Summary:
Amid China's comprehensive green transition, a structural conflict has emerged between the urgent need for high-carbon enterprises to decarbonize and their limited access to stable external financing. As a structural financial policy tool, green finance faces persistent challenges in identifying truly green or transitioning firms, a problem rooted in green information asymmetry. This asymmetry results in a mismatch: the supply of green credit is concentrated in fully green enterprises, while demand is concentrated in high-emission firms undergoing low-carbon transition. Addressing this allocation mismatch is critical to improving both the equity and efficiency of green financial resource distribution and supporting the real economy's decarbonization pathway. Artificial intelligence (AI), as a disruptive general-purpose technology, offers new potential to address this problem. Despite AI's proven utility in areas such as credit risk modeling and robot-advisory services, existing literature provides limited evidence on how AI can enable green finance—particularly under China's policy-driven structural financial system. Prior studies often emphasize the broad effects of fintech on the real economy while overlooking how fintech enables structural financial policies to have differentiated impacts on the real economy. To fill this gap, this paper starts from the perspective of enterprise loan cost and financing constraint and investigates whether and how AI adoption by commercial banks enhances green credit allocation by improving banks' ability to identify green or transitioning firms—what we refer to as their green recognition capability. This paper develops a theoretical model of banks' optimal loan pricing under green financial policy constraints and green information asymmetry, in which AI adoption influences banks' identification precision and thus their credit allocation behavior. The model's predictions are empirically tested using firm-level matched loan data for all A-share listed companies in China from 2013 to 2022. The results show that each 1% increase in a bank's AI adoption rate is associated with a 5.06 basis point reduction in the average loan interest rate markup, with a more pronounced reduction of 7.28 basis points for green firms, while the effect is statistically insignificant for non-green firms. From the perspective of financing constraints, each 1% increase in AI adoption is associated with a 1.61 basis point decrease in firms' financing constraints measured by the KZ index, driven mainly by a 1.40 basis point reduction for green firms. These findings suggest that AI adoption enables banks to better serve green enterprises without adversely affecting non-green firms, effectively reducing both loan costs and financing frictions for green firms. Crucially, this paper finds that the impact is significantly greater for firms with lower green information transparency, providing empirical support for the mechanism whereby AI strengthens banks' green recognition capability and enhances real economy outcomes. The effect is more pronounced for banks with faster loan growth and in regions with higher environmental spending. Based on these findings, three policy recommendations are proposed. First, promote the deep integration of AI and green finance by strengthening banks' green identification capabilities through unified green standards, transparent data infrastructure, and supportive regulatory frameworks. Second, accelerate the development of transition finance standards by leveraging AI to identify and support decarbonization in high-carbon sectors, thereby reducing banks' perceived risk premiums and improving credit access for transitioning firms. Third, adopt differentiated, region-and sector-specific policies to pilot AI-powered green finance reforms, enabling scalable policy experimentation and tailored support for diverse transition pathways. This paper contributes to the literature in three ways. First, it provides theoretical and empirical evidence on the feasibility and necessity of integrating AI with green finance, a topic at the frontier of interdisciplinary research between economics and artificial intelligence. Second, it expands the analysis of structural financial policy heterogeneity, demonstrating how targeted financial instruments, when empowered by technology, can yield differentiated real economy outcomes. From the perspective of artificial intelligence enabling green finance, it provides direct evidence for China's comprehensive green transformation. Third, it offers a novel perspective on addressing the financing gap for non-green firms during green transition, and provides actionable insights into how AI can help overcome green information asymmetries and improve the precision of financial support for decarbonization in high-emission sectors.
钟茜. 商业银行采纳人工智能有助于优化企业绿色转型中的信贷支持吗?[J]. 金融研究, 2025, 538(4): 57-74.
ZHONG Qian. Does AI Adoption by Commercial Banks Enhance Credit Support for Corporate Green Transition？. Journal of Financial Research, 2025, 538(4): 57-74.
