Echantillonnage de Gibbs et autres application econometriques des chaines merkoviennes

This survey provides an introduction to Markov Chain Monte Carlo (MCMC) sampling techniques and to their applications to Bayesian econometrics. In describing the Gibbs sampler and the Metropolis-Hastings algorithm, the emphasis is put on how these techniques can be put into practice; the theoretical foundations are outlined using the elementary properties of Markov chains. To illustrate the potential of MCMC techniques, we decribe several examples where their application has produced clear gains over classical methods of inference. Ce survol fournit une introduction aux techniques d’echantillonnage de type Markov Chain Monte Carlo (MCMC) et leurs applications a l’econometrie bayesienne. Par ce survol notre but n’est pas d’expliquer les fondements theoriques derriere les methodes de type MCMC, mais bien de faire un expose pratique des techniques qui s’y rapportent. Nous chercherons surtout a mettre en valeur la facilite et l’etendue des applications par l’utilisation d’exemples simples.(This abstract was borrowed from another version of this item.)(This abstract was borrowed from another version of this item.)