Enhanced risk management and green innovation are critical for enterprises to navigate the complex climate environment. Using panel data from Chinese listed firms (2009–2022), this study examines the relationship between peer effects in climate risk disclosure (CRD) and corporate green innovation (CGI) by applying machine learning techniques and employing a two-way fixed effects model. Our results reveal that corporate CRD exhibits both industry and regional peer effects that substantially promote CGI. From financing and governance perspectives, CRD peer effects stimulate CGI by alleviating financing constraints and enhancing executive environmental awareness, and market uncertainty diminishes the positive influence of CRD peer effects on CGI, and investor attention amplifies it. Peer effects from transit risk disclosure are more significant in driving CGI than those related to physical risk. Furthermore, CRD peer effects are more conducive to collaborative green innovation and have a stronger driving effect on CGI for high-carbon firms, capital-intensive industries, and transportation hub cities. This study examines the mechanisms through which corporate CRD influences CGI from a social interaction perspective, providing valuable insights for enhancing risk management and advancing sustainable development in emerging economies.
Green innovation is a key driver of sustainable development that has become a significant field in the contemporary global industrial revolution. The United Nations (UN) Trade and Development Technology and Innovation Report 2023 indicated that developing countries’ share of green technology exports is only half that of developed countries,1 which is attributed to stronger development needs, financial and technological constraints, and insufficient policy support. However, with the growing complexity of climate change and natural resource scarcity, risk management and green initiatives in developing countries have become increasingly urgent.
Public reporting of climate-related risks—both physical and transition-related—and corresponding responses to enterprises’ business operations and finances, which is known as climate risk disclosure (CRD), has an increasingly pivotal influence on national climate risk management frameworks (Carattini et al., 2022). In 2017, the Financial Stability Board’s Task Force on Climate-related Financial Disclosures established a global standard for climate-related disclosures, which has been overseen by the International Financial Reporting Standards Foundation since 2023, offering a valuable reference for countries’ CRD initiatives. Domestic firms can enhance their information transparency and brand image (Shao & Xue, 2024), and strengthen their risk management capabilities (Greenwood & Warren, 2022) by engaging in CRD, which contributes to aligning operational and environmental performance (Vestrelli et al., 2024). Therefore, efficiently leveraging CRD to drive green transformation has become a critical tool for nations to address the climate crisis and bolster competitiveness (Deng et al., 2024), particularly for developing countries facing severe pollution challenges.
Previous studies have predominantly focused on the economic ramifications of CRD, examining its influence on market reactions (Li et al., 2023; Shao & Xue, 2024), risk management (Greenwood & Warren, 2022), financing accessibility (Chalabi-Jabado & Ziane, 2024), and environmental governance (Wang et al., 2024). These studies are particularly related to the literature on green technologies (Tao et al., 2024). Scholars have widely acknowledged that extreme climate events such as abnormal temperatures can hinder corporate green innovation (CGI) output (He et al., 2024), and effective CRD can encourage firms to pursue green innovation (Gao et al., 2024; Ren et al., 2024). Within this field, some studies have applied signaling theory to examine the mechanisms by which CRD impacts CGI, with a focus on stakeholders’ assessment of investment risks in environmental practices (Tao et al., 2024). Other studies, grounded in legitimacy theory, have analyzed how corporate CRD mitigates potential operational risks and encourages green innovation (Deng et al., 2024; Gao et al., 2024). Nevertheless, by predominantly focusing on individual firm practices, existing literature has neglected interfirm interactions—specifically, how peer effects shape corporate decision-making. In fact, corporate strategic behavior is profoundly socially embedded, and is particularly evident in practices such as information disclosure, where decision-making and actions are systematically influenced by the practices of groups with similar characteristics (Leary & Roberts, 2014). This oversight has created a significant gap in understanding how peer firms’ CRD influences CGI. Addressing this gap is the primary motivation for our investigation.
Conceptually, peer effects are an intriguing and significant area of study that stem from a theoretical framework based on social interaction research (Bandura & Walters, 1963), which posits that individuals’ behavioral traits are shaped by typical role models in their environment, leading them to act accordingly. Many studies have demonstrated peer effects in various aspects of corporate decision-making such as financial policy (Leary & Roberts, 2014), research and development (R&D) investment strategy (Li et al., 2023; Kelchtermans et al., 2020), executive compensation (Gabaix & Landier, 2008), and stock buybacks (Grennan, 2019). However, the influence of peer effects in climate risk management has been underexplored. While Li et al. (2024) preliminarily identified significant industry peer effects in CRD, contemporary research has not yet explored the economic ramifications of these effects, particularly in the context of green innovation. From a theoretical integration perspective, institutional theory suggests that firms gain organizational legitimacy by imitating peers’ CRD practices and adopt CGI as a strategic response to institutional pressures (DiMaggio & Powell, 1983). Reputation theory asserts that firms tend to align their disclosure strategies with peers to maintain or enhance market reputation, and consolidate their environmental image by intensifying CGI efforts (Fombrun & Shanley, 1990). Information asymmetry theory indicates that the CRD peer effect mitigates information asymmetry, boosts investors’ confidence in firms’ green transition, and improves the financing environment for corporate innovation (Tseng et al., 2016; Tan et al., 2022). Collectively, these theories have systematically revealed the intrinsic correlations between CRD-based social interactions and CGI (Yang & Han, 2021). Based on this, we pose two questions. (i) Do CRD peer effects influence firms’ green innovation? (ii) What are the mechanisms that underlie the impact of CRD peer effects on CGI?
We investigate this underexplored yet crucial issue by examining China—the largest developing country—as a case study, with a twofold rationale. First, China is highly vulnerable to climate risks. The UN Intergovernmental Panel on Climate Change (IPCC) has warned that China could be severely affected by global heat and humidity without mitigation efforts (IPCC, 2023). Consequently, Chinese companies face heightened climate risks compared with other nations. This study intends to provide valuable insights to guide these companies in enhancing risk management strategies as a model for developing countries’ green transformation worldwide. Second, the Chinese government prioritizes green technological innovation, introducing various policies and plans, including the Guidance on Building a Market-Oriented Green Technological Innovation System. From a social interaction perspective, CRD peer effects could be a novel solution and policy direction for other emerging market economies.
We examine the impact of CRD peer effects on CGI using Chinese listed companies’ data from 2009 to 2022. Initially, we establish the fundamental premise by confirming the existence of industry and regional CRD peer effects, revealing that CRD peer effects significantly promote CGI. Regarding the underlying mechanisms, CRD peer effects enhance CGI by alleviating financing constraints and elevating executive environmental awareness. From information asymmetry and stakeholder interests perspectives, we conclude that market uncertainty diminishes the positive impact of CRD peer effects on CGI, while investor attention has the opposite effect. Specifically, we analyze the differences in CRD, innovation, and industry types, determining that the peer effects of transit risk disclosure significantly boost CGI. Furthermore, CRD peer effects promote collaborative innovation more strongly and have a more pronounced impact on CGI, specifically among high-carbon firms, capital-intensive industries, and firms in transportation hub cities.
