Climate-Related Renewable Energy Sources and Carbon Emissions : A Machine Learning-Based Investigation of Electricity Production in France

As a path forward to combat global climate change, the development of renewable energy share of electricity production remains the cornerstone of CO2 emissions reduction in the electric power sector. Considering the dependence of the availability and sporadicity of climate-related renewable energy (CRE) sources (wind, solar photovoltaics, and small-scale run-of-river hydroelectricity) on climate factors, the relationship between CO2 emissions and electricity production from these sources merits careful analysis. By means of a cutting-edge machine learning-based modeling technique (Extreme Gradient Boosting), this study characterizes the predictive impact of CRE electricity production on CO2 emissions in the French electric power sector over the 2013-2021 period. The results demonstrate that run-of-river hydroelectricity was the most important feature among the three CRE sources for predicting emissions, followed by wind energy. Empirical findings based on a counterfactual analysis also reveal that an increase in the share of energy from CRE sources would be associated with a statistically significant decrease in predicted emissions over the study period. Identification of the optimal mix of CRE sources for minimizing predicted emissions under four counterfactual scenarios of increased CRE production reaffirms the greater relative share of run-of-river hydroelectricity and wind energy within the mix. The findings of this research have two major implications for renewable energy development and management in France. First, they provide fresh quantifiable evidence on the conceptual premise that replacing carbon-intensive energy sources with renewable ones reduces CO2 emissions from electricity generation. Second, they cast a new light on the relative importance of each CRE source with regard to emissions reduction in the electricity sector