Responding to Climate Policy Risk Through the Dynamic Role of Green Innovation: Evidence from Carbon Information Disclosure in Emerging Markets

Abstract

This study investigates how firms in emerging markets respond to climate policy risk, with a particular focus on the dynamic role of green innovation in shaping carbon information disclosure. Using a difference-in-differences (DID) framework, we examine the impact of China’s 2018 carbon reporting policy, which represents an institutionally significant but non-mandatory regulatory intervention, on the disclosure behaviors of A-share listed firms from 2013 to 2022. The results show that the policy significantly increased firms’ attention to carbon information disclosure, especially among those with limited green innovation capacity. In contrast, firms with stronger innovation capabilities exhibited more stable disclosure practices, suggesting a buffering effect against regulatory uncertainty. Further analysis reveals that the moderating effect of green innovation changes over time, as innovation-oriented firms gradually adjust their disclosure strategies in response to evolving policy expectations. These findings highlight green innovation as a key internal resource that enables firms to strategically adapt to climate policy risks. This study contributes to the literature on climate risk management and corporate sustainability by providing empirical evidence on how dynamic capabilities shape disclosure outcomes and risk management strategies under changing regulatory conditions.

1. Introduction

Amid intensifying global efforts to address climate change, firms are increasingly expected to contribute to low-carbon transitions while navigating complex and evolving regulatory landscapes (Chithambo et al. 2020). Climate-related policy uncertainty has emerged as a significant factor influencing corporate strategy, particularly in relation to environmental disclosure practices (Shahrour et al. 2024). In response, policymakers have adopted diverse regulatory tools to facilitate low-carbon transitions and promote environmentally responsible corporate conduct (Kortetmäki and Huttunen 2023). Within this framework, carbon information disclosure has gained prominence as a critical mechanism for enhancing transparency, addressing information asymmetry, and responding to stakeholder demands for sustainability (Runyu et al. 2025). The strategic significance of carbon information disclosure is especially pronounced in carbon-intensive industries, which are subject to heightened regulatory scrutiny and reputational risk (Rohani 2023). As a result, understanding how firms adjust disclosure practices under evolving regulations and policy uncertainty has become a central issue in environmental governance and corporate risk management.

While the effects of mandatory carbon disclosure regimes have been extensively explored in developed countries, there is limited research on how firms in emerging markets respond to non-mandatory environmental policies (Fu et al. 2023). China represents a particularly relevant context, where the institutional framework for carbon information disclosure remains fragmented and underdeveloped (Borghei 2021). Unlike financial reporting, which is subject to uniform standards and public disclosure requirements, carbon information is often submitted privately to regulatory authorities, without mandatory public dissemination. This lack of standardized disclosure obligations results in considerable variation in reporting practices and introduces ambiguity in how firms interpret and respond to policy expectations. Despite the growing regulatory emphasis on environmental performance, empirical evidence on how Chinese firms react to such substantive but non-compulsory policy interventions remains scarce, highlighting a critical gap in the literature.

In the Chinese context, most firms are not obligated to publicly disclose carbon emissions or related information. While environmental regulations such as the 2018 directive require key carbon-intensive industries to conduct internal carbon accounting and submit verified carbon emission data to the Ministry of Ecology and Environment, these requirements are limited to government reporting and do not involve public disclosure through annual reports or other investor communications (Shen et al. 2020). Whether firms choose to include carbon information in their public disclosures is left to their own decision. Some firms highlight carbon information disclosure to demonstrate social responsibility or align with evolving policy expectations. These choices represent firms’ strategic responses to environmental and reputational risks under regulatory uncertainty. The lack of unified carbon disclosure standards results in substantial variation, as firms respond differently depending on their perceived risks and strategic priorities.

Although firms are not legally required to disclose carbon information, they increasingly face institutional pressure from regulatory bodies, financial markets, and civil society stakeholders (Guo et al. 2022). Non-mandatory policy shocks create implicit expectations for environmental accountability, particularly for firms operating in carbon-intensive industries. In response to such signals, some firms choose to emphasize carbon information disclosure as a means to demonstrate compliance orientation and environmental responsibility, whereas others take a more conservative stance, limiting disclosure to reduce exposure to regulatory or reputational risks (O’Dwyer and Unerman 2020). Whether these policy interventions lead to substantive changes in disclosure behavior remains an open empirical question, especially in transitional institutional contexts such as China, where regulatory enforcement varies considerably across regions and industries, and where firms differ widely in their capabilities and incentives to respond.

China provides a timely and relevant context for examining how firms respond to environmental policy interventions under evolving institutional conditions. In 2018, the Ministry of Ecology and Environment (MEE) issued the Notice on Strengthening the Annual Carbon Emission Report, Verification, and Emission Monitoring Plan Development Work (MEE 2018). This policy required enterprises in designated industries to conduct detailed internal carbon accounting and submit verified emissions data to regulatory authorities. Although the directive did not mandate public disclosure, it represented a non-mandatory environmental policy shock that reinforced the government’s increasing focus on carbon information disclosure. The policy specifically targeted eight carbon-intensive industries, including paper and paper products (C22), petroleum and coal processing (C25), chemical raw materials and products (C26), non-metallic mineral products (C30), ferrous and non-ferrous metal smelting and rolling (C31, C32), electricity production (D44), and transportation (G56). These industries form the core of China’s industrial economy and contribute a significant portion of national carbon emissions, making them particularly suitable for assessing the effects of regulatory policy on firms’ attention to carbon information disclosure.

This study conceptualizes corporate responses to environmental regulation as outcomes shaped by the interaction between external policy pressures and internal organizational conditions. Rather than viewing disclosure behavior as a direct reaction to regulatory signals, this perspective considers how firms’ internal resources, institutional positioning, and technological preparedness influence their capacity to respond. A more integrated understanding of carbon information disclosure requires attention to how different internal functions, such as regulatory compliance, green innovation, disclosure planning, and stakeholder engagement, collectively shape sustainability-related decisions and practices. This interaction reflects a dynamic, context-dependent process by which firms adapt to environmental policy in broader institutional and market settings.

Green innovation refers to the development and implementation of technologies, products, or processes that mitigate environmental impacts while supporting long-term economic objectives (Zhu et al. 2021). As a strategic capability, it enables firms to respond proactively to environmental regulations by aligning innovation efforts with sustainability goals (Hsu et al. 2021). Within the dynamic capabilities framework, green innovation allows firms to reconfigure internal resources, adapt routines, and integrate external policy expectations into their operations (Liu et al. 2021). Recent research also highlights the importance of corporate social responsibility (CSR) and green dynamic capability as key internal mechanisms that stimulate green innovation in manufacturing firms. For instance, Yuan and Cao (2022) show that CSR enhances both green product and process innovation by strengthening green dynamic capabilities, which mediate the relationship between CSR engagement and green innovation outcomes. In the context of carbon information disclosure, firms with stronger green innovation capacity may be better positioned to interpret and act upon non-mandatory environmental policy signals, thereby reducing their exposure to regulatory and reputational risks. This capability becomes especially relevant in institutional environments where regulatory enforcement is uneven or ambiguous. In such contexts, green innovation influences how firms perceive both policy risks and strategic opportunities, ultimately shaping their disclosure behavior.

Although the theoretical significance of green innovation is well established, empirical research has seldom explored how it moderates the relationship between non-mandatory environmental policy shocks and corporate carbon information disclosure (Liu et al. 2021). This research addresses a critical and underexplored question: how green innovation shapes firm responses to non-mandatory environmental policy in emerging markets with uneven institutional enforcement. In addition, limited attention has been given to the temporal dimension of this moderating effect. It remains unclear whether the influence of green innovation on disclosure behavior remains stable or shifts as firms gradually adjust to new regulatory expectations. Addressing these questions requires an empirical approach that captures both the interaction between policy interventions and firm-level innovation capacity and the evolution of this relationship over time.

To address these research gaps, this study investigates the influence of a non-mandatory environmental policy shock, specifically China’s 2018 carbon reporting regulation, on firms’ attention to carbon information disclosure. It also examines whether firms with stronger green innovation capacity respond differently to such regulatory changes, and whether the moderating role of green innovation evolves over time following the implementation of the policy. These questions are explored through a DID empirical strategy, using panel data from A-share listed firms in China covering the period from 2013 to 2022. This timeframe spans a decade of intensifying climate governance both globally and domestically and reflects China’s progressive policy developments in response to international climate commitments, including the Paris Agreement, as well as the implementation of the 2018 carbon reporting directive. It allows for the analysis of how firms adapted their disclosure behavior under shifting environmental expectations and regulatory signals. Carbon information disclosure is measured through textual analysis of annual reports, capturing the frequency of keywords as a proxy for firms’ attention to carbon information disclosure. Green innovation is operationalized as the natural logarithm of the number of green invention and utility patents filed by each firm, representing its technological capability in environmental sustainability.

