Introduction
With a growing consensus on global warming challenges in recent decades, green innovation has been deemed essential in dealing with environmental issues (Xie, Huo, & Zou, Reference Xie, Huo and Zou2019), enhancing corporate competitiveness, fostering sustainable transformation, and creating value for customers, shareholders, and the society as a whole (Chang, Reference Chang2011). Despite the considerable benefits of green innovation, its great costs, high risks, and uncertain returns (Falcone, Reference Falcone2020; Xie, Hoang, & Zhu, Reference Xie, Hoang and Zhu2022) may discourage corporate managers under performance pressures from investing in it (Laukkanen & Patala, Reference Laukkanen and Patala2014). As a critical driver for corporate transformation (David, Bloom, & Hillman, Reference David, Bloom and Hillman2007; Flammer, Toffel, & Viswanathan, Reference Flammer, Toffel and Viswanathan2021; Goranova & Ryan, Reference Goranova and Ryan2014), shareholder activism, by utilizing shareholder proposals and voicing environmental concerns, constitutes an emerging force in promoting green innovation (Cundill, Smart, & Wilson, Reference Cundill, Smart and Wilson2018; Lee & Lounsbury, Reference Lee and Lounsbury2011).
The power of shareholder activism may vary across different shareholder groups in different contexts (Mitchell, Agle, & Wood, Reference Mitchell, Agle and Wood1997). A wealth of literature has acknowledged the activist role of institutional investors as salient shareholders holding professional network resources and confirmed their significant impact on corporate green innovation (Jin, Monfort, Chen, Xia, & Wu, Reference Jin, Monfort, Chen, Xia and Wu2024; Wu, Chen, Li, Xu, Liu, & Wu, Reference Wu, Chen, Li, Xu, Liu and Wu2024) whereas the impact of minority shareholder activism on environmental governance remains largely disregarded. Scholars typically categorize minority shareholders such as retail investors as an insignificant group suffering from information asymmetry and limited voicing channels (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, Reference La Porta, Lopez-de-Silanes, Shleifer and Vishny2000; Li & Zhang, Reference Li and Zhang2023). In the traditional offline setting, they participate in shareholder activism through general meetings of shareholders, at which outcomes of proposal voting are mostly dominated by large shareholders. The failure to hear minority voices, however, can be detrimental to firms since it suppresses the diversity of external opinions and biases corporate decisions toward majority interests (Kastiel & Nili, Reference Kastiel and Nili2016). Notably, a nascent stream of research indicates that retail investors are more concerned with social and environmental issues that align with their individual well-being and moral values, which compared with financial issues are also more accessible to grassroots investors (Porzio, Salerno, & Stella, Reference Porzio, Salerno and Stella2023). Therefore, how to better leverage the insights of retail investors and to enhance their participation in environmental governance deserves more academic attention (Li, Wang, & Zhang, Reference Li, Wang and Zhang2022; Xu, Li, Jiang, Wu, & Zhang, Reference Xu, Li, Jiang, Wu and Zhang2022).
Fortunately, in the Internet Age, the emergence of diverse information technologies and digital platforms dramatically empowers retail investors to advance environmental activism in cyberspace. For example, in China, official interactive platforms established by two major stock exchanges provide retail investors with unique opportunities to present their environmental issues and suggestions directly to the board secretaries of listed firms. These two platforms are thus becoming battlefields for retail investors to launch online environmental activism and advocate green values. An emerging research stream has suggested that the emergence of online platforms facilitates information diffusion among retail investors, greatly reduces activism participation costs of minorities and also enhances their voicing power (Abbas, Zhou, Deng, & Zhang, Reference Abbas, Zhou, Deng and Zhang2018). In particular, Li, Qian, Wang, Wang, and Wang (Reference Li, Qian, Wang, Wang and Wang2022) have offered preliminary evidence that online environmental concerns raised by retail investors may promote corporate green innovation. Yet, prior literature has rarely tapped into the underlying mechanisms through which online environmental activism affects corporate attention allocation and green innovation decisions. Furthermore, although a nascent stream of literature highlights that digitalization has profoundly reshaped organizational decision making (Giustiziero, Kretschmer, Somaya, & Wu, Reference Giustiziero, Kretschmer, Somaya and Wu2023; Mithani, Reference Mithani2023), we know little about whether digital and non-digital firms regulated by different attention structures adopt distinct responses to online environmental activism.
Therefore, in this article, we adopt an attention-based view (ABV) (Ocasio, Reference Ocasio1997, Reference Ocasio2011) in order to better understand how companies make sense of and direct managerial attention to online environmental activism. We regard official interactive platforms as a new form of attention carriers that bridge environmental requests of retail investors and attention allocations of corporate managers. Through effective social mobilization, online environmental activism can easily scale up to a persistent collective movement, forcing firms to increase the amount of attention and corresponding resource commitments to environmental issues. Online environmental activism can also evoke extreme emotions and impose urgent reputational threats to firms, thereby pressing firms to place a higher priority on environmental issues. With adequate attention intensity and attention priority (Barreto & Patient, Reference Barreto and Patient2013; Eggers & Kaplan, Reference Eggers and Kaplan2009), firms are more likely to introduce green innovation as a substantial response.
Following the principle of situated attention of ABV (Ocasio, Reference Ocasio1997), we examine specific situations in which online environmental activism exerts a stronger influence on corporate attention reallocation. We posit that a large retail investor shareholding and a negative sentiment of environmental questions will positively moderate the impact of online environmental activism and green innovation respectively by increasing attention intensity and attention priority. Drawing on the principle of structural distribution of attention in ABV, we propose that digitalization may profoundly transform corporate attention structure, making digital firms more responsive to online activism. Specifically, the moderating effects of retail investor base and negative questioning sentiment (NQS) are stronger for highly digital firms. Empirical results based on a dataset of Chinese-listed companies from 2011 to 2018 support our arguments.
Our study contributes to the existing literature in three ways. First, we highlight that through social mobilization and collective emotion incitement, online environmental activism by retail investors can alter two dimensions of managerial attention allocation, i.e., attention intensity and attention priority, thereby able to elicit corporate green innovation. In doing so, this article offers new insights about the online patterns of shareholder activism from the ABV (Bergman & Roychowdhury, Reference Bergman and Roychowdhury2008; Hafeez, Kabir, & Wongchoti, Reference Hafeez, Kabir and Wongchoti2022). Second, drawing on the principle of situated attention in ABV, we uncover the boundary conditions of the relationship between online environmental activism and green innovation by contending that presence of a large retail investor base increases the intensity of managerial attention and NQS increases the priority of managerial attention (Deng, Huang, Sinha, & Zhao, Reference Deng, Huang, Sinha and Zhao2018). Third, we shed light on the heterogeneity of attention structure and responses to online activism across digital and non-digital firms, thus enriching a growing body of literature advocating that digitalization reshapes the way in which firms scan the environment and engage with stakeholders (Giustiziero et al., Reference Giustiziero, Kretschmer, Somaya and Wu2023; Mithani, Reference Mithani2023).
Theoretical Background
Enhanced Retail Investor Environmental Activism in the Digital Era
In recent decades, the escalating global warming challenges have triggered great sustainability concerns among firms and their stakeholders (Truong & Nagy, Reference Truong and Nagy2021; Wang, Ren, Li, Qiao, & Wu, Reference Wang, Ren, Li, Qiao and Wu2024). As corporate primary stakeholders, shareholders are also shifting their focus from financial performance to environmental performance of their investee companies (Chen, Meyer-Doyle, & Shi, Reference Chen, Meyer-Doyle and Shi2021; Goranova & Ryan, Reference Goranova and Ryan2014), and urging companies to adopt substantial green efforts through shareholder activism (Wu, Fang, Jacoby, Li, & Wu, Reference Wu, Fang, Jacoby, Li and Wu2022). Corporate green innovation – which is a vital green transformation strategy and a win–win solution for both economic development and environmental protection (Chang, Reference Chang2011; Li, Li, Khan, Khaliq, & Rehman, Reference Li, Li, Khan, Khaliq and Rehman2022) – has become a prevailing pursuit by shareholder activists. In general, green innovation refers to new green products and processes that are developed for energy saving, pollution prevention, waste recycling, and other aspects beneficial to the environment (El-Kassar & Singh, Reference El-Kassar and Singh2019; Zhang, Liang, Feng, Yuan, & Jiang, Reference Zhang, Liang, Feng, Yuan and Jiang2020). Green innovation can not only alleviate corporate environmental risks but also enhance production efficiency and market competitiveness, thus deemed essential for creating long-term value for shareholders (Chen, Reference Chen2008; Zhang et al., Reference Zhang, Liang, Feng, Yuan and Jiang2020).
