Forecasting government bond yields with large Bayesian vector autoregressions

► We use a large BVAR to forecast the term structure of interest rates. ► We optimally select the amount of shrinkage by maximizing the marginal likelihood. ► We evaluate the forecasts using trading schemes and portfolio allocation exercises. ► The proposed approach systematically outperforms a random walk forecast. ► We check the robustness of our results using a data based Monte Carlo simulation.