Juhee Bae, University of Glasgow
Project Description/Abstract
This paper studies the predictive performance of various factor estimations comprehensively, in a coherent forecasting framework, under the big data that consist of major U.S. macroeconomic and finance variables. 148 target variables are forecasted, using 7-factor estimation methods, with 11 decision rules that determine the number of estimated factors for forecasting. First, I find that the number of estimated factors used in forecasting is important. Incorporating more factors may not always provide better forecasting performance. Second, using consistently estimated number of factors may not necessarily improve predictive performance. Forecasts obtained by consistent estimators perform well, except for Partial Least Squares (PLS). The first PLS factor very often shows stronger forecasting performance than when the number of PLS factors is decided by consistently estimated number of total factors. Third, the 7 best forecasting performance of 7-factor estimation methods, chosen across different decision rules, tends to be very similar. However, there is a large difference in the forecasting performance across different decision rules, even when the same factor estimation method is used. Therefore, the choice of factor estimation method, as well as the decision rule for the number of factors, is crucial in forecasting practice. Finally, the first PLS factor tends to yield forecasting performance very close to the best result from the whole combinations of the 7- factor estimation methods and 11 decision rules. PLS estimates factors using not only predictors but also a target variable, which can explain the significant forecasting improvement of PLS.