Vol.:(0123456789) Journal of Business Ethics (2025) 200:867–896 https://doi.org/10.1007/s10551-024-05859-w ORIGINAL PAPER Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility Behavior? How Stakeholder Attention Shapes Responsiveness to Stakeholders Yichen Wang1 · Christopher Marquis2 Received: 6 July 2022 / Accepted: 1 November 2024 / Published online: 21 January 2025 © The Author(s) 2025 Abstract Does product market competition (PMC) promote or reduce firms’ corporate social responsibility (CSR) behavior? While some studies suggest that CSR is a differentiation strategy that leads to a positive relationship between PMC and CSR, oth- ers consider CSR a discretionary cost that firms in competitive markets should avoid. Drawing on instrumental stakeholder theory and research on organizational attention, we aim to clarify the extent to which CSR provides a competitive advantage for firms by exploring how different types of stakeholder attention—both between different stakeholder groups and within a specific stakeholder group—affect the PMC–CSR relationship. We test our framework using data on Chinese private (i.e., non-state-owned) companies on major Chinese stock exchanges from 2008 to 2017. Overall, our analyses highlight that stakeholder attention affects how firms assess the tension between the benefits and costs of CSR and that the contingent effect of different stakeholder attention varies within and between stakeholder groups. Keywords  Stakeholder attention · Corporate social responsibility · Product market competition · Emerging markets Introduction In recent decades, there has been an intense debate on the potential financial benefits of corporate social responsibility (CSR) (see Awaysheh et al., 2020; Margolis & Walsh, 2003; Orlitzky et al., 2003 for reviews). Recently, scholars have increasingly focused on this debate in the context of product market competition (PMC) since greater competition forces firms to optimize their decisions (Allen & Gale, 2000) and so could magnify the tension between CSR costs and ben- efits. However, conflicting results have been reported regard- ing this relationship. Some studies have suggested that firms can benefit from CSR under the pressure of PMC because it is a signal of reliability and comparative advantage that differentiates them from rivals (e.g., Declerck & M’Zali, 2012; Fernández-Kranz & Santaló, 2010; Flammer, 2015; McDonnell & King, 2013). Others have viewed CSR as a discretionary cost that should be avoided in competitive markets, allowing firms to achieve cost control, maximize profits, and ensure survival (Campbell, 2007; Duanmu et al., 2018; Shleifer, 2004). To better understand these conflicting findings, we develop a theoretical framework focused on how firms’ CSR behavior in response to increasing PMC is affected by the degree of stakeholder attention. We assume that, in order to accomplish the goal of differentiation, stakeholders must be aware of and reward firms’ CSR behavior (Godfrey et al., 2009; Madsen & Rodgers, 2015; McWilliams & Siegel, 2001), otherwise it would be viewed as an unnecessary expenditure. Inspired by instrumental stakeholder theory (Jones et al., 2018) and the literature on organizational atten- tion (Hoffman & Ocasio, 2001; Ocasio, 1997, 2011), we pro- pose that stakeholder attention is limited, which shapes the trade-off between benefits and costs of CSR. Furthermore, while prior research has examined stakeholders as a uni- fied group (Barnett, 2014; Harrison et al., 2010), we argue that the contingent role of stakeholder attention varies both between different stakeholder groups (e.g., the government, * Christopher Marquis c.marquis@jbs.cam.ac.uk Yichen Wang wangyichen@cuit.edu.cn 1 School of Management, Chengdu University of Information Technology, Chengdu, China 2 Judge Business School, University of Cambridge, Cambridge CB2 1AG, UK http://crossmark.crossref.org/dialog/?doi=10.1007/s10551-024-05859-w&domain=pdf 868 Y. Wang, C. Marquis institutional investors, financial analysts, and media) and within a particular stakeholder group (e.g., the central gov- ernment versus local governments, long-term institutional investors versus short-term institutional investors, female analysts versus male analysts, and traditional media versus new media). Our study examines the distinct interests of each group and how these different and even conflicting interests can shape attention, thereby affecting how firms address the tension between the benefits and costs of CSR in response to PMC. Figure 1 illustrates the proposed theo- retical framework. We test our arguments using all private firms [as opposed to state-owned enterprises (SOEs)] in China listed on major exchanges from 2008 to 2017. This is an ideal context for our study. Because the PMC process we articulate con- cerns competitive pressure, it is less applicable to Chinese SOEs because they typically have a monopoly status in their industries or regions (Branstetter & Feenstra, 2002). Fur- thermore, SOEs are also responsible for attaining noneco- nomic goals—such as public control over natural resources, regional policies, employment, and social issues (Goldeng et al., 2008), so CSR for SOEs can be viewed as a political task. Conversely, private firms are more market-driven and responsive to diverse stakeholders when considering CSR. During our study period, CSR was an emerging topic in China, and prior research has shown significant variation in awareness of and attitudes toward it among different stake- holders (Luo et al., 2017; Tang & Tang, 2012), which allows us to examine stakeholder attention in a more comprehensive and nuanced way than prior research. Our study puts forth a new theoretical framework for researching CSR behavior by uncovering how firms’ CSR behaviors are contingent on stakeholder attention. This enables us to identify the boundary conditions and under- lying stakeholder mechanisms that influence the extent to which CSR can be used strategically and instrumentally. Furthermore, while prior studies have explored the impact of corporate attention to stakeholders on firms’ decision- making, they have overlooked the role of stakeholder atten- tion. Our study offers novel evidence for the importance of stakeholders and contributes to a multi-stakeholder perspec- tive, and is particularly applicable when there are different or conflicting interests and demands of particular stakeholder groups. Stakeholder Attention and the Effect of Product Market Competition According to traditional strategic management theory, firms can achieve competitive advantage through cost leadership or differentiation (Porter, 1980, 1985). Therefore, when fac- ing pressure from more competitive conditions, firms typi- cally aim to sustain their competitive advantage by pursu- ing one of these two strategies. On the one hand, a cost leadership strategy provides consumers with products or services at lower prices by controlling overhead and dis- cretionary expenditures and exploiting economies of scale (Porter, 1985). This strategy can lead to above-average returns by enabling firms to lower their prices to match or beat their rivals (Phillips et al., 1983). Following this logic, studies have demonstrated that firms facing increasing PMC pay less attention to CSR as they incur discretionary costs. For instance, Duanmu et al. (2018) found a negative impact of PMC on environmental performance of Chinese firms between 2000 and 2005. Similarly, Lee et al. (2018) Government attention • Central government attention • Local government attention Product market competition Corporate social responsibility Institutional investor attention • Long-term institutional investor attention • Short-term institutional investor attention Financial analyst attention • Female financial analyst attention • Male financial analyst attention Media attention • New media attention • Traditional media attention Investment market attention H1a H1b H3a H3b H2a H2b Fig. 1   An illustration of the overall theoretical framework 869Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… demonstrated that Korean firms in more competitive prod- uct markets were less engaged in CSR activities during the 2010–2013 period, as CSR was viewed as an overinvestment by managers. Research has even shown that, when faced with stiff com- petition, firms may even engage in irresponsible behaviors to save costs. For example, Campbell (2007) argued that, in industries with high levels of competition, firms may even lower the quality and safety of their products, place undue pressure on their staff, and defraud consumers to gain a com- petitive advantage. In addition, Shleifer (2004) illustrated that unethical, cost-cutting behavior (e.g., corruption and employing children) is sometimes a consequence of market competition. Overall, there is a strong thread in the litera- ture showing that, when firms are in particularly competitive markets, they overwhelmingly focus on cost considerations and as a result are less likely to invest in CSR. In contrast, a differentiation strategy attempts to create unique customer value that enables a firm to build loyalty and command premium prices (Porter, 1985). This rationale is typically associated with the positive relationship between CSR and firm financial performance (Margolis & Walsh, 2003). Accordingly, studies that follow this logic have argued that firms in more competitive environments have a greater incentive to invest in CSR under the assumption that stakeholders find it to be valuable and will likely reward firms for their CSR performance (e.g., Declerck & M’Zali, 2012; Dupire & M’Zali, 2018; Fernández-Kranz & Santaló, 2010; Fisman et al., 2006; Flammer, 2015). For example, Fisman et al. (2006) presented a signaling model of corporate philanthropy and found that CSR may indicate that a firm’s products are reliable and can be trusted as high quality when consumers cannot check the quality before buying. Moreover, Flammer (2015) used trade liber- alization in the USA to establish a quasinatural experiment and explore the causal relationship between PMC and CSR, reporting that firms respond to higher competitive pressure from abroad by increasing their CSR behavior to preserve a better relationship with local stakeholders. Although there are conflicting findings on the relation- ship between PMC and CSR, the underlying issue is how firms act when faced with tensions or trade-offs between CSR costs and benefits. In other words, under the pressure of increasing PMC, firms may either use CSR to differenti- ate themselves from their rivals or reduce CSR to save costs, depending on whether the benefits exceed the costs of CSR behavior (McWilliams & Siegel, 2001). In our context of China, prior research has shown that different types of stakeholders have been increasingly aware of the importance of CSR and thus rewarded CSR behavior (e.g., Li & Lu, 2020; Tian et al., 2011). In such a context, our baseline expectations focus on whether firms are more concerned with how their actions differentiate them in the eyes of different