Forecasting in the presence of instabilities: How do we know whether models predict well and how to improve them

Abstract

This article provides guidance on how to evaluate and improve the forecasting ability of models in the presence of instabilities, which are widespread in economic time series. Empirically relevant examples include predicting the nancial crisis of 2007-2008, as well as, more broadly, uctuations in asset prices, exchange rates, output growth and ination. In the context of unstable environments, I discuss how to assess modelsforecasting ability; how to robustify modelsestimation; and how to correctly report measures of forecast uncertainty. Importantly, and perhaps surprisingly, breaks in modelsparameters are neither necessary nor su¢ cient to generate time variation in modelsforecasting performance: thus, one should not test for breaks in modelsparameters, but rather evaluate their forecasting ability in a robust way. In addition, local measures of modelsforecasting performance are more appropriate than traditional, average measures.