are illustrated in the contexts of time varyng and varyng generalizations of linear regression models. Examples of these models in Econometrics are provided and illustrated with Brazilian economic data.