China’s green credit policy originated with the 1995 “Circular on Issues Related to the Implementation of Credit Policy and Strengthening of Environmental Protection” issued by the People’s Bank of China (PBOC). However, these early efforts failed to garner sufficient attention, as economic growth took precedence over environmental concerns at the time (Zhang et al., 2023). A significant development occurred in 2007 when the former State Environmental Protection Administration (SEPA), PBOC, and former China Banking Regulatory Commission (CBRC) (The China Banking Regulatory Commission was established in 2003. In 2018, the China Banking Regulatory Commission and the China Insurance Regulatory Commission merged to form the China Banking and Insurance Regulatory Commission. In 2023, the National Financial Regulatory Administration was established on the basis of the China Banking and Insurance Regulatory Commission.) issued documents requiring commercial banks to support green production initiatives, marking the formal establishment of China’s green credit policy. Despite this, the policy still lacked strong enforcement and adequate support measures during this period (Zhang et al., 2021). In 2012, the CBRC issued the “Green Credit Guidelines”, which provided comprehensive directives for financial institutions, outlining specific supervisory roles in six areas, including management, policy systems, and information disclosure. This marked a significant milestone in formalizing China’s green credit practices (Liu et al., 2019).

For firms, green credit policies impose financing penalties on highly polluting companies, negatively affecting their debt financing (Song et al., 2021). Some studies suggest that green credit influences investment decisions and environmental performance, incentivizing firms to relocate polluting activities to avoid financial constraints, which results in no overall improvement in their environmental performance (Bartram et al., 2022). The impact on economic performance is debated. Some scholars argue that GCP restricts productive operations, although technological innovation can offset this through the Porter effect (Chen et al., 2022). However, there is no consensus on the Porter effect’s existence under the green credit policy (Egger et al., 2021).

The relationship between CSR and firm performance remains contested in the literature (Afrin and Rahman, 2024; Lopatta et al., 2024). Some argue that CSR enhances profitability and firm value by improving reputation, attracting investors, mitigating conflicts between shareholders and stakeholders and fostering innovation, encapsulated in the concept of “doing well by doing good” (Tetrault Sirsly and Lvina, 2019). However, the impact of CSR on company performance is not always positive. Some suggest that only firms with strong financial performance can afford CSR initiatives, which is referred to as “doing good by doing well” (Lopatta et al., 2024). Implementing CSR activities typically involves increased costs, such as environmental measures, charitable donations, and employee benefits, which may directly reduce short-term profits. Financial constraint theory suggests that companies may need to sacrifice other business activities to maintain CSR initiatives, leading to imbalanced resource allocation and thereby affecting short-term performance (Attig, 2024). Additionally, some studies indicate that management may use CSR as a tool for personal prestige or political gain rather than focusing on improving company performance (Gupta et al., 2021; Petrenko et al., 2016; Wang and Qian, 2011). Agency theory points out that there may be conflicts of interest between management and shareholders, with management potentially pursuing personal interests rather than company interests, leading to inefficient or ineffective implementation of CSR initiatives (Dmytriyev et al., 2021). Furthermore, CSR may conflict with the goal of maximizing shareholder value, especially when CSR projects do not yield direct profit returns, which may weaken shareholders’ financial returns and thus affect shareholder support and company market performance (Battilana et al., 2022). Overcommitting to CSR activities may also lead to overly high expectations from the market and customers. If the company fails to fulfill its commitments, it may face a trust crisis, affecting brand value and market performance.

Green credit policy, as an environmental policy tool, aims to channel financial resources toward environmentally friendly and sustainable projects, thereby promoting corporate environmental governance and corporate social responsibility. However, despite the original intent of green credit policy to encourage CSR, in some cases, the policy may inadvertently suppress corporate social responsibility performance. This phenomenon can be analyzed from multiple theoretical perspectives.

