Mapping Global Sustainable Development Regimes Through Governance, Ecological Risk, and Explainable Machine Learning

There are great differences between the sustainable development success stories in various nations; however, such differences are not justifiable on the grounds of the ability of governance institutions or ecological dangers. The purpose of this research is to understand whether there are different regimes of the interplay of governance and ecological danger and whether such regimes are somehow related to the heterogeneity of sustainable development success. In order to reveal the regimes of the nations that are defined in terms of the following variables: Governance Score, Water Risk, Food Insecurity, Impact of Natural Events, and Demographic Pressure, the PCA‐GMM framework will be used based on the data from 164 countries for 2024–2025. The following three different regimes of the interaction between governance and ecological risk were revealed, distinguished by the presence of different combinations of the variables of governance capacity, ecological risk, and socioeconomic pressure. Importantly, the regimes are not linear in terms of being organized from high to medium to low, since there may be situations where two countries demonstrate the same level of sustainable development but at the same time have very different combinations of governance and ecological variables, implying the presence of several distinct ways to achieve sustainable development. The outcomes of machine learning and SHAP analyses provide further evidence of nonlinear dependencies between the variables used, as three variables—Food Insecurity, Governance Score, and Demographic Pressure—are especially important.