A Bayesian analysis of normalized VAR models

Identified vector autoregressive (VAR) models have become widely used on time series data in recent years, but finite sample inference for such models remains a challenge. In this study, we propose a conjugate prior for Bayesian analysis of normalized VAR models. Under the prior, the marginal posterior of VAR parameters involved in identification can be either derived in closed form or simulated through Gibbs sampling. The method developed in the study is applied to a VAR of macroeconomic data.