Forecasting Global Recessions in a GVAR Model of Actual and Expected Output in the G7

The forecasting performance of a Global VAR model of actual and expected outputs in the G7 economies is compared with that of alternative models to judge the usefulness of modelling cross-country interdependencies and employing survey data. Both effects are found to be important in calculating density forecasts, in forecasting the occurrence of recessionary events deï¬ned at the national and G7-wide levels and, through a novel ‘fair bet’ exercise, in decision-making based on forecasts. The analysis argues for a nuanced approach to presenting output predictions, avoiding simple point forecasts and focusing on features of future growth dynamics relevant to decision-makers.