A dual-method approach integrating dynamic QCA and LightGBM-SHAP algorithms to uncover the configuration paths and key drivers of water resource green efficiency in China

• Published in: Ecological indicators Vol. 175; p. 113540
• Main Authors: He, Naiming, Ding, Rijia
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.06.2025; Elsevier
• Subjects: China; Configuration paths; Dynamic qualitative comparative analysis (QCA); environmental law; issues and policy; LightGBM; quantitative analysis; SHAP; Super-EBM model; sustainable development; sustainable technology; Technology-Organization-Environment (TOE) framework; water management; Water resource green efficiency; water shortages; China; SHAP; Technology-Organization-Environment (TOE) framework; Dynamic qualitative comparative analysis (QCA); Configuration paths; LightGBM; Water resource green efficiency; Super-EBM model
• ISSN: 1470-160X; 1872-7034
• DOI: 10.1016/j.ecolind.2025.113540

[Display omitted] •Super-EBM-GML model measures WRGE across time and space under high-quality development in China.•TOE framework identifies key technological, organizational, and environmental drivers of WRGE.•Dynamic QCA reveals three WRGE models with four configurations, reflecting temporal and regional differences.•To integrate qualitative and quantitative analysis, LightGBM-SHAP quantifies variable impacts on WRGE.•Findings offer new perspectives and differentiated pathways for regions worldwide facing water scarcity challenges. Water resource utilization is crucial for sustainable development, and enhancing water resource green efficiency (WRGE) is essential for addressing water scarcity. This study presents three key innovations: (1) It applies the super-efficiency epsilon-based measurement and global Malmquist–Luenberger index (Super-EBM-GML) model from the perspective of high-quality economic development to analyze the spatiotemporal characteristics of WRGE across 30 Chinese provinces from 2014 to 2022. (2) It uses the Technology-Organization-Environment (TOE) framework and dynamic qualitative comparative analysis (QCA) model to identify the key drivers of WRGE and regional variations. (3) It integrates machine learning (light gradient-boosting machine with Shapley additive explanations; LightGBM-SHAP) with QCA to quantify the impact of variables, combining qualitative and quantitative analysis. Key findings include the following: (1) WRGE showed an upward trend, with higher efficiency in the eastern and economically developed regions. Growth in the GML index was mainly driven by green technological progress (GTC). (2) Although no single necessary condition was found to drive WRGE, three model types and four configuration paths were identified: technology–environment-driven, environment-driven, and organization–environment-driven. (3) The most influential factors were digital economy development, followed by industrial structure rationalization and environmental regulation. This study provides key policy recommendations, including the promotion of green technology, the strengthening of regulations, the enhancement of policy resilience, the implementation of region-specific strategies, and the integration of the digital economy with water resource management, thus offering valuable insights for regions facing water scarcity.