Demand and supply-side determinants of electric power consumption and representative roadmaps to 100% renewable systems

Abstract

Abstract The need to transform electric power systems to a cleaner electricity mix to help save the climate is undisputable but existing transition pathways have become unreliable due to inadequate representation of both demand and supply-side determinants of electric power supply, as well as failure to inculcate volatility in electric power systems in the core transition modelling. This study inculcates demand and supply-side determinants to investigate the long- and short-run causal relationships between the electric power system, macroeconomic performance, demography, environmental quality, and capital formation, for Sweden. We also propose and implement econometric and two-stage attention-based machine learning models for volatility-consistent electric power forecasting. Using annual data spanning 1990-2018, the results suggest a long-run relationship exists between electricity generation and the independent variables. Empirical results for the volatility-consistent attention-based machine learning model predict that Sweden’s electric power demand in 2050 could reach ∼112TWh if conservation practices are implemented, and ∼146TWh if otherwise. If conservation practices are implemented, evidence from selected volatility-consistent simulations show that Sweden can provide 100% of all her electricity demand from cleaner sources by 2030. The findings depict that Sweden must implement stringent and radical policies to achieve its mid-century green electricity targets.