This study contributes to the literature on environmental regulation and corporate sustainability in several ways. First, it provides empirical evidence that non-compulsory yet institutionally significant environmental policies can influence firm behavior even in the absence of mandatory disclosure requirements. Second, it demonstrates that green innovation functions as a critical internal capability, shaping how firms interpret and respond to policy signals. Furthermore, it reveals that this moderating role of green innovation evolves over time, with innovation-oriented firms becoming increasingly proactive in their disclosure practices as the regulatory context matures. Methodologically, this study employs a difference-in-differences empirical strategy using panel data from A-share listed firms in China between 2013 and 2022. Carbon information disclosure is measured using textual analysis of annual reports, while green innovation is captured through the number of green invention and utility patents. This approach enables a robust assessment of the interaction between policy shocks and firm-level capabilities over time. Beyond its empirical and methodological contributions, this study also underscores the broader global significance of green innovation. It aligns with the United Nations Sustainable Development Goals (SDGs), particularly those related to climate action, responsible production, and innovation capacity building in emerging economies. By positioning green innovation as both a response to regulatory uncertainty and a pathway to sustainable growth, this study contributes to the ongoing conversation about corporate strategies for navigating environmental challenges in transitional institutional settings.

The remainder of the paper is structured as follows. Section 2 reviews the relevant literature and develops the research hypotheses. Section 3 outlines the research design, including data sources, variable definitions, and empirical strategy. Section 4 presents the main findings and robustness checks. Section 5 concludes with theoretical and policy implications and identifies directions for future research.

2. Literature Review

2.1. Environmental Policy Shocks and Carbon Information Disclosure

Environmental policy shocks have gained growing importance in corporate governance research, especially as climate regulations evolve from voluntary initiatives into more structured institutional frameworks (Park 2021). In regulatory contexts where disclosure mandates remain incomplete or evolving, firms often operate in environments characterized by ambiguous expectations and limited enforcement. Under such conditions, organizations must interpret broad regulatory signals and determine how, and to what extent, to respond through carbon information disclosure (Chopra et al. 2024). While these policies may not mandate public disclosure, they heighten institutional scrutiny and increase the visibility of carbon-related issues. As a result, policy signals can exert compliance pressure indirectly, even in the absence of binding mandates.

Compared to financial disclosures, which follow standardized formats and are subject to mandatory audits, carbon information disclosure in China lacks unified regulatory guidance and remains largely discretionary (Tang et al. 2023). Most listed companies are not legally obligated to disclose their carbon emissions in publicly available documents. Among those that do, disclosure styles vary considerably across firms, often reflecting strategic differences in how environmental narratives are presented (Tan et al. 2022). The absence of centralized requirements concerning the format, content, and level of detail contributes to significant variation across firms and industries (Zhang and Liu 2020). This regulatory incompleteness limits comparability and weakens the potential of carbon information disclosure as a tool for market-based environmental governance. It further enables firms to selectively interpret regulatory signals and tailor their disclosure strategies to fit internal priorities and perceived risks.

To further conceptualize these dynamics, this study draws on Scott (1995), who identifies three pillars that constitute institutional environments: regulatory, normative, and cognitive. The regulatory pillar encompasses formal rules and compliance mechanisms, such as government-issued carbon accounting directives. The normative pillar reflects the influence of social norms, stakeholder expectations, and legitimacy pressures from actors such as investors and media. The cognitive pillar captures shared beliefs and routines that shape how firms internalize environmental concerns and incorporate them into decision-making. Together, these institutional dimensions help explain how firms respond to non-compulsory policy signals, and why such responses vary based on internal capabilities such as green innovation.

Drawing on institutional theory, firm responses to regulatory change are shaped by a combination of coercive forces like government directives normative influences for industry norms, and mimetic behaviors such as peer imitation (Risi et al. 2023). Rather than passively complying, firms actively navigate institutional environments by aligning regulatory expectations with internal capabilities and external pressures. In the case of China’s 2018 carbon reporting policy, although public disclosure was not mandated, companies in carbon-intensive industries were required to carry out internal carbon accounting, develop emissions monitoring plans, and submit verified reports to environmental authorities. This regulatory action signaled a shift in institutional expectations, compelling firms to reconsider how they manage and communicate carbon issues. In response, some firms may seek to enhance their information disclosure as a way to strengthen legitimacy, prepare for future regulatory tightening, or demonstrate alignment with sectoral best practices. However, such responses are rarely uniform and often depend on a firm’s internal environmental governance structures, exposure to scrutiny, and perceived regulatory risks.

The 2018 policy introduced by the Ministry of Ecology and Environment presents a meaningful case through which to observe regulatory impact under transitional institutional conditions. Rather than imposing public disclosure obligations, the policy mandated internal carbon emission reporting for firms in carbon-intensive industries such as chemical production, metal smelting, electricity generation, and transportation. These industries account for a significant portion of China’s overall emissions and were therefore central to the policy’s objectives. While firms were not required to release their emissions data in annual reports or other public channels, the regulation created binding obligations for internal reporting and conveyed strong signals regarding the growing regulatory emphasis on carbon governance. Under these circumstances, firms may voluntarily increase carbon information disclosure as a strategic response, whether to build compliance credentials, appeal to investors, or position themselves for green financing opportunities (Steuer and Tröger 2022).

At the same time, disclosure may also carry risks, particularly for high-emission firms. Publicly highlighting carbon emissions may draw regulatory scrutiny, trigger critical media responses, or harm a firm’s reputation (Yadin 2023). Since the policy required reporting to government agencies but left public disclosure optional, companies retained flexibility in deciding how and whether to share carbon-related information with external audiences. This discretion gave rise to strategic variation in disclosure content, style, and visibility. Firms may emphasize certain aspects to project environmental responsibility while minimizing potential exposure to regulatory or reputational backlash. The resulting variation reflects how firms weigh competing concerns, including regulatory expectations, market perceptions, and legitimacy goals.

In this policy setting, carbon information disclosure functions as more than a compliance activity. It becomes a strategic tool for signaling environmental alignment, shaping corporate identity, and anticipating future regulatory developments. However, the behavioral effects of non-mandatory policies should not be presumed. If institutional pressure is weak or enforcement uneven, firms may ignore the policy altogether. In cases where expectations remain ambiguous, firms may resort to symbolic disclosures that lack substantive value. The overall impact of such policies thus depends not only on their content, but also on how they are interpreted, prioritized, and acted upon by firms in specific institutional contexts. This is particularly salient in emerging economies like China, where environmental policy implementation varies widely across industries and regions. These non-mandatory signals introduce strategic uncertainty, compelling firms to weigh potential compliance benefits against regulatory and reputational risks. Therefore, carbon information disclosure under such policy settings can be interpreted as a firm-level risk response strategy.

To evaluate whether non-compulsory regulatory interventions can produce observable behavioral shifts, this study examines whether firms in carbon-intensive industries increased their attention to carbon information disclosure following the introduction of the 2018 policy. Rather than assuming uniform behavioral responses, the analysis considers variation across firms and explores whether the regulation influenced disclosure patterns in measurable ways. This approach aims to advance understanding of how institutional pressures operate under transitional conditions and whether substantive change can emerge from non-mandatory policy shocks. Based on this, the following hypothesis is proposed:

H1. 

Environmental policy shocks positively and significant affect carbon information disclosure.

2.2. The Moderating Role of Green Innovation Between Environmental Policy Shocks and Carbon Information Disclosure

While environmental regulations may act as external drivers of corporate environmental behavior, firms’ actual responses often depend on their internal capacities (Zhang et al. 2020). One such capacity is green innovation, which refers to the development and implementation of environmentally sustainable technologies and processes. This organizational capability influences how firms interpret regulatory change and whether they choose to engage with evolving environmental expectations in a substantive way (Liu et al. 2025).

Firms with a high level of green innovation capability are generally more prepared to engage with regulatory demands (Zhao et al. 2025). They are more likely to possess the technical infrastructure, management systems, and knowledge resources required to monitor emissions and formulate disclosure strategies (Cheng and Shiu 2020). These internal advantages reduce the cost and complexity of complying with environmental policy and may increase firms’ willingness to respond to such policies in a proactive manner (Tu and Wu 2021). As a result, firms with greater green innovation output are often more inclined to participate in carbon information disclosure as part of a broader environmental commitment.

However, in policy contexts where disclosure is encouraged but not mandated, such as China’s 2018 carbon reporting policy, firms’ decisions are shaped not only by internal capacity but also by strategic assessments of risk and benefit. Even firms with advanced green innovation performance may choose to delay or limit disclosure if they perceive reputational, competitive, or regulatory risks (Wu et al. 2023). Concerns about revealing sensitive operational information or attracting excessive scrutiny can influence disclosure behavior. This suggests that green innovation, while enabling technical responsiveness, does not always lead to visible changes in disclosure practice (Yang et al. 2024b). Rather, it forms part of a broader strategic decision-making process shaped by institutional ambiguity, stakeholder expectations, and market considerations.