stakeholders when they decide to engage in CSR. For instance, Wang and Qian (2011) found that corpo- rate philanthropy, one important CSR behavior, can be used by Chinese firms to elicit positive stakeholder responses and obtain political access, thereby leading to different socio- political legitimacy and financial performance. Meanwhile, although Duanmu et al. (2018) pointed out that China expe- rienced an overall cost-oriented environment before 2008, firms pursuing differentiation strategy still acknowledged the benefits of sustainability. Since 2008, issues of legitimacy, reputation, customer loyalty, and employee morale that stem from CSR have only increased (Yin & Zhang, 2012). The Chinese Ministry of Commerce published its first Chinese reference book for pre- paring a CSR report in 2008 (Marquis & Qian, 2014). Sub- sequently, domestic third-party CSR rating agencies began evaluating Chinese firms’ CSR performance. Furthermore, the 2008 earthquake in Sichuan Province, China, triggered massive attention to CSR behavior (especially philanthropy) and active online discussions (Luo et al., 2016). In addition, the 2008 discovery that Chinese milk was contaminated by melamine, an industrial plastic chemical, led the Chinese public to gradually recognize the importance of firms operat- ing ethically and sustainably (Tian et al., 2011). Therefore, given our context and period, it is reasonable to expect the benefits of CSR behavior to outweigh the costs. Consequently, the baseline expectation on which we build the remainder of our hypotheses is that PMC is positively associated with a firm’s CSR behavior. Our core focus that examines how CSR activity may vary following PMC focuses on processes of stakeholder attention—that is, the degree to which stakeholders attend to the focal firm and especially its CSR behavior, which is essential to understand the differentiation effect of CSR. Existing studies have mainly focused on the role of corpo- rate characteristics such as product features, consumer diver- sity, and firm-level strategy (Duanmu et al., 2018; Flammer, 2015) while ignoring the features of external stakeholders. But potential stakeholder rewards are based on the implicit assumption that stakeholders are fully aware of the firm’s CSR behavior (Godfrey et al., 2009; McWilliams & Siegel, 2001). Thus, whether and to what extent stakeholders are able to pay attention to firms and consequently witness their corporate social initiatives is a precondition for CSR as a differentiation mechanism. Longstanding research in psychology and management has shown that attention is a scarce cognitive resource (Kah- neman, 1973). Attention is required to encode environmental stimuli and process ideas in conscious thought (Hirshleifer & Teoh, 2003). Owing to the enormous volume of informa- tion accessible in the environment and the limitations of information processing capability, selective attention is a necessary consequence (Ocasio, 2011). The attention-based 870 Y. Wang, C. Marquis view (ABV) provides a theory of organizational action and adaptation that demonstrates that firm behavior results from how firms channel and distribute their decision-makers’ attention (Ocasio, 1997; Ocasio et al., 2018). Drawing upon this perspective, many studies have explored the attention of corporations and their decision- makers (Bouquet & Birkinshaw, 2008; Cho & Hambrick, 2006; Tuggle et al., 2010). While few studies have examined stakeholder attention explicitly, some have suggested that individual and situational factors produce variation in stake- holder attention to firm misconduct (Barnett, 2014), other studies have examined related processes such as how atten- tion from specific stakeholders such as the media (Kölbel et al., 2017) and financial analysts (Qian et al., 2019) shapes firm performance outcomes. Furthermore, while much previous CSR literature has examined non-shareholder stakeholders as a combined group, assuming that customers, suppliers, employees, the community, governments, and other stakeholders share a similar utility function (e.g., Barnett, 2014; Harrison et al., 2010), to better consider stakeholder attention processes, our perspective alternatively focuses on the extent to which stakeholders’ utility functions may vary. Our intuition fol- lows significant anecdotal evidence that critical CSR issues are often complex and involve different and even conflicting stakeholder claims (Wang et al., 2020). Existing research on stakeholder multiplicity has also demonstrated that different stakeholders often view firms through their unique lenses and have varied interests and demands (Oliver, 1991; Zhao et al., 2017). A particular stakeholder may have been social- ized to value (or ignore) certain firm dimensions as part of developing an evaluative lens (Fisher et al., 2016)—for instance, investors have their own evaluation criteria (Chen et al., 2007; Hartzmark & Sussman, 2019) and cognitive styles (Barber & Odean, 2001). Furthermore, in China, incompatible institutional demands that arise from the cen- tral and local governments can exist, which also shows how the claims and orientations of stakeholders can vary (Luo et al., 2017; Marquis & Qiao, 2022). Tang and Tang (2012) highlighted the varying degrees to which different stake- holders consider firms’ engagement in CSR to be important. Accordingly, based on the above-mentioned literature, we contend that it is important to move beyond examining uni- fied stakeholder-group effects on the focal firm and consider that different stakeholders may have different attentional processes. More importantly, we additionally examine the complexity of stakeholder groups and explore the varied and even conflicting interests within the same broad stakeholder type. In the following subsections, we first develop a multi- stakeholder framework that integrates attention received from different stakeholder groups—government (including central and local governments), investment market (includ- ing institutional investors and financial analysts), and media (including traditional and new media)—to foreground their separate effects. We then compare the conflicting interests within a specific stakeholder group and explore whether their attention diversely shapes the PMC–CSR relationship. Government Attention Governments are essential in defining and regulating appro- priate conduct in the business sector (Luo et al., 2010). As governments maintain the power to allocate key resources, issue licenses and permits (for business entry), ratify pro- jects, grant subsidies, allow tax arrears, and provide access to infrastructure, firms must consider the role of govern- ments when developing strategies (Oliver, 1991). However, the state is not a monolithic entity, and the difference in pri- orities between a central government and local governments is common to a broad range of political systems (Marquis & Raynard, 2015). As a large and complex political economy, China exhibits a hierarchical and relatively autonomous relationship between the central government and lower-level governments such as a province or city. Although, in recent years, the central government has shifted from prioritizing economic growth to focus more on sustainable, balanced economic and social development, lower-level governments have maintained a higher priority on economic growth and largely downplayed social goals (Wang & Luo, 2019). This variation allows us to observe the effects of different and even conflicting governmental claims. During the study period, the Chinese central govern- ment increasingly emphasized the importance of CSR. For example, the 11th Five-Year Plan for National Economic and Social Development (2006–2010) articulated a national vision based on the principles of a harmonious society and scientific development. This policy was then strengthened in the next three 5-year plans as the Chinese government encouraged private enterprises to actively undertake their social responsibilities and participate in social good provi- sion and philanthropy. Furthermore, guidelines and recom- mendations on reporting corporate social and environmental activities have been issued continuously by other govern- mental actors; for example, major stock exchanges began doing so in 2006 (Marquis & Qian, 2014). Given the generally underdeveloped market institutions and the strong role of the central government in China, prior research has shown that firms must comply with its policies and institutional logic for legitimacy and associated resource benefits (Luo et al., 2017). To better maintain control over the economy, the central government also co-opts corporate leaders into the national political system (Lin, 2011). Thus, 871Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… in situations where firms receive more central government attention, the differentiation effect of CSR would be particu- larly pronounced. Nevertheless, since China’s market reforms, local govern- ments have maintained discretionary power in the economic realm and have continued to prioritize economic factors in decision making (Marquis & Qiao, 2022). For example, high GDP growth increases tax revenue, thus creating a pro-busi- ness incentive to promote growth (Jin et al., 2005). Further- more, GDP growth has been shown to be the most important criterion for evaluating and promoting local officials (Li & Zhou, 2005). Given the limited resources and attention of local governments, the continued focus on GDP might lead local governments to maximize short-term economic growth at the expense of social objectives and environmental sus- tainability (Xu, 2011). Thus, for firms with greater local government attention, a CSR strategy will not create ben- efits (or sufficient benefits) that exceed the cost. Therefore, there is less motivation and incentive to obtain competitive advantage via the differentiation effect of CSR, and avoiding CSR behavior is an optimal choice for firms. Such arguments on how there would be greater attention to CSR from a central government as compared with a local government thus lead to the following formal predictions: • Hypothesis 1a. The positive effect of PMC on CSR behav- ior is stronger for firms belonging to industries that receive greater central government attention. • Hypothesis 1b. The positive effect of PMC on CSR behavior is weaker for firms belonging to industries that receive greater local government attention. Investment Market Attention In addition to the government, investment market actors also emphasize CSR and sustainable development. For instance, Graves and Waddock (1994) and Cox et al. (2004) identified institutional investors’ CSR decision-making preferences such that they can spread private information from manag- ers to other shareholders and creditors and affect corporate governance (Chung et al., 2002). Ioannou and Serafeim (2015) reported that, since the 1990s, analysts (particularly high-status analysts) have increasingly recognized the stake- holder management function of CSR in their ratings. Luo et al. (2015) also