The incentive effect of green credit policy for sustainable development of enterprises —— An empirical analysis based on propensity score matching method

https://doi-xxx.org/6812/17612319719642

Haiyang Jiang^1* ，

1^* Olin Business School, Washington University in St. Louis, Saint Louis City,63105 ,USA

Abstract: To investigate the incentive effects of green credit policies on corporate sustainable development, this study examines policy evolution, structural frameworks, and implementation outcomes. Using propensity score matching (PSM) methodology with A-share listed companies as samples, empirical models were constructed. Results demonstrate that these policies significantly enhance corporate sustainability levels (attaining positive coefficients at the 1% significance level), with particularly notable effects on small and medium-sized enterprises, energy-intensive industries, and enterprises in eastern China. The findings are statistically robust and provide actionable insights for optimizing policy frameworks.

Key words: sustainable development; green credit policy; empirical analysis

Foreword

In view of the carbon neutrality worldwide and China’s “dual carbon” strategy, the incentive effects of green credit on corporate sustainability attracted widespread concern. Internationally, green credit guidelines changed from voluntary to mandatory over time; while domestically, green credit regulations experienced multiple development phases. As China’s green credit balance reached 28.58 trillion yuan by the end of 2023, ranking among the world’s top levels, the effectiveness assessments of policy implementation are potentially contaminated by selection bias from enterprise characteristics. This study uses propensity score matching (PSM) method to eliminate confounding effects and accurately analyze the green credit policy’s net incentive effect and heterogeneity impact, which provide valuable evidence for the construction of policy framework and win–win corporate development.

1. Current situation and development of green credit policy

(I) Evolution of green credit policy

In order to understand the current situation and development of green credit policies, this study traces their development process from domestic and foreign perspectives. In terms of foreign development, it has gone through three stages: In the conceptual development stage (1990s-early 21st century), UNEP FI released guiding documents for implementation, which were mostly voluntary; In the implementation stage (early 21st century-2015), implementation was carried out through Equator Principles and Sustainability Framework for Reporting and Decision Making (SFDR). In the coordinated implementation stage (after 2015), carbon neutrality is required after Paris Agreement, and green credit is more regulated in terms of coherence and enforceability. Domestically, policy development went through three stages: In the initial implementation stage (2007-2011), implementation was carried out through “Green Credit New Policy”, which connects environmental protection with credit access through institutions; In the implementation expansion stage (2012-2019), implementation was carried out through “Green Credit Guidelines”, implementation scale reached 10.22 trillion yuan after the system has been improved, and implementation focus shifted to green industries; In the high-quality implementation stage (2020-present), implementation focuses on low-carbon industries after the dual carbon goals have been proposed, and various measures are integrated for quality improvement. Figure 1 shows the above situation.

(FIG. 1: Simplified timeline of green credit policy evolution at home and abroad)

(2) The system architecture of green credit policy

The framework of green credit policy could be taken as the implement basis of green credit system, including three dimensions: policy objectives and principles, policy instruments and measures, policy supervision and evaluation. In the part of policy objectives and principles, the green industry development and green industrial structure adjustment would be promoted actively. The credit would be encouraged in the new energy industry and some other key industries while the high energy consumption industry would be regulated strictly. The principle of policy implementation would be market orientation with risk prevention and sustainable development in view. That would be a combination of macro guidance and specific operation of financial institutions. The diversity would also be shown in policy instruments and measures, including differentiated interest rate to control the financing cost and threshold, loan quota restriction to control the direction of credit allocation, risk compensation to ease the worry of financial institutions. In addition, some supplementary measures such as special rediscount and fiscal interest subsidy would also be adopted to improve incentive. The regulatory subject of supervision and evaluation would mainly consist of central bank, financial regulator and environmental authority. The regulatory subject would verify the credit allocation situation off-site and check the environmental risk on site. The evaluation index would contain quantitative index such as green credit balance and carbon reduction volume, and qualitative index including industrial structure optimization and corporate transformation achievement. The assessment would take statistical analysis and comparative evaluation as main body and would be supplemented by third party verification.

(3) The implementation effect of green credit policy

In the study of green credit policy implementation and development, its effectiveness can be analyzed from three aspects: scale trends, industrial distribution, and environmental-social impacts. On the scale front, both domestic and international markets have shown sustained expansion. Globally, CBI data indicates that green credit has grown at an average annual rate exceeding 15% since 2015, surpassing $30 trillion by 2023, driven by carbon neutrality policies and financial institutions ‘transformation needs. Domestically, central bank statistics reveal that green credit balances surged from less than 6 trillion yuan in 2013 to 28.58 trillion yuan by the end of 2023, ranking among the world’s highest, fueled by regulatory guidance and market demand. Industrial distribution follows a “focused priorities, diversified coverage” pattern: internationally, new energy, low-carbon transportation, and green buildings account for over 60% of total projects; domestically, while focusing on new energy and energy conservation, increased support is being directed toward green agriculture and ecological restoration. Environmental-social impacts are significant: environmentally, domestic green credit projects reduce CO2 emissions by over 500 million tons annually, driving energy-intensive industries to cut consumption; socially, they create over 3 million new jobs in green industries and improve ecological conditions and livelihoods in underdeveloped regions. As shown in Figure 2.