Firms with limited green innovation capacity are often perceived as lacking the tools or experience to respond meaningfully to environmental policy. They may face technical and organizational challenges in tracking emissions or communicating carbon-related information (Khanra et al. 2022). Nevertheless, these firms may respond more visibly to policy pressure under certain conditions. For example, when regulatory attention intensifies, disclosing carbon information can become a cost-effective way to demonstrate environmental awareness without requiring large-scale investments. In this case, disclosure serves as an accessible form of compliance signaling, particularly in institutional environments where formal enforcement remains inconsistent (Guo et al. 2023).

Prior studies have yet to fully account for the nuanced and sometimes counterintuitive relationship between green innovation and disclosure behavior. Although green innovation is often assumed to strengthen responsiveness to policy, this effect may vary depending on a firm’s strategic positioning, perceived risks, and the clarity of policy expectations. In transitional regulatory contexts, high levels of innovation do not always result in greater disclosure. In contrast, firms with limited green innovation may at times respond more actively in disclosure terms, using it as a symbolic or reputational tool in place of technical solutions. These complexities underscore the need for empirical examination of how green innovation modifies the relationship between policy shocks and disclosure outcomes. Based on this, the following hypothesis is proposed:

H2. 

Green innovation positively moderates the relationship between environmental policy shocks and carbon information disclosure.

2.3. The Dynamic Evolution of Green Innovation’s Moderating Effect

The impact of environmental regulatory interventions on corporate behavior is not static, but unfolds over time as institutions, market norms, and stakeholder expectations evolve (Gao and McDonald 2022). As firms encounter these shifting conditions, their evaluation of policy-related risks and opportunities also changes, prompting continuous adjustments in how they respond to external environmental pressures (Kobrin 2022). This suggests that the moderating influence of green innovation on policy-driven carbon information disclosure is likely to vary across different phases of policy implementation.

Building on dynamic capability theory, firms are understood as adaptive entities that modify internal structures and strategies in response to environmental changes (Yu et al. 2022). Green innovation, viewed as a dynamic organizational capability, can enhance a firm’s initial responsiveness to environmental regulation by facilitating the integration of sustainability objectives into operational and disclosure decisions (Qiu et al. 2020). During the early stages of a new policy, firms with higher levels of green innovation may be more likely to enhance their attention to carbon information disclosure. This responsiveness may be driven by their ability to interpret emerging regulatory signals, establish early legitimacy in sustainability domains, and differentiate themselves in markets oriented toward environmental performance (Hu et al. 2021).

However, this pattern may not apply universally. Firms with limited green innovation output may, under conditions of increasing institutional scrutiny, engage more actively in disclosure practices as a compensatory mechanism (Yang et al. 2024a). For these firms, providing visible carbon information may offer a relatively low-cost means of addressing regulatory expectations and signaling environmental awareness, even in the absence of underlying technical capacity.

Over time, as policy signals become embedded in sectoral routines and disclosure expectations begin to standardize, the role of green innovation may shift (Ren et al. 2023; Li et al. 2022). In earlier phases, innovation capability may be essential in shaping firm-level disclosure choices (Yuan and Cao 2022). But as disclosure becomes institutionalized, the distinctiveness of green innovation may diminish, reducing its marginal influence. At this point, it may operate less as a differentiating factor and more as one component of a broader compliance infrastructure.

Firm responses are also influenced by the evolving interactions between internal capabilities and external conditions, including stakeholder pressure, peer behavior, and regulatory enforcement. These changing dynamics suggest that the moderating effect of green innovation may not follow a consistent trajectory, but instead fluctuate in response to broader institutional developments (Yuan and Cao 2022; Lian et al. 2022). Static models are therefore insufficient to capture the full complexity of these relationships.

Although prior research has recognized the importance of green innovation, few studies have examined how its moderating role evolves over time. The dominant assumption has been one of temporal stability, overlooking the fluidity of both regulatory environments and internal firm resources. This oversight is particularly notable in emerging markets such as China, where institutional frameworks are still in transition and corporate responses remain highly adaptive. A dynamic perspective is essential for understanding how regulatory influence emerges, stabilizes, or weakens over time. Without accounting for temporal variation, assessments of green innovation’s impact risk misrepresenting the complex and evolving nature of firm-policy interactions.

Carbon information disclosure is embedded in this broader process of institutional transformation. Firms do more than react to policy, they also contribute to shaping new norms and standards for environmental conduct. Green innovation can function both as a mechanism for technical adaptation and as a means for influencing disclosure norms and expectations across industries. This dual role highlights the importance of analyzing not only whether green innovation moderates the relationship between regulatory pressure and disclosure, but also how this influence evolves across different stages of policy implementation and institutional development. Based on this, the following hypothesis is proposed:

H3. 

The moderating role of green innovation in the relationship between environmental policy shocks and carbon information disclosure changes dynamically over time.

3. Methodology

3.1. Research Design

3.1.1. Identification Strategy

This study employs a difference-in-differences (DID) strategy to identify the causal effect of a non-mandatory environmental policy shock on firms’ carbon information disclosure. The DID framework compares changes in disclosure behavior between a treatment group, consisting of firms likely to be affected by the policy, and a control group, consisting of firms less directly influenced, before and after the implementation of the policy.

The policy shock under investigation is the Notice on Strengthening the 2018 Annual Carbon Emission Report, Verification, and Emission Monitoring Plan Development Work, issued by China’s Ministry of Ecology and Environment (MEE). While the policy did not mandate public carbon information disclosure, it introduced substantially stricter internal requirements for carbon accounting, third-party verification, and emissions monitoring. These institutional pressures were primarily directed at enterprises in target carbon-intensive industries (MEE 2018). Although public disclosure was not mandated, the policy created a regulatory environment that heightened firms’ perception of climate-related policy risk, especially in carbon-intensive industries.

Accordingly, the treatment group is defined based on industry classification. Firms operating in eight target carbon-intensive industries which contain paper and paper products (C22), petroleum and coal processing (C25), chemical raw materials and products (C26), non-metallic mineral products (C30), ferrous and non-ferrous metal smelting and rolling (C31, C32), electricity (D44), and transportation (G56), which are considered subject to higher regulatory pressure and are therefore assigned to the treatment group. These industries were explicitly targeted in the 2018 policy and are recognized as key contributors to China’s overall carbon emissions.

This study constructed a dummy variable that takes the value of 1 if a firm operates in any of the eight target carbon-intensive industries, and 0 otherwise. This indicator captures the policy’s sectoral scope and enables the empirical model to distinguish between firms likely subject to the environmental intervention and those largely unaffected. The control group includes firms outside the targeted sectors, which are assumed to have experienced limited regulatory pressure in the post-policy period. By comparing the differential changes in carbon information disclosure between the treatment and control groups before and after the policy’s introduction, the DID design isolates the effect of the 2018 environmental policy shock under the assumption of parallel trends.

3.1.2. Model Specification

This study aims to evaluate the effect of the 2018 policy on firms’ carbon information disclosure, specifying the following baseline difference-in-differences (DID) model:

$${CID}_{i,t}=\alpha +{\beta }_1{(Treat}_i\times {Post}_t)+\gamma {Controls}_{i,t}+\sum Firm+\sum Year+{\epsilon }_{i,t}$$

(1)

In this model, ${CID}_{i,t}$ represents the carbon information disclosure from firm i to year t which measured by the frequency of keywords in annual reports. The key explanatory variable is the interaction term ${DID}_{i,t}$ ${(Treat}_i\times {Post}_t)$, ${Post}_t$ is a dummy variable equal to 1 if the year is 2018 or later, and ${Treat}_i$ equals 1 if the firm operates in eight target carbon-intensive industry as defined in Section 3.1.1. ${Controls}_{i,t}$ including Firm size (Size), Solvency (Lev), Liquidity (Liquid), Development ability (Growth), Profitability (Loss), Size of board (Board), Governance Capability (Indep), External monitoring ability (IShare), Financial Transparency (Big4), and Reliability (Opinion). This formula also added firm and year fixed effects. The coefficient ${\beta }_1$ captures the average treatment effect of the policy shock.

To test Hypothesis 2 (H2), this paper examines whether green innovation (GI) moderates the effect of the policy shock. The baseline model is extended to include GI, and an interaction term between the DID indicator and green innovation:

$${CID}_{i,t}=\alpha +{\beta }_1({Treat}_i\times {Post}_t)+{\beta }_2{GI}_{i,t}+{\beta }_2({Treat}_i\times {Post}_t\times {GI}_{i,t})+\gamma {Controls}_{i,t}+\sum Firm+\sum Year+{\epsilon }_{i,t}$$

(2)

Here, ${DID}_{i,t}$ $=({Treat}_i\times {Post}_t)$, and ${GI}_{i,t}$ denotes the level of green innovation, measured as the natural logarithm of one plus the number of green patents (invention and utility model) granted to firm i to year t. A positive and significant coefficient on the interaction term $({Treat}_i\times {Post}_t\times {GI}_{i,t})$ would indicate that firms with greater green innovation are more responsive to policy shocks, thereby providing evidence for the moderating effect of GI.