posited that analysts heed corporate social performance information and factor it into their recommen- dations for general investors. Accordingly, the investment market’s perceptions of CSR enables the market to reward CSR behavior. Similarly, not all investment market actors appreciate CSR to the same extent. Existing studies have found that, compared with long-term institutional investors, short-term institutional investors prefer short-term benefits over long- term gains and encourage more “myopic” behavior (Chen et al., 2007). In addition, CSR outcomes regarding short- term profits are often not necessarily realized (Kim et al., 2018). Therefore, long-term institutional investors may approve of firms’ CSR behavior to create long-term value whereas short-term institutional investors may not, which shapes firm goals (Gaspar et al., 2005; Kim et al., 2019). Furthermore, meta-analytic investigations provide empir- ical support for the idea that gender differences may exist in ethics and corporate responsibility (Kish-Gephart et al., 2010). Research has illustrated that males and females speak in different moral “voices” (Walker, 2006), and females seem to exhibit higher levels of empathy and lower levels of unethical decisions compared with males (Toussaint & Webb, 2005). Moreover, women in corporate leadership have been linked to greater CSR engagement (e.g., Bear et al., 2010; Marquis & Lee, 2013). Thus, we propose that female financial analysts may prefer and reward firms’ CSR behavior more than male financial analysts. Given our argument that the achievement of competitive advantage from CSR differentiation is contingent on the level of firm attention by stakeholders who reward CSR, we propose the following hypotheses: • Hypothesis 2a. The positive effect of PMC on CSR behav- ior is stronger for firms that receive greater attention from investment market stakeholders. • Hypothesis 2b. The contingent effect of attention from investment market stakeholders that are more apprecia- tive of CSR (e.g., long-term institutional investors and female financial analysts) is more profound than that of attention from investment market stakeholders that over- look CSR (e.g., short-term institutional investors and male financial analysts). Media Attention In addition to governments and the investment market, the public and society are crucial external stakeholders that firms recognize. Research shows that firms’ CSR behavior can enhance social reputation and image (Borghesi et al., 2014). We further consider public support for firms and examine how an important type of societal attention—the media—affects a firm’s responsiveness to competitive environments. Compared with other stakeholders, the media can work as an information intermediary that does not directly control resource flows to and from firms but rather how resource- wielding stakeholders perceive firms (Carroll & Hannan, 2000; Yu et al., 2008). The media decrease information asymmetry for investors, consumers, and other corporate 872 Y. Wang, C. Marquis stakeholders who lack direct interaction with firms and influ- ence public opinion. Thus, the media can profoundly shape corporate behavior, especially CSR (Siegel & Vitaliano, 2007; Zyglidopoulos et al., 2012). We argue that media attention accentuates the relationship between PMC and CSR behavior, suggesting that opportu- nities exist for firms with sufficient media attention to take advantage of this relationship and maximize the benefits of CSR behavior to attract public support, which in some cases may even represent inauthentic behavior (Marquis et al., 2016). For firms with insufficient media attention, the above- mentioned reputational and differentiation benefits do not occur. Even under the pressure of PMC, which highlights the importance of decisions for survival, firms may even cancel or suspend CSR behavior. Thus, when PMC intensifies, firms with greater media attention increasingly focus on obtaining a competitive advantage via this CSR differentiation strategy. Furthermore, unlike traditional media, new or online media have distinguishing features and advantages in infor- mation dissemination. Digitization has made the transfer of information nearly costless; thus, new media have spread wider and faster with the help of digital technology (Bar- nett et al., 2020). The development of digital technology has led to significant changes in people’s lives. The public can use smartphones to receive information anywhere and at any time, and the internet dominates how people obtain information. According to the World Bank, in 2000, there were only 367 million internet users globally (5.9% of the global population); by 2020 there were 4.7 billion users (59.6%). New digital communications (e.g., Facebook, Twit- ter, Weibo, and WeChat) have also enabled and accelerated knowledge-sharing processes (Zyglidopoulos et al., 2012), and the internet and social media have rendered informa- tion about CSR activities increasingly transparent (Capriotti, 2011). Therefore, we argue that new media attention covers a wider range than traditional media attention, resulting in more benefits of CSR behavior that exceed costs. Thus, we propose the following hypotheses: • Hypothesis 3a. The positive effect of PMC on CSR behav- ior is stronger for firms that receive greater media atten- tion. • Hypothesis 3b. The moderating effect of new media attention is more profound than that of traditional media attention. Methods This study encompasses all privately controlled Chinese firms listed on the Shanghai or Shenzhen stock exchanges from 2008 to 2017. Since 2008, CSR has gained increas- ing attention from stakeholders in China. Although the importance of CSR has gradually been realized in China, varying levels of awareness and attitudes regarding it is com- mon among different stakeholders. However, because CSR is still in early stages in China, we have the opportunity to explore how attention varies with different stakeholder CSR orientations. The data sources we drew on to study our hypotheses include archival data from the China Stock Market and Accounting Research (CSMAR) database, the Chinese Research Data Services (CNRDS) database, regional data from the National Bureau of Statistics of China, the China Civil Affairs Statistics Yearbook, and CSR ratings from Runling, a CSR rating agency (also known by its English acronym, RKS; http://​www.​rksra​tings.​com). The CSMAR database is the primary source of financial information for Chinese listed firms. Our ratings of the substance of a firm’s CSR behavior cover the 2008–2017 period (published from 2009 to 2018). After merging these databases and remov- ing firms with missing data—i.e., firms with less than 1 year of data and firms that were in unique financial circum- stances1—our sample includes 1677 unique firms and 8934 firm–year observations to explore the relationship between PMC and CSR. Dependent variables. The CSR score is an overall rating of CSR behavior evaluated by the RKS, a leading independ- ent CSR rating agency in China whose work has been used in prior research on Chinese firms’ CSR (Chen et al., 2023; Lau et al., 2016; Li & Lu, 2020; Marquis & Qian, 2014). To account for China’s specific context, the RKS adapts the KLD social index framework and standards of the Global Reporting Initiative (GRI 3.0) to create a 70-indicator rating system to examine the CSR behavior detailed in firms’ CSR reports. Specifically, these indicators fall into three major dimensions: social responsibility strategy and innovation (14 items), disclosure content (45 items), and technical sufficiency (11 items). Experts assessed firms along these indicators using an anchored scale of 0–4 with an interval of 0.5. The composite CSR score is a weighted average of scores along the three dimensions (30%, 50%, and 20%, 1  We dropped the following firms that were in unique financial cir- cumstances: (i) Firms designated as special-treatment (ST) or particular- transfer (PT) firms by the China Securities Regulatory Com- mission (CSRC) for abnormal financial performance in a year. (445 firms, 1511 observations); (ii) firms belonging to the financial industry, which has a spe- cial earnings structure (98 firms, 807 observations); (iii) insolvent firms (with a leverage ratio of more than 1) as they were in financial distress and outliers in financial data (136 firms, 340 observations). http://www.rksratings.com 873Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… respectively), ranging from 0 to 100.2 A higher CSR score indicates a higher level of CSR behavior. While the RKS measure has the advantages of precedence and comprehensiveness, its scores may not perfectly capture the substance of firms’ CSR behavior. As such, we tested the relationship between this score and a number of specific CSR activities, including donations, employment opportu- nities, environmental recognition, and sustainable strategy. The results, shown in Appendix 1, are all consistent with this variable and suggest that the RKS CSR score is an appropri- ate measure to proxy firms’ CSR behavior. Independent variables. We adopt a comprehensive index—the Herfindahl–Hirschman Index (HHI) following a scale of 100—to measure market concentration. In this cal- culation, we include same-industry SOEs to better measure competition for private firms, which face competitive pres- sure from other private firms as well as from government- protected firms (i.e., SOEs). When an industry contains a fixed number of firms, a lower HHI reflects a greater number of similarly sized firms, suggesting that competition would be more intense. In contrast, a higher HHI reflects a greater monopoly in the industry. Therefore, we use the inverse of the HHI as a proxy for the PMC to facilitate interpretation. As the inverse of HHI increases, the industry becomes more competitive. We define PMC as follows: where Si,j,t is the market share of firm i in industry j in year t ; Nj is the number of firms in industry j in year t ; and HHIj,t is the HHI of industry j in year t . An individual firm’s market share is calculated by dividing its net sales by the total sales of the entire industry. To highlight the robustness of this result, we also use another proxy to measure PMC in the robustness check section. Moderators. To test the hypotheses that predict the effect of stakeholder attention on the relationship between PMC and CSR and to show how CSR is channeled to obtain com- petitive advantage, we use different moderating variables. Owing to data limitations, we use indirect measures to cap- ture the attention from different external stakeholders, as is common in the literature. Five-year plans are the most important governmental documents in China. These plans are a series of economic and social development initiatives that map strategies for PMCj,t = −HHI j,t = − Nj ∑ i=1 S2 i,j,t , development, set targets, and launch reforms within a spe- cific timeframe (Yuan & Zuo, 2011). Therefore, we propose that the Chinese government would pay more attention to industries mentioned