This study makes several key contributions. First, building on a social interaction framework, we integrate financing constraints and upper echelons theories to examine the CRD–CGI link, bridging the literature on peer effects, climate finance, and corporate innovation. While existing research has separately examined CRD, CGI, and peer effects (Wang et al., 2022; Gao et al., 2024; Tao et al., 2024), it has not considered their critical intersection within climate risk management. Consequently, it remains unclear how CRD drives CGI through social mechanisms, or whether such corporate imitation accelerates green transition or devolves into symbolic compliance. By providing empirical evidence on peer effects, this study bridges the gap between CRD social interactions and innovation responses, demonstrating CRD’s multiplier effect. Furthermore, although peer effects’ financing mechanisms have been acknowledged (Tan et al., 2022), their specific pathway in CRD—particularly the central yet neglected role of executive cognition—remains unclear, and a complete explanation of how CRD transmits its influence on CGI is lacking. We delineate how financing constraints and executive environmental awareness mediate the CRD–CGI link, extending the peer effects framework in environmental management and strengthening the applicability of upper echelons theory to climate risk contexts.
Second, this study introduces an external context perspective by examining how market uncertainty and investor attention moderate CRD peer effects, delineating their boundary conditions. Previous research has largely overlooked these contextual contingencies, risking overestimation of disclosures’ efficacy. This oversight may result in policies that over-rely on disclosure tools, neglecting their potential ineffectiveness during market volatility or low investor attention. Empirically, we demonstrate that market uncertainty attenuates the effect, while investor attention amplifies it. This suggests to policymakers that the multiplier effect of CRD may be constrained without market stability and effective information dissemination. Therefore, this study confirms CRD peer effects and maps how their impact varies using external context, informing tailored policy in developing countries.
Third, by examining the heterogeneous effects of CRD, innovation, and industry types, this study provides specific, actionable insights for designing targeted climate risk policies in developing countries. Previous innovation research has focused on incremental and breakthrough innovations (Kline et al., 2019), neglecting the distinction between collaborative and independent approaches. Without this distinction, it is unclear whether firms develop green technology individually or through collaborative networks to address climate risks. Furthermore, compared with natural physical risks, increased green advocacy is more likely to strengthen the peer effects of transit risk disclosure, which encourages green innovation (Giglio et al., 2021). Notably, we conclude that high-carbon firms, capital-intensive industries, and firms located in transportation hub cities are more susceptible to the impact of CRD peer effects on CGI. These findings provide a nuanced basis for developing strategically targeted climate risk management and green innovation policies in emerging economies.
The remainder of this study is organized as follows. The theoretical analysis is presented in Section 2. Section 3 outlines the models and data sources. The empirical results and analysis are detailed in Section 4. Section 5 presents mediating and moderating effect tests, and heterogeneity analyses. Finally, Section 6 concludes the study and offers policy recommendations.
Theoretical analysisCRD peer effectsSocial learning theory posits that firms, guided by available information, tend to imitate or learn from industry peers’ CRD choices (Abrahamson & Rosenkopf, 1997). Firms in the same industry share similar business models, technical standards, and market environments, and are bound by common institutional norms, and their strategic CRD behaviors tend to exhibit systematic convergence, which is known as the industry peer effect (Lieberman & Asaba, 2006). From a competition-driven perspective, firms must continuously establish new competitive advantages to address competitive disadvantage or maintain positions in response to environmental changes (Feld & Zölitz, 2017). Firms face greater CRD uncertainty amid intense industry competition and often engage in competitive imitation to avoid falling into weak competitive positions (Lieberman & Asaba, 2006; Leary & Roberts, 2014), which is a defensive or offensive response to peer competition threats. Information asymmetry indicates that firms have incomplete information and denotes a highly uneven distribution of information (Haunschild & Miner, 1997). To reduce search costs and decision-making uncertainty, firms adopt industry peers’ CRD practices, thereby enhancing internal decision-making effectiveness (Tan et al., 2022). Enterprises perceived to hold more CRD-related information, share similar characteristics, and deliver fruitful decision outcomes become imitation benchmarks within the industry (Bikhchandani et al., 1998). In addition, firms’ CRD is constrained by internal industry codes of conduct, standards, and shared cognitive frameworks (DiMaggio & Powell, 1983). Such normative pressure drives firms’ CRD practices to converge with widely recognized industry standards, enabling them to gain organizational legitimacy, strengthen industry identity, and avoid cognitive uncertainties and evaluation risks that can arise from deviations from industry norms.
Similarly, firms in the same region share comparable locational advantages, resource endowments, labor supplies, policy environments and social networks, which encourages them to imitate and learn from one another’s CRD practices. First, information flow enables CRD imitation, and geographic proximity substantially reduces the marginal cost of information transmission (Baldwin & Okubo, 2006), making CRD practices more observable, comparable, and imitable. Furthermore, regional conduits such as labor mobility (Startz, 2016), supply chain collaboration networks, and industry associations serve as channels for CRD information dissemination, facilitating the diffusion of best practices within the region. Second, institutional isomorphism imposes normative pressure that drives CRD imitation and learning. Regional firms are subject to local institutional frameworks such as environmental regulations and industrial development policies (Dyer & Singh, 1998) and face similar legitimacy pressure regarding CRD practices. To address this pressure and comply with regional disclosure norms, firms adopt convergent CRD practices to ensure local institutional legitimacy (DiMaggio & Powell, 1983). Finally, regional social networks reinforce firms’ CRD convergence. Embedded in local information networks within relatively tight regional business ecosystems (Borgatti & Halgin, 2011), firms’ CRD performance directly shapes their business reputation and social evaluations. Therefore, firms tend to reference or even imitate locally network-central or institutionally influential peers’ CRD practices to avoid poor positioning in regional green rating systems. In summary, we formulate the following hypotheses:
H1a CRD exhibits an industry peer effect.
H1b CRD demonstrates a regional peer effect.