First, from the perspective of the resource dependence theory, a firm’s access to and allocation of resources heavily depend on the external environment, particularly financial resources (Chi et al., 2025). Green credit policies typically require firms to meet specific environmental standards and criteria when obtaining loans, which may lead to financial constraints for some firms. Firms with high capital needs that do not meet green credit criteria may redirect more resources toward complying with green finance requirements, thereby neglecting other aspects of CSR, such as employee welfare, social welfare, and corporate governance (Liu and He, 2021). Therefore, to secure loan support, firms may prioritize environmental compliance as their sole focus, sacrificing other dimensions of CSR.

Second, from the viewpoint of the stakeholder theory, green credit policies strengthen the monitoring of corporate environmental behavior by banks and financial institutions (Mahajan et al., 2023). However, in certain cases, firms may focus on short-term investments in environmental performance to obtain loans or avoid penalties, rather than fulfilling CSR obligations from a long-term perspective. This phenomenon is particularly evident in firms under financial pressure or those with tight profit margins (Almeida et al., 2024). These firms may view environmental compliance as the minimum requirement for dealing with financial regulation while neglecting other areas of CSR, such as labor rights protection and consumer rights. Thus, while green credit policies may lead firms to take action in the environmental domain, they may result in a narrow focus on environmental issues, overlooking broader CSR responsibilities.

Third, from the perspective of the agency theory, management may, due to external pressure from green credit policies, engage in short-term actions to meet the criteria of green credit, even if these actions are not part of the firm’s long-term strategy (Al-Faryan, 2024). In such cases, management may treat environmental projects as short-term signaling tools rather than long-term CSR commitments. This signaling behavior may lead to insufficient investment in other CSR areas, particularly when these areas are not directly addressed by green credit policies.

Finally, the implementation of green credit policy also increases compliance costs, especially for firms that are not competitive in the environmental field (Xu and Li, 2020). Financially constrained firms are more likely to prioritize short-term survival over long-term CSR investments since they are inclined to view social responsibility initiatives as discretionary when resources become scarce (Leong and Yang, 2021). In such cases, reduced investment in CSR can enable firms to allocate scarce funds to immediate operational needs, ensuring their survival in the marketplace. These firms may invest heavily in environmental compliance in the short term to meet policy requirements, but due to resource constraints, they may reduce their investments in other CSR areas, creating resource allocation conflict. This conflict may lead to a shift in priorities when fulfilling CSR, ultimately suppressing their performance in other CSR domains. Based on these insights, we propose the following hypothesis:

Hypothesis 1: The implementation of GCP exerts a negative direct effect on the CSR performance of heavily-polluting firm.

The institutional theory suggests that external pressures, such as regulatory mandates, shape firm behavior (Lewis et al., 2019). In post-socialist economies, institutional transitions have also influenced CSR adoption. For instance, East European firms’ responsible practices reflect both legacy constraints and new normative pressures (Koleva et al., 2010). In China, GCP serves as a regulatory tool that incentivizes heavily-polluting firms to align their resource allocation strategies with environmental regulations. By restricting financing for high-polluting firms, GCP encourages funding for environmentally compliant initiatives that support cleaner production and sustainability. However, this policy significantly constrains firms’ financial and strategic decision-making.

GCP amplifies financial pressures on heavily-polluting firms by increasing financing thresholds and costs. As a result, banks prioritize credit to green-compliant firms, which leaves high-polluting firms with limited external funding and reduced capital liquidity (Lei et al., 2023). Resource dependence theory highlights how this scarcity of external resources forces firms to reallocate funds toward meeting environmental compliance requirements to avoid penalties or further financing restrictions (Lin and Pan, 2024).

These financial constraints have cascading effects on firms’ total factor productivity (TFP). While stricter financing conditions push firms to optimize internal resource usage, GCP-mandated investments in green technologies and emission reduction projects often divert resources away from productivity-enhancing activities (Zhao et al., 2022). This reallocation causes a crowding-out effect, which reduces the marginal productivity of assets and negatively impacts TFP (Heil, 2019).