in these plans, and firms that belong to them are prioritized for development by the govern- ment such as through tax breaks, subsidies, and govern- ment procurement benefits. Every 5-year plan has unique characteristics because of the specific period in which it is composed and approved. The content and major targets of 5-year plans have changed significantly over time accord- ing to economic development and social growth conditions. During the 11th Five-Year Plan (2006–2010) and 12th Five-Year Plan (2011–2015) that were relevant in our study period, the prioritized aim of the Chinese central govern- ment was rebalancing economic and social development, mitigating social inequality, and conserving the environment (Li & Lu, 2020; Zhu & Lin, 2021). Thus, as one of the most essential ways for balanced and sustainable growth, CSR was also emphasized by the central government (Luo et al., 2017); in turn, firms in industries included in the above-men- tioned 5-year plans captured the most government attention and were most likely to be receptive to these ideas. However, during this period provincial governments still focused on gross domestic product (GDP) growth targets and largely ignore social goals as a result of local economic expansion or political promotion pressures (Luo et al., 2017). Thus, Central government attention is calculated by how many times a firm’s industry has been mentioned in the Chi- nese central government’s 5-year plans within a specific time frame (i.e., the 11th and 12th Five-Year Plan period). Local government attention is measured by the number of times the focal firm’s industry has been mentioned in the Chinese provincial government’s (where the firm is located) 5-year plans in relevant years.3 In addition, we use the sum of Cen- tral government attention and Local government attention to capture overall Government attention. We also consider the attention from institutional investors and financial analysts. Institutional investors use their exper- tise to follow and analyze focal firms and make investment decisions; however, they may only focus on a limited number of firms because attention is a scarce cognitive resource. Thus, we use the percentage of a firm’s total shares held by institutional investors to capture Institutional investor atten- tion. The more shares institutional investors hold, the more attention they pay to a firm. Additionally, we use the number of research reports issued by institutional investors on a par- ticular firm as a robustness test. Following Yan and Zhang (2009), we also divide institutional investors into short- and long-term based on their portfolio turnover over the past four 2  In 2012, RKS revised its evaluation criteria by adding an industry evaluation indicator to capture between-industry differences. The proportions were 30% (social responsibility strategy and innovation), 45% (disclosure content), 20% (technical sufficiency), and 5% (indus- try evaluation). We also examined the sample using only post-2011 data as a robustness check and obtained similar results. 3  Detailed information about industries included or mentioned by each provincial government is available from the authors on request. 874 Y. Wang, C. Marquis quarters, thereby using the percentage of total shares held by the short-term (long-term) institutional investors to capture Short-term (Long-term) institutional investor attention.4 For Financial analyst attention, following Gentry and Shen (2013), we measure the analyst coverage of each firm by using the natural logarithm of the number of analysts who issued earnings forecasts for the firm during the year.5 A higher value of this measure indicates a higher level of attention. In addition, we use the natural logarithm of the number of female (male) analysts who issued earnings fore- casts for a firm during the year as a proxy for Female (Male) financial analyst attention. Regarding the effect of media attention, we examine both traditional and online sources. Traditional media include newspapers and television. We use the natural logarithm of the number of financial news reports about a focal firm published by newspapers to assign a value to Traditional media attention in a given year. These data include news from eight mainstream financial newspapers commonly used in academic research on China: China Securities Journal, Shanghai Securities News, China Business News, 21st Cen- tury Business Herald, China Business Journal, Economic Observer, Securities Daily, and Securities Times. Fur- thermore, we consider financial news published on web- sites and measure the New media attention each firm has received using the natural logarithm of the number of online news articles about a focal firm in a given year. These data include news from the 20 most important mainstream finan- cial websites, including hexun.com, sina.com, eastmoney. com, finance.qq.com, money.163.com, finance.ifeng.com, and others. The distinguishing features and advantages of information dissemination via new or online media stand out because print newspapers have experienced a widespread decline in readership in recent years (Economist, 2020). The sum of Traditional media attention and New media attention is used as a proxy for overall Media attention. In addition, we conduct a robustness check to rule out potential concerns regarding the overlap between traditional and new media attention. Finally, we consider articles published on private financial and economic WeChat platforms and measure the Social media attention each firm has received using the natu- ral logarithm of the number of WeChat articles about a focal firm in a given year. Control variables. We control for additional variables that might affect a firm’s CSR decisions. We include Firm size as the natural logarithm of total assets. Firm age is the number of years that each firm has been listed. Firm resources and performance, which reflect firms’ feasibility to embrace a CSR strategy, are captured by two variables: ROA, which is the return on assets calculated as net income over total assets; and Slack resources, which is the ratio of the sum of cash flow from a firm’s operating, financing, and investing activities to its total assets. A firm’s governance also affects its investment in CSR (Chan et al., 2014). We use three variables to capture the impact of governance: TOP10 ownership, which is the sum of the share percentiles of the top 10 shareholders; Independent directors, which is depicted by the ratio of the number of inde- pendent directors to the total number of directors; and CEO duality, a dummy variable that measures whether the same 4  Each quarter, we first calculated the aggregate purchase and sale for each institutional investor. where Pi,t−1 and Pi,t are the share prices for stock i at the end of quar- ters t − 1 and t, and Sk,i,t −1, and Sk,i,t are the number of shares of stock i held by institutional investor k at the end of quarters t − 1 and t, respectively. CR_buyk,t and CR_sellk,t are institutional investors k’s aggregate purchases and sales in quarter t, respectively. Institutional investor k’s churn rate for quarter t is defined as   Next, we calculated each institutional investor’s average churn rate over the past four quarters as   where AVG_CRk,t is institutional investor k’s average churn rate for quarter t. Finally, we additionally calculated each institutional inves- tor’s annual average churn rate as where Y_AVG_CRk,t is institutional investor k’s annual average churn rate for year t and AVG_CRk,t,q is institution k’s average churn rate for quarter q of year t.   Given the annual average churn rate measure, we sorted all institu- tional investors each year based on Y_AVG_CRk,t. Those ranked above the sample mean (with the highest Y_AVG_CRk,t) are classified as short-term institutional investors, while those ranked below the sam- ple mean are classified as long-term institutional investors. CR_buyk,t = Nk ∑ i = 1 Sk,i,t > Sk,i,t−1 | | Sk,i,tPi,t − Sk,i,t−1Pi,t−1 − Sk,i,t−1ΔPi,t | | CR_sellk,t = Nk ∑ i = 1 Sk,i,t ≤ Sk,i,t−1 | | Sk,i,tPi,t − Sk,i,t−1Pi,t−1 − Sk,i,t−1ΔPi,t | | , CRk,t ≡ min � CR−buyk,t,CR−sellk,t � ∑Nk i=1 Si,i,tPi,t+Sk,i,t−1Pi,t−1 2 . AVG_CRk,t = 1 4 3 ∑ j=0 CRk,t−j, Y_AVG_CRk,t = 1 4 4 ∑ q=1 AVG_CRk,t,q, 5  Research reports in China are issued by either a team or an individ- ual analyst. For research reports issued by a team, we also obtained information about all relevant financial analysts. Thus, we calculated the total number of analysts who issued earnings forecasts for the listed firm during a given year. 875Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… individual holds the positions of chair and chief executive officer (CEO) to proxy executive power (Zhang et al., 2016). Chinese firms that expand outside China may be exposed to additional pressure from international audiences regard- ing reporting and transparency (Marquis & Qian, 2014). Therefore, we control for Foreign sales as a percentage of total sales. Moreover, firms’ product diversification levels determine the scope of stakeholders (Su & Tsang, 2015) and may affect firms’ CSR behavior. Therefore, we controlled for firms’ product diversification levels. Following Su and Tsang (2015), we adopt the entropy measure proposed by Jacquemin and Berry (1979) to calculate Firm diversifica- tion, defined as ∑ i Piln(1∕Pi) , where Pi is the percentage of firm sales in industry i. Previous studies have shown that corporate political con- nections may affect a firm’s strategic decisions. Political connection is measured by whether the chair has had experi- ence working in government before serving as a chair or has been a delegate to the national- and provincial-level People’s Congress or to the Chinese People’s Political Consultative Conference. In addition, we measure Regional development as the province’s GDP per capita (i.e., the province’s GDP divided by its population). Owing to the potential difference in CSR focus between firms listed on the Shenzhen and Shanghai stock exchanges (Marquis & Qian, 2014), we include a dummy variable, Shanghai, which equals 1 if the firm is listed on the Shang- hai Stock Exchange and 0 if it is listed on the Shenzhen Stock Exchange. In addition, firms with experience issuing CSR reports may be more likely to do so within a given year and more substantively; thus, we control for report- ing Experience, which is equal to 1 if the focal firm has issued a CSR report before and 0 otherwise. Furthermore, stock market regulations require some firms to issue CSR reports;6 however, as our exploratory analysis suggests, this requirement does not affect CSR behavior. Therefore, this can be an exclusion restriction in the Heckman selection model (Lennox et al., 2012). Required discloser is a dummy variable that equals 1 if a firm’s characteristics result in its being required to issue a CSR report in a given year and 0 otherwise. Finally, there is significant variation in the extent of CSR behavior across the 10 years of our study (2008–2017). To control for a potential time effect, we include year