As a critical form of non-financial information, CRD enhances corporate transparency, risk management, and strategic planning, subsequently attracting external investment and fostering innovation (Aggarwal & Dow, 2012). In the context of climate change, CRD reflects a firm’s stance toward climate-related risks and opportunities and may also drive green innovation as a core strategic response (Zhao, 2023). CRD mitigates information asymmetry between firms and stakeholders (Ilhan et al., 2023; Zhao, 2023), as companies signal their dedication to managing climate risks through CRD, diminishing stakeholders’ perceptions of risk. This can bolster their investment intentions (Adhikari & Zhou, 2021), channeling more funds into CGI initiatives. Concurrently, firms face normative government, social, and market pressure, which compel them to pursue CGI to meet legitimacy demands (Suchman, 1995; Pekovic & Vogt, 2021). Particularly with the implementation of policies such as China’s Measures for the Management of Enterprise Environmental Information Disclosure, CRD has become a key compliance requirement. To secure subsidies or avoid penalties, companies are increasingly introducing environmentally friendly equipment and advancing sustainable transitions (Adhikari & Agrawal, 2018). CRD also poses challenges to modern corporate governance by mitigating opportunistic behavior in corporate governance and fostering CGI by enhancing resource allocation efficiency (Xiao, 2013).
Beyond the direct influence of CRD on fostering CGI, its socially driven peer effect also significantly enhances CGI. Firms acquire advanced CRD practices through channels such as analyzing peer reports, industry exchanges, and supply chain networks. Enterprises systematically evaluate leading peers’ disclosure frameworks and risk models, selectively adapting these strategies to their own capabilities and objectives. This process of social learning lowers information and compliance costs, sharpens market opportunity recognition, and improves resource allocation—collectively advancing CGI. First, CRD peer effects alleviate resource constraints for CGI by narrowing the environmental information disclosure gap. Adopting comparable disclosure standards enables firms to gain stakeholder trust, secure more stable external resources, and reduce internal compliance and coordination costs, freeing up critical resources for CGI (Pfeffer & Salancik, 2015). Second, CRD peer effects enhance green innovation efficiency by improving external knowledge absorption and integration. By systematically tracking and analyzing peers’ climate strategies and practices, firms can identify and effectively adopt industry best practices, reduce uncertainty and trial-and-error costs in exploration, and integrate external knowledge into their innovation systems. This accelerates the iterative advancement of green technologies and management practices (Tseng et al., 2016; Xiong et al., 2024). Third, CRD peer effects facilitate proactive green innovation by mitigating normative and competitive pressures within the industry. As peer firms strengthen environmental disclosure, companies are incentivized to maintain compliance and reputation, prompting them to actively identify green opportunities and adjust their innovation focus. This fosters a virtuous cycle of green innovation at the industry level, ultimately enhancing overall environmental performance. Based on these arguments, we propose the following hypotheses:
H2a CRD has a significant positive effect on CGI.
H2b CRD peer effects promote CGI.
Financial constraints can hinder firms’ green innovation efforts (Yu et al., 2021). First, CRD peer effects enable firms to expand their financing channels. Firms’ alignment of CRD practices with those of peer enterprises facilitates risk sharing and benefit co-creation between firms (Cotter & Najah, 2013), which helps secure increased financing inputs and drives green innovation (Khan et al., 2016). From an investor’s perspective, corporate learning and CRD imitation publicize a company’s climate change initiatives, enhancing perceptions of its adaptive and mitigating capabilities, and bolstering investment confidence (Li et al., 2024). For creditors, CRD peer effects prompt firms’ active climate risk information disclosure, which enhances financial institutions’ risk assessment capabilities and mitigates credit risks (Nieto, 2019). Second, CRD peer effects reduce firms’ financing costs (Li et al., 2024). Firms can secure external incentives or evade penalties by emulating leaders within their peer group. Notably, mimicking firms that closely adhere to government policies is more likely to earn investor trust and approval, consequently lowering corporate financing costs (Long et al., 2020). Furthermore, CRD peer effects enable companies to cultivate social images that adhere to environmental regulations and embrace ecological responsibility (Milgrom & Roberts, 1982; Dhaliwal et al., 2011). This positive reputation fosters investor confidence in corporate profitability and debt repayment capacity, increasing potential financial support (Cheng et al., 2014). Other funding sources such as venture capital or specialized investment may also be more readily available for these firms. By leveraging these diverse financing channels, firms can ease financial constraints and substantially support green innovation (Cao et al., 2021). Therefore, we propose the following hypothesis:
H3a CRD peer effects incentivize CGI by alleviating financing constraints.
Upper echelons theory suggests that executives’ values and perceptions directly shape corporate innovation strategies (Hambrick & Mason, 1984). CGI can be influenced by managers’ self-perception and interpretation of the environment, particularly when facing external pressure from peer firms’ CRD. CRD peer effects indicate that the specific measures taken by other firms regarding climate risk management and environmental facilities allocation are clearly communicated (Li et al., 2024). The resulting peer “standard” provides an opportunity for managers to enhance their awareness of risks and benefits. First, CRD peer effects influence corporate management’s decision-making processes, garnering various stakeholders’ sustained interest. If executives do not have adequate environmental awareness, their reputation and the company’s interests could be at risk, and executives consciously elevate their environmental consciousness and actively promote CGI to avoid such risk (Gao et al., 2022). Naturally, under substantial external pressure, executives are compelled to embrace environmental philosophies to address the green demands of stakeholders and local governments. This trend significantly reduces self-interested decision-making and promotes green innovation aligned with sustainable development (Huang et al., 2024). Second, CRD peer effects provide executives with a low-cost learning channel. Managers can obtain vital green information from peer firms’ CRD practices, narrowing gaps in risk management and environmental construction. Research has indicated that executives with stronger environmental consciousness are more inclined to guide firms to adopt green innovation strategies (Sarkis et al., 2010). Therefore, we propose the following hypothesis (Fig. 1):
H3b CRD peer effects incentivize CGI by enhancing executive environmental awareness.
Prior to examining the impact of CRD peer effects on CGI, it is imperative to ascertain whether such peer effects actually exist (Manski, 2007). To do so, we construct the following models:
where CRDk,t denotes firm k’s CRD level in year t. ICRD−k,t−1 (CCRD−k,t−1) represents the industry (regional) peer effect of CRD in year t-1, measured by the average CRD level of other firms in firm k’s industry (city). Xk,t is the set of factors that affect corporate CRD. Unobservable variables for industries or regions are included in our peer effects calculation. Referencing Ren et al. (2023), we introduce year and region fixed effects (FEs) in Equation (1), and Equation (2) includes year and industry FEs. We also employ robust standard errors for t-statistics to mitigate heteroskedasticity. A statistically significant positive estimate of α1(β1) indicates the existence of CRD peer effects at the industry (regional) level.