To evaluate Hypothesis 3 (H3), this paper investigates whether the moderating effect of green innovation varies over time. This paper incorporates a set of interaction terms between Treat, GI, and individual year dummies for each post-policy year. The model is specified as follows:

$${CID}_{i,t}=\alpha +{\beta }_1{GI}_{i,t}+\sum _{k=2018}^{2022}{\beta }_k({Treat}_i\times {Post}_t\times {GI}_{i,t})+\gamma {Controls}_{i,t}+\sum Firm+\sum Year+{\epsilon }_{i,t}$$

(3)

In this formula, ${Year}_k$ is a dummy variable equal to 1 for year k, allowing for the estimation of time-varying moderating effects of GI in each post-policy year. This dynamic formulation provides insights into whether the influence of green innovation on policy-driven disclosure behavior strengthens, weakens, or stabilizes over time. A pattern of increasing ${\beta }_k$ coefficients would suggest that firms with higher GI capabilities are increasingly proactive in response to sustained regulatory pressure. All models are estimated using high-dimensional fixed effects regressions with firm and year fixed effects, and standard errors clustered at the firm level.

All regressions are estimated using firm and year fixed effects to control for unobservable heterogeneity, and robust standard errors clustered at the firm level to address potential autocorrelation and heteroskedasticity. The analyses are conducted using Stata 17.0.

To determine the most suitable estimation method (

Table A1. F Test, LM Test, and Hausman Test.

Test Type	Statistic	p-Value	Conclusion
F Test	387.85	0.000	FE model better than OLS model
LM Test	4795.26	0.000	RE model better than OLS model
Hausman Test	830.44	0.000	FE model better than RE model

), this study conducts an F test, an LM test, and a Hausman test to compare the effectiveness of ordinary least squares (OLS), random effects (RE), and fixed effects (FE) models. The results, reported in Table A1, indicate that the fixed effects model is the most appropriate choice. The F test yields a statistic of 387.85 (p = 0.0000), confirming that FE is superior to OLS in capturing firm-specific and time-specific variations. The LM test (4795.26, p = 0.0000) suggests that RE better accounts for firm heterogeneity than OLS. However, the Hausman test (830.44, p = 0.0000) strongly favors FE over RE due to the correlation between firm-specific effects and explanatory variables.

Based on these findings, this paper uses the fixed effects model to control for both firm-level and year-specific heterogeneity, reducing omitted variable bias and enhancing the robustness of the analysis. This choice strengthens the explanatory power of the empirical results, ensuring a more precise investigation of the relationships among the key variables in this study.

3.2. Variable Measurement

Table 1. Variable measurement.

Variables		Symbol	Descriptions
Dependent Variable	Carbon information disclosure	CID	Ln (1 + count of keyword frequency), based on textual analysis
Independent Variable	Policy Shock	DID	Treat × Post, Treat target carbon-intensive industries plus 1; Post years from 2018 onward plus 1 and 0 for pre-2018 years.
Moderator Variable	Green innovation	GI	Ln (1 + count of green inventions with independent applications and green utility models with independent applications)
Control Variable	Size	Size	Ln (total assets)
	Profitability	ROA	Return on Assets
	Solvency	Lev	Total liabilities/total assets
	Liquidity	Liquid	Current assets/current liabilities
	Development ability	Growth	Growth Rate of Net Profit, (Current Period Net Profit-Last Period Net Profit)/Last Period Net Profit
	Loss Status	Loss	Dummy: 1 if net income < 0
	Size of Board	Board	Ln (number of directors)
	Governance Capability	Indep	Ratio of independent directors
	External Monitoring Ability	IShare	Institutional shares/total shares
	Financial Transparency	Big4	Dummy: 1 if audited by Big 4
	Reliability	Opinion	Dummy: 1 if unqualified opinion

presents the definitions and measurement methods for all variables used in this study. The dependent variable, carbon information disclosure (CID) was measured using Python 3.10 based textual analysis of firms’ annual reports. Specifically, the frequency of predefined climate-related keywords is calculated for each firm-year observation. To address zero counts and improve distributional properties, one is added to the raw count, and the natural logarithm of this value is taken. This method captures the relative attention firms allocate to climate-related issues in their carbon information disclosures (Liu et al. 2025).

The independent variable policy shock (DID) is constructed as the interaction term between Treat and Post, where Treat is a dummy variable equal to 1 if the firm operates in a carbon-intensive industry (as defined in Section 3.1.1), and Post is a dummy variable equal to 1 for observations from 2018 onwards. This specification follows the standard difference-in-differences (DID) design to identify the effect of the 2018 environmental policy intervention.

The moderating variable green innovation (GI) is measured by the natural logarithm of the number of green invention patents with independent applications and green utility model patents with independent applications in the current year plus 1 (Li et al. 2022). This measure captures the firm’s green innovation output and reflects its technological capability in environmental sustainability.

A set of control variables is included to account for firm-level characteristics that may influence carbon information disclosure behavior (Liu et al. 2025; Hu et al. 2021; Li et al. 2022; Lu and Li 2023). These include firm size (Size), measured as the natural logarithm of total assets; profitability (ROA), calculated as net income divided by total assets; and loss status (Loss), coded as 1 if the firm reports a negative net profit. Financial characteristics include leverage (Lev), defined as total liabilities over total assets, and liquidity (Liquid), measured by the current ratio. Corporate governance variables include board size (Board), expressed as the logarithm of the number of directors, and board independence (Indep), measured by the proportion of independent directors. Institutional ownership (IShare) is measured as the shareholding percentage held by institutional investors. Financial disclosure quality is represented by a dummy variable (Big4), equal to 1 if the firm is audited by a Big Four accounting firm. In addition, audit opinion (Opinion) is set to 1 if the firm received an unqualified audit opinion. These controls align with the established literature on corporate governance and environmental disclosure, helping to isolate the effects of policy and innovation on carbon disclosure behavior.

3.3. Sample and Data Sources

This study utilizes a panel dataset of A-share listed firms on the Shanghai and Shenzhen Stock Exchanges, covering the ten-year period from 2013 to 2022. This time span includes both the pre- and post-policy periods surrounding the 2018 environmental regulatory intervention. Data are compiled from multiple authoritative sources: financial and firm-level variables are obtained from the China Stock Market and Accounting Research (CSMAR) database; corporate governance indicators are sourced from CNRDS; and green innovation data are drawn from the China National Intellectual Property Administration (CNIPA), based on patent applications.

A-share listed firms are selected because they are subject to uniform disclosure regulations and reporting standards under the supervision of the China Securities Regulatory Commission (CSRC) (Su and Alexiou 2023). These firms are more publicly visible, financially transparent, and more frequently targeted by environmental regulatory initiatives, including internal carbon accounting and climate risk reporting. Their institutional characteristics make them especially relevant for analyzing how firms respond to substantial but non-mandatory environmental policies. This focus also ensures data consistency and comparability, which is essential for conducting textual analysis of annual reports.

To ensure data quality and comparability, a rigorous filtering process is applied. Firms in the financial and utilities sectors are excluded due to their distinct regulatory environments and disclosure requirements. Additionally, firms classified as *ST (Special Treatment) or ST (Potential Special Treatment) are removed to exclude financially distressed companies that may exhibit abnormal disclosure behavior. Observations with missing values in key financial or disclosure variables are also eliminated. After applying these criteria, the final sample comprises 1753 firms over a ten-year period, offering a representative dataset of publicly listed companies in China, particularly those operating in target carbon-intensive industries.

All continuous variables are winsorized at the 1st and 99th percentiles to mitigate the influence of outliers and improve the robustness of regression estimates. The resulting dataset integrates detailed firm characteristics, financial performance metrics, green innovation indicators based on patent activity, and carbon information disclosure measures derived from textual analysis. This comprehensive dataset enables a robust empirical examination of how non-mandatory environmental policy shocks affect corporate carbon information disclosure and how green innovation moderates this relationship over time.

3.4. Descriptive Statistics and Correlation

Table 2. Descriptive statistics.