dummy variables in the analysis. To control for possible differences in CSR behavior across industries that may be under pressure from different sources, we include 16 industry dummies representing the 17 industry categories identified by the China Securities Regulatory Commission. We winsorize all continuous variables at the 1% and 99% levels. Estimation Method We use a two-stage Heckman selection model (Heckman, 1979) to correct for any sample selection bias because the CSR rating (CSR score) is contingent on whether a firm has issued a CSR report; only firms that issue a CSR report have a CSR score. In such analysis, parameter estimates from a first-stage probit model based on information representing all firms are incorporated into the second stage. The first stage of the Heckman process involves esti- mating the likelihood of CSR behavior, which differs from that predicted by various firm and regional factors. Required discloser is included as a covariate in the first- stage regression only (6.72% of the sampled firm–years are compulsorily required to disclose their CSR behavior). The likelihood of CSR behavior is estimated by applying a probit model to the entire sample, including firms in the main and control groups. In total, 1677 unique firms and 8934 firm–year observations are included in the first stage of the Heckman process. We calculate an adjustment term, the inverse Mills ratio, using a first-stage probit regres- sion. The ratio is then included as a control variable in the main second-stage equation (see Heckman, 1979) that examines the relationship between PMC and CSR behav- ior using a sample of privately controlled Chinese-listed firms that issued CSR reports between 2008 and 2017. This sample includes 355 unique firms, corresponding to 1722 observations. For industry-level concentration variables, the main source of variation is cross-sectional because it is rare for firms in China to change industries. If we include firm fixed effects in the regression, the effect from PMC will be absorbed. Thus, Chen et al. (2022) suggest using industry–year fixed effects when the main source of vari- ation is cross-sectional. As our key independent variable (PMC) is an industry-level, time-varied measurement, we use a broader industry sector7 to interact with year to miti- gate any concerns about omitted variables correlated with an industry-based structure and vary within industries and years. We also use 1-year lagged independent and control variables to lower concerns about additional endogeneity and reverse causality in our model. The following equation is used to test our hypotheses: 6  In May 2008, the Shanghai Stock Exchange issued a policy requir- ing firms in the corporate governance group, firms listed on foreign stock exchanges such as the New York Stock Exchange and Hong Kong Stock Exchange, and firms in the financial industry to issue CSR reports. On the Shenzhen Stock Exchange, firms belonging to the SZSE 100 Index are required to issue CSR reports. 7  The broader industry sector includes three categories, while we cal- culated the PMC with 17 specific industries. 876 Y. Wang, C. Marquis where PMC is product market competition, and X is a set of control variables expected to influence CSR behavior. IMR is the inverse Mills ratio of the first-stage model. Sector*Year refers to the dummy variables used to control for broader industry sector–year fixed effects,8 while λ connotes the dummy variables used to control for industry fixed effects, and � is an error term. Although not reported, all our results still hold when run with standard ordinary least-squares models (i.e., without correcting for selection; results are available upon request). Results Table 1 presents the descriptive statistics and correlations for each variable in the second-stage model used to test the hypotheses. The first-stage descriptive statistics and correla- tions are substantively similar and not included because of page length requirements, but are available upon request. The correlation between PMC and CSR score is positive (0.181). We checked variance inflation factors (VIFs) and found that the mean VIF was 1.20, and the highest VIF was 1.73 (Firm age), which is below the rule-of-thumb cutoff of 10. Therefore, multicollinearity does not pose a serious concern. Overall, only 18.95% of the sampled firm–years issued CSR reports between 2008 and 2017. Among these firms, the average CSR score was 36.900, with a maximum score of 89. Table 2 presents the results of government attention. Model 1 is our baseline results, the second-stage results of the Heckman model, which examine the relationship between PMC and CSR. The model also contains all the control variables. Firm size, Firm age, Top 10 ownership, CEO duality, Shanghai, Firm diversification, and Experi- ence significantly influence CSR score. The coefficient of PMC is positive and significant ( �= 0.175,p < 0.01 ), which is in line with prior research and our baseline expectation. This shows that, when firms face more intense PMC, they increasingly engage in CSR to achieve competitive advan- tage by differentiating themselves. The economic magni- tude of this effect is relatively modest, which is not particu- larly surprising given that the effect is measured over the entire firm sample. In line with our prediction, we expect the significance to vary across firms depending on external CSR scorei,t+1 = � + �1PMCj,t + �2Xi,t + �3IMRi,t + Sector ∗ Yearjt + �j + �i,t, stakeholder attention; that is, we suspect that the effect of PMC will be stronger among firms with greater attention from particular stakeholders. Model 2 in Table 2 tests hypothesis 1a, whereas models 3 and 4 in Table 2 test hypothesis 1b. The interaction of PMC and Central government attention in model 2 displays the expected signs and is significant ( 𝛽 = 0.020, p < 0.05 ). To assess economic magnitude, when holding other variables at the sample means, with low central government attention (i.e., 4.459, 1 s.d. below the mean), there is a difference of −4.274 in the CSR score between firms with maximum versus minimum PMC. In contrast, with high central govern- ment attention (i.e., 18, maximum), there is a difference of 10.993 in the CSR score between firms with maximum ver- sus minimum PMC. This is approximately a 338% increase, thus supporting hypothesis 1a. Furthermore, in model 3, although the interaction between PMC and Local govern- ment attention is not significant ( p > 0.1 ), it becomes nega- tive, suggesting that attention from local governments that prioritize GDP growth and ignore social issues will not pro- mote firms’ CSR behavior to obtain a differentiation effect under PMC pressure. In addition, we simultaneously con- sider the role of central and local government attention in model 4. The interaction of PMC and Central government attention is positive and significant ( 𝛽 = 0.036, p < 0.01 ), while the interaction of PMC and Local government atten- tion is negative and significant ( 𝛽 = −0.011, p < 0.01 ). Thus, hypothesis 1b is strongly supported. In addition, we explore the contingent role of the overall Government attention in model 5. The interaction between PMC and Government attention is not significant. This is consistent with the con- flicting effects of Central government attention and Local government attention. Table 3 presents the results for investment market atten- tion. Model 1 in Table 3 captures the role of institutional investor attention. The coefficient of the interaction term PMC × Institutional investor attention also has the expected positive sign and is significant ( 𝛽 = 0.252, p < 0.01 ). When holding other variables at the sample means, with low institutional investor attention (i.e., 0.153, 1 s.d. below the mean), there is a difference of 5.274 in the CSR score between firms with maximum versus minimum PMC, whereas with high institutional investor attention (i.e., 0.690, 1 s.d. above the mean), there is a difference of 12.903 in the CSR score between firms with maximum versus minimum PMC. Furthermore, we divide institutional investor atten- tion into long-term and short-term. In model 2, the inter- action between PMC and Long-term institutional investor attention is positive and significant ( 𝛽 = 0.237, p < 0.01 ). However, Short-term institutional investor attention has no discernible effect on the PMC–CSR relationship in model 3. When we include these two interaction terms in model 4, 8  Year fixed effects are absorbed by industry sector–year fixed effects. 877Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… only long-term, institutional investor attention significantly amplifies the promoting effect of PMC on CSR. Model 5 captures the effect of financial analyst attention on CSR behavior under competitive pressure. The coef- ficient of the interaction term, PMC × Financial analyst attention, has the expected positive sign and is significant ( 𝛽 = 0.139, p < 0.10 ). When holding other variables at the sample means, with low financial analyst attention (i.e., 1.037, 1 s.d. below the mean), there is a difference of 1.587 in the CSR score between firms with maximum versus mini- mum PMC whereas, with high financial analyst attention (i.e., 3.149, 1 s.d. above the mean), there is a difference of 18.136 in the CSR score between firms with maximum versus minimum PMC. Interestingly, the marginal effect of PMC on firms’ CSR behavior is positive but becomes negative when the value of financial analyst attention is less than 0.835 (= 0.116/0.139). That is, for firms that receive sufficient attention from financial analysts (the top 73.1% of firms), the differentiation effect of CSR is greater than the cost effect of CSR. However, for the minority of firms that receive little financial analyst attention (i.e., the remaining 16.9%), PMC reduces CSR behavior. This is presumably because, as these firms garner less attention, they reduce costs by not prioritizing CSR behavior and instead focus on cost-minimizing strategies to obtain competitive advantage. The result reveals that stakeholder attention can reconcile opposing assessments between PMC and CSR. Moreover, we split financial analysts on the basis of their gender and consider the attention from female and male financial analysts. In model 6, the coefficient of the interaction between PMC and Female financial analyst attention has the expected positive sign and is significant ( 𝛽 = 0.034, p < 0.10 ). However, the coefficient of the inter- action between PMC and Male financial analyst attention Table 1   Descriptive statistics and correlation matrix for the Heckman second-stage model Because of page length requirements, only attention from different major stakeholder groups is reported. Other correlations are available upon request Variable Mean SD 1 2 3 4 5 6 7 8 9 10 1. CSR score 36.900 10.570 1 2. PMC −8.522 6.843 0.181 1 3. Government attention 29.800 19.460 0.101 0.120 1 4. Institutional investor attention 0.359 0.261 0.005 −0.062 −0.028 1 5. Analyst attention 8.602 10.720 0.171 0.017 0.031 0.138 1 6. Media attention 8.455 1.964 