We next model the relationship between confirmed CRD peer effects and CGI as follows:
where GInnok,t represents firm k’s green technology innovation. If γ1 and δ1 are significantly positive, it implies that industry (regional) CRD peer effects significantly incentivize firms to engage in green technology innovation.VariablesDependent variable
We measure CGI using granted green patent counts, consistent with the established practice of employing patent data to assess technological innovation (Wu et al., 2022). First, our classification follows the World Intellectual Property Organization’s (WIPO) 2020 International Patent Classification (IPC) Green Inventory, whereby green patents are identified by primary IPC codes that align with any of seven eco-domains, encompassing alternative energy, energy efficiency, waste management, greenhouse gas mitigation, transport abatement, sustainable agriculture/forestry, or water conservation. Second, we compile all granted invention patents filed by A-share listed firms from the China National Intellectual Property Administration and retain only those with primary IPC codes that match one of the WIPO green categories. We focus on granted patents—as opposed to applications—as they represent innovations that have passed substantive examination and reflect firms’ actual innovative capacity more accurately. Third, to address right-skewness and zero-inflated counts, we construct the CGI variable (GInno) as the natural logarithm of one plus the number of granted green invention patents (Kline et al., 2019).
Independent variablesText analytics and machine learning are widely used to extract keywords from corporate disclosures (Sautner et al., 2023). We adopt this approach to measure corporate CRD. Initially, we derive a set of seed words related to climate risk information from corporate annual reports. Specifically, we refine the word set provided by Li et al. (2024) to select seed words that align with Chinese semantics through translation and textual analysis. In the second step, we expand our initial set of seed words to account for diverse Chinese expressions. Referencing LeCun et al. (2015), we identify and integrated the top 10 semantically similar words into the seed word set.2 Finally, to derive proxy metrics for corporate CRD, we compute the ratio of the frequency of CRD-related terms to the total word count in the annual reports.3 Additionally, following Leary & Roberts (2014) and Ren et al.(2023), we use peer firms’ average CRD, excluding the firm, to proxy for industry (ICRD) and regional (CCRD) CRD peer effects.
Fig. 2 illustrates the relationship between corporate CRD, ICRD, CCRD, and GInno. We find that firms’ increased CRD encourages CGI, aligning with Gao et al. (2024). The figure reveals a pronounced incentive effect wherein CRD by peer firms within the same industry and region significantly spurs firms’ CGI. This finding elucidates the factors influencing CGI, demonstrating the influence of environmental regulation and industrial cleanliness. Since the China Securities Regulatory Commission mandated corporate disclosure, corporate disclosure related to climate risk has increased.4 Peer firms’ CRD can motivate firms to adopt green practices such as climate-friendly production through imitation and learning. It also serves as a regulatory signal, driving firms to meet environmental targets via CGI.
Relationships between CRD, peer effects, and CGI.
Note: The solid line represents the relationship between CRD and CGI, the short dashed line represents the relationship between the mean CRD and CGI of peer firms at the industry level, and the long dashed line represents the relationship between the mean CRD and CGI of peer firms at the regional level. Fig. 2 shows that all relationships show a significant upward trend.
In alignment with previous literature, we control for significant factors affecting corporate CRD and CGI. Firm growth and performance are strongly correlated with innovation (Kemp et al., 2003; Walker et al., 2015; Rajapathirana & Hu, 2018). To account for this, we introduce the total asset growth rate (Tagr), Tobin’s Q (Tobin), and return on net worth (Roe) into our model. Furthermore, as larger firms generally have a greater capacity to undertake R&D, we control for firm size (Size). Overall, survival rates and cash holdings indicate a firm’s risk resilience (Pástor & Pietro, 2003). Additionally, weaker solvency implies higher risk for creditors when extending loans, which is closely associated with firms’ business risk. Therefore, we control for firm age (Age), cash holdings (Cash), and gearing (Lev). As executives significantly affect a firm’s decision-making (Camelo-Ordaz et al., 2005), we control for executive compensation (Salary), considering the presence of compensation contracts with management as a key factor. Finally, we control for corporate ownership (SOE), coded as 1 for state-owned enterprises (SOEs) and 0 for others.
Data sourceThis study uses data on companies listed in China’s Shanghai and Shenzhen A-share markets from 2009 to 2022. The CGI indicator is derived from the China National Intellectual Property Administration’s green patent application data. The remaining firm-level data are sourced from the China Stock Market & Accounting Research Database (CSMAR)5 and Chinese-listed companies’ annual reports.6 Investor attention indicator data are from China’s Baidu search engine, which houses a database of over 100 billion items, enabling investors to locate the key information they seek.
Upon obtaining the raw data, we processed it as follows. (i) We excluded data from ST, ST*, PT firms with abnormal financial performance and from financial and insurance firms. (ii) We excluded data from firms with significant missing key variables. (iii) To ensure that firms have a visible target for imitation and learning, we excluded data from industries (or regions) with only one listed firm. (iv) To address potential distortions from outliers, we apply 1 % bilateral winsorization to all continuous variables. After processing, we obtained 22,260 observations for 2535 listed firms across 73 industries from 2009 to 2022. The firms are located in 156 cities across 31 provinces in China, providing a highly representative sample.
Descriptive statisticsTable 1 provides the descriptive statistics. GInno ranges from a minimum of 0.0000 to a maximum of 4.2047, with a standard deviation of 0.8990. This range indicates significant heterogeneity across the sample data, reflecting substantial differences in green innovation across firm types. The variations noted above correspond to the heterogeneous traits of China’s industrial landscape, encompassing both technology- and capital-intensive service sectors alongside low-end, labor-intensive manufacturing industries. This diversity results in significant disparities in firms’ innovation capabilities. The average values of ICRD and CCRD are 0.1594 and 0.1590, respectively, with minimum and maximum values of 0.0590 and 0.5527 for ICRD, and 0.0477 and 0.3195 for CCRD. This reveals that the mean CRD value at the regional level is lower than that at the industry level. One possible reason is that firms within the same industry maintain closer connections and typically face more homogeneous market conditions, regulatory policies, and environmental factors. In addition, control variables exhibit strong relevance to Chinese firms’ contemporary development trajectories, demonstrating consistency with recent empirical findings (Ren et al., 2023).
Descriptive statistics.
Exploring CRD peer effects is a key foundation for our research. Fig. 3 illustrates the CRD peer effect analysis, spanning both industry and regional considerations. Notably, the coefficients for ICRD in the left panel are significantly positive. Similarly, although the coefficients of CCRD are significantly lower, their confidence intervals do not include zero. This demonstrates significant peer effects on CRD at both industry and regional levels. As the CRD level of other firms within the same industry and region rises that of the individual firm also exhibits a notable upward trend. Therefore, H1a and H1b are confirmed as valid.
Existence test for peer effects in CRD.
Note: In Fig. 3, the dots represent the estimated coefficients of the explanatory variables and the line segments represent the corresponding 90 % confidence intervals.
Columns (1) and (2) of Table 2 demonstrate that CRD has a significantly positive effect on CGI, confirming its influence on enhancing enterprises’ sustainable practices. This aligns with present-day scenarios and theoretical insights. First, CRD is integrated into the financing system to bolster CGI by extending greater financial support (Birindelli et al., 2022). Second, CRD can mitigate firms’ agency problem and information asymmetry (Gao et al., 2024), promoting CGI to adapt to the complex climate environment.