	CID	DID	GI	Size	ROA	Lev	Liquid
Mean	1.339	0.084	0.921	22.589	0.038	0.436	2.176
SD	1.081	0.278	1.204	1.298	0.056	0.196	1.927
25%	0	0	0	21.665	0.013	0.282	1.135
50%	1.386	0	0	22.416	0.034	0.431	1.588
75%	2.079	0	1.609	23.357	0.064	0.583	2.450
Min	0	0	0	20.193	−0.170	0.064	0.366
Max	3.989	1	4.812	26.250	0.202	0.862	12.422
	Growth	Loss	Board	Indep	IShare	Big4	Opinion
Mean	0.140	0.106	2.130	37.699	45.803	0.072	0.978
SD	0.336	0.308	0.196	5.506	23.912	0.258	0.146
25%	−0.033	0	1.946	33.330	28.212	0	1
50%	0.090	0	2.197	36.360	47.786	0	1
75%	0.236	0	2.197	42.860	64.565	0	1
Min	−0.508	0	1.609	33.330	0.534	0	0
Max	1.818	1	2.639	57.140	90.973	1	1

presents the descriptive statistics for the key variables used in this study, including carbon information disclosure (CID), policy treatment (DID), green innovation (GI), and control variables. The mean of CID is 1.339, with a standard deviation of 1.081, which suggests significant variation in firms’ attention to carbon information disclosure. The median CID value is 1.386 indicates that while some firms actively disclose carbon information, others have minimal or no disclosure, as reflected in the minimum value of 0.

The DID variable has a mean of 0.084, reflecting that approximately 8.4% of the sample firms are subject to the policy treatment, which belong to target carbon-intensive industries after 2018. The mean of GI is 0.921, with a maximum of 4.812, indicating that while some firms exhibit substantial green innovation output, a significant proportion have no green innovation output. This is further reflected in the 25th and 50th percentiles, both of which are zero, suggesting considerable heterogeneity in firms’ green innovation capability.

Among the control variables, the mean of Size is 22.589, with values ranging from 20.193 to 26.250, which suggests moderate variation in firm scale. ROA has an average of 3.8%, though it ranges from −17.0% to 20.2%, which indicates that some firms experience significant financial distress. Lev averages 43.6%, while Liquid has a mean of 2.176, which suggests variability in firms’ capital structures and financial stability.

Regarding corporate governance characteristics, the mean of Board is 2.130, while Indep account for an average of 37.7% of the board. IShare has a mean of 45.8%, which reflects substantial institutional investor participation. In terms of audit-related variables, only 7.2% of firms are audited by a Big Four firm (Big4 = 1), while 97.8% receive an unqualified audit opinion (Opinion = 1). The descriptive statistics suggest that firms in the sample exhibit heterogeneous levels of carbon information disclosure, financial performance, and governance structures, providing a strong foundation for empirical analysis.

Figure 1 visually represents these correlations using a heatmap, where color intensity indicates the strength and direction of correlations (red for negative correlations, blue for positive correlations).

The results show that DID is positively correlated with CID (r = 0.356), suggesting that firms affected by the 2018 policy shock tend to disclose more carbon-related information. Green innovation (GI) is also positively correlated with CID (r = 0.271), implying that firms with greater green patent activity are more likely to engage in carbon information disclosure.

Among the control variables, Size is moderately correlated with CID (r = 0.298) and DID (r = 0.124), indicating that larger firms are more likely to be subject to policy interventions and engage in voluntary disclosure. Lev exhibits a weak positive correlation with CID (r = 0.124) but a strong negative correlation with ROA (r = −0.314), reflecting the typical trade-off between financial leverage and profitability. Liquid is negatively correlated with CID (r = −0.174), Size (r = −0.360), and Lev (r = −0.655), which suggests that firms with higher liquidity ratios tend to be smaller and less leveraged.

Corporate governance variables, such as Indep and IShare, show weak correlations with CID, which indicates that corporate governance mechanisms may have a limited direct impact on carbon information disclosure decisions. Big4 is positively correlated with GI (r = 0.171) and CID (r = 0.063), suggesting that firms with better audit quality may also engage more in environmental transparency.

The correlation heatmap in Figure 1 provides a visual representation of these relationships. The overall pattern suggests no severe multicollinearity issues, as no correlations exceed the commonly accepted threshold of 0.7.

In order to assess multicollinearity, a Variance Inflation Factor (VIF) test was conducted (

Table A2. Multicollinearity test.

Variables	VIF	1/VIF
Lev	2.45	0.408677
Size	2.27	0.44118
ROA	2.06	0.484388
Liquid	1.77	0.563495
Loss	1.71	0.58366
Board	1.62	0.616246
Indep	1.49	0.671496
IShare	1.37	0.730055
GI	1.23	0.81584
Big4	1.19	0.838157
Growth	1.1	0.907018
Opinion	1.05	0.954946
DID	1.03	0.969206
Mean VIF	1.57

), confirming that all VIF values remain below 10, Mean VIF is 1.57. These results ensure the robustness and reliability of the regression model.

4. Result

4.1. Baseline Regression Analysis

Table 3. Baseline regression analysis.

	(1)	(2)	(3)	(4)	(5)	(6)
	CID	CID	CID	CID	CID	CID
DID	1.3851 ***	0.2200 ***	0.2086 ***	0.2088 ***	0.2077 ***	0.2080 ***
	(0.0405)	(0.0354)	(0.0357)	(0.0357)	(0.0357)	(0.0356)
Size			0.0784 ***	0.0768 ***	0.0689 **	0.0726 **
			(0.0283)	(0.0284)	(0.0286)	(0.0285)
ROA			0.3632 **	0.1915	0.1626	0.1491
			(0.1495)	(0.1973)	(0.1981)	(0.1982)
Lev			−0.0820	−0.0809	−0.0708	−0.0681
			(0.0982)	(0.0982)	(0.0984)	(0.0984)
Liquid			0.0017	0.0018	0.0012	0.0011
			(0.0065)	(0.0066)	(0.0066)	(0.0066)
Growth				0.0066	0.0022	0.0008
				(0.0164)	(0.0165)	(0.0165)
Loss				−0.0370	−0.0378	−0.0349
				(0.0259)	(0.0259)	(0.0259)
Board					−0.0281	−0.0230
					(0.0889)	(0.0890)
Indep					0.0016	0.0016
					(0.0024)	(0.0024)
IShare					0.0016 *	0.0016 *
					(0.0009)	(0.0009)
Big4						−0.1371 **
						(0.0684)
Opinion						0.0746
						(0.0470)
Year FE	NO	YES	YES	YES	YES	YES
Id FE	NO	YES	YES	YES	YES	YES
N	17,530	17,530	17,530	17,530	17,530	17,530
Adj. R2	0.1266	0.6987	0.6996	0.6996	0.6997	0.7

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the baseline regression results examining the impact of environmental policy shocks (DID) on corporate carbon information disclosure (CID). The analysis covers a ten-year panel dataset from year 2013 to 2022 of 1753 firms, ensuring statistical reliability. The regression models progressively incorporate firm and year fixed effects, firm characteristics, financial performance indicators, and governance factors, controlling for potential confounding variables that could bias the estimated effect of DID on CID. Across all specifications, DID remains consistently positive and significant, confirming that the 2018 environmental policy shock significantly increases firms’ attention to carbon information disclosure, supporting Hypothesis 1.

Column (1) presents the univariate regression results without controlling for firm- or time-fixed effects. The coefficient of DID (β = 1.3851, p < 0.01) suggests a strong positive correlation between policy exposure and carbon information disclosure. However, this estimate likely suffers from omitted variable bias, as firms affected by the policy may have different baseline disclosure behaviors due to regulatory pressure, stakeholder scrutiny, or firm characteristics.

Column (2) introduces firm and year fixed effects, substantially altering the results. The DID coefficient drops to 0.2200 (p < 0.01), though it remains statistically significant. The adjusted R² increases sharply to 0.6987, indicating that unobserved firm-specific and macroeconomic factors previously confounded the estimates. This substantial change highlights the necessity of controlling for firm-level heterogeneity and time trends, which may otherwise distort the measured impact of policy shocks on carbon information disclosure.

Columns (3) to (5) sequentially introduce control variables to ensure that the observed effect of DID on CID is not driven by omitted firm-specific characteristics. The coefficient of DID remains highly significant across all models, stabilizing around 0.2077 to 0.2086 (maintain p < 0.01), confirming the robustness of the policy effect.

Column (6) presents the full model, where DID (β = 0.2080, p < 0.01) remains significant even after accounting for all firm-level characteristics, governance factors, and financial indicators. The adjusted R² of 0.7 suggests that the model explains a substantial proportion of the variation in corporate carbon information disclosure, providing strong empirical support for the causal link between policy shocks and the attention level of carbon information disclosure.

These findings confirm Hypothesis 1, aligning with institutional theory, which posits that regulatory pressures compel firms to adopt legitimacy-seeking behaviors. In the context of China’s evolving environmental policies, the results suggest that government-led interventions effectively drive corporate transparency on carbon-related issues, reinforcing the role of regulatory mechanisms in shaping sustainable business practices.