0.291 0.134 0.002 0.114 0.435 1 7. Firm size 21.630 1.032 0.274 0.145 0.020 0.194 0.359 0.478 1 8. Firm age 7.379 5.818 −0.015 0.068 −0.156 0.147 −0.117 0.185 0.297 1 9. ROA 0.044 0.051 0.019 −0.027 0.030 0.118 0.398 0.127 0.039 −0.158 1 10. Slack resources −0.004 0.094 0.009 0.027 −0.080 0.059 0.068 0.093 0.187 0.203 0.060 1 11. Top 10 ownership 57.340 15.380 0.111 −0.027 0.080 0.302 0.170 0.004 0.029 −0.432 0.257 −0.096 12. Independent directors 0.374 0.053 −0.008 0.022 0.003 −0.080 −0.013 0.043 −0.042 −0.016 −0.019 0.010 13. CEO duality 0.330 0.470 0.020 0.008 0.064 −0.101 0.025 0.013 −0.097 −0.161 0.025 −0.057 14. Foreign sales 0.084 0.260 0.086 0.069 0.120 −0.106 −0.002 −0.018 0.019 −0.082 −0.015 0.038 15. Shanghai 0.263 0.440 −0.125 0.001 −0.079 0.249 −0.046 0.023 0.224 0.401 −0.067 0.095 16. Regional development 10.900 0.449 0.196 0.128 0.057 −0.162 0.114 0.103 0.117 −0.111 0.043 −0.011 17. Experience 0.159 0.366 0.133 0.008 −0.001 0.079 0.214 0.244 0.334 0.167 0.092 0.065 18.Political connection 0.312 0.463 0.019 −0.046 0.001 0.127 0.028 0.081 0.092 0.054 −0.019 −0.029 19. Firm diversification 0.405 0.421 0.019 0.001 −0.114 0.007 −0.078 0.029 0.150 0.238 −0.097 0.061 Variable Mean SD 11 12 13 14 15 16 17 18 19 11.Top 10 ownership 57.340 15.380 1 12. Independent directors 0.374 0.053 0.023 1 13. CEO duality 0.330 0.470 0.075 0.115 1 14.Foreign sales 0.084 0.260 0.037 0.047 0.049 1 15. Shanghai 0.263 0.440 −0.215 −0.065 −0.152 −0.177 1 16. Regional development 10.900 0.449 0.118 0.056 0.117 0.166 −0.115 1 17.Experience 0.159 0.366 −0.064 0.026 −0.043 −0.034 0.137 0.023 1 18.Political connection 0.312 0.463 −0.001 −0.018 −0.051 −0.096 0.051 −0.118 0.072 1 19.Firm diversification 0.405 0.421 −0.174 −0.013 −0.030 −0.037 0.158 −0.007 0.059 0.012 1 878 Y. Wang, C. Marquis Table 2   PMC, government attention, and CSR behavior Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 Model 5 CSR scores CSR scores CSR scores CSR scores CSR scores PMC 0.175*** −0.165 0.315*** −0.194 0.312 (0.022) (0.129) (0.106) (0.148) (0.211) Central government attention 0.366 0.484* (0.230) (0.247) PMC × Central government attention 0.020** 0.036*** (0.007) (0.009) Local government attention −0.130*** −0.168*** (0.037) (0.019) PMC × Local government attention −0.007 −0.011*** (0.004) (0.001) Government attention −0.100* (0.047) PMC × Government attention −0.004 (0.005) Firm size 3.189*** 3.192*** 3.198*** 3.221*** 3.184*** (0.544) (0.543) (0.538) (0.537) (0.539) Firm age −0.310* −0.314* −0.308* −0.311* −0.310* (0.147) (0.150) (0.151) (0.153) (0.151) ROA 7.333 7.069 7.281 6.663 7.446 (4.538) (4.405) (4.788) (4.490) (4.802) Slack resources 2.799 2.712 2.931 2.797 2.969* (1.715) (1.762) (1.665) (1.730) (1.668) Top 10 ownership 0.046* 0.047* 0.043 0.043* 0.043* (0.026) (0.026) (0.025) (0.024) (0.024) Independent directors −6.263 −6.086 −6.460 −6.327 −6.413 (4.090) (4.151) (4.090) (4.126) (4.121) CEO duality −0.830* −0.852* −0.849* −0.897* −0.836* (0.455) (0.466) (0.448) (0.466) (0.448) Foreign sales 0.746 0.743 0.912 0.938 0.879 (0.626) (0.632) (0.612) (0.625) (0.603) Shanghai −3.237*** −3.238*** −3.303*** −3.306*** −3.299*** (0.431) (0.423) (0.466) (0.457) (0.464) Regional development 0.727 0.759 0.549 0.585 0.560 (2.412) (2.390) (2.364) (2.359) (2.345) Experience 5.600*** 5.662*** 5.603*** 5.656*** 5.622*** (1.248) (1.268) (1.268) (1.287) (1.269) Political connection 0.767 0.755 0.696 0.670 0.709 (0.560) (0.544) (0.594) (0.570) (0.597) Firm diversification 1.456** 1.479** 1.539** 1.604** 1.511** (0.594) (0.585) (0.629) (0.613) (0.627) Inverse Mills ratio 2.648*** 2.678*** 2.623*** 2.647*** 2.631*** (0.813) (0.825) (0.832) (0.843) (0.833) Constant −47.911** −52.793** −44.060** −51.626** −44.026** (21.117) (21.392) (20.049) (21.625) (18.912) Industry fixed effects Yes Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Yes Number of observations 1722 1722 1722 1722 1722 Adjusted R2 0.282 0.283 0.285 0.288 0.284 879Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… Table 3   PMC, investment market attention, and CSR behavior Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 CSR scores CSR scores CSR scores CSR scores CSR scores CSR scores CSR scores CSR scores PMC 0.055 0.078** 0.124*** 0.040 −0.116 0.030 0.071 0.014 (0.039) (0.031) (0.024) (0.034) (0.131) (0.039) (0.061) (0.043) Institutional investor attention 5.028*** (0.961) PMC × Institutional investor atten- tion 0.252*** (0.077) Long-term institutional investor attention 3.956** 3.798** (1.662) (1.690) PMC × Long-term institutional inves- tor attention 0.237*** 0.220*** (0.051) (0.039) Short-term institutional investor attention 18.549*** 17.856*** (5.361) (5.699) PMC × Short-term institutional investor attention 0.699* 0.611 (0.366) (0.381) Financial analyst attention 2.025** (0.682) PMC × Financial analyst attention 0.139* (0.067) Female financial analyst attention 0.288*** 0.086 (0.091) (0.104) PMC × Female financial analyst attention 0.034*** 0.025* (0.008) (0.014) Male financial analyst attention 0.147** 0.125** (0.054) (0.054) PMC × Male financial analyst atten- tion 0.011 0.005 (0.007) (0.008) Firm size 3.078*** 3.125*** 3.121*** 3.057*** 2.817*** 3.182*** 2.935*** 2.980*** (0.565) (0.555) (0.554) (0.567) (0.624) (0.533) (0.584) (0.587) Firm age −0.343** −0.333** −0.312** −0.334** −0.277 −0.311* −0.291* −0.291* (0.146) (0.151) (0.144) (0.146) (0.161) (0.149) (0.152) (0.151) ROA 6.865 7.734 2.944 3.200 2.730 8.100 4.791 5.462 (4.258) (4.548) (5.832) (5.824) (7.465) (5.288) (5.845) (5.543) Slack resources 2.469 2.723 1.790 1.709 2.451* 2.480 2.572* 2.211 (1.709) (1.668) (1.525) (1.483) (1.194) (1.491) (1.387) (1.527) Top 10 ownership 0.021 0.028 0.052* 0.034 0.050* 0.049* 0.051* 0.052* (0.033) (0.038) (0.025) (0.037) (0.025) (0.026) (0.026) (0.026) Independent directors −5.997 −6.129 −5.864 −5.794 −5.321 −4.633 −5.324 −4.630 (4.048) (4.045) (4.279) (4.219) (4.147) (4.031) (4.144) (4.037) CEO duality −0.902* −0.870* −0.992** −1.032** −0.783* −0.789* −0.731 −0.797* (0.443) (0.450) (0.427) (0.425) (0.390) (0.412) (0.418) (0.439) Foreign sales 1.005 1.001 0.533 0.787 0.565 0.609 0.602 0.570 (0.650) (0.665) (0.642) (0.695) (0.629) (0.626) (0.616) (0.617) Shanghai −3.707*** −3.543*** −3.360*** −3.657*** −3.128*** −3.253*** −3.119*** −3.193*** 880 Y. Wang, C. Marquis is indistinguishable from zero in model 7. Additionally, when considering female and male financial analyst atten- tion simultaneously in model 8, only female financial analyst attention can significantly amplify the promoting effect of PMC on CSR. In summary, hypothesis 2 also receives strong support. Table 4 tests hypothesis 3a and 3b. Model 1 adds the inter- action between PMC and Media attention to test hypothesis 3a. The coefficient of the interaction exhibits the expected positive sign and is significant ( 𝛽 = 0.083, p < 0.05 ). When holding other variables at the sample means, with low media attention (i.e., 6.491, 1 s.d. below the mean), there is a differ- ence of −1.141 in the CSR score between firms with maxi- mum versus minimum PMC, whereas with high media atten- tion (i.e., 10.419, 1 s.d. above the mean), there is a difference of 17.238 in the CSR score between firms with maximum versus minimum PMC. It is not surprising that, if firms do not receive sufficient attention from the media, their CSR behavior is less likely to be rewarded, and so the benefits of CSR behavior would not exceed the costs. However, the benefits of CSR would increase with an increase of media attention and gradually outweigh the costs. Therefore, the promoting effect of PMC on CSR is not permanent and can exhibit a dampening effect when lacking sufficient attention from some essential stakeholders such as the media. Model 2 in Table 4 tests hypothesis 3b by considering traditional media attention. We add the interaction between PMC and Traditional media attention, and the coefficient of the interaction exhibits the expected positive sign and is significant ( 𝛽 = 0.137, p < 0.05 ). When holding other vari- ables at the sample means, with low traditional media atten- tion (i.e., 2.618, 1 s.d. below the mean), there is a difference of 1.503 in the CSR score between firms with maximum versus minimum PMC. In contrast, with high traditional media attention (i.e., 5.195, 1 s.d. above the mean), there is a difference of 21.406 in the CSR score between firms with maximum versus minimum PMC. Similarly, because the coefficient of PMC is negative when traditional media attention is lower than the threshold of 2.423 (0.332/0.137), the promoting effect of PMC on CSR becomes a dampen- ing effect, suggesting that, if firms do not receive sufficient attention from traditional media, their best course of action is to reduce costs and avoid CSR investment. Model 3 in Table 4 tests hypothesis 3b by considering new media (online media) attention. Similarly, the coef- ficient estimate of PMC × New media attention is posi- tive and significant at the 5% level ( 𝛽 = 0.160, p < 0.05 ), suggesting that the positive effect of PMC on firms’ CSR behavior is enhanced by new media attention. However, the coefficient estimate of PMC is negative and significant at the 5% level. The marginal effect of PMC on firms’ CSR behavior is positive but becomes negative when the value of new media attention is less than 4.006 (= 0.641/0.160). The evidence is consistent with, but also extends, our theory that PMC positively contributes to CSR behavior by obtain- ing sufficient attention from new media (for the top 92% of Table 3   (continued) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 CSR scores CSR scores CSR scores CSR scores CSR scores CSR scores CSR scores CSR scores (0.365) (0.311) (0.514) (0.376) (0.548) (0.481) (0.491) (0.446) Regional development 0.827 0.784 0.832 0.886 0.711 0.697 0.713 0.759 (2.311) (2.347) (2.374) (2.316) (2.279) (2.418) (2.328) (2.371) Experience 5.863*** 5.760*** 5.862*** 6.004*** 5.752*** 5.624*** 5.785*** 5.807*** (1.266) (1.257) (1.316) (1.326) (1.342) (1.244) (1.296) (1.244) Political connection 0.719 0.714 0.903 0.849 1.019* 0.679 0.902 0.795 (0.569) (0.544) (0.528) (0.510) (0.521) (0.574) (0.544) (0.500) Firm diversification 1.562** 1.506** 1.565** 1.612** 1.717** 1.414** 1.623** 1.534** (0.624) (0.607) (0.615) (0.629) (0.612) (0.574) (0.598) (0.620) Inverse Mills ratio 2.746*** 2.693*** 2.839*** 2.879*** 2.796*** 2.678*** 2.764*** 2.786*** (0.850) (0.828) (0.847) (0.862) (0.853) (0.818) (0.828) (0.801) Constant −47.920** −47.874** −49.743** −49.489** −44.193** −49.505** −44.491** −46.772** (21.293) (21.643) (21.665) (22.201) (19.977) (20.872) (20.457) (20.462) Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Number of observations 1722 1722 1722 1722 1722 1722 1722 1722 Adjusted R2 0.285 0.283 0.287 0.288 0.294 0.287 0.287 0.289 Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 881Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… Table 4   PMC, media attention, and CSR behavior Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 CSR scores CSR scores CSR scores CSR scores PMC −0.559* −0.332 −0.641** −0.491** (0.290) (0.209) (0.278) (0.192) Media