In this paper, we provide evidence that fat tails and stochastic volatility can be important in improving in-sample fit and out-of-sample forecasting performance. Specifically, we construct a VAR model where the orthogonalised shocks feature Student’s t distribution and time-varying variance....