Evaluating the use of realtime data in forecasting output levels and recessionary events in the USA

Abstract

The paper proposes a modelling framework and evaluation procedure to judge the usefulness of real-time datasets incorporating past data vintages and survey expectations in forecasting. The analysis is based on `meta models obtained using model-averaging techniques and judged by various statistical and economic criteria, including a novel criterion based on a fair bet. Analysing US output data over 1968q4-2015q1, we find both elements of the real-time data are useful with their contributions varying over time. Revisions data are particularly valuable for point and density forecasts of growth but survey expectations are important in forecasting rare recessionary events.