Abstract
Urgent climate action, especially carbon emissions reduction, is required to achieve sustainable goals. Therefore, understanding the drivers of and predicting CO2 emissions is a compelling matter. We present two global modeling frameworks—a multivariate regression and a Random Forest Regressor (RFR)—to hindcast (until 2021) and forecast (up to 2035) CO2 emissions across 117 countries as driven by 12 socioeconomic indicators regarding carbon emissions, economic well-being, green and complexity economics, energy use and consumption. Our results identify key driving features to explain emissions pathways, where beyond-GDP indicators rooted in the Economic Complexity field emerge. Considering current countries’ development status, divergent emission dynamics appear. According to the RFR, a −6.2% reduction is predicted for developed economies by 2035 and a +19% increase for developing ones (referring to 2020), thus stressing the need to promote green growth and sustainable development in low-capacity contexts.
Costantini, Lorenzo