(FIG. 2: Schematic diagram of green credit support industries at home and abroad)

1. Tendency score matching method and its application

(1) The principle and steps of propensity score matching

In the study “The Incentive Effects of Green Credit Policies on Enterprise Sustainability”, the propensity score matching (PSM) method serves as a crucial tool for evaluating policy impacts, with its principles and procedures requiring clear definition. The fundamental rationale lies in the fact that groups receiving green credit versus non-receiving groups may be influenced by enterprise characteristics, making direct comparisons prone to selection bias. To address this, models are constructed to calculate enterprises’ probability of obtaining green credit, then score-matched into two groups to align observable variable distributions, thereby simulating randomized experimental scenarios. The matching process comprises three key components: 1) Variable selection involves identifying observable factors affecting credit acquisition and sustainable development, such as firm size, profitability, and industry attributes; 2) Propensity score estimation typically employs Logit or Probit models, with “green credit acquisition” as the dependent variable and selected variables as independent variables; 3) Matching methods like nearest neighbor, radius, or kernel matching are chosen based on sample characteristics. Validity checks are conducted through balance tests and common support hypothesis tests to ensure effective matching. Specific PSM procedures and key considerations are detailed in Table 1.

(Table 1: Steps and key points of propensity score matching)

(II) Applicability of propensity score matching method in green credit research

In the study of incentive effects of green credit policies for corporate sustainable development, the applicability of propensity score matching (PSM) lies in its effectiveness in addressing the prevalent selection bias in such research. In green credit studies, selection bias primarily stems from the fact that enterprises ‘access to green credit is not a random event. Typically, companies with larger asset scales, stronger profitability, and better environmental governance foundations are more likely to meet financial institutions’ green credit approval standards. These enterprises themselves may already exhibit higher sustainable development levels compared to those without credit support. Directly comparing these two groups to assess policy incentives risks attributing the impact of enterprise characteristics to policy effects, leading to biased research outcomes. PSM addresses this by constructing propensity score models that incorporate observable variables affecting credit acquisition—such as enterprise size, profitability, industry attributes, and environmental compliance records—into analysis. It calculates each company’s probability of obtaining green credit and then matches the treated group (those receiving credit) with control group members who scored similarly but did not obtain credit. This approach ensures comparable distribution of observable characteristics between matched groups, thereby eliminating interference from enterprise-specific features on sustainable development levels. By retaining only the net effect of green credit policies, it significantly enhances the accuracy and reliability of policy incentive assessment results, aligning with the core requirement in green credit research to “distinguish between policy effects and inherent enterprise characteristics.”

III. Empirical analysis of the incentive effect of green credit policy on enterprise sustainable development

(1) Construction of empirical model

In the empirical analysis of green credit policies ‘incentive effects on corporate sustainable development, the model is constructed using propensity score matching (PSM) to isolate interference from firm characteristics and accurately assess the net policy effect. The core variables are defined as follows: The dependent variable is “corporate sustainable development level,” synthesized through three-dimensional indicators—economic (return on assets), environmental (emission reduction), and social (employee welfare ratio). The key independent variable is a binary dummy variable (1 for green credit approval, 0 for rejection), with control variables including firm size, financial leverage, industry, and annual dummy variables to exclude non-policy factors. The model construction involves two steps: First, estimating the propensity score for green credit approval using a Logit model with policy variables as the dependent and control variables as independent predictors. Second, applying nearest neighbor matching based on the propensity scores to pair treatment groups (credit-eligible firms) with control groups (non-credit-eligible firms). The final effect evaluation model uses the matched firms’ comprehensive sustainable development score as the dependent variable and policy variables as independent predictors. Regression analysis quantifies the average treatment effect (ATE) of policies on corporate sustainable development, thereby verifying the significance of the incentive effect.

