Comparative analysis of generalized additive model and quantile regression model in assessing renewable energy's role in emissions mitigation in West Africa

Abstract

West Africa experiences rising carbon emissions driven by rapid urbanization, population growth, and continued dependence on fossil fuels. Despite increasing interest in renewable energy, regional analyses often fail to capture the nonlinear and distributional effects of different energy sources on emissions. Addressing this gap, we investigates how six renewable energy sources influence emissions across varying pollution levels using long-term data (1990–2100) and advanced modeling techniques. Generalized Additive Models (GAM) and Quantile Regression (QR) were employed to assess both average and distribution-specific impacts. The GAM model demonstrated strong predictive performance (R² = 0.996), with deviance explained and error metrics supporting its reliability, indicating that solar and wind energy reduces emissions significantly, particularly in high-pollution scenarios. QR findings showed biomass becomes more effective at higher emission quantiles, while hydropower and nuclear consistently reduce emissions across all quantiles. Additionally, evidence of GDP-emission decoupling at higher quantiles suggests that economic growth can align with emissions reductions. The findings reveal asymmetric impacts across the emissions spectrum, offering a theoretical extension to energy transition models through a dual heterogeneity framework. These insights highlight the importance of context-specific strategies like targeting wind and solar in high-emission areas, promoting biomass in rural zones, and expanding hydropower and nuclear where feasible, informing a more nuanced and effective decarbonization pathways in West Africa.