Dynamic analysis of China's load capacity factor: The roles of energy heterogeneity and economic globalization through ARDL and machine learning methods

This study investigates the dynamic impacts of energy heterogeneity, economic globalization, and gross domestic product on China's Load Capacity Factor (LCF) within the context of its dual carbon goals. Using annual data from 1993 to 2021, the study integrates econometric and machine learning approaches to examine dynamic relationships and forecast future energy trends. The Autoregressive Distributed Lag (ARDL) model is used for empirical analysis, while the Gated Recurrent Unit (GRU) model supports forecasting. The robustness of the ARDL results is verified through cointegration tests, stability diagnostics, and Fully Modified Least Squares (FMOLS) estimation. The long-run results indicate that fossil fuel consumption has the strongest negative impact on LCF, with a coefficient of −1.110, while renewable energy consumption also negatively affects LCF with a coefficient of −0.137. In contrast, nuclear energy consumption, economic globalization, and gross domestic product contribute positively, with coefficients of +0.077, +0.564, and +0.322, respectively. In the short run, nuclear energy consumption (+0.116), renewable energy consumption (−0.275), and gross domestic product (+0.524) exhibit significant effects, whereas fossil fuel consumption and economic globalization primarily influence long-term dynamics. GRU forecasts project a substantial transformation of China's energy structure by 2030: the fossil fuel consumption share is expected to fall to 54.95 %, while nuclear and renewable energy shares are projected to rise to 7.96 % and 37.09 %, respectively. LCF is projected to stabilize overall, with a potential turning point in 2027, followed by a gradual upward trend. These findings underscore the imperative of transitioning toward a lower-carbon energy system to enhance LCF and meet sustainability goals.