Key points are not available for this paper at this time.
Promoting green innovation in energy-intensive industries is of strategic significance for achieving sustainable economic and social development. However, the role of government subsidies in improving green innovation efficiency (GIE) remains unclear. This study employs a combined approach—network DEA, regression analysis, Bootstrap method, and qualitative comparative analysis (QCA)—to examine the complex effects of government subsidies on GIE in energy-intensive sectors. Results show that, from a net effect perspective, high levels of government subsidies have a significant negative impact on GIE. This adverse effect is mitigated by market competition. In contrast, configurational analysis reveals that high government subsidies are not a necessary condition for achieving high GIE. One configuration leads to high efficiency under high-subsidy conditions, while two distinct pathways emerge under non-high subsidy regimes, indicating equifinality. These findings suggest that the relationship between government subsidies and GIE is not deterministic but depends on the interaction of multiple conditions—including market competition, and human resource. Rather than adopting uniform subsidy policies, policymakers should implement differentiated and context-sensitive strategies, strengthen market mechanisms, and optimize policy bundling. By combining net effect analysis and configurational effect analysis, this study provides a more comprehensive understanding of how government subsidies influence GIE. It advances analytical methods in green innovation research and offers empirical support for refining subsidy policies in energy-intensive sectors. • Regression analysis and QCA are utilized to explore the role of government subsidies. • Market competition mitigates the negative impact of government subsidies on GIE. • Government subsidies are not a necessary condition for achieving high GIE. • Achieving high GIE depends on the configuration of multiple influencing factors.
Peng et al. (Sat,) studied this question.