In order to ensure the validity of the difference-in-differences (DID) estimation, it is essential to verify the parallel trend assumption, which requires that before the policy intervention, the treatment and control groups followed similar carbon information disclosure trends. Figure 2 presents an event study analysis, plotting the estimated coefficients for the pre- and post-policy periods to assess whether the disclosure behaviors of treated and control firms were comparable before 2018.

The results indicate that before 2018 (pre-treatment period), the estimated coefficients fluctuate around zero and are statistically insignificant, confirming that treated and control firms exhibited no systematic differences in carbon information disclosure trends before the policy shock. This supports the parallel trend assumption, suggesting that any post-policy divergence in disclosure behavior is attributable to the regulatory intervention rather than pre-existing trends.

After the policy implementation in 2018, there is a sharp and sustained increase in CID among treated firms, with statistically significant positive coefficients in 2019 and beyond. This result indicates that the 2018 regulatory policy effectively stimulated increased corporate attention to carbon information disclosure, aligning with the theoretical expectations of institutional pressure and compliance incentives.

Additionally, 2017 is excluded from the analysis due to the potential anticipation effect, where firms may have adjusted their disclosure behavior in response to regulatory expectations before the formal policy implementation. Dropping 2017 ensures that the pre-treatment period reflects firms’ natural disclosure patterns, preventing policy-driven adjustments from contaminating the control group. Therefore, this paper drops the pre1 (2017) variable to avoid multicollinearity (Bayman and Dexter 2021).

Overall, the event study results provide strong empirical support for the validity of the DID framework, confirming that the observed post-policy increase in carbon information disclosure is causally attributable to the policy shock rather than pre-existing disclosure trends or external confounders.

4.2. The Moderating Role of Green Innovation (GI) on Policy Shocks and Carbon Information Disclosure

Theoretically, green innovation is expected to enhance firms’ responsiveness to environmental regulations by equipping them with the technological capabilities and institutional readiness necessary for sustainability reporting (Chin et al. 2022). Firms with stronger green innovation capacity may voluntarily engage in carbon information disclosure as a strategic response to regulatory and market expectations, while less innovative firms may rely more on external regulatory pressure to adjust their disclosure behaviors (Tao et al. 2024). Consequently, if green innovation plays a positive moderating role, firms with higher levels of green innovation should exhibit a stronger response to policy shocks in terms of increased carbon information disclosure compared to those with lower innovation levels.

Table 4. Moderation effect of green innovation on policy shocks and carbon information disclosure.

	(1)	(2)	(3)	(4)
	CID	CID	CID	CID
DID	0.2200 ***	0.2080 ***	0.2028 ***	0.1752 ***
	(0.0354)	(0.0356)	(0.0356)	(0.0448)
Size		0.0726 **	0.0580 **	0.0580 **
		(0.0285)	(0.0285)	(0.0285)
ROA		0.1491	0.1792	0.1764
		(0.1982)	(0.1968)	(0.1969)
Lev		−0.0681	−0.0634	−0.0626
		(0.0984)	(0.0981)	(0.0981)
Liquid		0.0011	0.0011	0.0011
		(0.0066)	(0.0065)	(0.0065)
Growth		0.0008	0.0014	0.0013
		(0.0165)	(0.0165)	(0.0165)
Loss		−0.0349	−0.0323	−0.0329
		(0.0259)	(0.0257)	(0.0257)
Board		−0.0230	−0.0250	−0.0248
		(0.0890)	(0.0884)	(0.0883)
Indep		0.0016	0.0015	0.0014
		(0.0024)	(0.0024)	(0.0024)
IShare		0.0016 *	0.0015 *	0.0015 *
		(0.0009)	(0.0009)	(0.0009)
Big4		−0.1371 **	−0.1408 **	−0.1416 **
		(0.0684)	(0.0672)	(0.0672)
Opinion		0.0746	0.0695	0.0699
		(0.0470)	(0.0468)	(0.0468)
GI			0.0422 ***	0.0393 ***
			(0.0098)	(0.0103)
DID_GI				0.0223
				(0.0211)
Constant	1.3209 ***	−0.4387	−0.1336	−0.1312
	(0.0030)	(0.6618)	(0.6591)	(0.6593)
Year FE	YES	YES	YES	YES
Id FE	YES	YES	YES	YES
N	17,530	17,530	17,530	17,530
Adj. R2	0.6987	0.7	0.7006	0.7006

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the regression results examining whether green innovation moderates the effect of environmental policy shocks on corporate carbon information disclosure. Column (1) shows the baseline difference-in-differences (DID) model. DID (β = 0.2200, p < 0.01) is significant at 1% level, indicating that firms subject to the 2018 policy intervention significantly increased their carbon information disclosure levels.

In Column (2), DID (β = 0.2080, p < 0.01) is significant at the 1% level, confirming the robustness of the baseline result. Among control variables, Size (β = 0.0726, p < 0.05) and IShare (β = 0.0016, p < 0.1), suggesting that larger firms and those under greater investor scrutiny are more inclined attention to carbon information. Interestingly, Big4 (β = –0.1371, p < 0.05), possibly reflecting a more conservative approach to voluntary environmental reporting under stricter audit supervision.

In Column (3), GI (β = 0.0422, p < 0.01) is significant at the 1% level, suggesting that firms with greater green innovation capability are more likely to engage in carbon information disclosure, even in the absence of direct regulatory pressure.

However, in Column (4), DID × GI (β = 0.0223), but is statistically insignificant. This implies that, across the full sample, green innovation does not significantly strengthen the effect of policy shocks on disclosure behavior.

This finding diverges from previous expectations and suggests the presence of heterogeneity in firms’ responses. One possible explanation is that firms with higher green innovation were already disclosing climate-related information prior to the policy shock, thus exhibiting limited marginal change in response to the new regulation. By contrast, firms with lower or no green innovation capacity may have responded more noticeably to the policy, either to compensate for their technological gaps or to signal environmental legitimacy under increasing institutional pressure.

Furthermore, these results point to a non-uniform moderating effect of green innovation and suggest the need to further explore whether the effect varies across firms with different levels of innovation capacity. Accordingly, grouping analysis based on green innovation groups will be conducted.

To explore this possibility, the sample is divided into three distinct groups based on firms’ green innovation levels. Given that a substantial proportion of firms report no green innovation output, these firms are categorized as NO_GI (GI = 0). Among firms with positive green innovation, the sample is further divided using the 50th percentile of the GI distribution. Firms with GI values between 0 and 1.609438 are classified as LOW_GI, while those with GI values above 1.609438 are categorized as HIGH_GI. This classification captures meaningful variation in firms’ technological engagement with sustainability and enables more refined analysis of heterogeneity in policy responsiveness (Hao et al. 2022).

Table 5. Heterogeneous effects of policy shocks on carbon information disclosure across different levels of green innovation.

Group	No_GI	Low_GI	High_GI
	(1) CID	(2) CID	(3) CID
DID	0.3024 ***	0.0957	0.0750
	(0.0509)	(0.0622)	(0.0716)
Size	0.0353	−0.0123	0.0815
	(0.0360)	(0.0518)	(0.0600)
ROA	0.0763	0.3295	0.0076
	(0.2519)	(0.4025)	(0.4652)
Lev	−0.0813	0.0711	0.1164
	(0.1260)	(0.1895)	(0.2523)
Liquid	−0.0041	0.0313 **	0.0485 *
	(0.0072)	(0.0131)	(0.0262)
Growth	0.0004	0.0113	−0.0110
	(0.0199)	(0.0390)	(0.0441)
Loss	−0.0224	−0.0247	−0.0571
	(0.0328)	(0.0566)	(0.0591)
Board	0.1079	−0.0485	0.0600
	(0.1074)	(0.1751)	(0.1753)
Indep	0.0047	−0.0017	−0.0014
	(0.0030)	(0.0050)	(0.0046)
IShare	0.0017	0.0003	0.0020
	(0.0011)	(0.0015)	(0.0019)
Big4	−0.1855 **	−0.3037 *	−0.0519
	(0.0885)	(0.1844)	(0.0853)
Opinion	0.0564	0.0959	0.0938
	(0.0639)	(0.0971)	(0.0980)
Constant	−0.2149	17,018	−0.5250
	(0.8112)	−12,021	−13,915
Year FE	YES	YES	YES
Id FE	YES	YES	YES
N	8813	4094	3956
Adj. R2	0.6688	0.7047	0.7373

Standard errors in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the regression results from the sub-sample analysis across these three groups. This approach facilitates a more nuanced understanding of how the effect of the 2018 environmental policy shock on carbon information disclosure differs depending on firms’ underlying levels of green innovation.

The results reveal notable heterogeneity. In the NO_GI group (Column 1), the coefficient of the policy shock (DID) is positive and highly significant (β = 0.3024, p < 0.01), suggesting that firms without any green innovation output responded most strongly to the policy intervention by increasing their carbon information disclosure. This result indicates that even in the absence of technological innovation, regulatory signals can induce meaningful behavioral change, possibly because these firms relied more heavily on disclosure as a symbolic or compensatory mechanism.