It should be noted that different types of shareholders, ranging from retail investors to institutional investors, may exert varying influences on target firms’ green innovation through environmental activism (Mitchell et al., Reference Mitchell, Agle and Wood1997; Ryan & Schneider, Reference Ryan and Schneider2003). A considerable body of literature has recognized that majority shareholders such as institutional investors often generate greater impacts when undertaking environmental activism due to their concentrated ownership stakes, rich information channels, professional network resources, and abundant on-site communication opportunities with management (Aluchna, Roszkowska-Menkes, Kamiński, & Bosek-Rak, Reference Aluchna, Roszkowska-Menkes, Kamiński and Bosek-Rak2022; Flammer et al., Reference Flammer, Toffel and Viswanathan2021). In this vein, previous research has widely documented the positive influence of institutional investor activism on firms’ green innovation (Jin et al., Reference Jin, Monfort, Chen, Xia and Wu2024; Wu et al., Reference Wu, Chen, Li, Xu, Liu and Wu2024).
In contrast, the impact of retail investor activism on corporate green innovation remains typically disregarded (Li, Qian, et al., Reference Li, Qian, Wang, Wang and Wang2022). Despite of their dominance in China's capital markets (Titman, Wei, & Zhao, Reference Titman, Wei and Zhao2022; Xu, Yin, & Lou, Reference Xu, Yin and Lou2022), retail investors are usually perceived as a trivial constituent of shareholder activists with negligible voicing power (La Porta et al., Reference La Porta, Lopez-de-Silanes, Shleifer and Vishny2000; Li & Zhang, Reference Li and Zhang2023). In the traditional offline setting, retail investors suffer from information asymmetry, lack of resources, and limited voicing channels (Flammer et al., Reference Flammer, Toffel and Viswanathan2021). They have to overcome high entry barriers and participation costs to engage in shareholder activism through general meetings of shareholders. For example, in China, only when retail investors hold over 3% of shares in a firm are they allowed to submit shareholder proposals to the boardFootnote 1, and voting on these proposals is mostly dominated by large shareholders. Moreover, due to geographic dispersion, retail investors are unlikely to connect with each other and make coordinated activism actions in the offline context. In such circumstances, retail investors tend to passively ‘free ride’ on the activism of majority shareholders or even avoid taking activism initiatives (Kastiel & Nili, Reference Kastiel and Nili2016). As a result, retail investors struggle to capture managerial attention and their demands are usually ignored by management (Kastiel & Nili, Reference Kastiel and Nili2016; Lin, Liao, & Xie, Reference Lin, Liao and Xie2023).
Yet, ignoring minority voices is potentially detrimental to corporate governance since it may lead to over-concentration of managerial attention on majority interests and reduce the diversity of external opinions that are essential for balanced decision making (Gompers & Kovvali, Reference Gompers and Kovvali2018; Hafeez et al., Reference Hafeez, Kabir and Wongchoti2022). Notably, scholars have begun to notice that retail investors have a peculiar taste for corporate environmental and social issues (De Villiers & Van Staden, Reference De Villiers and Van Staden2010; Hartzmark & Sussman, Reference Hartzmark and Sussman2019) that align with their personal beliefs and moral values, for example environmental pollution (Yao, Pan, Wang, Sensoy, & Cheng, Reference Yao, Pan, Wang, Sensoy and Cheng2023), environmental disclosures (De Villiers & Van Staden, Reference De Villiers and Van Staden2010) and certain socio-political events (Brownen-Trinh & Orujov, Reference Brownen-Trinh and Orujov2023). Distinct from financial issues that require professional knowledge, these issues are often closely related to individual well-being and are thus more accessible to grassroots investors (Porzio et al., Reference Porzio, Salerno and Stella2023). For example, to mitigate their investment risks and normatively behave as good citizens, retail investors tend to directly vent their frustrations on corporate environmental violations and press the invested firm to adopt green innovation (Li, Qian, et al., Reference Li, Qian, Wang, Wang and Wang2022). In this regard, enhancing retail investor environmental activism remains an unsettled issue of significant social value.
With the advent of the digital era, the emergence of online interactive platforms and digital media has greatly empowered minority shareholders (Mousavi & Gu, Reference Mousavi and Gu2019). Retail investors can access more information at much lower costs and publicly voice their opinions through diverse online channels, including online search engines (Hafeez et al., Reference Hafeez, Kabir and Wongchoti2022; Li, Li, et al., Reference Li, Li, Khan, Khaliq and Rehman2022), social networks, and online stock forums (Jiang, Liu, & Yang, Reference Jiang, Liu and Yang2019). Also, the Internet connects retail investors from different regions into an important force in sharing common beliefs, making collective activism moves, and reshaping public opinion (Gao, Li, & Wang, Reference Gao, Li and Wang2021). This is particularly evident in China, where retail investors have been provided with a legitimate online channel to convey their demands. Since 2010, two major stock exchanges have established official interactive platforms, i.e., Interactive Easy Platform and SSE E-Interactive Platform, on which retail investors can propose their environmental issues directly to board secretaries of listed firms. Moreover, regulatory authorities explicitly require companies to respond promptly to such investor inquiriesFootnote 2 and incorporate the quality of their replies into the rating of corporate information disclosure.Footnote 3 This ensures that online retail investor activism garners adequate attention from companies. In addition, all the Q&A records are released to the public in a timely manner, making retail investors’ requests likely to spread to a larger scope of audience (Li, Qian, et al., Reference Li, Qian, Wang, Wang and Wang2022; Wu, Gao, Chan, & Cheng, Reference Wu, Gao, Chan and Cheng2022). With the above advantages, these two platforms have become a main focus for Chinese retail investors to launch online environmental activism in their invested firms. The number of environmental questions on the platforms has increased dramatically, from 4,826 in 2011 to 17,728 in 2019.
However, we know little about how such online activism by retail investors directs managerial attention to environmental issues and promotes corporate green innovation. A handful of financial studies indicate that retail investor engagement with listed firms through the two interactive platforms does indeed alleviate their information costs, thereby leading to greater market liquidity, a reduced bid-ask spread, and higher productivity for Chinese-listed firms (Lee & Zhong, Reference Lee and Zhong2022). To our knowledge, there are limited studies that explore the impact of retail investor activism on corporate environmental governance. Li, Qian, et al. (Reference Li, Qian, Wang, Wang and Wang2022) show that green concerns promote firm green innovation by increasing media attention and reducing financing constraints. However, the underlying mechanisms that how the online activism of retail investors is interpreted by company management and thus affects corporate green innovation decisions have not yet been fully explained.
Research Framework from an ABV
In this article, we adopt an ABV to explain how companies make sense of and respond to online environmental activism by retail investors. The ABV conceptualizes firms as attention distribution systems in which managerial decisions hinge on ‘noticing, encoding, interpreting, and focusing of time and effort on’ specific issues and solutions (Ocasio, Reference Ocasio1997; Zhong, Ma, Tong, Zhang, & Xie, Reference Zhong, Ma, Tong, Zhang and Xie2021). Since firms with limited attentional resources are typically overwhelmed by a range of heterogeneous issues simultaneously (Bouquet & Birkinshaw, Reference Bouquet and Birkinshaw2008), in order to achieve efficient attention allocation, they have to take the complexity and urgency of issues into account to determine attention intensity (Fiske & Taylor, Reference Fiske and Taylor1991) and attention priority (Barreto & Patient, Reference Barreto and Patient2013).
Such dimensional decomposition of attention focus is in line with the widely-employed Eisenhower matrix, which considers importance and urgency as two dimensions to guide task planning (Gray, Reference Gray2021). Attention intensity refers to the amount of attention that corporate managers devote to an issue (Fiske & Taylor, Reference Fiske and Taylor1991; Kahneman, Reference Kahneman1973); it determines the amount of resource commitment and efforts that will be expended in addressing an issue (Eggers & Kaplan, Reference Eggers and Kaplan2009; Yadav, Prabhu, & Chandy, Reference Yadav, Prabhu and Chandy2007). Attention intensity depends largely on the complexity of an issue, which is manifested as the scale of involved entities and stakeholders, the scope of influence it may yield, and the persistence of the crisis (Eggers & Kaplan, Reference Eggers and Kaplan2009). Attention priority, from the temporal dimension, refers to the order of precedence in sequential attention allocation (Barreto & Patient, Reference Barreto and Patient2013). When confronted with multiple important issues that all demand managerial attention, firms tend to place a higher attention priority on urgent issues with destructive power and adopt prompt actions (Hu, Reference Hu2012).
In our study, if online environmental activism generates large-scope and persistent impacts that require greater commitment to curb, firms need to allocate a large amount of attention. On the other hand, if online activism triggers urgent and destructive threats, it demands corporate attention priority to ensure the adoption of immediate actions. We seek to elaborate on how online environmental activism elicits substantial green innovation responses, rather than symbolic or postponed responses, through increasing the intensity and priority of managerial attention. In this vein, we compare online and traditional patterns of retail investor activism and corresponding attentional processes of target firms (see Table 1).
Comparison of minority shareholder activism in online and traditional contexts
The rules of situated attention and structural distribution of attention in ABV indicate that attention allocation depends on characteristics of external situations and corporate inherent attention structures (Ocasio, Reference Ocasio1997). However, we still lack the understanding of in which circumstance online environmental activism is more likely to evoke intense and immediate attention. Furthermore, since digitalization has profoundly reshaped organizational operations and decision making processes (Giustiziero et al., Reference Giustiziero, Kretschmer, Somaya and Wu2023; Mithani, Reference Mithani2023), it is important to determine whether digital and non-digital firms are regulated different attention structures and thus adopt distinct responses to online activism.
