New model predicts economic crises with unprecedented accuracy, changing forecasting game.
Time-varying VAR models are important for predicting economic crises, but current methods have limitations. A new model was created that improves on existing ones by allowing for more flexible dynamic parameters. An efficient estimation algorithm was developed using a combination of traditional and new techniques. By studying the Great Recession in four major economies, it was found that using the new model significantly improves forecast accuracy compared to the traditional random walk approach.