The possibility of machine learning algorithms to explain long-run economic growth
Abstract: The empirical investigation of economic growth has been one of the most researched topics in economics. Most recently, machine learning algorithms that can handle nonlinearities, discontinuities and other issues inherent with traditional linear approaches have been proposed to be able to more accurately describe the empirical determinants of economic growth. We investigate the determinants of economic growth using three state-of-the-art machine learning algorithms (Support Vector Regression Machines, Gaussian Process Boosting and Long Short-Term Memory Neural Networks) together with a more conventional regression algorithm (Generalized Additive Model), on 28 European countries between 1950-2019 and 56 explanatory variables. The machine learning algorithms diverge in their prediction patterns and estimated variable importance, and do not provide a compelling improvement over traditional methods.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)