Assessing Institutional Determinants of Money Laundering Risk: A Machine Learning Approach Based on Economic Freedom Indicators

Abstract

Money laundering remains one of the key challenges of the global financial system, and identifying its institutional determinants is a relevant task for both scholars and regulators. This paper examines the relationship between the components of the Economic Freedom of the World index (EFWI) and the Basel AML Index (BAMLI) using a cross-country sample of 156 countries. Unlike existing studies that focus on individual institutional factors or limited groups of countries, this work is the first to apply a comprehensive approach combining correlation analysis with four machine learning algorithms (Ordinary Least Squares regression (OLS), Ridge regression, Gradient Boosting, and Random Forest) with 10-fold cross-validation. The findings reveal a statistically significant negative association between the level of economic freedom and money laundering risk. The “Legal System and Property Rights” sub-index is the dominant predictor: its correlation with BAMLI is r = −0.780 (p < 0.001), and its contribution to the Random Forest model’s predictive power is 67.5%. The EFWI components collectively explain approximately 52% of the variation in BAMLI (R² ≈ 0.53). The consistency of results across linear and non-linear models indicates the stability and predominantly linear nature of the detected association. The findings underscore the pivotal role of legal institutional quality in preventing financial crime and may serve as a basis for developing differentiated AML strategies that account for the institutional context of individual countries.