Sustainability

Climate risk has become an emerging source of financial instability as global climate governance tightens and the low-carbon transition accelerates. Using a balanced panel of 33 Chinese A-share listed commercial banks from 2014 to 2023, this study examines the association between a composite climate-risk indicator and the risk-weighted assets (RWA) ratio. The comprehensive climate risk index (CCRI) combines branch-weighted physical risk exposure with the length-adjusted intensity of transition risk information in banks’ annual reports. The results show that a higher CCRI is associated with a higher RWA ratio, indicating higher regulatory risk density in banks’ asset portfolios. The finding is robust to alternative dependent variables, the exclusion of key policy years, alternative CCRI weighting schemes, and a lagged CCRI specification. Further analyses show that the loan-to-asset ratio (LTA) is positively associated with the RWA ratio, while liquidity weakens the positive CCRI–RWA association and heterogeneity emerges across coastal exposure, provision coverage, and long-term physical climate risk exposure. The findings inform bank asset allocation, liquidity management, and climate-related prudential supervision.

Digital technology investment expands manufacturers’ technical resources, but its sustainability implications depend on whether those resources become embedded in operating routines. This conversion remains underexamined. Drawing on socio-technical systems theory and the natural-resource-based view (NRBV), this study examines automated quality control (AQC) as an operational pathway linking digital technology investment breadth (DTIB) to lower energy-cost intensity, with managerial human capital—represented empirically by top-manager educational attainment—as a proposed first-stage boundary condition. The empirical analysis uses 3534 manufacturing establishments from nine World Bank Firm-level Adoption of Technology (FAT) survey contexts and estimates sampling-weighted models with design-based inference and stratified bootstrap tests. DTIB is positively associated with AQC adoption, and AQC adoption is associated with lower energy-cost intensity. The pooled DTIB association with lower energy-cost intensity is positive in the primary weighted model but is sensitive to survey weighting, the exclusion of India, and cross-country heterogeneity, indicating that the association is context dependent. The interaction with top-manager educational attainment is positive in the primary linear probability model, but its strength varies across nonlinear and extended-control specifications. The overall statistical indirect association is positive. The between-group difference in conditional indirect associations is positive in the primary bootstrap analysis but is sensitive to the confidence-interval method. These cross-sectional findings emphasize the operational conversion of digital resources while treating the proposed managerial boundary condition as suggestive rather than conclusive.

This study investigates a three-dimensional electrooxidation (3D-EO) process utilizing steel slag as a particle electrode for the treatment of mature landfill leachate. First, having been used as a particle electrode the steel slag were characterized by X-ray diffraction (XRD), X-ray fluorescence (XRF), Brunauer–Emmett–Teller (BET), energy-dispersive X-ray spectroscopy (EDS), and Fourier transform infrared spectroscopy (FTIR) analyses. Then, the performance of the conventional two-dimensional electrooxidation (2D-EO) process was compared with that of the 3D-EO process incorporating the particle electrode. While the chemical oxygen demand (COD) removal efficiency was 36.7% in the 2D-EO process, it was 50.4% in the 3D-EO process. Five different anode materials were tested in each process, where Ti/IrO₂ was determined as the optimum anode for both. The process operating parameters were modeled utilizing machine learning (ML) algorithms. Among the evaluated models, the XGBoost algorithm demonstrated the highest predictive accuracy, yielding high R² values coupled with low mean absolute error (MAE) and root mean square error (RMSE) values. The optimal operating parameters were identified as follows: initial pH = 5, particle electrode dosage = 2 g/L, applied current = 2 A, and reaction time = 120 min. The removal efficiencies for COD, UV₂₅₄, and total organic carbon (TOC) at the optimal conditions reached 83.0%, 85.3%, and 51.8%, respectively, requiring a specific energy consumption of 84.4 kWh/kg COD. The 3D-EO process reduced the inert COD fraction from 80.0% to 65.5%, improving both the biochemical oxygen demand/chemical oxygen demand (BOD₅/COD) ratio from 0.1 to 0.4 and the soluble COD fraction from 83.1% to 95.2%. Phytotoxicity assessments indicated that the treated effluent required a 75% dilution to reach a safe, non-toxic threshold (GI > 70%). Overall, the 3D-EO process emerges as a promising and highly effective technology for mature landfill leachate treatment, successfully complemented by the Extreme Gradient Boosting (XGBoost) algorithm.