Sustainable Decarbonization Under Renewable Energy
Penetration: A Hybrid Fixed Effects and Machine Learning
Framework for Multi-Country Panel Evidence
Citra Dewi1,∗
1Universitas Lampung, Indonesia
Email: citra.dewi@eng.unila.ac.id
Abstract
Realizing the carbon reduction capabilities of deploying renewable energy. is core to the constructive
plan of effective climate policy in heterogenous national. contexts. Even though there is an
accumulating corpus of panel econometric and machine learning. literature dealing with this relationship,
methodological inconsistencies and limited geographic scope leave important empirical
questions unanswered. This paper put forward a mixed analytical model combining a within-group
Fixed Effects. country-clustered standard errors estimator and a Random Forest ensemble. model to
measure the combined effect of renewable energy penetration, economic growth, energy consumption
and reliance on fossil fuels per capita carbon. emissions. Findings affirm that the growth of
renewable energy has a statistically significant impact. strong and economically significant negative
impact on carbon intensity, which remains. following the elimination of country-specific unobserved
heterogeneity. Economic structure and energy efficiency are shown to be co-dominant determinants,
highlighting. that the energy transition is not decoupled of larger structural. transformation. Articulated
income-group and regional heterogeneity issues. single-coefficient policy prescriptions, which
propose decarbonization. plans have to be aligned to the national development levels. The machine
learning complement validates econometric variable rankings and proves. good cross-country generalisability
with country-stratified. cross-validation.
Keywords: Renewable energy; Sustainable development; Carbon emissions; Fixed effects; Random
forest; Panel data; Green technology; Energy transition