Description

This article develops a large Bayesian VAR with more than 100 variables for the Chilean economy, as Banbura, Giannone and Reichlin (2010). We show that, when the degree of shrinkage is set in relation to the cross-sectional dimension of the sample (bayesian shrinkage), the forecasting performance of a VAR can be improved by adding macroeconomic variables and sectoral information. The results show that the large bayesian VAR compares favorably with some univariate models. It further examines the impulse responses to a monetary shock, as well as to some sectoral shocks