Hypotheses Development
Online Environmental Activism and Corporate Green Innovation
Using an ABV, we propose that the online environmental activism of retail investors can promote corporate green innovation by enhancing both the intensity and priority of managerial attention to environmental issues.
First, through effective social mobilization in the digital environment, online environmental activism can easily scale up to a collective movement and trigger pervasive influence in cyberspace (Luo, Zhang, & Marquis, Reference Luo, Zhang and Marquis2016). This forces target firms to increase attention intensity on environmental issues. Official interactive platforms in China enable retail investors to directly propose their environmental requests to top management teams (Li, Qian, et al., Reference Li, Qian, Wang, Wang and Wang2022), which ensures that their voices are clearly heard by corporate managers, and also that requests are seen by their peer investors and the public. Retail investors that jointly hold shares of a firm frequently interact to identify investment risks of the firm in online communities. While environmental voicing on interactive platforms by one retail investor may initially seem insignificant, it can easily spread to a broad investor group through spontaneous information transmissions (Wang, Xing, Wei, Zheng, & Xing, Reference Wang, Xing, Wei, Zheng and Xing2020), yielding a contagion effect in cyberspace (Earl & Schussman, Reference Earl and Schussman2004).
By subscribing, giving ‘likes’, retweeting, or commenting on an environmental question, investor peers can follow up and engage in ongoing online environmental activism at a low cost (called ‘clicktivism’). Such pervasive information dissemination can even evoke widespread public discussions on alleged environmental misdeeds of the target firm, generating profound social impacts. The Internet and digital platforms geographically connect scattered retail investors into a sizable monitoring force to self-organize online activism in a decentralized approach (Luo et al., Reference Luo, Zhang and Marquis2016) and exert significant influence on corporate environmental practices. Moreover, the features of a decentralized coordination structure in virtual space and nondirective information dissemination make online environmental activism unpredictable and more challenging for target firms to deal with.
As online environmental concerns of retail investors escalate into a large-scale and persistent campaign through social mobilization, firms are motivated to allocate intensive managerial attention to environmental issues and accordingly adopt green innovation as a substantial response. As a win–win solution for both economic development and environmental protection (Chang, Reference Chang2011; Li, Li, et al., Reference Li, Li, Khan, Khaliq and Rehman2022), corporate green innovation is considered as a substantial and profound green transformation strategy (Sariol & Abebe, Reference Sariol and Abebe2017; Wu, Monfort, Jin, & Shen, Reference Wu, Monfort, Jin and Shen2022). By investing considerable resources and efforts in green innovation, firms can signal their commitment to meet online environmental demands as well as their sufficient capabilities of achieving green technology breakthroughs to bring sustainable value for investors (Chang, Reference Chang2011; Wu et al., Reference Wu, Chen, Li, Xu, Liu and Wu2024). Moreover, through internal research and development, target firms can drastically renew their technological paradigm and reshape organizational processes toward green development, thereby fully eliminating internal sources of environmental risks. If target firms paid little attention to online activism and merely took symbolic environmental actions (such as ‘greenwashing’ or empty rhetoric without providing a viable solution), it would be difficult to stand up to public scrutiny and thus fail to appease online activists (Lyon & Montgomery, Reference Lyon and Montgomery2013).
Second, compared with traditional activism, online environmental activism is more likely to spark negative collective emotions within a short time; it therefore imposes urgent and damaging reputational threats to target firms, thereby forcing them to place a higher attention priority on environmental issues. Compared with institutional investors who hold large ownership stakes, retail investors often hold a much lower individual shareholding ratio in a given firm (Lee, Reference Lee2010; Li & Zhang, Reference Li and Zhang2023). This weaker economic interdependence means that retail investors have more liberty to publicly condemn target firms’ environmental misdeeds and vent their discontent (Zhang & Luo, Reference Zhang and Luo2013).
Negative emotions can spread explosively on the Internet, which plays a decisive role in the convergence of negative public opinion on the target firms. The unique features of online communication, such as anonymity and the loosening of social restraints, can induce an online ‘disinhibition effect’, whereby Internet users often act impulsively and opt for high-arousal emotional expressions (Asker & Dinas, Reference Asker and Dinas2019). In a ‘post-truth’ era, the higher level of emotional intensity in online messages increases the likelihood that opinion polarization will occur within a short time (Kwon & Cho, Reference Kwon and Cho2017; Sunstein, Reference Sunstein2002). Due to the high relevance of environmental issues to individual well-being, environmental concerns posted by retail investors can easily trigger intensive online attacks and extreme public emotions against target firms, thus exposing the firms to legitimacy threats and severe reputation risks (Wu, Jin, Monfort, & Hua, Reference Wu, Jin, Monfort and Hua2021).
Rapidly rising reputational pressure increases the urgency and destructiveness of retail investors’ environmental requests, which compels companies to allocate a higher attention priority on environmental issues and make a prompt response. When they are confronted with multiple important issues simultaneously, decision-makers tend to prioritize the handling of urgent issues and destructive threats (Sullivan, Reference Sullivan2010). In the environment of online public opinion, every move a company makes is subjected to intense scrutiny by the public. Any symbolic response with no substantial resource commitment is likely to provoke an even greater public backlash and condemnation.
In the traditional context, even if a firm considers green innovation to be an effective solution for addressing environmental issues, it may prioritize other cost-effective projects and delay green innovation investment that entails great sunk costs, high innovation risks, and uncertain returns. Conversely, when perceiving urgent reputational threats induced by online activism, firms are more inclined to pay immediate attention to environmental issues and accelerate the development of green innovation in order to resolve the crisis. By introducing green innovation, target firms can establish a public image of embracing green technologies and fulfilling social responsibilities (Wu et al., Reference Wu, Chen, Li, Xu, Liu and Wu2024), thus eliminating the devastating impacts of extreme emotions and negative opinions from the online audience.
To summarize, online activism can enhance the intensity and priority of managerial attention to environmental issues, leading to substantial responses from target firms. Thus, we posit the following hypothesis:
Hypothesis 1 (H1): Online environmental activism on official interactive platforms promotes corporate green innovation.
The Moderating Role of Retail Investor Base and NQS
Based on the principle of situated attention in ABV, we further investigate situations when online environmental activism exerts stronger influence on target firms’ attention allocation and thus promotes corporate green innovation to a larger extent (Brielmaier & Friesl, Reference Brielmaier and Friesl2023). First, as retail investors who publicly present environmental requests on official interactive platforms are primary actors in online environmental activism, how much attention their activism draws from target firms depends on whether they are perceived as a stakeholder group of influence. In this article, we propose that the intensity of managerial attention to online environmental activism is enhanced by larger aggregate shareholding of the retail investor group.
Due to limited attentional resources, managers who confront diverse stakeholder demands tend to allocate more attention to prominent stakeholders who enjoy power and influence (Uysal, Yang, & Taylor, Reference Uysal, Yang and Taylor2018). When the aggregate shareholding of retail investors constitutes a substantial part of ownership stakes in a firm, the collective identity of retail investors as a shareholder group and their economic power becomes salient to the company. In such circumstances, firms with larger retail investor bases are more inclined to actively participate in online interactions on these platforms (Huang, Liu, Shi, & Ying, Reference Huang, Liu, Shi and Ying2024) and value every single environmental demand or suggestion from the retail investor group. Overall, a large retail investor base enhances the salience of online environmental activism to the firm, thus eliciting a larger amount of managerial attention on green development.
Moreover, a bigger retail investor base that a firm holds indicates a larger scale that online activism may expand to, which calls on corporate substantial attention and efforts in curbing its potential wide impact. For retail investors who often face information deficiency, interactive platforms serve as a primary information source (Lee & Zhong, Reference Lee and Zhong2022; Wang & Li, Reference Wang and Li2023). As Q&A records are timely released on these platforms and fully accessible for the public, online environmental concerns from a handful of retail investors can induce explosive information transmission and extensive discussion among the entire retail investor group who jointly invests in the target firm. Huang and Ying (Reference Huang and Ying2022) also suggests that online interaction on interactive platforms is crucial for retail investors to exchange and assimilate information, especially when firms hold large retail investor bases. In this regard, firms with large retail investor bases are more likely to encounter rapid scale expansion of online activism. As a result, they tend to allocate greater attention to newly emerged online environmental concerns and adopt prompt green innovation response to contain the continuing impact of such activism.
In addition, online environmental activism can essentially induce a large-scale ‘herding effect’ which leads the retail investor group to divest themselves of the firm's shares. Prior finance literature contends that retail investors typically exhibit herding investment behavior, i.e., they tend to follow opinions and decisions of peer investors (Gupta & Shrivastava, Reference Gupta and Shrivastava2021), especially in the Internet era. Online environmental concerns can greatly erode retail investors’ positive evaluations of the firm and lead to investment withdrawal. When retail investors hold a considerable number of shares in total, the herding effect is more detrimental to the stock market performance of a firm (Balp, Reference Balp2018).
