Exploring the Twin Deficits Hypothesis Through Machine Learning: A New Approach to Economic Forecasting

This study investigates the Twin Deficits Hypothesis using machine learning, analyzing data from over 50 countries from 1990 to 2020 with models like Gradient Boosting Machines (GBM), Neural Networks, and Random Forest. It demonstrates that machine learning surpasses traditional models in forecasting the twin deficits, highlighting the role of government spending, national income, and factors like technological innovation and political stability. The findings suggest machine learning's significant potential in economic analysis and policy guidance, pointing to future research directions including real-time data integration and the study of new economic trends.