Machine learning-based analysis of economic efficiency disparities and transition drivers between high- and low-carbon industries in China

基于机器学习的中国高碳产业与低碳产业经济效率差异及转型驱动因素分析

阅读:3

Abstract

In the context of global climate change, understanding economic efficiency disparities between high-carbon and low-carbon industries is crucial for advancing low-carbon transitions and improving carbon governance. This study examines heterogeneity in corporate carbon emission management and economic performance across Chinese industries and identifies key drivers of firms' transformation capacity. Using a panel dataset of 633 listed enterprises from eight industries in China over 2010-2021, we classify firms into high- and low-carbon groups based on their emissions profiles and benchmark four machine-learning models-Random Forest, XGBoost, LightGBM, and Decision Tree-to capture nonlinear relationships and evaluate the relative importance of environmental and financial indicators. Random Forest delivers the best performance, achieving a classification accuracy of 95.7% (rounded) and strong discriminatory ability (AUC = 0.989). Feature-importance results consistently show that carbon emissions are the most influential variable, followed by total liabilities and total assets, while profitability-related indicators (e.g., operating revenue and gross profit margin) also contribute to distinguishing firms' carbon profiles and performance differences. Overall, high-carbon enterprises appear to face greater transition barriers due to higher abatement cost exposure and tighter balance-sheet constraints, whereas low-carbon firms may be better positioned to benefit from policy incentives and market opportunities. These findings highlight the pivotal role of financial health in enabling low-carbon transformation and underscore the need for differentiated policy design. Policy implications include targeted transition finance and more flexible allowance allocation mechanisms for high-carbon enterprises, alongside continued incentives for technological innovation and market expansion in low-carbon sectors. JEL CLASSIFICATION: Q56; G30; C55; Q43; L60.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。