The impact of CRD and its peer effects on CGI.
Note: ***, **, and * represent significance at the 1 %, 5 %, and 10 % levels, respectively. Robust standard errors are reported in parentheses. The same applies to the following tables.
Columns (3)–(6) of Table 2 present the results of CRD peer effects’ impact on CGI. At the industry level, the coefficients of ICRD in Columns (3) and (4) are 1.5342 (p < 0.01) and 1.4499 (p < 0.01), respectively, indicating that CRD peer effects significantly enhance CGI within the same industry. Likewise, regional CRD peer effects encourage firms to pursue CGI, validating H2b (Table 2, Columns (5) and (6)). In a broader context, our findings align with Ren et al. (2023), who examined the impact of digital transformation peer effects on corporate environmental performance. However, our study uniquely investigates the relationship between peer effects and CGI through a risk management lens, enriching the literature on environmental economics and business management.
Endogenous analysisThe potential endogeneity between CRD peer effects and CGI related to a two-way causality is a critical concern for our study. To address this concern, we adopt the instrumental variable (IV) approach. Previous studies have employed peer firms’ average idiosyncratic equity return shocks (PIERS) as an IV to examine peer effects (Leary & Roberts, 2014; He & Wang, 2020). PIERS is favored in such research due to its resistance to direct firm manipulation and exclusion from market influences. To closely align our analysis with the thematic focus on climate risk, we add the degree of topographic relief in the area where each firm is located to our IVs. This fixed geographic condition, indicating substantial elevation variations, has been linked to more complex climatic environments. Areas with greater topographic relief are more prone to natural disasters and environmental emergencies (Pankratz et al., 2023), prompting managers to disseminate climate risk information with greater regularity. We construct two IVs (IV1 and IV2), creating interaction terms between the industry (region)-specific PIERS and the regional topographic relief index of the area where the firm operates.
Columns (1) and (2) of Table 3 presents the results based on the industry CRD peer effect.7 The first-stage regression yields a statistically significant positive relationship between ICRD and IV1, meeting the correlation condition. In the second stage, the coefficient for ICRD is significantly positive at the 5 % level. Columns (3) and (4) of Table 3 shift the focus to the regional CRD peer effect, revealing a significant positive relationship between CCRD and IV2. The second-stage regression yields a coefficient for CCRD of 8.2679 (p < 0.01). These results indicate that industry and regional CRD peer effects significantly spur CGI, validating the robustness of our baseline results.
2SLS regression results.
To address omitted variable bias concerns, which can skew the estimation of our regression model, we employ a diagnostic test proposed by Altonji et al. (2005). This test quantifies the severity of the impact of omitted unobservable variables by calculating the ratio |βF/(βR−βF)|, where βF represents the unconstrained regression coefficient, encompassing all control variables, and βR denotes the constrained regression coefficient, which includes a subset of the control variables. A larger ratio indicates smaller omitted variable impact on the coefficient estimates. Table 4 presents the results of the tests for omitted unobservable variables.8
Omitted variable test.
Columns (1)–(3) present the industry level results. The coefficients for ICRD in Column (1) show a relatively narrow range, varying between 1.3095 and 1.5847, indicating minimal influence from control variables on the relationship between ICRD and GInno. The minimum value of |βF/(βR−βF)| in Column (3) is 10.3301, and the maximum value reaches 323.3280, both of which exceed the threshold value of 1. This indicates that the effect of unobservable variables would need to be at least 10.3301 times that of the observable variables to alter the existing outcomes. Regional analysis results in Columns (4) to (6) reveal that coefficients for CCRD vary slightly between 0.2893 and 0.3956. The minimum value of |βF/(βR−βF)| in Column (6) is 5.4787, indicating that the impact of unobservable variables must be at least 5.4787 times that of the observable variables to significantly alter the existing results. This finding confirms the reliability and stability of our baseline estimates.
Entropy balancingTo address potential sample selection bias and attrition that may compromise the accuracy of propensity score matching, we employ the entropy balanced matching method (McMullin & Schonberger, 2020), presenting the results in Columns (1) and (2) of Table 5. Our findings indicate that peer effects of CRD at industry and regional levels significantly enhance CGI, with ICRD yielding a coefficient of 1.4277 (p < 0.01) and CCRD having a coefficient of 0.4114 (p < 0.01).
Entropy balancing, exogenous shocks excluded, and other robustness tests.
Next, to isolate the effects of other relevant policies, we exclude their influence in our analysis. We consider stringent environmental regulatory policies, encompassing the 1998 Acid Rain and Sulfur Dioxide Control Zones, which set stricter emissions limits and encouraged local firms to engage in CGI. Additionally, the 2013 Air Pollution Prevention and Control Action Plan integrated air quality objectives into government performance evaluations, further incentivizing local firms’ green innovation (Cheng et al., 2023). We also control for digital innovation policies. The 2012 Smart City pilot Initiative facilitated information technology development, providing a solid digital base for CGI (Yang et al., 2024). Similarly, the 2015 National-level Big Data Comprehensive Experimental Zone Policy expedited data market development and improved enterprises’ generation and sharing of green knowledge. The results after excluding the aforementioned policy interventions are presented in Columns (3) and (4) of Table 5. The coefficient for ICRD is 1.3882 (p < 0.01), and the coefficient for CCRD remains significantly positive.
Other robustness testsTo address the challenge of a high incidence of zero values in our green innovation dataset, we first employ a Poisson pseudo-maximum likelihood regression model, which is particularly adept at accommodating excess zeros (Gourieroux et al., 1984). Furthermore, considering the profound disruptions to business operations caused by the COVID-19 pandemic, we exclude data from 2020 and subsequent periods. We also recompute climate risk word frequencies using alternative annual reports—specifically, those containing Management Discussion & Analysis (MD&A) sections—released by listed companies. Columns (5)–(10) in Table 5 present the corresponding regression estimates. The results confirm that the core findings of our study remain robust.
Further discussionMediating effect testsFinancing constraintsInter-firm learning and imitation of CRD practices can diminish investors’ uncertainty risks and ameliorate firms’ financing constraints. Such practices can also alleviate the pressure of legitimacy faced by enterprises, establish a positive corporate image, and reduce the costs associated with financing. By expanding the avenues for financing and lowering these costs, firms are better positioned to obtain the financial backing needed to pursue CGI (Nieto, 2019). Therefore, we propose that CRD peer effects may alleviate financing constraints, stimulating CGI. Referencing Fazzari et al. (1988), we use the FC index to measure financing constraints (FC).
