| 摘要: |
| 科学有效地评价企业的碳信息披露质量并探寻影响碳信息披露的主要因素，有助于进一步提升企业的碳信息披露质量水平、发挥企业在我国如期实现碳达峰及碳中和进程中的积极作用。基于碳信息披露的四个维度构建制造业企业碳信息披露质量的评价体系，综合主成分分析法和熵权法测度我国制造业企业2018—2020年真实的碳信息披露质量水平及变化趋势，有效降低评价指标的维数，避免单一熵权法权重失真情况，并通过随机森林回归识别各个指标对企业碳信息披露质量的重要性。研究发现：（1）主成分和熵权法组合可有效降低信息量少的“干扰”维度对企业碳信息披露的影响，能客观真实地评价我国制造业企业的碳信息披露质量，从而获得比传统单独采用单一方法更有效的评价结果。（2）2018—2020年，我国制造业企业的碳信息披露质量整体呈现上升趋势且2020年的碳信息披露质量水平提升最大。（3）我国制造业企业的碳信息披露质量整体偏低，不同企业间的碳信息披露质量存在较大差异。（4）企业的研发投入、总资产、总负债等因素对企业碳信息披露质量有较大影响。 |
| 关键词: 制造业 碳信息披露 主成分分析 随机森林 |
| DOI：10.7515/JEE222074 |
| CSTR： |
| 分类号: |
| 基金项目:国家社会科学基金项目（20XJL013） |
|
| Research on quality evaluation and influencing factors of corporate carbon information disclosure |
|
GUO Sidai, YUAN Zihan, LEI Gaowen
|
|
School of Economics and Management, Southwest University of Science and Technology, Mianyang 621010, China
|
| Abstract: |
| Background, aim, and scope Scientific evaluation of enterprise carbon information disclosure quality (ECIDQ) and explore its influencing factors are crucial to achieving carbon peak and carbon neutrality in China as scheduled. The enhancement of ECIDQ is not only an important basis for improving the carbon trading system, but also an important way to promote corporate emission reduction. Materials and methods Based on the four dimensions of carbon information disclosure, including disclosure carrier, carbon governance, carbon business and carbon performance, an evaluation index system of ECIDQ is constructed, and the methods of comprehensive principal component analysis, entropy weight method and random forest model are integrated to study the real level and change trend of carbon information disclosure quality of manufacturing enterprises in China from 2018 to 2020, and the importance of each index. Results (1) The combination of principal component and entropy weight method can effectively reduce the impact of interfering dimensions with less information on ECIDQ, and can objectively and truly evaluate the ECIDQ of China’s manufacturing enterprises, so as to obtain more effective evaluation results than the traditional single method alone. (2) From 2018 to 2020, the overall ECIDQ of China’s manufacturing enterprises showed an upward trend, and reaching a peak in 2020. (3) The overall ECIDQ of China’s manufacturing enterprises was low, and there were great differences in different enterprises. (4) Factors such as research and development (R&D) investment, total assets, and total liabilities have a great impact on ECIDQ. Discussion ECIDQ has been greatly improved in 2020, most likely due to that Chinese government put forward the goals of carbon peak and carbon neutrality in 2020, most companies disclose potential opportunities and risks in their social responsibility reports and annual reports. The score of ECIDQ with the highest quality of carbon disclosure remains unchanged, indicating that the content of corporate carbon disclosure is consistent and comprehensive. In the analysis of the influencing factors of ECIDQ, based on the root node splitting frequency and node purity increase, it is found that the asset dimension has the highest impact on ECIDQ, while the management dimension has the lowest impact. Conclusions From 2018 to 2020, the overall ECIDQ of China’s manufacturing enterprises was low, but showed an upward trend. R&D investment, total assets, and total liabilities have a great impact on ECIDQ. Recommendations and perspectives In the future, it is necessary to continue to improve the incentive mechanism for carbon information disclosure, promote the increase of enterprise innovation investment, rationally and efficiently allocate enterprise assets through multiple channels, and improve the quality of corporate carbon information disclosure. |
| Key words: manufacturing carbon information disclosure principal component analysis random forest |
