Factor-Augmented Forecasting in Big Data

Abstract

This paper evaluates the predictive performance of various factor estimation methods with big data. Extensive forecasting experiments are examined, using 7 factor estimation methods with 11 decision rules that determine the number of factors. First, the number of estimated factors used in forecasting is important. Incorporating more factors tends to diminish predictive accuracy. Second, using a consistently estimated number of factors may not necessarily improve forecasting performance. The greatest predictive gain often derives from decision rules that do not estimate consistently the true number of factors. Third, the best forecasting performances of 7 factor estimation tend to be similar when an appropriate decision rule is applied. However, forecasting performance varies widely across different decision rules, even when the same factor estimation method is used. Finally, the first partial least squares factor tends to yield forecasting performance very close to the best result from all the possible alternatives.

Keywords: Factor Models, Forecasting, Partial Least Squares (PLS)

JEL Classification: C22, C23, C53