The coefficient for ICRD in Column (1) of Table 6 is significantly negative, indicating that the industry CRD peer effect eases corporate financing constraints. Column (2) of Table 6 reveals that while the coefficient for ICRD remains significantly positive, the coefficient for FC is significantly negative, implying that the industry CRD peer effect substantially mitigates firms’ financing constraints, bolstering their CGI. Columns (3) and (4) of Table 6 demonstrate that regional CRD peer effects also alleviate financial constraints. Therefore, H3a is validated. This confirms the significance of CRD peer effects for financing. Wang et al. (2025) demonstrated that climate-disclosing firms mitigate information asymmetry and yield systematic valuation premiums, particularly under high inherent risk, which enhances financing access and bolsters financial support for CGI.
Mediating effect tests (1): Financing constraints.
Peer firms share CRD insights with firms through social interactions, generating beneficial information spillovers that shape executive awareness of risks and returns (Gao et al., 2022). Firms environmental awareness is enhanced through this low-cost learning channel, improving corporate reputations and promoting sustainable development. Individuals use embedded language to convey their thoughts in social contexts, with frequently used words revealing their inner cognition. To examine this, we employ text analysis to measure executive environmental awareness. We use listed companies’ MD&A reports, selecting a series of keywords across green competitive advantage cognition, corporate social responsibility cognition, and external environmental pressure perception dimensions (Khalid et al., 2024). We assess executive environmental awareness (EA) by calculating the ratio of keywords’ frequency to the total word count in MD&A reports.
Column (1) of Table 7 shows that the industry CRD peer effect has a significant positive impact on executive environmental awareness. In Column (2), the coefficients for ICRD and EA are significantly positive, indicating that the industry CRD peer effect enhances firms’ environmental awareness, which stimulates CGI. This channel is also evident for the regional CRD peer effect on CGI. Therefore H3b is verified. This conclusion broadly aligns with Wang et al. (2024), who identify executives’ environmental experience as a key mechanism through which CRD influences reduced carbon emissions, demonstrating the significance of managerial environmental orientation as a pathway for green transition.
Mediating effect tests (2): Executive environmental awareness.
To validate the mediating effects, we employ a bias-corrected bootstrap method with 1000 resamples (Table 8). The results indicate that financing constraints and executive environmental awareness are significant mediators for CRD’s industry peer effects, with all 95 % bias-corrected confidence intervals for direct and indirect effects excluding zero. This finding holds for regional peer effects wherein, while the direct effect is insignificant, the significant indirect pathway through executive environmental awareness demonstrates complete mediation. The consistent results across contexts verify the appropriateness of our mediating variable selection and validate the proposed mediation mechanisms.
Results of the Bootstrap test for mediating effects (N=1000).
Note: BC CI=bias-corrected confidence interval. Bootstrap replications=1000.
Market uncertainty instills a sense of potential crisis in decision makers, deterring risky actions and impeding firms’ innovation process. It also increases the likelihood of innovation failure, reducing the propensity for corporate innovation (Wang et al., 2023). Such uncertainty can diminish resource allocation efficiency in external markets, which is detrimental to enterprises’ ability to attract talent, capital, and other essential resources through CRD. Market uncertainty also exacerbates issues such as information concealment, tampering, and under-reporting, which makes it difficult for firms to secure external financing (Myers & Majluf, 1984) and can encroach upon existing funds allocated for CGI. So we contend that market uncertainty can suppress the impact of CRD peer effects on CGI. Referencing Ghosh and Olsen (2009), we quantify firms’ market uncertainty (MU) using the coefficient of variation in sales revenue adjusted for industry.
Column (1) of Table 9 shows that the coefficient of ICRD × MU is −0.2531 (p < 0.01), and Column (2) reveals that the coefficient of CCRD × MU is −0.1976 (p < 0.01), indicating that market uncertainty significantly weakens the promotional effect of CRD peer effects on CGI at industry and regional levels. In other words, when external market uncertainty is high, firms can find it challenging to reduce information asymmetry and attract innovation funding by simply imitating their peers’ CRD practices.
Moderating effect tests: Market uncertainty and investor attention.
Limited investor attention theory indicates that focus on a firm is scarce (Andrei & Hasler, 2015), potentially creating reward externalities for CRD. This drives firms to enhance their reputations and capital market visibility to secure additional funding (Manski, 2000). Amid variable climate conditions such external attention can also place significant pressure on firms engaged in CRD convergence, compelling them to embrace a green transition and pursue sustainable development. Therefore, we posit that investor attention strengthens the impact of CRD peer effects on CGI. Considering that web search volume is a direct barometer of investor interest in a company’s stock, we adopt the methodology from Da et al. (2011) to gauge investor attention (IA). This approach can accurately quantify this indicator and effectively avoid irrelevant information obtained when searching with the stock abbreviation as the keyword.
Columns (3) and (4) of Table 9 show that the coefficients for ICRD × IA and CCRD × IA are 0.0710 (p < 0.01) and 0.0379 (p < 0.01), respectively. These findings demonstrate that CRD industry and regional peer effects become more influential in promoting CGI as investor attention increases. In summary, investor attention intensifies the relationship between CRD peer effects and CGI. By focusing on CRD convergence, firms can attract valuable external attention, gain approval in the capital markets, and enhance CGI.
Heterogeneity analysisTypes of CRDHaving established the positive impact of CRD peer effects on CGI, it is crucial to recognize that varying types of CRD peer effects may have distinct influences on this process. As the academic community currently differentiates climate risks into physical and transit risks (Giglio et al., 2021),9 we categorize CRD into physical and transit risk disclosures. After calculating the peer effects of physical and transit risks, we re-estimate model (1).
Columns (1) and (3) of Table 10 indicate that physical risk-based peer effects do not drive CGI, whereas Columns (2) and (4) yield opposing results. Such discrepancies may be due to inherent climate risk differences. Physical risks are highly correlated with extreme weather and natural disasters, characterized by suddenness, unpredictability, and regional specificity. Even when imitating peer firms in disclosing such information, firms struggle to respond accurately in a short time frame. This results in more cautious CGI investment. However, transit risks, which are associated with policy costs and market operational risks during the green transition process (Giglio et al., 2021), can be effectively perceived by firms. Transit risks are closely related to the industry and regional attributes, and peer effects of such risk disclosure can reduce corporate cost constraints and market risks, increasing investment in green R&D and promoting sustainable development.
Heterogeneity analysis: Types of CRD.
Independent green innovation is often marked by technical challenges and protracted R&D timelines, and corporate innovation frequently emerges in clusters (Jaffe et al., 1993). The Chinese government emphasizes the significance of industry–university–research partnerships in local governance. Beyond autonomous green innovation, companies also engage in technological collaborations with other firms, universities, research institutions, and other entities to pursue joint green innovation initiatives. We categorize CGI into independent and cooperative forms. Based on patent application data, a green patent solely applied for by a firm is classified as an independent green patent, whereas one applied for in conjunction with other organizations is deemed a collaborative green patent. By computing the logarithm of the count of independent (cooperative) green patent applications incremented by one, we derive the respective metrics for GInno_In (GInno_Co).
