Machine learning for regularized survey forecast combination: Partially-egalitarian LASSO and its derivatives |
| |
Institution: | 1. Queen Mary, University of London, United Kingdom;2. Warwick Business School, University of Warwick, United Kingdom;3. Bank of England, United Kingdom;4. City University and Centre for Macroeconomics, United Kingdom;1. KU Leuven, Belgium;2. Vrije Universiteit Amsterdam, The Netherlands;3. Tinbergen Institute, The Netherlands;4. University of Sydney, New South Wales, Australia;5. Econometric Institute, Erasmus University Rotterdam, The Netherlands;1. BEA, United States;2. Bank of Canada, Canada;1. The Bank of Korea, 55 Namdaemunno, Jung-Gu, Seoul 100-794, Republic of Korea;2. Department of Economics, Rutgers University, 75 Hamilton Street, New Brunswick, NJ 08901, USA |
| |
Abstract: | Despite the clear success of forecast combination in many economic environments, several important issues remain incompletely resolved. The issues relate to the selection of the set of forecasts to combine, and whether some form of additional regularization (e.g., shrinkage) is desirable. Against this background, and also considering the frequently-found good performance of simple-average combinations, we propose a LASSO-based procedure that sets some combining weights to zero and shrinks the survivors toward equality (“partially-egalitarian LASSO”). Ex post analysis reveals that the optimal solution has a very simple form: the vast majority of forecasters should be discarded, and the remainder should be averaged. We therefore propose and explore direct subset-averaging procedures that are motivated by the structure of partially-egalitarian LASSO and the lessons learned, which, unlike LASSO, do not require the choice of a tuning parameter. Intriguingly, in an application to the European Central Bank Survey of Professional Forecasters, our procedures outperform simple average and median forecasts; indeed, they perform approximately as well as the ex post best forecaster. |
| |
Keywords: | Forecast combination Forecast surveys Shrinkage Model selection LASSO Regularization |
本文献已被 ScienceDirect 等数据库收录! |
|