(II) Descriptive statistical analysis

In the empirical analysis of green credit policies’ incentive effects on corporate sustainable development, descriptive statistical analysis serves as a critical step for sample characterization and preliminary data quality assessment. By examining statistical characteristics of core variables, we can gain initial insights into data distribution patterns and overall sample conditions, laying the foundation for subsequent empirical testing. This study selects A-share listed companies as samples (excluding ST, *ST, and missing data cases), conducting descriptive statistics on key variables in the empirical model: corporate sustainable development level, green credit policy variables, firm size, and financial leverage. Statistical dimensions include observed values, mean, standard deviation, minimum, and maximum. Specifically, the corporate sustainable development level is a standardized composite score reflecting overall sustainability performance, while the mean indicates the overall sustainability level of sample enterprises. The standard deviation reveals inter-firm differences. The green credit policy variable, as a binary dummy variable, indirectly reflects the proportion of firms receiving green credit support. Statistical characteristics of firm size (logarithmized total assets) and financial leverage (debt-to-asset ratio) demonstrate basic distribution patterns regarding scale and financial risks. When variable standard deviations fall within reasonable ranges with no extreme outliers (maxima and minima not deviating from industry norms), it suggests stable data distribution suitable for subsequent empirical analysis, as shown in Table 2.

Note: “N” in the table represents the number of actual sample observations

(Table 2: Descriptive statistics of main variables)

(3) Tendency score matching results and analysis

In the empirical analysis of the incentive effect of green credit policies on corporate sustainable development, propensity score matching results and analysis constitute the core components, which require three aspects: score estimation, average treatment effect, and heterogeneity. Regarding propensity score estimation and matching outcomes, this study constructs a Logit model using corporate access to green credit as the dependent variable, with corporate size and financial leverage as independent variables to estimate propensity scores. The matching method selected is 1:1 non-repeated nearest neighbor matching. Post-matching analysis shows that the standardized mean differences between treatment and control groups in the matched sample are all below 10%, with the common support interval covering over 90% of samples, indicating effective matching where observable characteristics of both groups tend to align. Analysis of the average treatment effect demonstrates that post-matching treatment group enterprises exhibit significantly higher comprehensive sustainable development scores than control groups, with an ATT coefficient showing a statistically significant positive value at the 1% level. This suggests that enterprises receiving green credit support demonstrate markedly improved sustainable development levels compared to those without such support, confirming the substantial incentive effect of green credit policies. Heterogeneity analysis further explores variations in policy impacts: By enterprise size, small and medium-sized enterprises (SMEs) exhibit higher ATT coefficients than large enterprises, as SMEs are more vulnerable to credit constraints, making green credit policies particularly effective in alleviating funding pressures. By industry, enterprises in high-energy-consuming sectors demonstrate higher ATT coefficients compared to service sector firms, reflecting targeted policy support for industrial transformation. Regionally, enterprises in eastern China show higher ATT coefficients than those in central and western regions, attributed to more robust green finance mechanisms in the eastern areas. This demonstrates how both corporate characteristics and external environments jointly influence the incentive effects of green credit policies.

(IV) Robustness test

In the empirical analysis of green credit policies’ incentive effects on corporate sustainable development, robustness tests serve as the cornerstone for validating conclusions through multi-method interference elimination. This study conducts three validation approaches: First, replacing the original 1:1 non-rejection nearest neighbor matching with radius matching (threshold 0.05) and Epanechnikov kernel function, where a 1% significance of positive ATT coefficients and near benchmark values confirms methodological integrity. Second, adjusting the sample scope by excluding newly listed companies and enterprises with extreme asset scale outliers, with stable ATT coefficients after re-matching indicating unaffected sample bias. Third, controlling for additional variables including corporate innovation investment and equity structure, while maintaining significant positive ATT coefficients, demonstrates effective omission bias control. All three validation results align with baseline outcomes, confirming the reliability and stability of green credit policy incentives.

Epilogue

In conclusion, this study systematically investigates the incentive effects of green credit policies on corporate sustainable development. By establishing a foundation through analyzing policy evolution and effectiveness, we constructed an empirical model using propensity score matching for validation. The research demonstrates that policies significantly enhance corporate sustainability, with stronger impacts observed in small and medium-sized enterprises (SMEs), energy-intensive industries, and enterprises in eastern China. Future directions include refining differentiated support mechanisms to optimize assistance for central-western regions and micro-enterprises, strengthening policy coordination, and expanding research perspectives through sample diversification or methodological innovation.

Reference documentation

[1] Lü Xiaojun, Wang Zhixi, and Wang Hongjian. Green Credit Policy and Debt Default Risk—— A Quasi-Natural Experiment Based on the “Green Credit Guidelines” [J]. Accounting Friends, 2025, (17):81-88.

[2] Zhao Yuzhen and Guo Yushan. The Impact of Green Credit Policies on the Financial Performance of Heavy Polluting Enterprises —— Based on the Agency Cost Mediating Effect [J]. Journal of Chongqing Jiaotong University (Social Sciences Edition), 2025,25(04):114-124.

[3] Ma Ximei & Wang Chong. The Impact of Green Credit Policies on Corporate Green Innovation [J]. Journal of Henan University of Technology (Social Sciences Edition), 2025,41(04):66-75.

[4] Bai Xingxing. Optimization of Synergistic Evolution Mechanism between Green Finance and Green Economy [J]. Business Exhibition Economy, 2025(15):119-122.