When Darkness Pollutes: Grid Electricity Deficits, Dirty Energy Substitution, and Carbon Emissions in West Africa

Abstract

West Africa confronts one of the world's most acute electricity crises, yet its environmental consequences remain poorly understood. Where grid electricity is absent or unreliable, households and firms substitute towards private diesel generators and solid biomass fuels, both markedly dirtier than grid supply. This paper introduces and formally tests the dirty substitution hypothesis for 13 West African countries (2000-2022), using a constructed electricity deficit index (the gap between installed capacity and output per capita) to test whether larger deficits raise CO2 emissions through dirty substitution. We apply a three-layer empirical strategy: CS-ARDL and Augmented Mean Group (AMG) estimators for cross-sectional dependence and slope heterogeneity; method-of-moments quantile regression (MMQR) of Machado and Santos Silva to test whether the effect varies across the emissions distribution; and a machine learning ensemble (Random Forest, Gradient Boosting) validating variable importance out of sample. Results reveal a bounded rather than uniform relationship: the deficit index is significantly and negatively associated with emissions at the lower and median quantiles, consistent with severe deficits being met by rationed energy consumption, but this weakens and loses significance toward the upper quantiles, where income and generator capacity are highest. Granger causality tests indicate emissions lead the deficit index rather than the reverse, consistent with economic growth jointly driving both. Renewable energy share carries the expected negative sign throughout but never reaches conventional significance. These findings reframe West Africa's electricity access challenge as conditional rather than uniformly climate-relevant, with the emissions consequences of grid failure concentrated among the region's poorest-served economies, with implications for how ECOWAS energy transition targets and multilateral climate finance are sequenced.