attention 1.495*** (0.340) PMC × Media attention 0.083** (0.035) Traditional media attention 2.477*** 2.079*** (0.555) (0.507) PMC × Traditional media attention 0.137** 0.097 (0.062) (0.070) New media attention 2.706*** 0.434 (0.651) (0.546) PMC × New media attention 0.160** 0.061* (0.057) (0.032) Firm size 2.240*** 2.298*** 2.474*** 2.270*** (0.371) (0.359) (0.463) (0.382) Firm age −0.300** −0.306** −0.297* −0.300** (0.132) (0.129) (0.140) (0.133) ROA 2.214 2.380 3.731 2.482 (5.598) (5.311) (5.757) (5.696) Slack resources 2.542* 2.641* 2.508* 2.561* (1.209) (1.299) (1.255) (1.217) Top 10 ownership 0.054* 0.053* 0.053* 0.055* (0.027) (0.026) (0.028) (0.029) Independent directors −7.421* −7.580* −6.737 −7.469* (3.812) (3.823) (3.904) (3.827) CEO duality −1.083** −1.107** −0.969* −1.121** (0.478) (0.484) (0.453) (0.474) Foreign sales 0.631 0.702 0.575 0.675 (0.589) (0.575) (0.618) (0.588) Shanghai −2.909*** −2.940*** −2.971*** −2.893*** (0.435) (0.451) (0.415) (0.424) Regional development 1.027 1.088 0.866 1.014 (2.478) (2.467) (2.465) (2.429) Experience 5.675*** 5.547*** 5.792*** 5.636*** (1.242) (1.264) (1.197) (1.185) Political connection 0.757 0.740 0.776 0.716 (0.615) (0.622) (0.587) (0.595) Firm diversification 2.004** 1.944** 1.912** 1.917** (0.676) (0.684) (0.656) (0.719) Inverse Mills ratio 2.726*** 2.667*** 2.771*** 2.714*** (0.758) (0.770) (0.751) (0.726) Constant −42.647* −41.158 −46.476* −40.993* (23.780) (24.467) (21.973) (23.263) Industry fixed effects Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Number of observations 1722 1722 1722 1722 Adjusted R2 0.308 0.309 0.300 0.309 882 Y. Wang, C. Marquis firms) and, consequently, receiving rewards. For these firms, the CSR differentiation effect is greater than the CSR cost effect. However, for a minority of firms that receive little new media attention (i.e., the remaining 8%), results suggest that PMC reduces CSR behavior. The change in the marginal effect of PMC on firms’ CSR behavior (i.e., from a promot- ing effect to a dampening effect) supports our argument that stakeholder attention can reconcile opposing assessments between PMC and CSR. Finally, we compare the moderating effects of tradi- tional and new media attention in model 4, which includes both PMC × Traditional media attention and PMC × New media attention. The results show that only the coefficient of PMC × New media attention is positive and significant ( p < 0.1 ), indicating that the moderating effect of new media attention is more profound than that of traditional media attention. In summary, hypotheses 3a and 3b are strongly supported. Figure 2 illustrates how PMC interacts with different stakeholder attention. We place three types of stakeholder attention—attention from the government (including central and local governments), the investment market (including institutional investors and financial analysts), and the media (including traditional and new media)—on the x-axis in dif- ferent panels and use a line to indicate the effect of stake- holder attention on the relationship between PMC and CSR. As shown in panels a and c–g, the relationship between PMC and CSR is stronger in firms that receive more attention from these stakeholders, as the slopes are positive, while in panel b, local government attention blunts the promoting effect of PMC on CSR, as the slope is negative. Furthermore, panels a and d–g, also show that the accentuating effect of PMC on CSR behavior can become a dampening effect when firms do not receive sufficient central government attention, analyst attention, and media attention (including both traditional and new media attention) because the lines cross 0. Robustness Checks We also perform several robustness tests to ensure the valid- ity of our main results. As previously mentioned, we argue that privately controlled firms in China are more likely than SOEs to engage in CSR behavior in competitive environ- ments to obtain an advantage and differentiate themselves from rivals. Privately controlled firms are more likely to rely on recognition and rewards from CSR behavior, and they are more sensitive to efficiency and profitability concerns. Thus, we conduct robustness checks on the relationship between PMC and CSR behavior in SOEs to show that our private firm sample is the most suitable for examining our hypoth- eses. Model 1 in Table 5 presents the second-stage results of the Heckman model for the sample with only SOEs. We find that the coefficient of PMC in model 1 is negative and significant, suggesting that PMC does not incentivize SOEs’ CSR strategies because they more commonly follow gov- ernment orders in choosing their CSR strategies. Model 2 presents the second-stage results of the Heckman model for both the SOEs and private firm samples. The coefficient of PMC in model 2 is positive and significant (p < 0.05), imply- ing that our baseline results still exist but become weaker when including SOEs. RKS presents three different dimensions of the CSR score; thus, we examine how the different CSR dimensions react to PMC and stakeholder attention. We consider the dimensions of social responsibility strategy and innovation (model 1 in Table 6), disclosure content (model 2), and tech- nical sufficiency (model 3), respectively. For the dimensions of social responsibility strategy and innovation as well as disclosure content, which reflect firms’ CSR behavior or per- formance more effectively, we find that the baseline results and all interaction terms9 remain positive and statistically significant (p < 0.01), with the exception of the technical sufficiency dimension, which only focuses on the technical aspects of CSR reports. Furthermore, we use an alternative to the RKS composite measure which considers only the dimensions of social responsibility strategy and innovation as well as disclosure content together in model 4 and find the results are still robust. In addition, we use an alternative proxy for CSR behavior or performance (i.e., CSR (Hexun)) in model 5. We obtain CSR rating data from the Hexun web- site (www.​hexun.​com), China’s first vertical financial portal website. Hexun’s CSR evaluation system is based on firms’ CSR and annual financial reports, ranging from 0 to 10010 and these rating have been used by other scholars to repre- sent firms’ CSR performance (e.g., Gong et al., 2021; Xue et al., 2022). The results of model 5 in Table 6 show that our baseline is robust to using an alternative measure of CSR. One potential concern is that just examining listed firms could lead to inaccurate PMC calculations. To mitigate this concern, we use the annual number of all types of firms in each industry, PMC (all), as an alternative measure of PMC. We obtain these data from China’s National Bureau of Sta- tistics, which has included information about the number of firms at the industry level since 2013.11 The results in model 6 of Table 6 show that the positive relationship between PMC and CSR is robust when an alternative measure of PMC is used. 9  The results are available from the authors. 10  Almost 97% of our sampled firm–year observations have Hexun CSR rating; thus, we did not use the two-stage Heckman selection model. 11  This annual survey does not include firms in the agriculture, for- estry, animal husbandry, and fishery industry (CSRC industry code: A), financial industry (CSRC industry code: J), or comprehensive industry (CSRC industry code: S). http://www.hexun.com 883Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… To alleviate the potential concern that firms issuing CSR reports are better performing and may engage in more CSR, we split our sample into firms with high and low financial performance on the basis of the sample median of finan- cial performance. Model 1 in Table 7 indicates the high- performance firm subsample, and model 2 indicates the low-performance firm subsample. We estimate the baseline specification for these subsamples separately. The coef- ficients of PMC in models 1 and 2 are both positive and significant (p < 0.01). In addition, we introduce a dummy variable, Better performance, which equals 1 if the firm’s financial performance is above the sample median and 0 otherwise. We include Better performance and the interac- tion term of PMC and Better performance in our baseline specification in model 3 of Table 7. The coefficient of this interaction term is not statistically significant, suggesting that the estimated effect of PMC is unlikely to be affected by firms’ performance. Bigger firms are typically seen as more visible, and thus stakeholders may be more aware of their activities. To illus- trate the influence of firm size, we also conducted an addi- tional robustness check. Specifically, we split our sample into firms above and below median firm size; model 1 in Table 8 represents the larger-size firm subsample, and model 2 the small-size firm subsample. The coefficient of PMC in model 1 is bigger and more significant ( � = 0.469 , p < 0.01) than that in model 2 ( � = 0.058 , p < 0.05). We also included the interaction term of PMC and Firm size in our baseline specification in model 3 of Table 8. The coefficient of the interaction between PMC and Firm size is also positive and significant. These results also are generally supportive of our conclusions that the promoting effect of PMC on CSR is more profound when stakeholders pay more attention to the firms. Panel (a) Panel (b) Panel (c) Panel (d) Panel (e) Panel (f) Panel (g) Average marginal effects with 95% Cls. -. 4 -. 2 0 .2 A ss o ci at io n b et w ee n P M C a n d C S R 0 18 Central Government Attention Morderating Effect of Central Government Attention -. 2 0 .2 .4 .6 A ss o ci at io n b et w ee n P M C a n d C S R 0 39 Local Government Attention Morderating Effect of Local Government Attention 0 .0 5 .1 .1 5 .2 .2 5 .3 .3 5 A ss o ci at io n b et w ee n P M C a n d C S R 0 9 Institutional Investor Attention Morderating Effect of Institutional Investor Attention -.4 -.2 0 .2 .4 .6 As so ci at io n be tw ee n PM C a nd C SR 0 3.7 Analyst Attention Morderating Effect of Analyst Attention -1 -. 