The decline in TFP, coupled with financial constraints, further weakens firms’ ability to meet their CSR commitments. With limited resources, firms are forced to prioritize short-term operational needs over long-term CSR goals. Time preference theory explains this behavior, by suggesting that firms favor investments with immediate returns, often at the expense of CSR initiatives (Cohen et al., 2020). As a result, heavily-polluting firms may scale back or delay CSR expenditures, viewing them as non-essential under constrained capital conditions.

In summary, GCP disrupts firms’ resource allocation by tightening financial constraints, which in turn reduces TFP and weakens CSR performance. This double mediation channel illustrates how GCP simultaneously influences firms’ financial and operational decisions and, ultimately, restricts their social and environmental contributions. Fig. 1 illustrates the theoretical framework of the mediating effect. From this theoretical foundation, we formulate the following hypothesis:

Fig. 1.

Theoretical framework diagram of the mediating effect. CSR, corporate social responsibility.

Hypothesis 2: Financial constraints mediate the effect of GCP on CSR. Specifically, the implementation of GCP increases financial constraints, which negatively affect TFP. The reduction in TFP, together with the increased financial constraints, ultimately leads to a decline in CSR performance.

Given the unique ownership structures of Chinese firms, the variability in regional green financial development, and the diversity of industry structures, this study examines firm ownership (SOE), the regional green economic development index (GREEN), and industry concentration (HHI) as moderators of the GCP–CSR relationship.

State-owned enterprises (SOEs) are a dominant organizational form in China. They benefit from low capital costs due to preferential access to loans and government subsidies, which insulates them from market discipline and makes them less responsive to environmental accountability measures. SOEs also pursue objectives beyond profitability, such as social stability, employment maintenance, and regional development (Xu and Li, 2020). Under the GCP, heavily-polluting SOEs may neglect environmental standards in favor of fulfilling other government objectives. This is consistent with the theory of institutional multiplicity, which suggests that firms facing competing institutional logics may engage in symbolic rather than substantive CSR (Brammer et al., 2012). Moreover, SOEs tend to have lower managerial autonomy and weaker stakeholder pressure compared to private firms. In the absence of strong accountability from shareholders or capital markets, their CSR practices are often compliance-oriented and reactive, making them more vulnerable to suppression when GCP imposes regulatory or reputational risks that they do not perceive as financially consequential. These financial advantages may also reduce their incentives to voluntarily expand CSR engagement.

Regional disparities in green finance development further shape how firms respond to GCP. In regions with well-developed green finance systems, resources are primarily directed toward firms with strong environmental performance (Gao et al., 2023). This allocation can restrict financing for heavily-polluting firms, suppressing their ability to invest in CSR. Such firms face higher financing costs and capital shortages, which weaken their competitiveness and may crowd out CSR (Ragazou et al., 2025). In contrast, firms in weaker financial systems may find ways to bypass institutional constraints, but this can exacerbate symbolic CSR responses. Local variation in green finance also affects market perceptions of the “green firm” label. In regions with robust financial systems, lenders and stakeholders impose stronger “green reputation” pressures (Khalid et al., 2024). High-polluting firms may respond outwardly to these pressures, yet given the gap between their current production models and green transformation, they often resort to symbolic rather than substantive CSR.

Industry concentration likewise influences CSR engagement. Resource dependence theory suggests that firms in highly concentrated industries may be less responsive to external demands because they control critical resources (Zhu et al., 2025). Dominant firms often possess greater bargaining power in negotiating debt covenants and may secure favorable terms despite green credit restrictions, thereby weakening policy incentives to improve CSR (Claver-Cortés et al., 2019). Reduced competition further lowers incentives to adopt substantive CSR. Moreover, firms in concentrated industries are more likely to leverage their market power to resist external pressures, opting for incremental rather than large-scale environmental investments (Kok and Ligthart, 2014). Consequently, heavily-polluting firms in highly concentrated industries may adopt a slower CSR approach under GCP, delaying substantive green transformation.