Column (1) of Table 11 reveals that the coefficient of ICRD is 0.8124 (p < 0.01), while in Column (2), the coefficient of ICRD is 1.3447 (p < 0.01). In terms of the magnitude of the coefficients, this peer effect has a stronger promotional impact on collaborative green innovation. Similarly, the regional peer effect of CRD also has a more pronounced influence on collaborative green innovation. These results are closely associated with the fact that firms are embedded in social networks. Firms facing similar climate risk threats are more likely to establish collaborative innovation networks, resulting in more frequent patent collaboration, technology transfer, and factor mobility among companies (Li et al., 2024).
Corporate environmental attributesDifferent firms exhibit varying sensitivities to climate risks. Compared with low-carbon emissions firms (e.g., in education), high-carbon emissions firms (e.g., in electricity and steel) are exposed to more significant climate-related risks. Particularly under environmental regulations and social pressure, high-carbon firms face substantial cost pressure during green transitions (Yao et al., 2024). Such firms can reduce cost burdens and mitigate uncertainty risks more effectively by learning from peer practices in the CRD context (Demerjian et al., 2012), opening opportunities for CGI. Furthermore, these firms are further motivated to pursue green transition to avoid market elimination. We measure corporate carbon emissions intensity as the ratio of carbon emissions to operating revenue, using the sample mean as the threshold to classify firms with a ratio above the mean as high-carbon firms, and those below as low-carbon firms.
Columns (1) and (2) of Table 12 show that the industry CRD peer effect significantly enhances CGI for both groups. However, the Fisher combination test shows an empirical p-value of 0.000, indicating the presence of inter-group coefficient differences. This means that the industry peer effect of CRD is more likely to stimulate high-carbon enterprises’ CGI. Similarly, the regional CRD peer effect also exhibits this difference, indicating that high-carbon firms that face tighter environmental regulations and investor focus, are pushed toward CGI to maintain market share and avoid obsolescence. CRD peer effects can reduce costs and risks, and significantly increase investment and the potential for innovation success by sharing green innovation knowledge.
Heterogeneity analysis: Corporate environmental attributes and industrial capital intensity.
CGI in capital-intensive industries (e.g., chemicals) faces significant uncertainties due to high costs, long cycles, and technological complexity (Rogge & Hoffmann, 2010), and the peer effect in CRD provides critical information spillovers and risk benchmarks. Observing peers’ CRD enables capital-intensive firms to better assess the financial, regulatory, and market prospects of CGI, lowering decision and search costs. Furthermore, strong supply chain and cluster ties in such industries amplify competitive imitation and normative pressure as public commitments to transition through CRD push firms to invest more in green innovation to avoid falling behind in industry restructuring (Acemoglu et al., 2016). Building on Lu & Dang (2014) and following the Guidelines for Industry Classification of Listed Companies (2012 Revision), we classify the full sample into capital-intensive and non-capital-intensive industries.
Columns (5) to (8) of Table 12, demonstrate that the industrial and regional peer effects of CRD are more conducive to enhancing capital-intensive industries’ CGI—a finding that passes the Fisher combination test. This can be attributed to capital-intensive industries facing higher transition costs and greater uncertainties. The observable actions and commitments of peers within the CRD framework provide critical informational and normative benchmarks, enabling firms to assess risks, reduce search costs, and synchronize investments within coupled industrial networks, amplifying the marginal effect on CGI.
Regional transportation characteristicsWe contend that the network connectivity and resource agglomeration of transportation hub cities amplify the impact of CRD peer effects on CGI. Dense transportation networks lower information barriers, enhancing information spillover and learning efficiency from peers’ CRD disclosures. As focal points for regional resources, these hubs also intensify competitive imitation, driving firms to invest more in CGI to secure resources and avoid losing ground in the regional industrial landscape. Following the Medium and Long-term Railway Network Plan (2016), we categorize sample cities into transportation hub and non-hub city groups for empirical examination.
The regression results in Table 13 reveal that industrial and regional CRD peer effects exert a significantly stronger positive influence on CGI for firms located in transportation hub cities, which is confirmed by between-group coefficient tests. This indicates that the locational advantages of transportation infrastructure can amplify the governance effect of peer CRD, positioning hub cities as critical arenas for green innovation synergy and transition competition.
Heterogeneity analysis: Regional transportation characteristics.
This study demonstrates that CRD generates peer effects at industry and regional levels, promoting CGI by alleviating financing constraints and enhancing executive environmental awareness. Our findings extend the theory of informational peer effects (Seo, 2021) to climate governance and demonstrate how peer influence shapes CGI strategies in China’s institutional context.
While confirming the pervasiveness of imitation in corporate disclosure, our results reveal its distinctive expression in the domain of climate governance. The findings corroborate earlier evidence of peer effects in corporate disclosure decisions (Seo, 2021; Li & Ding, 2024), reinforcing the perspective that firms are influenced by peers’ practices. In contrast, compared with disclosure studies in areas such as corporate product quality (Wang et al., 2023), key audit matters (Wu et al., 2025), and shadow banking activities (Liu et al., 2025), CRD peer effects reflect firms’ collective response to systemic environmental risks more prominently. Within China’s dual-carbon policy context, such imitation represents a collective response to policy signals and a strategic approach to mitigating compliance risks under institutional pressure.
These findings should be interpreted within China’s unique institutional context. The peer effects in CRD exhibit a dual-driven pattern that encompasses industry and regional dimensions and that reflects the institutional logic of China’s matrix governance system. The industry effect demonstrates strategic alignment under centralized policies, while the regional effect arises from divergent local enforcement and competitive dynamics (Li et al., 2024). This dual perspective advances peer effect theory by transcending the single-level analyses that characterized earlier research.
More importantly, while previous studies have documented peer effects in Chinese firms’ CRD (Gu & Shen, 2024; Li et al., 2024), their economic consequences remained underexamined. San et al. (2025) focused exclusively on audit demands, leaving core technological domains unexplored. Our study bridges this gap by establishing that CRD peer effects primarily promote substantive CGI, particularly collaborative approaches. This contrasts with the crowding-out effects of certain environmental policies (Xie & Wang, 2025), indicating that Chinese firms prioritize substantive innovation over symbolic disclosure when addressing climate risks—a critical dimension of long-term competitiveness. Our findings indicate that improving disclosure systems may guide industrial green transition more effectively than distortionary fiscal subsidies, particularly among high-carbon firms (Song et al., 2024).
