Machine Learning-Based Modeling of Economic Growth and Governance Quality: The MENA region

Abstract

This study investigates the predictive performance of various machine-learning models in forecasting economic growth across the MENA region. Four approaches were compared: Ordinary Least Squares (OLS), Random Forest (RF), Gradient Boosting Machine (GBM), and Support Vector Regression (SVR). The dataset was divided into training (70%) and testing (30%) subsets to assess the robustness and generalization capacity of the models. Model accuracy was evaluated using the Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE). The results indicate that the Random Forest model provides the highest predictive accuracy (MAPE = 0.0192), outperforming traditional econometric approaches. Variable importance analysis highlights that corruption, government effectiveness; political stability, regulatory quality, and rule of law significantly influence economic growth. These findings confirm the relevance of non-parametric methods in capturing complex and nonlinear relationships between governance indicators and economic performance. Moreover, the results emphasize the crucial role of institutional quality as a structural determinant of growth, consistent with institutional and endogenous growth theories. The study concludes that machine learning models, particularly ensemble methods, offer robust and complementary tools for economic forecasting and policy analysis in emerging economies.