On this basis, we propose the following hypotheses:

Hypothesis 3a: State ownership strengthens the negative relationship between GCP and CSR of heavily-polluting firms.

Hypothesis 3b: Higher levels of regional green financial development strengthen the negative relationship between GCP and CSR of heavily-polluting firms.

Hypothesis 3c: Higher industry concentration strengthens the negative relationship between GCP and CSR of heavily-polluting firms.

To evaluate the impact of China’s 2012 Green Credit Policy (GCP) on CSR performance, we employed a difference-in-differences (DID) model using data from Chinese listed firms between 2010 and 2020. The data were sourced from the China Stock Market and Accounting Research (CSMAR) and Hexun databases. To ensure data integrity and alignment with the study’s objectives, we refined the sample through the following steps in line with Lin and Pan (2024): (1) Exclusions: Financial companies, Special Treatment (ST) and Particular Transfer (PT) companies were excluded. In the Chinese context, ST firms are publicly listed companies flagged by stock exchanges for financial instability or operational challenges, while PT firms face even more severe financial or operational issues. (2) Data Completeness: Firms with missing data were removed from the sample. The distribution of the top 20 industries by proportion after excluding financial, ST, and PT companies is shown in Appendix Table 8. Appendix Table 9 reports the ratio of the total value and average of total assets and total revenue in the sample after excluding certain observations, compared to the corresponding metrics before the exclusion, to assess whether the remaining sample still represents the overall population. (3) Outlier Mitigation: All continuous variables were winsorized at the 1% and 99% levels to reduce the influence of extreme outliers. The final dataset consists of 28,001 observations from 3627 listed firms, including 891 heavily-polluting and 2736 non-heavily-polluting firms.

Table 2A,2B,2C summarizes the key variables, comparing treatment (heavily-polluting firms) and control groups. The treatment group exhibits distinct characteristics in the pre-policy period: on average, they have higher leverage (mean leverage 43.4% vs. 40.8% for controls), slightly lower profitability (mean ROE 6.3% vs. 7.1%), and notably lower mean CSR scores (0.296 vs. 0.303). These differences reflect the nature of heavy-polluting industries—they tend to be capital-intensive with higher debt and historically less CSR engagement. Importantly, the parallel trends assumption appears to hold: Fig. 2 (the dynamic CSR trend plot) shows that CSR trajectories for treated and control firms were similar prior to 2012, diverging only after the policy. Correlation coefficients of Table 2B indicate no multicollinearity concerns, as all values are below 0.5. Furthermore, Table 2C displays the variance inflation factor (VIF) values, which range from 1 to 3, confirming that multicollinearity does not pose a problem for our analysis (Bouchmel et al., 2024).

Fig. 2.

Dynamic effects of the Green Credit Policy (GCP) on CSR. Note: 2012 is used as the baseline year (t = 0); all reported estimates are relative to this reference.

A key identification assumption for applying the difference-in-differences (DID) method is that the treatment and control groups follow a common trend before the implementation of the policy. This requirement is known as the parallel trend assumption. Based on this, we conduct a test using an event study approach, with the following model specification (Eqn. 2):

(2)$$CS{R}_{it}=\alpha +{\sum }_{k=\mathrm{–}2,k\ne 0}^{k=8}{\beta }_k\times D_{i,k}+{\delta }_i+{\mu }_t+{\epsilon }_{it}$$

In this model, $D_{i,k}$ denotes a set of dummy variables indicating the k-th year relative to the start of the Green Credit Policy. The other variables are defined in the same way as in Eqn. 1. The policy implementation year (k = 0) is used as the baseline period. The coefficients reflect the differences in CSR changes between the treatment and control groups in year k, relative to the policy implementation year. If the estimates for the pre-policy period (k 0) include zero within 95 percent confidence intervals, this indicates that the coefficients are not significantly different from zero, and the parallel trends assumption holds. As shown in Fig. 2, the pre-policy coefficients are not statistically different from zero. This indicates that, prior to the introduction of the Green Credit Policy, the treatment and control groups did not show significant differences in CSR trends. Therefore, the parallel trends assumption is supported.

