Educational Losses and Damages Attributable to Anthropogenic Climate Change: End-to-End Attribution Evidence from South Africa

Climate change threatens education through various channels. These include extreme temperature effects on brain performance, extreme temperature and precipitation effects on agriculture affecting incomes, which in turn affects education funding and food security.  Unlike other sectors where climate impacts are visible, educational losses remain poorly quantified and seldom attributed to anthropogenic climate change. This study provides the first and robust end-to-end attribution analysis of climate change impacts on educational outcomes in a developing country context, quantifying educational losses attributable to human-induced climate change and associated economic losses and damages in South Africa. We employ an integrated three-step methodological framework that integrates econometric modeling, climate emulation, and impact attribution science. First, we estimate causal climate-education relationships using panel fixed-effects regression with rich National Senior Certificate (NSC) examination data (2008-2023) merged with the Learner Unit Record Information and Tracking System (LURITS). We have acquired these datasets through collaborations with the Department of Education. Unlike previous studies using aggregated data, LURITS tracks individual learners from grade 9 to 12 across four complete cohorts (2017-2023), enabling precise allocation of cumulative climate exposure across all schools attended. This eliminates measurement errors from assumptions about school mobility. We specify temperature and precipitation as daily bins capturing extreme exposure within 15km buffers around each school, controlling for student-teacher ratios, school characteristics, and fixed effects. We will perform heterogeneity analysis across school quintiles and urban-rural locations. We will also explore whether agricultural channels amplify impacts through food security and income disruptions affecting school attendance and cognitive performance.  Second, we generate counterfactual climate scenarios using IIASA's Rapid Impact Model Emulator (RIME). RIME interpolates global warming levels to produce grid-cell-level climate impact drivers under factual (with anthropogenic forcing) and counterfactual (without anthropogenic forcing) scenarios, requiring less computational power while maintaining methodological rigour for attribution analysis. This enables robust comparison of educational outcomes under observed versus counterfactual climate conditions. Third, we apply estimated coefficients from the impact model to both factual and counterfactual distributions of climate variables. The difference in predicted exam performance and dropout rates provides estimates of educational losses attributable to anthropogenic climate change. We extend attribution to specific historical emitter groups using RIME-X transformations following recent methodological advances in pollutant-source attribution. Economic valuation converts standard deviation losses into years of schooling lost and lifetime wage impacts using established education-earnings literature. Such papers include the World Bank paper that provided conversion estimates: one standard deviation loss in test score is equivalent to 5.75 years of schooling lost (Evans and Yung, 2019). This research will produce policy-relevant evidence for loss and damage discourse, adaptation prioritization, and climate justice frameworks.  Our findings will inform efficient allocation of climate finance, provide evidence for climate-related litigation, and highlight intergenerational consequences of disrupted human capital formation in climate-vulnerable populations. This paper exemplifies interdisciplinary integration of econometrics and climate science to quantify anthropogenic contributions to socioeconomic losses, advancing both attribution methodologies and empirical evidence for global climate justice discourse.