In contrast, the DID coefficient for the LOW_GI group (β = 0.0957, p > 0.1) is smaller and not significant, and the effect further diminishes in the HIGH_GI group (β = 0.0750, p > 0.1). These results suggest that firms with greater green innovation output were less responsive to the policy shock in terms of disclosure behavior. One possible explanation is that high-GI firms had already adopted proactive disclosure practices prior to the policy change, thus exhibiting limited incremental response. Additionally, firms with stronger innovation capabilities may have chosen to signal environmental performance through technological outputs rather than increased disclosure.

In addition, Figure 3 illustrates the heterogeneous effects of the 2018 environmental policy shock (DID) on carbon information disclosure (CID) across firms with different levels of green innovation capability. The Y-axis represents the estimated coefficient of DID on CID, while the X-axis categorizes firms into three groups: No_GI group, Low_GI group, and High_GI group.

The results show that firms without green innovation output exhibit the strongest and most statistically significant response to the policy shock, with a point estimate close to 0.30 and a relatively narrow confidence interval. This suggests that even without technical capabilities, such firms may rely on disclosure as a visible, symbolic response to external policy pressure. By contrast, firms in the Low_GI and High_GI groups demonstrate smaller and statistically insignificant disclosure responses. The wider confidence intervals for these groups indicate greater variability and reduced precision in the estimated effects. This may imply that firms with more innovation capability are either less sensitive to non-mandatory regulatory signals or may already have internal systems in place, thus dampening the need for visible shifts in disclosure behavior.

Overall, findings from Table 5 and Figure 3 challenge the conventional expectation that green innovation uniformly enhances policy responsiveness. Instead, they indicate that firms with no green innovation, those typically viewed as less environmentally capable, were the most reactive to regulatory pressure, possibly due to reputational concerns or a perceived need to comply symbolically in a shifting institutional environment.

This heterogeneity underscores the importance of considering firms’ baseline innovation status when evaluating the moderating role of green innovation. The observed pattern also raises further questions about whether the effect of green innovation is constant over time, which motivates the subsequent analysis of its dynamic moderating effect.

4.3. The Dynamic Moderating Effect of Green Innovation (GI) on Policy Shocks and Carbon Information Disclosure

In order to further explore whether the moderating role of green innovation evolves over time, this section estimates a dynamic difference-in-differences specification by interacting green innovation with year-specific policy shock indicators. Figure 4 illustrates the estimated coefficients for these year-by-year interactions (DID × GI), capturing the temporal pattern of green innovation’s moderating effect on the relationship between environmental policy shocks and carbon information disclosure from 2018 to 2022.

The results reveal a clear dynamic trajectory in Figure 4 and

Table A3. Dynamic moderating effect of green innovation (yearly interaction coefficients).

Year	Coefficient (β)	Std. Error	p-Value
2018	0.0635	0.0171	0.000
2019	0.0757	0.0210	0.000
2020	0.1216	0.0187	0.000
2021	0.1289	0.0217	0.000
2022	0.0728	0.0258	0.005

. In 2018, the first year of the policy implementation, DID_GI_2018 is already positive and statistically significant (β = 0.0635, p < 0.01), suggesting that innovation-capable firms responded more strongly to the initial regulatory signal. This effect strengthens in the following years, peaking in DID_GI_2021 (β = 0.1289, p < 0.01), before declining slightly in DID_GI_2022 (β = 0.0728, p < 0.01), though it remains significant throughout the period.

This pattern suggests that green innovation plays a dynamic role in enabling firms to respond to evolving regulatory expectations. In the early stage, firms with higher green innovation output may be better equipped to absorb new compliance requirements and align with policy goals. Over time, as disclosure norms become more institutionalized and widely adopted, the marginal impact of green innovation may taper due to a convergence in disclosure behaviors across firms, regardless of innovation levels.

These findings support the hypothesis that the moderating effect of green innovation is not static, but changes over time in line with shifts in institutional pressure, firm learning, and policy implementation. They underscore the importance of adopting a temporal perspective when assessing how internal capabilities interact with external regulatory forces in shaping corporate environmental behavior. For completeness, Table A3 reports the exact coefficient estimates, standard errors, and p-values for each post-policy year. The results provide consistent evidence of statistically significant and time-varying interaction effects between policy shocks and green innovation.

To further justify this modeling choice, we retained year-specific interaction terms rather than aggregating post-policy years into broader time phases. This structure allows for a more detailed observation of the year-by-year evolution in the moderating effect of green innovation. It reveals how firms dynamically adapt their disclosure behaviors as policy implementation progresses and institutional expectations stabilize. While grouping years into early and late phases (2018–2019 and 2020–2022) could offer a more simplified summary, such aggregation may obscure the temporal heterogeneity in firm responses, especially during a period of regulatory transition. The annual specification thus provides richer insights into how green innovation influences policy responsiveness across different stages of institutional change, which is central to the objectives of this study.

4.4. Robustness Checks

This paper validates the robustness of the findings; however, several additional tests were conducted. A lagged dependent variable approach was applied to mitigate potential endogeneity concerns arising from reverse causality or omitted variable bias. Specifically, the carbon information disclosure (CID) variable was lagged by one period and re-estimated within the DID framework. The results remain qualitatively consistent and statistically significant, reinforcing the causal interpretation of policy shocks on carbon information disclosure.

Figure 5 shows a placebo test, performed to rule out spurious correlations. The mixed placebo test, which randomly assigns treatment status, confirms that the estimated DID effects do not emerge under artificial treatment assignments, indicating that the observed relationship is driven by actual policy shocks rather than unobserved confounders. The results, as illustrated in Figure 5, show that the true DID estimate (red line) lies outside the distribution of randomly generated estimates, further strengthening the causal claim.

In addition, the parallel trend assumption is reassessed by dropping the year 2017 from the sample to avoid potential anticipation effects. As shown in Figure 5, the pre-treatment coefficients remain statistically insignificant, while the post-treatment effects exhibit a clear upward trend, confirming that the treatment and control groups followed similar disclosure trends before the policy intervention.

Collectively, these robustness checks confirm that the baseline findings are not driven by sample selection bias, omitted variables, or anticipation effects, reinforcing the validity of this study’s conclusions.

5. Discussion and Conclusions

This study examines how non-mandatory environmental policy interventions affect corporate carbon information disclosure and how this relationship is influenced by firms’ green innovation capabilities. Using firm-level panel data and a DID design, the analysis reveals that the 2018 policy intervention led to significant changes in disclosure behavior. These effects were not uniform across firms but varied according to differences in innovation capacity and changed over time, reflecting the interaction of regulatory pressure, organizational readiness, and institutional adjustment. This view is consistent with prior research suggesting that internal capabilities play a role in how organizations adapt to changing policy environments (Stratu-Strelet et al. 2023).

Carbon information disclosure emerges from the combined influence of multiple organizational and environmental factors. Rather than being driven solely by policy, disclosure is shaped by firms’ technological capabilities, internal reporting routines, and stakeholder expectations. Although the 2018 regulation did not require public disclosure, it altered the regulatory landscape by mandating internal carbon accounting and verified reporting to authorities. This shift prompted some firms to initiate new reporting practices, while others, especially those with existing innovation infrastructure, adapted more gradually. These varied responses reflect firms’ positions within broader institutional contexts and their differing capacities to interpret and act on regulatory signals.

The 2018 policy issued by the Ministry of Ecology and Environment required firms in carbon-intensive industries to submit verified emissions data, implement monitoring systems, and establish internal accounting protocols. Although public disclosure was not mandated, the policy introduced new behavioral expectations that reshaped how firms processed and communicated carbon-related information. By embedding internal reporting norms and third-party verification mechanisms, it influenced disclosure practices through institutional change rather than through legal compulsion.

The results show that firms with limited green innovation capacity responded more visibly and immediately to the policy. These firms may have relied on disclosure as a lower-cost means of demonstrating environmental commitment, particularly when lacking the resources to implement substantive environmental technologies. For such firms, emphasizing carbon topics in public reporting may serve reputational goals and help satisfy stakeholder expectations without necessitating operational transformation.

In contrast, firms with stronger green innovation capabilities displayed more measured responses. Their existing internal systems may have already aligned with policy expectations, reducing the need for additional disclosure. In some cases, these firms may have withheld further information to protect competitive advantages or avoid increased scrutiny, suggesting that innovation capacity alone does not guarantee stronger disclosure outcomes, and effects depend on firms’ strategic assessments and perceived regulatory risks.

Over time, the influence of green innovation on disclosure behavior changed. As policy expectations stabilized and external demands became more standardized, differences between high- and low-innovation firms began to narrow. Initially, reactive firms appeared to converge toward established norms, while firms with greater innovation resources adjusted their disclosure strategies in line with evolving stakeholder expectations and industry benchmarks. The diminishing marginal effect of green innovation in later years suggests a process of institutional stabilization, where disclosure becomes routine and the scope for strategic differentiation declines.