Although Fig. 2 shows a slight upward movement in CSR levels during the later post-policy years, this change is not statistically significant. The confidence intervals in those periods overlap with earlier negative effects, suggesting that the observed rebound may reflect natural fluctuation rather than a meaningful reversal of the policy impact. Thus, the Green Credit Policy’s negative effect on CSR persists throughout the study period.

Table 3 presents the results of the baseline regression model. Column (1) shows the results without control variables, while Column (2) includes all controls. In both cases, the interaction term treat $\mathrm{\times }$ post is significantly negative, with a coefficient of –0.002 (p 0.01) in Column (1). This indicates that GCP implementation reduced the average CSR performance of heavily-polluting firms by 0.2%. After adding controls in Column (2), the coefficient remains significantly negative, confirming the robustness of the findings and providing strong support for Hypothesis 1. Columns (3) to (7) examine the impact of GCP on specific CSR dimensions, including shareholder, employee, supplier, environmental, and social responsibilities. The results indicate significant negative effects on shareholder, employee, and environmental responsibilities, while the effects on supplier and social responsibilities are insignificant.

Among the control variables, higher leverage (Lev) and fixed assets ratios (Fix) are negatively associated with CSR. Firms with higher leverage face increased debt burdens, limiting funds available for CSR. Similarly, capital-intensive firms with higher fixed asset ratios may prioritize operational investments over CSR activities. Other control variables, such as firm size and profitability metrics, show varying effects on CSR dimensions.

To examine the double mediating effect of the GCP on CSR, we employ a model that explores the sequential relationships between GCP, WW, TFP, and CSR. The analysis builds on methodologies proposed by Marco-Lajara et al. (2023), utilizing three equations to capture the interconnected effects. To better control for the potential influence of firm life cycle and industry characteristics on total factor productivity (TFP), we follow Breuer and DeHaan (2024) by including a “life cycle $\times$ industry” interaction fixed effect in the double mediation model (Eqns. 3,4,5).

(3)$$W{W}_{it}={\beta }_0+{\beta }_{1}trea{t}_i+{\beta }_{2}pos{t}_t+{\beta }_{3}trea{t}_i\times pos{t}_t$$ $$+ {\beta }_{4}Control{s}_{it}+{\delta }_i+{\mu }_t+{\epsilon }_{it}$$

(4)$$TF{P}_{it}={\theta }_0+{\theta }_{1}trea{t}_i+{\theta }_{2}pos{t}_t+{\theta }_{3}trea{t}_i\times pos{t}_t$$ $$+ {\theta }_{4}W{W}_{it}+{\theta }_{5}Control{s}_{it}+{\delta }_i+{\mu }_t+{\epsilon }_{it}$$

(5)$$CS{R}_{it}={\phi }_0+{\phi }_{1}trea{t}_i+{\phi }_{2}pos{t}_t+{\phi }_{3}trea{t}_i\times pos{t}_t$$ $$+ {\phi }_{4}W{W}_{it}+{\phi }_{5}TF{P}_{it}+{\phi }_{6}Control{s}_{it}+{\delta }_i$$ $$+ {\mu }_t+{\epsilon }_{it}$$

The variable definitions of $CS{R}_{it}$, treat, post, $Control{s}_{it}$, ${\mu }_t$, and ${\epsilon }_{it}$ are the same as Eqn. 1. WW denotes the financing constraint, and TFP is total factor productivity. ${\delta }_i$ represents the “life cycle $\times$ industry” interaction fixed effect.

