UN MODELO MACROECON OMICO BVAR DE PREDICCI ON PARA LA ECONOM IA VENEZOLANA

In this paper we consider and compare VAR bayesian models as a tool for analysis and decision-making in Venezuelan economy. We estimate three kinds of bayesian autoregressive vectors (BVAR): the rst one with the Minnesota prior, the second one simulates a posteriori densities using the gibbs sampling, and the third one a time-varying coecient model. We forecast the main variables of economic policy: GDP, ination, demand of money, private consumption and private imports. Then we compare the predictive performance through time horizons. The results suggest that the predictive performance of bayesian VAR models, is higher than traditional