Energy-System performance under external dependence: Composite measurement, channel heterogeneity, and dynamic panel evidence

Education and R&D are frequently treated as broad capacity inputs to energy transition, but their empirical role is difficult to interpret when composite outcomes overlap with the regressors or when persistence and common shocks are ignored. Using World Bank Environment, Social and Governance (ESG) Data, this study constructs an outcome-only Energy-System Performance Index (ESPI) from electricity access, import security, renewable-electricity output, renewable-energy consumption, and resource protection for 46 World Bank economy aggregates over 1999–2020. Education and R&D expenditure are excluded from the index and modeled separately. The five outcomes are jointly observed for 942 of 1012 potential aggregate-years; incomplete predictor reporting leaves 455 observations across 31 aggregates in the baseline fixed-effects model. Baseline estimates associate education with lower ESPI (−0.1041, p = 0.0272) and R&D with higher ESPI (0.3234, p = 0.0219), while R&D is also associated with lower net energy-import dependence (−0.1478, p = 0.0172). The static ESPI coefficients are specification-sensitive, but all four reported dynamic estimators imply positive ESPI persistence (0.4640–0.9110); the split-panel-jackknife interval includes one, so stationarity is not uniform. Component-equality tests reject common education and R&D slopes (both p < 0.001), showing that the predictor associations are heterogeneous across energy outcomes. Persistence-adjusted confidence intervals include zero, and alternative weighting and leave-one-component-out checks show that composite coefficients depend on index construction. The contribution is therefore a measurement- and identification-disciplined framework for integrated energy-system performance, with the clearest recurring channel linking R&D to lower external energy dependence.