The results in Table 4 show that GCP significantly increases WW, as indicated by the positive coefficient on the interaction term $trea{t}_i\times pos{t}_t$ ($\beta$₃ = 0.0033, p 0.01, Column 1). Higher WW values then negatively affect TFP (–0.0976, p 0.01, Column 2), underscoring the adverse influence of financial constraints on productivity. Column 3 further demonstrates that both WW and TFP shape CSR performance: increased WW directly reduces CSR ($\phi$₄ = –0.0163, p 0.01), while TFP has a positive effect ($\phi$₅ = 0.0050, p 0.05). Taken together, these findings confirm that GCP indirectly suppresses CSR performance by increasing financing constraints, which subsequently reduce productivity and, in turn, CSR engagement. This provides strong empirical support for Hypothesis 2, showing that financial constraints and TFP operate as sequential mediators in the relationship between GCP and CSR.

Fig. 3 illustrates these relationships visually, mapping the pathways through which GCP affects financial constraints, TFP, and, eventually, CSR performance. The mediation pathway highlights the ripple effects of GCP, in which pressures from regulators not only generate financial strain but also influence firms’ productivity and social responsibility efforts.

Fig. 3.

Path of double mediating effect. ** and *** indicate significance at the 5% and 1% levels, respectively.

Table 5 examines how firm ownership (SOE), the regional green economic development index (GREEN), and industry concentration (HHI) moderate the relationship between GCP and CSR. The interaction terms are all negative and statistically significant: posttreat $\mathrm{\times }$ SOE (–0.0027, p 0.05, Column 1), posttreat $\mathrm{\times }$ GREEN (–0.0132, p 0.10, Column 2), and posttreat $\mathrm{\times }$ HHI (–0.0347, p 0.01, Column 3). These results suggest that the negative impact of GCP on CSR performance is more pronounced for state-owned enterprises, regions with advanced green financial infrastructure, and industries with high levels of concentration. The findings validate Hypotheses 3a, 3b, and 3c, demonstrating that the moderating factors of ownership type, regional financial development, and industry structure influence the degree to which GCP impacts CSR performance.

The relationship between GCP and CSR varies depending on a firm’s TFP. TFP reflects a firm’s ability to adapt to changes, such as market demand fluctuations, policy shifts, and technological advancements. Firms with different TFP levels exhibit distinct investment behaviors, which may influence how GCP impacts their CSR performance (Zhong et al., 2022).

To investigate this, we applied a threshold regression model based on Wang C and Wang L (2023), using TFP as the threshold variable (Eqn. 6). This model identifies whether the GCP–CSR relationship changes when firms’ TFP surpasses specific thresholds.

(6)$$CS{R}_{it}={\lambda }_0+{\lambda }_{1}DI{D}_{it}I(TFP\le {\gamma }_1)+{\lambda }_{2}DI{D}_{it}I$$ $$+ {\delta }_i+{\mu }_t+{\epsilon }_{it}$$

Where I is an indicator function that equals 1 when the condition is satisfied and 0 otherwise. ${\gamma }_1$, ${\gamma }_2$… ${\gamma }_n$ are determined through testing, and the other parameters maintain the same economic interpretation as in previous sections. Through 500 bootstrap resampling iterations, a single threshold was found significant at the 5% level, with a value of 0.9298 (Panel A, Table 6). This indicates that TFP significantly moderates the effect of GCP on CSR.

The results in Panel B of Table 6 show that when TFP is below 0.9298, the coefficient for the GCP’s effect on CSR is –0.0143 (p 0.05), indicating a stronger negative impact. However, for firms with TFP above this threshold, the absolute value of the coefficient decreases to –0.0094 (p 0.10), suggesting a weaker negative effect.

Two explanations can account for this phenomenon. First, increased TFP improves the efficiency of a firm by allocating capital, labor, and resources, which lowers production costs and enables companies to absorb additional costs associated with GCP without significantly affecting profitability (Lee and He, 2022). Second, a higher TFP makes a firm more competitive in the market, prompting it to enhance its CSR practices to attract consumers and investors (Peng et al., 2021).

