Does One Size Fit All? A Machine Learning-Based Analysis of the Heterogeneous Nature of Economic Freedom

Economic freedom is widely discussed as an institutional condition associated with national prosperity in contemporary economic discourse. However, traditional studies in the literature generally treat this concept as a homogeneous structure, largely ignoring institutional heterogeneity across countries. This article presents an integrated analytical framework that combines clustering, machine learning, and explainable artificial intelligence (SHAP) techniques to analyze the multidimensional nature of economic freedom. In this study, data from 174 countries covering the period 2017–2025 were analyzed. Then, countries were organized into five empirically derived trajectory-based clusters according to their dynamic development patterns using the k-means algorithm. According to the Extra Trees algorithm, which showed the highest performance in the comparison, and the SHAP analysis findings, financial freedom, investment freedom, and property rights most strongly discriminate among the economic freedom profiles. The results go beyond the traditional ‘free’ and ‘non-free’ dichotomy by revealing a trajectory-based empirical classification that captures structural differences in level, direction, and volatility. Furthermore, the multi-criteria decision-making (MCDM) methodologies and scenario stages are interpreted as diagnostic exercises that show how rankings respond to alternative weighting priorities rather than as estimates of the causal effects of reforms. The empirical results provide policymakers with a strategic diagnostic guide identifying relative institutional vulnerabilities that align with their countries’ existing capacities.