In summary, this study extends peer effect theory to CRD, demonstrating its distinct capacity to drive CGI within China’s institutional framework. By reframing climate disclosure from mere informational imitation to a substantive transition mechanism, our findings provide novel theoretical and empirical insights into how Chinese firms use social interactions to achieve innovative breakthroughs amid complex climate challenges.
Conclusions and implicationsConclusionsAs the complexity of climatic conditions expands, green innovation is crucial for firms to retain market positions and competitiveness. Grounded in the behavioral science of social interactions, this study explores whether CRD peer effects can spur firms to pursue CGI. We begin by confirming the presence of peer effects in corporate CRD and their influence on CGI. Our results reveal that CRD exerts industry and regional peer effects, incentivizing firms to engage in CGI. We also demonstrate that financing constraints and executive environmental awareness are the primary channels through which CRD peer effects influence CGI. Specifically, CRD peer effects ease firms’ financing constraints and enhance executive environmental awareness, encouraging firms to engage in CGI.
Our findings also demonstrate that market uncertainty weakens the positive impact of CRD peer effects on CGI, while investor attention has the opposite effect. The results also indicate that peer effects from transit risk disclosures have a stronger influence on CGI than those from physical risk disclosures. Furthermore, we classify CGI into independent and collaborative types, finding stronger CRD peer effects on collaborative CGI. Finally, we observe that CRD peer effects are particularly conducive to promoting green innovation for high-carbon firms, capital-intensive industries, and transportation hub cities.
Policy implicationsOur findings provide valuable insights for stakeholders, including government officials and business managers. Economic downturns due to climate change are frequent (Somanathan et al., 2021), making it essential to enhance the disclosure system for climate and environmental information. Considering the negative impacts of climate risks, local governments should actively guide companies to adopt preventive and mitigating measures. Additionally, to sustain CGI momentum, it is vital to learn from peer firms’ experiences and raise managers’ environmental awareness. Therefore, firms should pay attention to the strategic decisions of their industry or regional peers and refine internal business strategies based on the peer group experiences. Rational responses to peer firms’ strategic decisions are the main path for ensuring healthy operation and innovative development, particularly for enhancing market competitiveness. Considering the complexity of climatic conditions and the frequent occurrence of pollution, it is crucial for management to raise environmental awareness and integrate sustainable development into production. Companies should also actively collaborate and innovate with external partners, leveraging green initiatives and transit risks to drive research. In particular, high-carbon industries should seize opportunities to optimize their product lines and management through green innovation to advance sustainable transformation.
Limitations and future researchWhile scholars have examined the relationship between disclosure and corporate innovation, research on the social interaction and climate risk dimensions has remained limited. This study uses Chinese data to reveal that CRD peer effects foster firms’ green innovation. However, our findings raise some noteworthy questions. Are these finding applicable to firms in other countries such as economically diverse developed nations or geographically distinct African countries? What contributes to any similarities or differences in these results? These questions warrant further in-depth investigation in future research. We concentrate on how CRD peer effects influence firms’ green innovation. Future studies could also explore whether CRD at industry or regional levels affects firms’ financial performance, environmental efficiency, and other business outcomes.
Data availabilityData is available upon request.
CRediT authorship contribution statementJie Peng: Writing – review & editing, Writing – original draft, Software, Methodology, Conceptualization. Jianmin Dou: Supervision, Investigation, Funding acquisition, Formal analysis, Conceptualization. Liping Li: Writing – review & editing, Writing – original draft, Supervision, Formal analysis.
This work was supported by Natural Science Foundation of China [71974120]; Shanghai Office of Philosophy and Social Science [2024BJL001]; Fundamental Research Funds to the Central Universities [CXJJ-2024-304; CXJJ-2025-313].
While the Technology and Innovation Report 2023 predicts that the green technology market could surpass $9.5 trillion by 2030, the majority of economic benefits will accrue to developed countries. For example, developed countries’ green technology exports rose from $60 billion in 2018 to over $156 billion in 2021. In contrast, developing countries’ exports only grew from $57 billion to approximately $75 billion in the same period.
The specific seed and expanded word sets for the CRD are presented in Appendix Tables A1 and A2.
We multiply this indicator by 100 to obtain economically significant estimated coefficients. The rationale of the CRD indicator is presented in Appendix Fig. A1 and A2.
In August 2007, the China Securities Regulatory Commission issued the Circular on Regulating Disclosure of Information by Listed Companies and the Behavior of Relevant Parties, establishing mandatory disclosure requirements. This resulted in a significant rise in the number of Chinese listed companies reporting on climate risks.
CSMAR, which covers a range of indicators for Chinese-listed companies, has been widely adopted by the academic community.
These annual reports summarize a company’s operations, management, and production activities over the past year, including extensive information on climate risks and environmental protection.
To verify the validity of our IVs, we conduct identification and weak IV tests. In the identification test, we report the Kleibergen–Paap rk LM statistic, and the weak IV test reports the Cragg–Donald Wald F statistic. p-values for the Kleibergen–Paap rk LM statistic are reported in brackets []. When the p-value is less than 0.01, the hypothesis of IV under-identification is rejected at the 1% level. The critical F-value at the 10% level of distortion given by Stock & Yogo (2005) is reported in {}. When the F-value is greater than 16.38, the hypothesis of “weak instrumental variables” is rejected at the 10% level.
Column (1) of Table 4 is the regression coefficient of ICRD on GInno after sequentially accumulating the corresponding control variables, e.g., after controlling for Tagr, the regression coefficient of ICRD on GInno is 1.5377, and further controlling for Cach, the regression coefficient of ICRD on GInno is 1.5847, and column (2) is the corresponding t-value. Column (3) then shows the value of |βF/(βR - βF)|, where βF is the regression coefficient of ICRD controlling for all variables (i.e., 1.4499), and βR is the regression coefficient after sequentially accumulating the corresponding control variables. Columns (4)–(6) show the regression results of CCRD on GInno. Note that |βF/(βR - βF)| is calculated based on the original coefficients, not the coefficients after retaining 4 bits; therefore, the values differ from the manual calculation.
Referencing Giglio et al. (2021), we posit that physical risk refers to the economic losses caused by extreme weather and natural disasters due to climate change. This includes economic losses caused by sudden-onset extreme weather and natural disasters due to climate change such as the risks posed by heatwaves, earthquakes, floods, wildfires, and other sudden calamities, as well as economic losses caused by long-term natural disasters due to climate change such as the long-term, gradual climatic risks associated with excessive rainfall, high humidity, and cold conditions. Transit risk pertains to the policy costs and market operational risks that accompany the societal shift toward a low-carbon economy and zero emissions. The specific word sets are presented in Appendix Table A2.
articles
- Download PDF
- Bibliography
- Additional material