These findings highlight the moderating role of TFP in the GCP-CSR link. They emphasize the importance of productivity improvements in mitigating the policy’s negative effects on CSR performance and suggest that supporting productivity enhancements can make GCP implementation more effective.

To verify the reliability of our findings, we conducted several robustness checks.

Placebo tests. We reassigned “fake” policy years before 2012 and re-estimated the DID model. As shown in Fig. 4, placebo estimates are centered around zero and statistically insignificant, confirming that the observed CSR decline is attributable to the 2012 Green Credit Policy rather than spurious shocks.

Fig. 4.

Placebo test.

Propensity score matching (PSM). We matched treated and control firms based on pre-policy characteristics to mitigate selection bias. Post-matching results remain consistent with our baseline estimates: the negative impact of GCP on CSR persists and is of similar magnitude (Table 7, Column 1). Detailed covariate balance statistics and kernel density plots (showing improved overlap between treatment and control groups) are provided in Appendix Table 13 and Fig. 5.

Fig. 5.

Kernel density distribution before and after matching.

Concurrent policy shocks. We further accounted for major contemporaneous events that could confound the analysis, such as the 2015 Environmental Protection Law, the establishment of green finance pilot zones in 2016, and the COVID-19 outbreak in late 2019. Regression results (Table 7, Columns 2–4) show that excluding these events does not materially alter our findings.

Overall, these robustness checks reinforce the validity of our results. Across placebo tests, matched samples and controls for concurrent shocks, the evidence consistently indicates that the Green Credit Policy reduced CSR among heavily-polluting firms.

This study offers several important insights into how an environmental financial policy can reshape corporate behavior in a transition context. First, by focusing on heavily-polluting firms, we highlight an under-examined segment in the green finance literature. Prior works often looked at greener firms or average effects. We show that firms most in need of reform—the worst polluters—may paradoxically pull back on certain positive behaviors (CSR) when faced with financial strictures. This contributes to the literature on institutional voids and firm strategy in emerging markets (Meyer and Peng, 2016) by illustrating a context where policy intent (to improve environment) meets firm reality (resource constraints), producing unintended outcomes. Second, we contribute to the micro-foundations of sustainable finance by unpacking the mechanisms—financial constraints and productivity—through which policy affects firm outcomes. While studies like Aghion et al. (2016) emphasize innovation’s role in climate policy and others like Attig (2024) focus on finance–CSR links, we bring these threads together, demonstrating a connected causal chain. This integrative perspective enriches both CSR theory and environmental economics by showing how financial and operational dimensions jointly mediate policy impact. Third, methodologically, our introduction of a threshold regression to identify a TFP turning point is novel in this domain. It provides a quantitative way to test conditional theories (like Porter’s) within a single policy evaluation. We empirically establish that beyond a certain productivity level (~0.93 in our sample), the firm’s response to GCP changes nature—offering a rare concrete estimate that future research can build on or test in other contexts. Finally, our findings on SOEs add to institutional theory by evidencing how state ownership can moderate policy effectiveness in emerging markets. We provide a nuanced narrative: despite their financial advantages, SOEs might underperform on voluntary CSR under environmental financial pressure, due to institutional multiplicity and weaker external oversight. This underscores that governance structure is a critical lens when evaluating sustainability policies (Aguilera et al., 2006).

Our results carry practical implications for policymakers and financial regulators in China and beyond. For China: The Green Credit Policy, while well-intentioned and partly successful (firms did cut pollution investments as other studies show), appears to have side effects that need addressing. The fact that CSR fell suggests that heavy polluters might be neglecting social and governance aspects (like workplace safety, community engagement) when financially strained. Regulators should be aware that sustainability is multidimensional; focusing on one aspect (environment) in isolation may inadvertently harm others (social responsibility). Policy refinement is warranted: Our threshold finding implies a one-size-fits-all approach could be recalibrated. For lower-productivity firms, complementary support could help them comply with environmental mandates without sacrificing CSR. This might include subsidized green loans for SMEs, technical assistance to improve process efficiency, or phased targets that give them time to adjust. For higher-productivity firms, on the other hand, the government could tighten standards, as they can better absorb the costs and maybe turn them into innovation opportunities. By tailoring GCP intensity to firm capability, policymakers can mitigate negative externalities on CSR. Additionally, monitoring and incentives can be broadened: if banks (perhaps in collaboration with stock exchanges) also monitor CSR performance or if credit terms slightly reward comprehensive sustainability performance (not just environmental metrics), firms would have motivation to not let their CSR lapses too much. Banks could integrate Environmental, Social, and Governance (ESG) scoring that covers social factors into their credit risk models.

