Working Papers, Department of Economics, Rutgers, The State University of New Jersey 2001,14
Inference on the long-run properties of a Vector Autoregression (VAR) consisting wholly of I(1) variables are made using Bayesian methods. In particular, the implications on the forecast and impulse response function distributions of directly estimating and restricting the drift parameters of variables in a VAR are studied. It is found that the forecast and impulse response distributions are sensitive to the prior distribution placed over the drift parameters.
Forecasts Impulse Response Function Markov chain Monte Carlo Vector Autoregression