To sum up, if a firm with a larger retail investor base confronts online activism, it suffers to a greater extent from the spreading activism impact and divestment threats from retail investors. Accordingly, the firm tends to allocate more attention to retail investors’ online environmental voices and strives to prevent online panic among a large retail investor group. Green innovation as a substantial response can help the company effectively curb the wide influence of online activism and consolidate financial support from a large retail investor group. Therefore, we propose that:
Hypothesis 2 (H2): The positive effect of online environmental activism on corporate green innovation is stronger if firms have a larger retail investor base.
In addition to the prominence of retail investors as an activist group, whether target firms promptly attend to online environmental activism also depends on how environmental issues are framed (Dutton, Ashford, O'Neill, & Lawrence, Reference Dutton, Ashford, O'Neill and Lawrence2001). The framing of issues refers to the way in which issues are presented (Liu & Mair, Reference Liu and Mair2023) to ‘promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation’ (Entman, Reference Entman1993; Leung, Xue, & Wen, Reference Leung, Xue and Wen2019). It is often manifested as subtle linguistic and syntactical changes in discourse (Tang & Gray, Reference Tang and Gray2021). Previous research has revealed that different framing of the same issue – for example, in a passive tone or a positive tone – can induce differential emotional responses and attentional processes in information recipients (Druckman & McDermott, Reference Druckman and McDermott2008). Following this idea, we focus on the ‘tone’ that retail investors utilize when raising environmental issues, and propose that the priority of managerial attention to online environmental activism is enhanced by NQSs.
When launching online environmental activism, some retail investors provide constructive suggestions to companies in a polite manner while others choose to express negative sentiments and urge immediate corporate attention. By using a negative tone while pointing out environmental risks in a firm's operations, or accusing a company of environmental violations (Wu, Gao, et al., Reference Wu, Gao, Chan and Cheng2022), retail investors convey the severity of corporate environmental issues to their investor peers and to a broader public audience. Online environmental activism thus can evoke collective emotions and trigger stronger public pressure on the firm (Deng et al., Reference Deng, Huang, Sinha and Zhao2018). Social psychology studies have almost unanimously concluded that individuals are typically more responsive to negative stimuli than to positive clues (Fu, Tang, & Chen, Reference Fu, Tang and Chen2020), and such negativity bias in attention allocation (Smith et al., Reference Smith, Larsen, Chartrand, Cacioppo, Katafiasz and Moran2006) is particularly evident in cyberspace. Since individual emotions are easily heightened by seditious and even uncivil discourse with intense emotions, negative information exhibits greater contagiousness and penetrates faster into online communities (Rozin & Royzman, Reference Rozin and Royzman2001). In this regard, online questioning with strong negative sentiments is more likely to trigger a strong emotional resonance among the public audience (Asker & Dinas, Reference Asker and Dinas2019; Kramer, Guillory, & Hancock, Reference Kramer, Guillory and Hancock2014) and trigger intensive cyber-campaigns or even destructive consequences for a firm within a short time.
Spurred by the negative framing of environmental issues, an online backlash and increased scrutiny of the firm can rapidly come to dominate the virtual public sphere with the growing consensus about corporate unethical environmental behaviors and management defects, thus amplifying the reputational threats of online environmental activism. Such a negative spotlight propels corporate managers to reschedule their sequence of attention allocation and place a higher attention priority on environmental issues; that, in turn, accelerates corporate concessions to activists’ demands (Zhang & Luo, Reference Zhang and Luo2013). Hence, target firms are more likely to introduce green innovation to demonstrate their commitment to fulfill their social responsibilities, restore their public image in a timely fashion (Chen, Reference Chen2008), and try to reverse the negative public impression. Thus, we suggest that:
Hypothesis 3 (H3): The positive effect of online environmental activism on corporate green innovation is stronger if the sentiment in environmental questions is negative.
Heterogeneous Responses Across Digital and Non-Digital Firms
The rule of structural distribution of attention in ABV indicates that situated attention of decision-makers is further regulated by organizational attention structures. Considering the pervasiveness of digital platforms and technologies, managers have to adapt their attention structure and develop new sense-making routines to understand and navigate through extended digital networks. Extant literature has indicated that digitalization not only transforms the way of how firms coordinate business processes and interact with their stakeholders (Elia, Margherita, & Passiante, Reference Elia, Margherita and Passiante2020) but also introduces drastic changes to their attention allocation patterns (Hanelt, Bohnsack, Marz, & Antunes Marante, Reference Hanelt, Bohnsack, Marz and Antunes Marante2021). On the one hand, from the cognitive perspective, highly digital firms usually have formed a shared digital identity within the organization (Bouncken & Barwinski, Reference Bouncken and Barwinski2021), which propels them to foster a more inclusive culture and place greater attention on minority voices, especially those from online channels. On the other hand, from the operational aspect, digitalization streamlines organizational architecture into flatter structures empowered by digital intelligence (Volberda, Khanagha, Baden-Fuller, Mihalache, & Birkinshaw, Reference Volberda, Khanagha, Baden-Fuller, Mihalache and Birkinshaw2021) wherein the efficiency of attention reallocation and alignment is enhanced particularly in the face of external threats. In this vein, we explore whether the moderating effects of retail investor base and NQS vary across highly digital and non-digital firms with disparate attention structures.
First, a shared digital identity establishes a strong cognitive foundation for digital firms to place greater amount of attention on online voices and treat a broad base of retail investors as a prominent online group who may yield considerable social impacts. Digital firms often have developed diverse digital touchpoints (Lemon & Verhoef, Reference Lemon and Verhoef2016), including mobile applications, E-commerce channels, intelligence supply chain management platforms, online shareholder meetings, investor forums, and official social media accounts, to pursue business growth and enhance connectivity with different types of stakeholders (Nambisan, Wright, & Feldman, Reference Nambisan, Wright and Feldman2019). As highly embedded in a digitalized value chain, digital firms have typically formed a shared digital identity that attributes cognitive focus to the dynamics of online contexts among corporate decision-makers and actors (Vial, Reference Vial2019). Under the aegis of such a digital identity, digital firms usually grasp a more comprehensive and deeper understanding of the power of the Internet in mobilizing resources (Giustiziero et al., Reference Giustiziero, Kretschmer, Somaya and Wu2023), aggregating dispersed grassroots into a formidable force, altering public opinion, and reshaping corporate image (Matarazzo, Penco, Profumo, & Quaglia, Reference Matarazzo, Penco, Profumo and Quaglia2021). Accordingly, when confronted with online environmental activism, digital firms are more inclined to recognize a large retail investor base as a prominent online activist group and allocate greater attention to their environmental voices.
Furthermore, as the ever-expanding digital footprint enhances information diffusion and corporate visibility in cyberspace (Jin et al., Reference Jin, Monfort, Chen, Xia and Wu2024), for digital firms with a larger retail investor base, an increased volume of heated debates, online scrutiny, and even boycotts among the retail investor group is likely to ensue from the rise of online activism. Therefore, to prevent online concerns from permeating a large retail investor base and maintain a positive digital identity in the eyes of a broad range of online audience, digital firms have to increase attention intensity and resource commitment to address online environmental activism. They can leverage various digital tools, including automatic detection of social media content, online reputation evaluation algorithms, and artificial intelligence-assisted customer interaction techniques (Jin et al., Reference Jin, Monfort, Chen, Xia and Wu2024), to detect the diffusion path of online environmental activism and negative public opinion. By announcing green innovation on those digital channels contaminated by environmental activism, for instance on investor interactive platforms, digital firms can effectively clear up online environmental concerns and appease the retail investor group.
In contrast, since non-digital firms have not yet experienced the enabling effects of digitalization, when attacked by online environmental activists, they tend to treat a large retail investor base as grassroot individuals who have only trivial influence from a traditional cognitive view. Due to the lack of universal digital touchpoints and online channels supported by solid digital infrastructures and sophisticated big data analytical techniques, they possess limited information processing capacity (Wiesböck, Hess, & Spanjol, Reference Wiesböck, Hess and Spanjol2020) and thus often fail to keenly capture online environmental voices and perceive their potential ripple effects spreading through the large retail investor group. In such circumstances, despite a substantial retail investor base, non-digital firms are unlikely to heighten attention intensity on online environmental activism. Based on the foregoing, we posit that:
Hypothesis 4 (H4): The positive moderating effects of a large retail investor base are stronger for highly digital firms.
Compared with their non-digital counterparts that operate mainly in an offline mode, digital firms who serve a broader range of online audience are more susceptible to the widespread dissemination of negative environmental questioning (Fu et al., Reference Fu, Tang and Chen2020). Digital firms typically provide digital products or services and maintain stakeholder relationships through online channels such as social media, digital platforms, and branded mobile applications (Constantiou & Kallinikos, Reference Constantiou and Kallinikos2015; Lemon & Verhoef, Reference Lemon and Verhoef2016). Yet, these abundant digital touchpoints make negative environmental information targeting on digital firms spread faster and yield more devastating impact. For example, a public backlash triggered by negative sentiments may induce a breakdown of trust between primary partners and customers, thus causing severe reputational damages to digital firms. Therefore, driven by greater urgency to prevent emotional contagion in cyberspace, digital firms have to place a higher attention priority on pacifying negative environmental questioning. From the operational perspective, we propose that with flatter organizational structures empowered by digital intelligence, highly digital firms can efficiently reallocate and align attention across different levels of management, thus exhibiting greater agility in adjusting attention priority to pacify negative sentiments that may surge from online environmental activism in a short timeframe.
