Corporate environmental performance prediction in China: An empirical study of energy service companies

JOURNAL OF CLEANER PRODUCTION

卷:266

Article

Zheng, Saina¹  He, Chenhang²  Hsu, Shu-Chien¹  Sarkis, Joseph³  Chen, Jieh-Haur⁴ 

[1] Hong Kong Polytech Univ, Dept Civil & Environm Engn, Hong Kong, Peoples R China

[2] Hong Kong Polytech Univ, Dept Comp, Hong Kong, Peoples R China

[3] Worcester Polytech Inst, Foisie Business Sch, Worcester, MA 01609 USA

[4] Natl Cent Univ, Dept Civil Engn, Taoyuan, Taiwan

关键词: Corporate environmental performance;    Extreme gradient boosting (XGBoost);    Energy service company (ESCO);   

DOI  :  10.1016/j.jclepro.2020.121395

来源: Elsevier

【 摘 要 】

Businesses are constrained by and dependent upon nature and institutional context. The global climate crisis has put pressure on and increased firm sensitivity to environmental issues. Predicting corporate environmental performance can help plan for environmental impact mitigation by adjusting organizational practices. Lack of environment-related information makes it difficult to make such predictions. This paper tends to provide useful insights into corporate environmental performance among firms to facilitate better environmental management for both government and firms. A theoretical framework informed by the natural-resource-based view (NRBV) of the firm and institutional theory is used to identify variables for predicting corporate environmental performance. Five dimensions including institutional context, governance capability, information management capability, system capability, and technology-related capability, populated with 14 variables are used to empirically investigate the relationship of these variables with corporate environmental performance. Using 1100 data points on energy service companies (ESCOs) from 2011 to 2015 in mainland China, the Extreme Gradient Boosting (XGBoost) algorithm, a statistical nonlinear machine learning approach, is utilized to predict corporate environmental performance. The results demonstrate that the XGBoost model can be effective for ESCO environmental performance prediction, with satisfactory prediction accuracy. This study also adopted the SHapley Additive exPlanations (SHAP) values for model interpretation, indicating that total assets, amount of proactive environmental costs, proportion of technicians and number of patents contribute most to corporate environmental performance. Several policies and environmental strategies for improving corporate environmental performance in the ESCO industry are derived from this analysis. (C) 2020 Elsevier Ltd. All rights reserved.

【 授权许可】

Free