[1] 杜兴强、谢裕慧和曾泉,2024,《绿色金融政策抑制了企业的环境违规吗？——基于绿色金融改革创新试验区的一项准自然实验》,《金融研究》第5期,第132～149页。 [2] 杜克锐、李旻旸、王思泉和林伯强,2024,《煤炭价格冲击与新能源业务扩张叠加下的企业融资约束》,《经济研究》第12期,第4～20页。 [3] 方颖和郭俊杰,2018,《中国环境信息披露政策是否有效:基于资本市场反应的研究》,《经济研究》第10期,第158～174页。 [4] 黄益平和邱晗,2021,《大科技信贷:一个新的信用风险管理框架》,《管理世界》第2期,第12～21页。 [5] 蒋涛、吴卫星和宫迪,2017,《政治风险会影响贷款定价吗》,《管理评论》第9期,第3～14页。 [6] 刘锡良和文书洋,2019,《中国的金融机构应当承担环境责任吗？——基本事实、理论模型与实证检验》,《经济研究》第3期,第38～54页。 [7] 刘婷婷,2016,《利率市场化、特许经营权与贷款定价》,《宏观经济研究》第3期,第62～72页。 [8] 刘莉亚、余晶晶、杨金强和朱小能,2017,《竞争之于银行信贷结构调整是双刃剑吗？——中国利率市场化进程的微观证据》,《经济研究》第5期,第131～145页。 [9] 李雅婷和陈济,2023,《高碳资产绿色转型的关键》,《中国金融》第15期,第34～35页。 [10] 马骏,2022,《〈G20转型金融框架〉及对中国的借鉴》,《中国金融》第23期,第21～23页。 [11] 潘冬阳、陈川祺和Michael Grubb,2021,《金融政策与经济低碳转型——基于增长视角的研究》,《金融研究》第5期,第1～19页。 [12] 宋全云、李晓和钱龙,2019,《经济政策不确定性与企业贷款成本》,《金融研究》第7期,第57～75页。 [13] 苏冬蔚和刘子茗,2023,《绿色金融改革是否影响业绿色绩效与漂绿风险》,《国际金融研究》第4期,第74～85页。 [14] 王馨和王营,2021,《绿色信贷政策增进绿色创新研究》,《管理世界》第6期,第173～188页。 [15] 文书洋、刘浩和王慧,2022,《绿色金融、绿色创新与经济高质量发展》,《金融研究》第8期,第1～17页。 [16] 王红建、张科和李青原,2023,《金融科技的经济稳定器作用:金融加速器理论的视角》,《经济研究》第12期,第4～21页。 [17] 魏志华、曾爱民和李博,2014,《金融生态环境与企业融资约束——基于中国上市公司的实证研究》,《会计研究》第5期,第73～95页。 [18] 姚加权、张锟澎、郭李鹏和冯绪,2024,《人工智能如何提升企业生产效率？——基于劳动力技能结构调整的视角》,《管理世界》第2期,第101~116+133+117~122页。 [19] 易行健和周利,2018,《 数字普惠金融发展是否显著影响了居民消费——来自中国家庭的微观证据》,《金融研究》第11期,第101～116页。 [20] 俞中、佟孟华和赵江山,2025,《排污许可监管、绿色金融创新与企业信贷融资》,《世界经济》第5期,第89～127页。 [21] 严成樑、赵扶扬和牛欢,2024,《环境目标责任制、环境治理与内生经济增长》,《经济研究》第4期,第133～152页。 [22] 战明华和应诚炜,2015,《利率市场化改革、企业产权异质与货币政策广义信贷渠道的效应》,《经济研究》第9期,第114～126页。 [23] 张希良、黄晓丹、张达、耿涌、田立新、范英和陈文颖,2022,《碳中和目标下的能源经济转型路径与政策研究》,《管理世界》第1期,第35～51页。 [24] 张一林、郁芸君和陈珠明,2021,《人工智能、中小企业融资与银行数字化转型》,《中国工业经济》第12期,第69～87页。 [25] 朱光顺和魏宁,2023,《负责任的银行贷款与上市公司ESG表现》,《数量经济技术经济研究》,第4期,第133～151页。 [26] Babina, T., A. Fedyk, A. He, and J. Hodson, 2024, “Artificial Intelligence, Firm Growth, and Product Innovation”, Journal of Financial Economic, 151, 103745. [27] Fraisse, H., and M. Laporte, 2022, “Return on Investment on Artificial Intelligence: The Case of Bank Capital Requirement”, Journal of Banking ＆ Finance. 138(5). 106401. [28] Gambacorta, L., 2008, “How Do Banks Set Interest Rates？”, European Economic Review, 52(5), pp.792～819. [29] Haenlein, M., and A. Kaplan, 2019, “A Brief History of Artificial Intelligence: On the Past, Present, and Future of Artificial Intelligence”, California Management Review, 61(4), pp.5～14. [30] Kaplan, S., and L. Zingales, 1997, “Do Investment-cash Flow Sensitivities Provide Useful Measures of Financing Constraints？”, Quarterly Journal of Economics, 112, pp.169～215.