First, digitalization reshapes corporate architecture toward flatter and more flexible structures that foster the emergence of agile and cross-functional teams (Maedche, Reference Maedche2016). Such streamlined organizational architecture enables more efficient scanning and collective sense making of environmental clues (Joseph & Ocasio, Reference Joseph and Ocasio2012), which leads to more swift adjustment of attention allocation sequence to external dynamics and newly emerged crises (Vial, Reference Vial2019). As such, digital firms can rapidly shift their attention priority toward negative emotions and public opinion polarization that is evident in online environmental activism. This fast attention shifting allows digital firms to adopt prompt responses before further reputational losses are incurred.
Second, assisted by various digital intelligent technologies, digital firms are more capable of achieving intra-organizational attention integration, which facilitates the consensus formation and rapid development of green innovation solutions in response to environmental crises induced by negative investor sentiments. The adoption of real-time communication tools, machine learning and other digital intelligent techniques can greatly promote information dissemination and opinion convergence across different organizational levels to help managers make informed collective decisions at a rapid pace (Lee, Reference Lee2017). In this regard, it takes less time and fewer efforts for digital firms to align organization-wide attention toward negative sentiments triggered by online environmental activism and establish a shared understanding on the necessity of green innovation among senior managers and frontline R&D managers in a timely manner (Song, Fisher, & Kwoh, Reference Song, Fisher and Kwoh2019). Therefore, digital firms are more likely to launch green innovation in a shorter lead time and recover far sooner from the tarnished public image than non-digital firms do.
On the contrary, non-digital firms, confined by hierarchical structure and traditional organizational routines that entail higher coordination costs, typically place attention priority on offline production, selling, and other issues that emerge in the physical environment. As a result, they may fail to perceive the urgent and disruptive threats induced by negative emotions in cyberspace, thus exhibiting delayed attention adjustment and lower responsiveness to online environmental activism. Based on the above discussion, we propose that:
Hypothesis 5 (H5): The positive moderating effects of negative questioning sentiment are stronger for highly digital firms.
We present the overall research framework in Figure 1.
Research framework
Methods
Data
China is an exciting laboratory for testing the impact of online environmental activism from retail investors on green innovation. With the rapid development of digital platforms, online interactions between investors and publicly listed firms in China is gaining increasing academic attention (Huang & Ying, Reference Huang and Ying2022) while the impact on corporate green innovation remains rarely explored. Moreover, as an efficient strategy to address environmental issues and establish competitiveness in the emerging green sector, green innovation is crucial for China to achieve its dual carbon targets and accelerate technological catch-up in the green sector.
Our initial sample consists of Chinese firms listed on the Shanghai and Shenzhen Stock Exchanges from 2011 to 2018. Financial and governance data on these firms mainly come from the China Stock Market & Accounting Research (CSMAR) database. We collect on corporate green patents and transcripts of institutional investor on-site interviews from the China Research Data Service Platform (CNRDS) database. We use the Python program to collect environmental inquiries on the Interactive Easy Platform and the SSE E-Interactive Platform from their official websites (Lee & Zhong, Reference Lee and Zhong2022). Specifically, we obtain unbalanced panel data from the CSMAR database for 3,775 listed companies with complete basic operational data from 2011 to 2018, resulting in a total of 26,442 samples. We follow the general practice in the literature to screen our initial sample (Li, Wang, et al., Reference Li, Wang and Zhang2022; Wu et al., Reference Wu, Chen, Li, Xu, Liu and Wu2024). First, we eliminate listed firms in financial and insurance industries due to the considerable differences in ESG criteria and green practices between these two industries and others (Chen, Chen, & Jebran, Reference Chen, Chen and Jebran2021). Second, we delete firms that did not disclose detailed data on relevant variables to ensure the quality of the data. Third, we exclude firms labeled for special treatment (ST/*ST/PT) by stock exchanges since these firms are not normally traded in the stock market (Huang, Mbanyele, Wang, Song, & Wang, Reference Huang, Mbanyele, Wang, Song and Wang2022). These steps ultimately left us with obtain 13,795 firm-year observations. To avoid the influence of extreme values, all variables except dummies are winsorized at the top and bottom 1%.
Measures
Dependent variable
Green innovation
Following the literature (Amore & Bennedsen, Reference Amore and Bennedsen2016; Ho, Shen, Yan, & Hu, Reference Ho, Shen, Yan and Hu2023), we use the number of green patent applications to measure corporate green innovation. As patent output requires considerable R&D investment and patent filing costs, green patent applications can reflect corporate substantial efforts in addressing environmental issues. Considering the development of green patents is a lengthy process, we measure green innovation as the number of green patent applications in one-year forward (year t + 1) (Cui, Dai, Wang, & Zhao, Reference Cui, Dai, Wang and Zhao2022). The relevant data are obtained from the Green Patent Research Database (GPRD) on the CNRDS. This database selects green patents from all the patents filed by Chinese publicly listed firms following the IPC Green Inventory list published by the World Intellectual Property Organization (WIPO).
Independent variables
Online environmental activism
We construct two variables to measure online environmental activism using the following steps. First, referring to the existing literature (Li, Qian, et al., Reference Li, Qian, Wang, Wang and Wang2022), we use the Python web crawler technologies to retrieve the questions proposed by retail investors on the SZSE Interactive Easy Platform and the SSE E-Interactive Platform. Second, drawing on Chen, Kahn, Liu, and Wang (Reference Chen, Kahn, Liu and Wang2016), we employ text analysis to identify all of the questions that contain environmental-related keywords.Footnote 4 Third, we use two variables to measure online environmental activism. OEA1 is calculated by the natural logarithm of the number of environmental questions targeting a firm i in year t plus one. OEA2 is measured as the number of environment-related questions divided by the total number of questions that a company i received in year t and multiplied by 100 for interpretation convenience.
Retail investor base
Following prior research (Brav, Cain, & Zytnick, Reference Brav, Cain and Zytnick2021), we measure the retail investor base by the aggregate shareholding proportion of retail investors in a company.
Negative questioning sentiment
We use text analysis to construct the NQS. After obtaining the text of environment-related questions, we correct common typos in the text and delete stop words. We then segment the text data based on Chinese Natural Language Processing (NLP) and tokenize the text data with the Natural Language Toolkit (NLTK). Next, similar to Zhang, Wei, Wang, and Liao (Reference Zhang, Wei, Wang and Liao2018), we use the emotional dictionary of informal terms that include emojis to identify positive and negative words. Finally, two methods are utilized to construct the indicators of NQS. NQS1 is the number of negative words minus the number of positive words divided by the total words in environmental questions targeting a firm in each year. NQS2 is the number of negative words minus the number of positive words divided by the total number of positive and negative words in environmental questions targeting a firm in each year. The higher the value of these two indices, the greater the negative sentiment of environment-related questions presented by retail investors.
Digitalization
Referring to the existing literature (Shang, Raza, Huo, Shahzad, & Zhao, Reference Shang, Raza, Huo, Shahzad and Zhao2023; Wu, Fu, & Kong, Reference Wu, Fu and Kong2022), we use the text analysis method to calculate the frequency of digital-related keywords in the annual reports published by listed companies as a proxy indicator for corporate digitalization level. We first collect the annual reports of all A-share listed companies on the Shanghai and Shenzhen Stock Exchanges using the Python web crawler and employing the Java PDFbox library to extract all the text. Next, following Wu, Hu, Lin, and Ren (Reference Wu, Hu, Lin and Ren2021), we identify a total of 76 keywords of digitalization, which are presented in Table A2 in the Appendix. We then count the number of digitalization-related words in each annual report for each firm. To avoid the influence of extreme values, we take the logarithm of the word frequency to measure the digitalization level of each firm in each year.
Control variables
We control for a series of variables that may have a significant impact on green innovation, including corporate financial position and governance structure (Chen & Yang, Reference Chen and Yang2019; Zhou, Gao, & Zhao, Reference Zhou, Gao and Zhao2017). For financial position, leverage (Lev) is the ratio of total liabilities to total assets. Return on equity (Roe) is the ratio of net income to average shareholders’ equity. Cash holdings (Cash) are equal to cash and cash equivalents divided by the difference between total assets and cash and cash equivalents. Tobin's Q is the ratio of market value to replacement cost. For corporate governance structure, asset size is the size of the total assets in annual reports and is transformed into its natural logarithm to reduce the potential effects of extreme values. The R&D ratio refers to the proportion of total R&D expense to total sales revenue of a firm in the same year. Executive size is the natural logarithm of the number of senior managers in a firm. Bnum is the natural logarithm of the number of board meetings in a year. Listed years are measured as a natural logarithm of the time gap between the current year and the year of an IPO. In addition to retail investor activism, the literature has widely confirmed that institutional investor activism also serves as a critical driver for corporate green innovation (Wu, Gu, Liu, & Liu, Reference Wu, Gu, Liu and Liu2023). Thus, we measure institutional investor environmental activism (IIEA) by the number of environment-related words in the Q&A part of institutional investor on-site interviews, which indicates the intensity of majority shareholder environmental advocacy toward firms. We introduce three sets of dummies to control for year, industry, and province fixed effects.
