An analysis of carbon emissions across eight East African countries over nearly three decades finds conditional evidence that emissions may begin to fall with rising incomes at a far lower development threshold than the classic environmental Kuznets Curve literature typically reports. The study, published in Discover Energy, examined a balanced panel of Burundi, Djibouti, Eritrea, Ethiopia, Kenya, Rwanda, Tanzania and Uganda from 1994 to 2022, using 232 country-year observations drawn from World Bank and IMF databases.
The research tested the Environmental Kuznets Curve hypothesis, which posits that environmental degradation first rises with economic development and then declines beyond a threshold income level. Using fixed-effects, system GMM and decoupling elasticity estimations, the author found that the quadratic relationship between GDP per capita and CO2 emissions per capita was negative and statistically significant in the fixed-effects specifications, with an estimated turning point of approximately 2,846 constant 2015 US dollars. The 95 percent confidence interval, calculated with the Fieller method, ranged from 2,100 to 4,200 dollars, a wide band that the study itself flags as indicating substantial uncertainty.
Formal testing strengthened the case only partially. The Lind–Mehlum test for an inverted U-shape rejected the null hypothesis of a monotonic relationship at the 5 percent level, with a test statistic of 1.95 and a p-value of 0.029. Yet the evidence was not robust across every specification: in the pooled ordinary least squares specification shown in the paper’s scatter plot, the quadratic term was significant, but the author repeatedly cautions that the turning point estimate is imprecise, particularly given the limited number of observations above the threshold. The estimated turning point is substantially lower than typical EKC estimates for CO2 in developed economies, which the paper places at 15,000 to 25,000 dollars, though it aligns with recent evidence from lower-middle-income countries.
Country-level results revealed that the inverted U-shape is far from universal. Only five of the eight countries exhibited the theoretically predicted sign pattern of a positive linear and negative quadratic coefficient. Kenya, Tanzania and Uganda showed significant quadratic relationships with plausible turning points between 3,200 and 4,100 dollars, while Burundi and Eritrea displayed monotonic negative relationships that the author attributes more plausibly to deindustrialization and conflict-related economic contraction than to environmental policy success. Ethiopia presented a turning point near 3,069 dollars. R-squared values across the country models ranged from 0.28 for Djibouti to 0.84 for Kenya.
Renewable energy consumption emerged as negatively associated with emissions, with a fixed-effects coefficient of −0.018, meaning each percentage point increase in the renewable share was associated with a reduction of roughly 0.018 metric tons of CO2 per capita, or about 7.2 percent at the regional mean of 0.25 tons per capita. But the study devotes considerable attention to a measurement problem: the World Development Indicators renewable energy variable includes traditional biomass such as fuelwood and charcoal, which constitutes 70 to 90 percent of household energy in most East African countries and carries high black carbon emissions. The author argues this likely attenuates the estimated coefficient for genuinely clean energy sources.
A sensitivity analysis excluding the biomass-dominated countries of Burundi, Rwanda and Uganda produced a larger coefficient of −0.026, still significant at the 1 percent level, suggesting the aggregate figure understates the association with modern renewables such as hydro, geothermal, solar and wind. The paper explicitly warns against policy confusion, noting that recommendations based on the aggregated coefficient could incorrectly imply that expanding traditional biomass is climate-neutral. It recommends that all policy language specify modern renewables rather than renewables broadly.
The dynamic analysis pointed to strong emissions persistence. System GMM estimation, using collapsed instruments and two-step robust standard errors, yielded a persistence coefficient of 0.58, significant at the 1 percent level, lower than the 0.68 obtained from the fixed-effects ADL specification, confirming that short-panel Nickell bias had inflated the earlier estimate. The author interprets this as consistent with, but not proof of, carbon lock-in, noting that lock-in is a specific mechanism involving sunk capital, infrastructure and institutional inertia that cannot be directly tested with the available data. Hansen J and AR(2) diagnostics were reported as satisfactory, but the study stresses that with only eight countries, a favorable Hansen test does not validate the instruments, and the 3.5-to-1 instrument-to-country ratio is a limitation rather than a justification.
Decoupling analysis revealed striking heterogeneity in country-specific CO2–GDP elasticities, ranging from −0.42 for Ethiopia to 0.79 for Kenya, with Tanzania at 0.68. Ethiopia’s negative elasticity, comparable to figures observed in Sweden, Germany and the United Kingdom during the 2010s but at a per capita income of roughly 883 dollars rather than 35,000 to 50,000 dollars, was described as remarkable but not unprecedented. The author cautions that the negative elasticity may reflect structural change, economic disruption or measurement issues, and that a decomposition analysis would be required to attribute the observed decoupling to renewable energy investment. Kenya and Tanzania’s positive elasticities indicate relative decoupling, where emissions grow more slowly than GDP, but not the absolute decoupling in which emissions decline during growth. The study also found that tree-cover loss from Global Forest Watch data was positively and significantly associated with emissions, while forest area, a poor proxy for carbon stocks, was excluded from the main specifications.
Several robustness checks supported the core findings. Alternative emissions metrics, outlier exclusion, cubic specifications, Driscoll-Kraay standard errors addressing significant cross-sectional dependence, a common correlated effects estimator, CS-ARDL estimation with a turning point of 2,912 dollars, a dynamic panel threshold model with a threshold of 2,712 dollars, and results using observed data only without interpolation all yielded qualitatively similar results. Bootstrap confidence intervals of 1,000 replications placed the turning point between 2,200 and 4,000 dollars, confirming that the wide interval reflects genuine imprecision rather than an artifact of asymptotic assumptions.
The author frames the overall contribution carefully: the study does not resolve the EKC debate for East Africa but demonstrates that the evidence is conditional, model-dependent and limited by data quality. All results are described as statistical associations rather than causal effects, with the instrumental-variable strategy presented as exploratory only because its exclusion restriction is not credible. Policy implications include targeted investment in modern renewable energy, forest carbon governance that targets carbon stocks and deforestation rates rather than forest area, and country-specific decoupling strategies. The persistence findings imply that each year’s delay in low-carbon infrastructure decisions carries long-lived consequences, since current policy choices will shape emissions trajectories for decades even if incomes eventually cross the estimated threshold.
Subject of Research: Environmental Kuznets Curve analysis of carbon emissions, renewable energy and decoupling in East Africa from 1994 to 2022
Article Title: The environmental Kuznets Curve and renewable energy in East Africa 1994 to 2022 a panel analysis
Article References: Dumo, W. M. (2026). The environmental Kuznets Curve and renewable energy in East Africa 1994 to 2022 a panel analysis. Discover Energy, 6(1), Article 92. https://doi.org/10.1007/s43937-026-00211-x
Image Credits: AI Generated
DOI: 10.1007/s43937-026-00211-x
Keywords: environmental Kuznets Curve, East Africa, carbon emissions, renewable energy, decoupling, system GMM, CO2 emissions, economic growth, forest carbon, panel data, emissions persistence, Ethiopia
News Source: Sloane Callahan. (October 28, 2026). East Africa’s emissions may bend at lower incomes than expected, study finds. Scienmag.
