An Algorithmic Crystal Ball: Forecasts-Based on Machine Learning

Abstract

Forecasting macroeconomic variables is key to developing a view on a country's economic outlook.Most traditional forecasting models rely on fitting data to a pre-specified relationship between inputand output variables, thereby assuming a specific functional and stochastic process underlying thatprocess. We pursue a new approach to forecasting by employing a number of machine learningalgorithms, a method that is data driven, and imposing limited restrictions on the nature of the truerelationship between input and output variables. We apply the Elastic Net, SuperLearner, andRecurring Neural Network algorithms on macro data of seven, broadly representative, advanced andemerging economies and find that these algorithms can outperform traditional statistical models,thereby offering a relevant addition to the field of economic forecasting.

Keywords: Information technology, Economic forecasting, Forecasting models, Machine learning, forecasts, forecast errors, Forecasting and Other Model Applications, Neural Networks and Related Topics

JEL Classification: C53, C45