Model
Considering that the number of green patents is a count variable that takes non-negative integral values, ordinary linear regression can generate systematic bias. Thus, to examine the impact of online environmental activism on green innovation, following N'Guessan (Reference N'Guessan2021), we employ the Poisson pseudo-maximum likelihood model with high-dimensional fixed effects, which is a recently improved Poisson model that can well accommodate heteroscedasticity problems and allow for faster computation when controlling for the multiple sources of heterogeneity (Correia, Guimarães, & Zylkin, Reference Correia, Guimarães and Zylkin2020), such as year, industry, and region fixed effects. We employ the Poisson model with robust Huber-White standard errors, which allows us to adjust for over-dispersion and produce consistent estimates (Mawdsley, Paolella, & Durand, Reference Mawdsley, Paolella and Durand2023). A Stata command for HDFE-Poisson estimation called ‘PPMLHDFE’ is implemented.
Results
Main Results
The descriptive summary and correlation matrix for the main variables are presented in Table 2. The correlation coefficients between explanatory variables are all below 0.6, indicating no strong correlation between the variables. We also calculate the Variance Inflation Factor (VIF) of the main variables, which have values far below the threshold 10, indicating that our analysis does not suffer from multicollinearity issue.
Summary statistics and correlation matrix of main variables
Notes: On the left is the Pearson correlation coefficient and on the right is the Spearman correlation coefficient. N = 13,795. * represents significant correlation at the level of 5%.
Table 3 shows the regression results of main effect and moderating effects of retail investor base. After including all control variables, online environmental activism is significantly and positively associated with green innovation (β OEA1 = 0.049, p < 5% in Model 1; β OEA2 = 0.010, p < 1% in Model 2), which supports H1. Since our model is nonlinear, we estimate marginal effects to better understand the economic significance of the main effect. It shows that one unit increase in OEA1, i.e., additional 1.72 environment-related question posted by retail investors on the online interactive platforms (e1–1 ≈ 1.72), leads to 3.9% increase in the number of green patent applications in the following year. In addition, 1% increase in the proportion of environment-related questions is associated with 0.8% increase in the number of green patents applications. The impact of control factors is also in line with prior literature.
The main effect and moderating effects of retail investor base
Notes: ***, ** and * represent significant correlation at the level of 0.1%, 1% and 5%, respectively. Clustered standard errors are shown in parentheses.
Model 3 and 4 in Table 3 presents the moderating effects of the retail investor base. The coefficient of the interaction term between retail investor base and online environmental activism is positive and significant (β OEA1×Retail investor base = 0.185, p < 5%; β OEA2×Retail investor base = 0.029, p < 1%). Moreover, the average marginal effect of OEA1 on green innovation increases from 0.014 (insignificant) at a low level of retail investor base (Mean – 1SD) to 0.078 (p < 1%) at a high level of retail investor base (Mean + 1SD). Similarly, the marginal effect of OEA2 also becomes larger at a high level of retail investor base (MA Low retail investor base = 0.004, insignificant; MA High retail investor base = 0.014, p < 0.1%). We also plot the marginal effects of online environmental activism at different levels of retail investor base in Figure 2. The positive effect of online environmental activism on corporate green innovation becomes stronger when retail investor base is larger, thus supporting H2.
Marginal effects of online environmental activism at different levels of retail investor base
Table 4 presents the moderating effects of NQS on the relationship between online environmental activism and corporate green innovation. The coefficients of two-way interaction term are positive and significant (β OEA1×NQS1 = 0.135, p < 1%; β OEA1×NQS2 = 4.523, p < 0.1%; β OEA2×NQS1 = 0.010, p < 10%; and β OEA2×NQS2 = 0.195, p < 10%). The marginal effects of OEA1 are insignificant (MA Low NQS1 = −0.016, insignificant; MA Low NQS2 = −0.038, insignificant) at a low level of NQS and turn significant and positive (MA High NQS1 = 0.106, p < 1%; MA High NQS2 = 0.130, p < 0.1%) at a high level of NQS. The marginal effects of OEA2 also becomes stronger at a high level of NQS (MA Low NQS1 = 0.002, insignificant; MA Low NQS2 = 0.003, insignificant; MA High NQS1 = 0.012, p < 1%; MA High NQS2 = 0.011, p < 1%). We also depict the marginal effects of online environmental activism at different levels of the NQS in Figure 3. All the above results confirm H3.
The moderating effects of negative questioning sentiment
Notes: ***, **, * and + represent significant correlation at the level of 0.1%, 1%, 5%, and 10%, respectively. Clustered standard errors are shown in parentheses.
Marginal effects of online environmental activism at different levels of negative questioning sentiment
Table 5 shows whether digitalization enhances the moderating effects of retail investor base. The coefficient of three-way interaction term of online activism, retail investor base, and digitalization is positive and significant (β = 4.320, p < 5% in Model 1; β = 0.620, p < 1% in Model 2), which supports H4. We also estimate the marginal effect of online activism at different levels of retail investor shareholding and also different levels of digitalization, which is significant and positive only at both a high level of retail investor base and a high level of digitalization (MA High retail investor base, high digital = 0.173, p < 0.1% for OEA1; MA High retail investor base, high digital = 0.023, p < 0.1% for OEA2). We also plot the marginal effects of online environmental activism at different levels of retail investor base and digitalization in Figure 4. It shows that a large retail investor base insures the positive impact of online environmental activism only for highly digital firms.
The moderating effects of retail investor base considering digitalization
Notes: ***, **, * and + represent significant correlation at the level of 0.1%, 1%, 5%, and 10%, respectively. Clustered standard errors are shown in parentheses.
Marginal effects of online environmental activism at different levels of retail investor base and digitalization
The models in Table 6 examine whether digitalization strengthens the moderating effects of NQS. When we measure online activism and questioning sentiment with different methods, the three-way interaction terms are all positive and significant. The marginal effect of online activism is significant and positive only at both a high level of NQS and a high level of digitalization. We depict marginal effects of online environmental activism at different levels of NQS and digitalization in Figure 5. The above results suggest that the moderating effects of NQS are stronger for highly digital firms, which lends support for H5.
The moderating effects of negative questioning sentiment considering digitalization
Notes: ** and * represent significant correlation at the level of 1% and 5% respectively. Clustered standard errors are shown in parentheses.
Marginal effects of online environmental activism at different levels of negative questioning sentiment and digitalization
Robustness Tests
To increase the credibility of our empirical results, we conduct a series of robustness tests. Due to space limitations, we present the results of robustness checks in the Appendix. We adopt alternative measure for the dependent variable and different regression models. In addition, as a common specification problem in empirical analysis, endogeneity may affect our results. We use five methods to address it, including sensitivity analysis, inference to replacement (RIR) analysis, instrumental variable (IV) analysis, the two-stage Heckman selection model, and entropy balancing.
Discussion
Our study contributes to the literature in three ways. First, by integrating the literature of online opinion dissemination and the attention-based theory of corporate responses, we propose that through effective social mobilization and collective emotion incitement, online environmental activism can enhance both the intensity and priority of managerial attention to environmental issues, which underpin corporate substantial responses. Although some recent studies have suggested that online green concerns can direct managerial attention to environmental issues (Li, Qian, et al., Reference Li, Qian, Wang, Wang and Wang2022), how online concerns diffuse in cyberspace, alter public opinion, and ultimately reshape corporate attention allocation has not yet been fully explained.
In our study, we argue that owing to costless social mobilization on the Internet (Wang et al., Reference Wang, Xing, Wei, Zheng and Xing2020), environmental voices from a small group of retail investors can easily scale up to a large movement with contagion effects (Earl & Schussman, Reference Earl and Schussman2004) and push firms, especially those with a large retail investor base, to allocate a greater amount of attention to the issue. Moreover, by leveraging high-arousal emotional expressions, such as environmental questioning with a strong negative sentiment, online environmental activism can trigger extreme emotions and opinion polarization, which urges firms to place a higher attention priority on environmental issues. Both heightened attention intensity and priority are conducive to corporate resource commitment and their prompt adoption of green innovation. Thus, we unveil a nuanced picture of how corporate attention allocation adapts with dynamically evolving public opinion in cyberspace, thus improving the explanatory power of ABV in the online setting.