These patterns of convergence and early differentiation can also be observed in other institutional contexts. Comparing our findings with prior research conducted in emerging economies helps situate this study within a broader theoretical and empirical landscape. These findings are consistent with prior research on voluntary environmental disclosure in emerging economies, where regulatory ambiguity often leads to diverse corporate strategies. For instance, studies on Brazil and India have found that firms in carbon-intensive industries exhibit selective disclosure behaviors based on their perceived legitimacy risks and stakeholder demands (Kouloukoui et al. 2021). Our observation that low-innovation firms responded more visibly in the short term aligns with these patterns, suggesting that disclosure is often used as a reputational buffer in high-risk policy environments. In addition, the evolving role of green innovation as a moderator resonates with recent studies on dynamic capabilities. Prior research has shown that firms in developing markets leverage innovation not only to enhance compliance capacity but also to adjust disclosure strategies as institutional expectations shift over time (Liu et al. 2025). Our results contribute to this literature by demonstrating that innovation-driven differentiation may be more salient during early regulatory phases, while convergence occurs as disclosure norms stabilize. Taken together, these comparisons support the relevance of our findings beyond the Chinese context and illustrate how institutional theory can help explain firm behavior under non-compulsory policy shocks in similar transitional environments.

Overall, the findings indicate that the relationship between environmental policy shocks, innovation capacity, and carbon information disclosure is dynamic and context-dependent. Firms interpret and respond to policy signals based on their internal structures, strategic orientations, and institutional environments. Green innovation does not uniformly strengthen responsiveness but operates as a conditional influence shaped by time and context. As disclosure norms consolidate, the strategic value of innovation-driven variation may decrease, giving way to convergence around shared standards. These results highlight the need to consider how external regulatory change and internal organizational adaptation interact over time and underscore the importance of incorporating temporal and institutional dimensions into analyses of corporate environmental behavior.

5.1. Managerial Implications

The findings of this study provide important insights for corporate operations in emerging markets. In these markets, firms face unique challenges, including regulatory uncertainty, fragmented enforcement of environmental policies, and evolving market dynamics. Carbon information disclosure is more than a compliance requirement. It reflects a firm’s overall environmental strategy and its capacity for innovation. For firms in emerging markets like China, the evolving regulatory environment presents both challenges and opportunities. Managers should consider how to use green innovation not only to improve internal efficiency but also to distinguish their firms in markets that increasingly prioritize sustainability.

For firms with robust green innovation capabilities, this study suggests that maintaining basic levels of disclosure may be insufficient in an evolving market context. Firms with strong technological capabilities should leverage their green innovations to proactively enhance their carbon information disclosures. This can improve their visibility in green finance markets, attract eco-conscious investors, and strengthen relationships with stakeholders who are increasingly sensitive to environmental impact. The competitive advantage provided by green innovation can thus play a crucial role in shaping long-term sustainability positioning.

For firms with limited green innovation capacity, the results highlight the importance of using carbon information disclosure as a low-cost mechanism to demonstrate policy responsiveness. Although these firms may lack the technological capacity for full environmental transformation, disclosure can still demonstrate their commitment to sustainability goals. Managers should avoid relying on symbolic reporting and instead focus on gradually improving the consistency and credibility of their disclosures over time.

Moreover, both highly innovative firms and those with limited innovation capacity should regard carbon information disclosure as part of broader risk mitigation strategies. By aligning disclosure with policy expectations and stakeholder demands, firms can better manage regulatory uncertainty, reputational exposure, and strategic positioning in low-carbon transitions.

5.2. Policy Implications

This study also provides meaningful insights for policymakers, particularly those in emerging markets where regulatory frameworks are still evolving. The 2018 policy intervention, which mandated internal carbon reporting but did not require public disclosure, demonstrates how non-compulsory policies can effectively shape firm behavior. This finding highlights the value of flexible policy frameworks in emerging markets, where institutional environments often lack the consistency seen in more developed economies. Policymakers should recognize that non-mandatory regulatory interventions, when viewed as credible and forward-thinking, can push firms to take visible actions towards sustainability.

To further enhance the impact of such policies, regulators should provide more detailed guidance on carbon information disclosure. While policy signals are important, clearer definitions of reporting requirements, standardized data collection methods, and a consistent format for disclosures will help firms in emerging markets with limited resources that better align with environmental goals. Regulatory clarity not only improves data comparability but also helps integrate carbon disclosure into market mechanisms, such as green finance.

Moreover, policymakers should support firms with limited green innovation capacity by providing practical tools, such as standardized reporting templates and digital platforms. For firms that are still developing their green innovation capabilities, additional capacity-building initiatives, such as technical training and third-party consulting services, would be invaluable in helping them transition from basic reporting to more meaningful environmental disclosures. These measures would not only improve the quality of disclosure but also prevent firms from resorting to symbolic compliance, ensuring that carbon disclosure serves as an effective governance tool.

5.3. Limitations and Future Research

This study provides significant insights into how firms in emerging markets respond to environmental policy interventions. However, this analysis centers on a single regulatory change in China, which, although highly relevant, does not account for the broader regulatory landscape in emerging markets. Different regions within China may experience distinct regulatory enforcement, and firms in other emerging economies may face even more fragmented institutional environments. Future research could expand the focus to examine multiple policy interventions across different emerging markets to capture the diversity of regulatory responses and the role of local institutional contexts in shaping firm behavior.

In terms of measuring green innovation, this study relies on patent data, which may not fully capture the breadth of green innovation efforts. While patents provide an objective measure of technological output, other forms of innovation, such as R&D investment or internal environmental initiatives, are equally important but often unquantified. Future studies should consider broader indicators of green innovation, such as R&D investment and environmental certifications, to more fully capture how firms in emerging markets pursue sustainability.

The empirical sample is also limited to A-share listed firms, which are typically subject to more standardized governance structures and have higher visibility in the market. The responses of non-listed firms or small- and medium-sized enterprises (SMEs) may differ from those of large listed firms. Future studies could examine a broader set of firms, including SMEs or non-listed firms, to explore whether the patterns observed in this study hold across different organizational types.

This study treats the moderating effect of green innovation as a continuous variable, assuming that the influence of innovation capacity on disclosure is linear. However, it is possible that green innovation has a non-linear impact, or that there are threshold effects depending on the level of innovation capacity or the stage of innovation development. Future research could explore these dynamics using non-linear models or interaction thresholds, offering more nuanced insights into how green innovation influences corporate responses to environmental policy.

Future studies could also benefit from exploring firm perceptions and qualitative dimensions of disclosure behavior, such as the motivations behind decisions to disclose or withhold carbon information. Including these factors could enrich our understanding of the underlying mechanisms that drive corporate responses to environmental regulation.

5.4. Conclusions

This study investigates the influence of environmental policy shocks on corporate carbon information disclosure, with a particular focus on the dynamic moderating role of green innovation. Using panel data from Chinese A-share listed firms over the period 2013–2022 and employing a difference-in-differences (DID) approach, the findings show that regulatory signals can significantly shape firm behavior, even without mandatory enforcement mechanisms. However, the extent of this influence depends heavily on firms’ green innovation capacities and their evolving interpretations of policy expectations.

This study finds that firms with stronger green innovation capabilities are initially less reactive to policy shocks, possibly due to their existing internal systems aligning with regulatory expectations. Over time, however, as disclosure norms become institutionalized, these firms increase their engagement with carbon information disclosure. Conversely, firms with lower levels of green innovation exhibit more immediate responses, often using disclosure as a strategic tool to manage reputational risks or comply with evolving institutional expectations.

The dynamic nature of green innovation’s moderating role is a key contribution of this study. In the early stages of policy implementation, firms with higher green innovation capacity may be better equipped to respond to regulatory changes, thus amplifying their engagement with disclosure practices. However, as the regulatory environment stabilizes and disclosure practices become more standardized, the role of green innovation in shaping firm behavior begins to diminish, as firms across the board converge toward similar disclosure practices.

This dynamic adjustment process underscores the need for a temporal perspective in understanding how internal capabilities interact with regulatory pressures and evolving climate-related risks. It challenges the assumption that green innovation always leads to stronger corporate responses and suggests that the impact of innovation evolves as firms adapt to changing institutional and market conditions. In emerging markets like China, where institutional frameworks are still in flux, this dynamic perspective is crucial for understanding how firms respond to environmental policies and how these responses change over time.

These findings provide important theoretical and practical insights for policymakers and corporate managers in emerging markets. They emphasize the importance of considering both regulatory signals and firm-level innovation capacities when designing policies aimed at enhancing corporate environmental governance. Moreover, this study highlights the need for flexible policy frameworks that can adapt to the evolving nature of corporate responses in transitional institutional contexts.