For transition/emerging economies: Our comparative discussion pointed out that Eastern European countries like Poland, Hungary, as well as nations like Brazil or India, are experimenting with green finance policies. Our study suggests a caution: ensure firms in these contexts are not overburdened to the point they neglect other responsibilities. It might be beneficial for these countries to simultaneously promote frameworks for CSR or to channel some green finance into capacity-building (like tech upgrades that boost productivity) so that firms are more resilient. International development agencies and financial institutions (e.g., World Bank, IFC) should take note that tying lending to green outcomes could have trade-offs; they should encourage holistic sustainability assessments.

For business managers: The findings highlight the importance of efficiency and stakeholder management in navigating policy shocks. Firms that invested in productivity (through innovation, better management) were less adversely affected in their CSR by GCP. This signals to managers that building a robust, efficient operation not only improves competitiveness but also provides slack to maintain CSR even when new regulations hit. Managers in high-pollution sectors should, paradoxically, double down on CSR communication and integration during challenging times, to avoid reputational damage from being seen as retracting their social commitments. Those in SOEs, in particular, might need to proactively maintain CSR to counteract the complacency that can arise from their safety nets.

While this study provides valuable insights, it has several limitations. First, the sample is limited to Chinese listed firms through 2020. Data beyond 2020 were not fully available (and 2021–2022 saw COVID-19 disruptions and new green policies). Future research could extend the timeframe as data emerge to see if the negative CSR effect persists, diminishes, or reverses with longer adjustment or with newer policy refinements (China updated green credit guidelines in 2020–2021). Second, our focus on listed firms means the findings may not directly generalize to private or smaller firms not captured in Hexun’s database. These unlisted companies might react differently or even more strongly (since they often have even tighter finances and lower CSR baselines). Future studies could use survey methods or case studies to examine GCP impacts on SMEs or privately-held polluters. Moreover, a limitation of our approach is the inherent difficulty in objectively measuring CSR. Although we use a widely recognized index (Hexun), sustainability metrics such as CSR or ESG are known to vary significantly across data providers, methodologies, and underlying assumptions. These measurement discrepancies can influence empirical findings and may obscure deeper behavioral patterns. Recent studies have discussed the fragmentation and subjectivity of ESG/CSR metrics in the literature (Clément et al., 2025; Drago et al., 2024; Kotsantonis and Serafeim, 2019). Future research may explore cross-validating CSR through a combination of text analysis, case interviews, and third-party surveys to provide a more comprehensive and objective portrayal of corporate social responsibility practices. Finally, our identification, like all DID analyses, assumes no other concurrent shocks systematically differ by treatment group. We addressed major ones (e.g., COVID, other green policies) as best as possible, but we cannot rule out all confounding events. Researchers might supplement our quantitative approach with qualitative research (interviews with firm managers, regulators) to validate the causal stories—understanding, for example, how decision-makers prioritized budgets under GCP constraints.

Despite these limitations, our study opens several avenues. One is exploring synergies between green finance and CSR incentives: for instance, do firms that had strong CSR cultures cope better with GCP because of stakeholder goodwill or innovation mindset? Another is cross-country analysis: comparing China’s experience with say, an Eastern European country’s green credit roll-out to see if similar patterns emerge, which could strengthen external validity of our conclusions.