Second, this article shows the importance of minority voicing on social issues and the contributing role of retail investors in corporate governance by demonstrating the positive impact of retail investors’ environmental voicing on green innovation in an emerging market context. Prior literature has widely shown the effectiveness of majority shareholder activism (DesJardine, Shi, & Marti, Reference DesJardine, Shi and Marti2023; Shi, Connelly, Hoskisson, & Ketchen, Reference Shi, Connelly, Hoskisson and Ketchen2020; Shi, Xia, & Meyer-Doyle, Reference Shi, Xia and Meyer-Doyle2022), while the voices of minority shareholders are often disregarded (Kastiel & Nili, Reference Kastiel and Nili2016; Lin et al., Reference Lin, Liao and Xie2023). But such ignorance of minority voicing, especially on social issues, may lead to over-concentration of majority interests and reduce the diversity of external opinions in decision making (Gompers & Kovvali, Reference Gompers and Kovvali2018; Hafeez et al., Reference Hafeez, Kabir and Wongchoti2022).
Particularly in the context of emerging countries such as China that is in the economic catching up phrase, most firms may prioritize economic agendas and lack incentives to invest in risky and costly green innovation (Laukkanen & Patala, Reference Laukkanen and Patala2014). Under such circumstances, grassroots retail investors who are more concerned with individual well-beings and public health, constitute an indispensable force in promoting green development. Although existing studies have provided some preliminary support for the idea that that online minority shareholder activism can lead to information efficiency and greater liquidity in capital markets (Lee & Zhong, Reference Lee and Zhong2022), exactly how such activism alleviates social issues remains largely underexplored. In this study, we show that online environmental activism by retail investors can successfully promote green innovation; we thus add to a nascent stream of literature on leveraging minority voicing to accelerate social transformation and create green value (Wang & Li, Reference Wang and Li2023; Yao et al., Reference Yao, Pan, Wang, Sensoy and Cheng2023).
Third, our study confirms that attention distribution and responses to online activism vary across digital and non-digital firms, thus adding to a growing body of literature advocating that digitalization introduces drastic changes to organizational architecture and reshapes the way firms scan the environment and engage with stakeholders (Giustiziero et al., Reference Giustiziero, Kretschmer, Somaya and Wu2023; Mithani, Reference Mithani2023). Our results suggest that the moderating effects of both retail investor base and NQS are stronger for highly digital firms. Since digital firms that rely heavily on information technologies typically have a deep understanding of the power of the Internet in mobilizing resources and generating profound social impacts (Matarazzo et al., Reference Matarazzo, Penco, Profumo and Quaglia2021), they tend to perceive a large retail investor base as a prominent group rather than as dispersed individuals, thereby devoting adequate attention and resources to pacify online concerns from this group. Digital firms also exhibit greater agility in tackling urgent crises that may develop as a result of the spread of negative emotions. Thus, they are more capable of swiftly adjusting attention sequence and prioritizing environmental concerns. Overall, we provide a better understanding of why firms with digital architecture are more responsive to online attacks in the Internet age.
These findings carry practical implications for both retail investors and for companies. Retail investors should make full use of the Internet and digital technologies to advance shareholder activism and proactively convey their demands to firms. Official interactive platforms – such as the Interactive Easy Platform and the SSE E-Interactive Platform in China – provide individual investors with a legitimate channel to actively engage in corporate governance, increasing their salience as environmental activists and ensuring their voices are heard by the top management of listed firms. Through information dissemination in online communities, retail investors can share the identified investment risks of the firm, mobilize their investor peers and even the general public, to take collective actions and scale their activism up. Retail investors can also strategically devise their rhetoric to highlight the urgency of their online requests and evoke stronger emotional reactions in cyberspace, so as to compel firms to take immediate action to change the status quo.
Companies should pay more attention to the voices of retail investors when making sustainable investment decisions, since potential insights from minority shareholders may complement the mainstream views of institutional investors. To encourage retail investors to get involved in corporate governance, companies are advised to actively participate in online interactions with retail investors. In addition to maintaining the relationship with majority investors via on-site visits, firms should be aware of the pervasive influence that the online activism of retail investors may produce and strive to prevent its potentially negative impact on corporate reputations. For example, companies can set up a regular information division to monitor and pacify online voices within a short time. Alternatively, a monitoring position can be established such as a chief sustainability officers (CSOs) as an internal attention carrier to channel managerial attention to social issues (Fu et al., Reference Fu, Tang and Chen2020). Moreover, digital transformation is essential for companies to effectively deal with short-term disruptions and achieve long-term sustainable growth. Specifically, companies should integrate advanced digital technologies such as machine learning, big data analytics, and artificial intelligence into production management, organizational operations, and innovation activities. By replacing traditional technologies with advanced digital technologies, companies can optimize resource allocations both internally and externally, and enhance their responsiveness to online environmental activism.
Limitations and Suggestions for Future Research
Our study is not without limitations, but these suggest opportunities for future research. First, our measurement may fail to capture some of the nuances of online environmental activism. We use the number and the proportion of environmental questions to measure the overall intensity of online environmental questioning, but future research is needed to differentiate these questions by categories. Different categories of questions may reflect distinct concerns about environmental protection, thus inducing different types of corporate environmental endeavors.
Second, we focus only on questions raised by retail investors, without considering the answers provided by managers (Duan, Li, Park, & Wu, Reference Duan, Li, Park and Wu2023; Wu, Gao, et al., Reference Wu, Gao, Chan and Cheng2022). Communication is a two-way interaction between investors and companies, and the quality of answers may indicate the extent to which companies value retail investors’ concerns and suggestions. Future research can measure the quality of corporate replies, so as to provide a more complete picture of environmental activism on online platforms.
Third, although we explore how firms with varying levels of digitalization respond differently to online environmental activism, it cannot comprehensively reflect firm heterogeneity. To unveil heterogenous responses across different types of firms, future studies can explore other corporate features, such as firm size, financial standing, reputation, corporate governance structure.
Finally, we argue that retail investors can share their environmental concerns with their investor peers and mobilize inactive investors and even those with opposite values into activist group through information dissemination and online social interactions. Future research can employ data mining and social network analysis to unveil how green concerns from a handful of investors diffuse among the whole investor group. In doing so, we can empirically investigate the mechanisms of social mobilization and emotion incitement that underpin the success of online environmental activism.
Conclusion
In the digital era, minority shareholders can leverage online platforms to advance shareholder activism, thereby enjoying greater power in directing managerial attention to their specific interests. In this study, drawing on an ABV, we investigate whether, and if, online environmental activism by retail investors elicits substantial corporate environmental commitment. Our empirical results – based on a sample of 13,795 firm-year observations of Chinese-listed companies over 2011 to 2018 – confirm that online environmental activism is positively associated with corporate green innovation. We find that the effect of online environmental activism on corporate green innovation is stronger for firms with a larger retail investor base. In terms of the framing of online activism, our results suggest that online environmental questions with a negative sentiment are more likely to induce corporate green innovation. Furthermore, the moderating effects of both retail investor base and NQS are stronger for highly digital firms.
Data Availability Statement
The data that support the findings of this study are openly available in the Open Science Framework at https://osf.io/2xhyk/?view_only=56d7e6833d074c0daa347ea9c6af2619 Part of the derived data is available from the corresponding author on request.
Acknowledgements
This study was funded by the Key Project of National Social Science Foundation (23AGL016).
Appendix
Vocabulary Lists
For the determination of feature words for corporate digitalization, we follow the study of Wu, Hu, et al. (Reference Wu, Hu, Lin and Ren2021) and include a total of 76 keywords across five dimensions: artificial intelligence technology, blockchain technology, cloud computing technology, big data technology and big data technology application. Digitalization-related vocabulary is presented in the following table.
Robustness Tests
Substitution for the Dependent Variable
To ensure the credibility of our empirical results, we conduct a series of robustness tests. One important concern is that the empirical results might differ when using different measures for green innovation. Therefore, we replace the dependent variable with different measures of green innovation variables.
First, we replace green patents with low-carbon patents as dependent variable and re-test the model since low-carbon patents are viewed as important categories of green innovation in prior literature (Sun, Chen, Yang, Ying, & Niu, Reference Sun, Chen, Yang, Ying and Niu2022). As shown in Table A2, online environmental activism has a significant positive impact on corporate low-carbon innovation after controlling for other factors (the coefficient of OEA1 = 0.059, p < 5%; the coefficient of OEA2 = 0.011, p < 0.1%), which is consistent with our main results.
Second, there are considerable differences between green invention patents and green utility patents. In general, the granting of an invention patent demands a higher level of novelty and originality and occurs after a vigorous process called substantive examination. Utility patents meet lower standards (e.g., they have fewer technical merits) than invention patents do (Ma, Hu, Shen, & Wei, Reference Ma, Hu, Shen and Wei2021). Thus, we conduct analyses based on these two patent categories respectively. Table A3 shows that online environmental activism has a significant positive impact on green invention patents (the coefficient of OEA1 = 0.042, p < 10% in Model 1, and the coefficient of OEA2 = 0.010, p < 1% in Model 2). Online environmental activism also has a significant positive impact on green utility patents (the coefficient of OEA1 = 0.060, p < 5% in Model 3, and the coefficient of OEA2 = 0.008, p < 5% in Model 4).
