Has modelsforecasting performance for US output growth and ination changed over time, and when?

We evaluate various models’relative performance in forecasting future US output growth and in‡ation on a monthly basis. Our approach takes into account the possibility that the models’relative performance can be varying over time. We show that the models’ relative performance has, in fact, changed dramatically over time, both for revised and real-time data, and investigate possible factors that might explain such changes. In addition, this paper establishes two empirical stylized facts. Namely, most predictors for output growth lost their predictive ability in the mid-1970s, and became essentially useless in the last two decades. When forecasting in‡ation, instead, fewer predictors are signi…cant (among which, notably, capacity utilization and unemployment), and their predictive ability signi…cantly worsened around the time of the Great Moderation. Keywords: Output Forecasts, In‡ation Forecasts, Model Selection, Structural Change, Forecast Evaluation, Real-time data. Acknowledgments: Barbara Rossi gratefully acknowledges …nancial support from a SAS Forecasting Research Grant. The Grant supported research on new methods for forecasting and model evaluation. This paper shows the empirical relevance of such techniques in the presence of time-varying relative forecasting performance. We thank seminar participants at the 2007 Midwest Econometrics Group Conference, the Macro-Money seminar at UNC-Chapel Hill and the EMSG at Duke University for comments. J.E.L. Codes: C22, C52, C53