.png){kind=icon}
.png){kind=icon}
Return
Predicting corporate environmental violations in China: Evidence from machine learning based on the Fraud Triangle Theory
.png){kind=icon}
PRE
AI
B
M
Y
Z
DOI：10.1016/j.jclepro.2026.149151.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Existing studies on corporate environmental violations mainly examine determinants from separate perspectives, lacking an integrated framework to explain their complex behavioral drivers. To address this limitation, this study integrates the Fraud Triangle Theory (FTT) with interpretable machine learning to develop a theory-driven framework for predicting and explaining corporate environmental violations using Chinese A-share listed companies. The results show that LightGBM model achieves the best predictive performance, and SHAP analysis reveals nonlinear effects, suggesting that environmental violations are driven by complex behavioral processes rather than simple linear relationships. Among the three FTT dimensions, pressure exhibits the greatest importance than opportunity and rationalization, indicating that external institutional and economic pressures dominate environmental violation risks. Further analysis reveals that the FTT-based risk structure is context-dependent. By integrating behavioral theory with interpretable machine learning, this study provides a theory-driven and interpretable framework for understanding the complex drivers of environmental violations, offering new insights into their underlying behavioral mechanisms.
Keywords:
Environmental violation
Machine learning
Fraud Triangle Theory
Journal
IF:
10
Papers:
4.6W
Citations:
36.8W
Organization
Cited Papers