Third, since granted green patents can confer exclusive rights and the competitive advantages to companies (Ho et al., Reference Ho, Shen, Yan and Hu2023), we choose the number of green patents granted to each company as a proxy variable for green innovation. According to prior literature, it usually takes around 2 to 3 years from patent application to patent grant in China (Yang, Reference Yang2008). Therefore, we measure green innovation with the logarithm number of green granted patents in year t + 1, t + 2, t + 3 respectively for robustness check. As shown in Table A4, online environmental activism has a significant positive impact on granted green patents with different time lags, consistent with our previous results.
Alternative Measure for IIEA
In the main analysis, we include the number of environment-related words appearing on the Q&A part of offline investor records to control for the potential impact of IIEA. To enhance the robustness of results, we also use the proportion of environment-related words to the total number of words in the Q&A part of on-site interviews to measure IIEA. As results in Table A5 indicate, indicate after controlling for the impact of IIEA in a proportion-based approach, there still exists a significant positive association between online environmental activism and corporate green innovation (the coefficient of OEA1 = 0.044, p < 10% in Model 1, and the coefficient of OEA2 = 0.010, p < 1% in Model 2), supporting H1.
Substitution for the Regression Models
We rely on Poisson pseudo-maximum likelihood regressions (PPML) with multi-way fixed effects to test a series of hypotheses in the main analyses. To avoid the possible influence of different regression methods on research results, we replace the PPMLHDFE regression model with three different regression models to re-test the main hypothesis. The empirical results are shown in Table A6. First, we use the panel data Tobit model (xttobit). Model 1 shows that online environmental activism has a significant positive impact on corporate green innovation (the coefficient of OEA1 = 0.011, p < 1%). Second, we employ linear regression models with multi-way fixed effects (reghdfe) (Correia, Reference Correia2016). Model 2 shows that our main hypothesis still holds (the coefficient of OEA1 = 0.047, p < 5%). Third, since not all companies apply for green patents, we redefine green innovation as a dummy variable that equals 1 when the firm applied for at least one green patent, and 0 otherwise. We thereby use panel data Probit model (xtprobit) to re-test the hypothesis. As Model 3 suggests, the impact of online environmental activism on green innovation is significant and positive (the coefficient of OEA1 = 0.078, p < 0.1%), confirming the robustness of our results.
Dealing with the Endogeneity Issue
As a common specification problem in empirical analysis, endogeneity may affect our results. We use five methods to address it.
Sensitivity Analysis
First, we conduct the sensitivity analysis to test potential bias in the results due to omitted variables. Oster (Reference Oster2019) develops an approach to to examine the impact of omitted variable bias by using bounds on R max and δ to develop a set of bounds for β. Here, δ represents the selection proportionality, measuring the strength of the correlational relationship between observable variables and the variable of interest relative to the correlational relationship between unobservable omitted variables and the variable of interest. R max indicates the maximum goodness of fit of the regression equation if the unobservable omitted variables could be observed. Following Oster (Reference Oster2019), we set δ = 1 as an appropriate bound and R max takes 1.3 times the current regression goodness of fit. The results pass the robustness test if β*(R max, δ) falls within the 95% confidence interval of the estimated parameters. By using psacalc command, we find that for OEA1, β*(R max, δ) = 0.024 falls within the 95% confidence interval of the estimated parameters [0.015, 0.110]. For OEA2, β*(R max, δ) = 0.006 falls within the 95% confidence interval of the estimated parameters [0.002, 0.014]. This indicates that after controlling for omitted variable bias, our main results still hold. Besides, we calculate the value of δ that makes the estimated β = 0. For OEA1, δ = 1.557 > 1 and for OEA2, δ = 3.513 > 1, suggesting that unobservable factors must be more than 1.557 or 3.513 times as important as observable and controlled factors to produce no treatment effect. Overall, the sensitive analysis indicates that unobservable factors are unlikely to drive our main results.
RIR Analysis
To further examine the omitted variable bias we have conducted a robustness of RIR analysis (Frank, Reference Frank2000), which is equivalent to the impact threshold of a confounding variable (ITCV) analysis but more appropriate for nonlinear model (Roccapriore & Pollock, Reference Roccapriore and Pollock2023). The RIR provides insight into the percentage of requisite observations for which the estimated causal effect is driven entirely by an omitted variable, not by the treatment, to invalidate findings. By using konfound command, we conduct the RIR analysis and report the results in the Appendix. It suggests that 47.93% (6,612) of observations would have to be replaced with cases for which there is an effect of 0 to invalidate the observed impact of OEA1 on green innovation. In addition, 46.11% (6,361) of observations for OEA2 would need to be entirely overturned. These numbers are much higher than thresholds accepted in prior work (e.g., Busenbark, Lange, & Certo, Reference Busenbark, Lange and Certo2017). Overall, the results of the RIR analysis indicate that it is unlikely that an omitted variable is driving the results. Please refer to pages 56 and 57 for detailed information.
IV Analysis
We use the most common method of IV analysis to check for robustness. We introduce two exogenous variables as IVs which are highly correlated with the independent variable, but not with stochastic disturbance terms. At the industry level, IV1 is the average number of environmental questions and the proportion of environmental questions from other enterprises in the same industry and in the same year. At the provincial level, IV2 is the average number of environmental questions or the proportion of environmental questions from other enterprises in the same province and in the same year. We use two-stage least square (2SLS) for the analysis. We mainly report the outcomes of the second stage in Table A7. The online environmental activism measured by the number of online questions has a significant positive impact on corporate green innovation (the coefficient of OEA1 = 1.908, p < 10% in Model 1, and the coefficient of OEA2 = 0.013, p > 10% in Model 2), which lends further support for our main hypothesis.
The Two-Stage Heckman Selection Model
As noted, our goal is to examine the impact of online environmental activism on corporate green innovation. However, such a relationship may suffer from sample selection bias. This is because online environmental questioning behavior is partly determined by corporate inherent features. Firms that highlight sustainability and value the voice of minority shareholders may be more likely to attract the attention of retail investors and therefore receive more environmental questions online. This means that firms are self-selected into online environmental activism. We therefore use the two-stage Heckman selection model (Heckman, Reference Heckman1979) to correct for sample selection bias.
The first stage of the Heckman model predicts the likelihood of a firm receiving environmental questions on the interactive platforms on various factors, which is estimated by a Poisson pseudo-maximum likelihood regression model with multi-way fixed effects for the entire sample of firms. We transform the independent variable into a dummy variable of online environmental activism in the first-stage model. This variable is coded as 1 if a firm receives at least one environmental question from retail investors on interactive platforms in the current year, and 0 otherwise. Specifically, we include the corporate characteristic variables with a one-period lag and calculate an adjustment term, the inverse Mills ratio. The inverse Mills ratio from the first-stage regression is then incorporated as a control variable into the second-stage equation, which examines the relationship between online environmental activism and corporate green innovation using a smaller sample of firms who have received environmental questions on their platforms. We report the outcomes of the second stage in Table A8. Consistent with the previous results, the online environmental activism has a significant positive impact on corporate green innovation (the coefficient of OEA1 = 0.040, p < 5% in Model 1, and the coefficient of OEA2 = 0.006, p < 5% in Model 2).
Entropy Balancing
It is possible that firms with higher levels of green innovation may draw more attention from retail investors on interactive platforms, so our research design might suffer from the endogeneity problem of ‘reverse causality’. Traditional propensity score matching (PSM) suffers from the issue of sample loss, and it is often difficult to jointly balance all covariates when estimating propensity scores in practice, especially in high-dimensional data with possibly complex mechanisms. We therefore adopt entropy balancing developed by Hainmueller and Xu (Reference Hainmueller and Xu2013) to address such endogeneity issue. Entropy balancing allows users to achieve a high degree of covariate balance by imposing a possibly large set of balance constraints including the first, second, and higher moments of the covariate distributions as well as interactions. Table A9 shows that based on an entropy-balanced sample, there is still a positive and significant relationship between online environmental activism and corporate green innovation (the coefficient of OEA1 = 0.068, p < 1% in Model 2, and the coefficient of OEA2 = 0.009, p < 1% in Model 4), which confirms our main prediction.
Digitalization-related vocabulary
Substitution for the dependent variable (low-carbon patents)
Substitution for the dependent variable (green invention and utility patents)
Substitution for the dependent variable (granted green patents with 1- to 3-year lag)
Alternative measure for IIEA (the ratio of environment-related words)
Substitution for the regression models
IV Analysis
Two-stage Heckman selection model
Results based on an entropy-balanced sample
Lanhua Li (lanhuali@zjut.edu.cn) is an assistant researcher at China Institute for Small and Medium Enterprises, Zhejiang University of Technology. She received her PhD from Zhejiang University, and her research interests are in the field of technology innovation and intellectual property management. She has published research articles in well-recognized international journals.
Yuxin Zhang (zhangyx0219@163.com) is a PhD candidate at School of Management, Zhejiang University of Technology. Her research interests focus on CSR Management and sustainable development of SMEs.
Bao Wu (wubao@zjut.edu.cn) is a full Professor and the Dean of School of Management, Zhejiang University of Technology. His field of research is green innovation, sustainable development, and ESG activism. He has published in Journal of Business Research, Technological Forecasting and Social Change, The British Accounting Review, among other leading journals, and eight of his papers are ranked into Top 1% ESI highly cited papers.