5 0 .5 1 A ss o ci at io n b et w ee n P M C a n d C S R 2 15 Media Attention Morderating Effect of Media Attention -.5 0 .5 1 As so ci at io n be tw ee n PM C a nd C SR .5 7 Traditional Media Attention Morderating Effect of Traditional Media Attention -.5 0 .5 1 As so ci at io n be tw ee n PM C a nd C SR 2.2 7.7 New Media Attention Morderating Effect of New Media Attention Fig. 2   Graphic illustration of moderating effects 884 Y. Wang, C. Marquis Furthermore, we also tested alternative measures of institutional investor attention, financial analyst attention, and media attention. We use the number of research reports issued on a particular firm to capture attention by insti- tutional investors and divide it into long- and short-term attention. To better measure financial analyst attention related to CSR, we identify the key theme of each earnings forecast with the help of a textual analysis. CSR theme atten- tion, which is the natural logarithm of the number of analysts who issued earnings CSR- and sustainability-themed fore- casts for the firm, is used as an alternative measurement. For media attention, we use Social media attention, the natural logarithm of the number of WeChat articles about a focal firm in a given year, to capture the media attention. The results are summarized in Table 9. Models 1–4 suggest that, when using an alternative measure of institutional investor attention, the promoting effect of PMC on CSR is still more profound for firms with more institutional investor attention. Additionally, compared with short-term institutional investor attention, only long-term institutional investor attention can significantly amplify this promoting effect of PMC. Model 5 also shows that our findings remain robust to alternative measures of financial analyst attention. Moreover, model 6 indicates that the interaction term PMC × Social media attention remains positive and significant (p < 0.05), indicat- ing that social media attention strengthens the relationship between PMC and CSR. We also focus on issues related to endogeneity. Our reported baseline may be driven by reverse causality or unobserved variables despite using a one-period lagged index of PMC and controlling for industry and category–year interaction fixed effects in all analyses. Table 5   Robustness checks: relationship between PMC and CSR behavior for the SOE-only sample and the sample including both SOEs and privately controlled firms Robust standard errors clustered at the industry level are in parenthe- ses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Only SOEs SOEs and private firms PMC −0.046*** 0.026** (0.013) (0.010) Inverse Mills ratio 0.926 2.320* (0.560) (1.098) Constant −71.937*** −79.850*** (23.669) (19.429) Control variables Yes Yes Industry fixed effects Yes Yes Industry category–year fixed effects Yes Yes Number of observations 3374 5036 Adjusted R2 0.355 0.333 Table 6   Robustness check: alternative CSR measure and alternative PMC measure Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 CSR (strategy) CSR (disclosure) CSR (technical) CSR (strat- egy + disclo- sure) CSR (Hexun) CSR score PMC 0.049*** 0.097*** 0.006 0.145*** 0.019** (0.010) (0.006) (0.007) (0.015) (0.009) PMC (all) 1.953* (1.070) Inverse Mills ratio 0.799** 1.287*** 0.379*** 2.086*** 3.003*** (0.357) (0.302) (0.095) (0.638) (0.613) Constant −8.122 −13.506 −3.353 −21.629 −28.146*** −120.591*** (3.4376) (5.2477) (1.6831) (16.876) (6.762) (35.762) Control variables Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Yes Yes Number of observations 1679 1679 1679 1679 10,080 967 Adjusted R2 0.375 0.223 0.341 0.270 0.378 0.195 885Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… We identify China’s substantial revision to the Catalogue of Industries for Guiding Foreign Investment in 201112 as an important exogenous shift that may have affected PMC. This revision greatly reduced restrictive foreign investment projects and market access to primary, secondary, and ter- tiary industries (Kong et al., 2020). Specifically, the revi- sion scope of the Catalogue in 2011 was relatively large, and several industries were included such as power, heat, gas, and water production and supply; wholesale and retail; leasing and business services; scientific research and tech- nical services; water conservancy, environment and public facilities management; education; health and social work; and some specific projects in manufacturing. Finally, for- eign direct investment (FDI) increased by 42% on average during this revision.13 The relaxation of foreign investment regulations has facilitated the entry of foreign competitors into local markets (Hu & Wang, 2020; Jiang & Lu, 2018; Wang & Zhang, 2019). In addition, firms cannot accurately predict the specific time and scope of the Catalog revision, and the revision objective is not related to the firms’ CSR behavior. Therefore, the 2011 revision provides a sharp exogenous shift in PMC that Chinese companies face from foreign rivals. Following recent studies (e.g., Hu & Wang, 2020; Jiang & Lu, 2018; Wang & Zhang, 2019), we use this quasi-natural experiment to mitigate endogeneity concerns. The indicator variable Subject to revision denotes firms in industries that experienced a relaxation during this revision. After revision indicates the period after revision. Following extant studies (e.g., Jia et al., 2019), we estimate the fol- lowing fixed-effects difference-in-differences (diff-in-diff) regression: where Subject to revisioni × After revisiont is the diff-in- diff estimator that equals 1 if firm i belongs to the industry in which foreign investment regulations were relaxed after the 2011 revision and 0 otherwise. As in our main analysis, Xi,t is a set of control variables expected to influence CSR behavior, Tt refers to the dummy variables used to control for year fixed effects, �i is used to control for firm fixed effects, and �i,t is an error term. Therefore, Subject to revisioni and After revisiont are omitted because of firm and year fixed effects. CSR scorei,t+1 = � + �1Subject to revisioni ∗ After revisiont + �2Xi.t + �i + Tt + �i,t, Table 7   Robustness check: considering firm performance Robust standard errors clustered at the industry level are in parenthe- ses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 CSR score CSR score CSR score PMC 0.201*** 0.139*** 0.185*** (0.027) (0.023) (0.031) Better performance 0.684 (0.662) PMC × Better performance −0.017 (0.069) Inverse Mills ratio 2.273** 3.133*** 2.623*** (1.051) (0.877) (0.792) Constant −25.144 −72.819*** −47.046* (21.312) (20.312) (22.159) Control variables Yes Yes Yes Industry fixed effects Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Number of observations 1050 672 1722 Adjusted R2 0.237 0.359 0.282 Table 8   Robustness check: influence of firm size Robust standard errors clustered at the industry level are in parenthe- ses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 CSR score CSR score CSR score PMC 0.058** 0.469*** −3.933*** (0.027) (0.034) (0.788) PMC × Firm size 0.187*** (0.037) Inverse Mills ratio 1.748** 2.623*** 2.743*** (0.728) (0.817) (0.730) Constant 7.125 −95.661*** −79.220*** (17.916) (28.677) (25.598) Control variables Yes Yes Yes Industry fixed effects Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Number of observations 825 897 1722 Adjusted R2 0.263 0.344 0.296 12  Although the 2011 revision was not the only revision to the Cata- logue of Industries for Guiding Foreign Investment, other revisions in 1997/2002/2004/2007 are not applicable to our sample period. Fol- lowing Wang and Zhang (2019), we compared the content of each revision and found that the 2011 revision was substantial. In addi- tion, we examined the causal relationship between the 2011 revision (Subject to revision ×After revision) and product market competition, and found a positive and significant relationship between Subject to revision×After revision and PMC. 13  We calculated the growth in FDI amount by using the change in the national average amount of FDI 3 years before and after 2011, since there was another revision of the Catalogue in 2007 (4 years before 2011). 886 Y. Wang, C. Marquis In models 1 and 2 of Table 10, the coefficients of the interaction between Subject to revision and After revision are positive and statistically significant ( p < 0.01 ) regardless of whether control variables are included; That is, when firms are in industries that experience unexpected revisions of for- eign investment regulations, thereby increasing competition in those industries post-revision, they will implement more CSR behavior. As such, the reported baseline is robust. Furthermore, we conducted two placebo tests to address the concern that our results could be driven by chance. The first placebo test artificially moves the event time 2 years prior to the actual event year and repeats the baseline diff- in-diff analysis in model 1 of Table 11. The second placebo test artificially assigns the treatment and control group status in our sample and repeats the baseline diff-in-diff analysis in model 2 of Table 11. We did not find any significant results in either test. Overall, these placebo tests suggest that our main results were unlikely to be driven by chance. Figure 3 reveals the relationship dynamics between the unexpected revision in 2011 and CSR behavior, suggesting the satisfac- tion of the parallel trend assumption. In addition, to further rule out endogeneity concerns, we examine how exogenous shocks that unexpectedly shape attention from various stakeholders in China affect our results. China Central Television (CCTV), the official media of the Chinese government, has disclosed irresponsi- ble corporate social on March 15—international consumer rights day—every year since 1991, in what is known as Table 9   Robustness check: alternative measures of institutional investor attention, financial analyst attention, and media attention Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 CSR score CSR score CSR score CSR score CSR score CSR score PMC 0.163*** 0.159*** 0.168*** 0.163*** −0.043 0.147*** (0.024) (0.026) (0.023) (0.024) (0.076) (0.032) Institutional investor attention 0.828 (0.473) PMC × Institutional investor attention 0.025* (0.014) Long-term institutional investor attention 9.418*** 10.530*** (2.550) (3.020) PMC × Long-term institutional investor attention 1.529** 1.750** (0.637) (0.749) Short-term institutional investor attention 0.724* 0.539 (0.406) (0.441) PMC × Short-term institutional investor attention 0.016 −0.014 (0.013) (0.014) CSR theme attention 1.195*** (0.350) PMC × CSR theme attention 0.071** (0.030) Social media attention 1.695*** (0.532) PMC × Social media attention 0.017** (0.008) Inverse Mills ratio 2.702*** 2.673*** 2.682*** 2.700*** 2.802*** 2.768*** (0.818) (0.813) (0.825) (0.820) (0.874) (0.768) Constant −45.655** −48.188** −45.326** −45.673** −46.378** −41.163* (20.240) (20.924) (20.375) (20.211) (19.611) (19.859) Control variables Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Yes Yes Number of observations 1722 1722 1722 1722 1722 1722 Adjusted R2 0.283 0.283 0.283 0.284 0.297 0.288 887Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… the 315 Evening Party.14 After this date, affected and other related firms (here, we consider other firms in the same industry) receive more attention from different stakehold- ers. For example, Chaqi Caiye, a firm that produces sauer- kraut in Hunan Province, was exposed for its dirty produc- tion process on 15 March 2022. As shown in Fig. 4, the Baidu search volume of “Chaqi Caiye” was 0 prior to 15 March, but increased to 108,346 on 16 March 2022. After this disclosure, Kangshifu, a famous instant noodles firm, terminated its partnership with Chaqi and began investigat- ing food safety. On 16 March, there was scrutiny from offi- cial regulators, and some related local government officials were punished. Meanwhile, the public search for the word “sauerkraut” (Suan Cai in Chinese) increased significantly, and the Baidu search volume of “sauerkraut” (Suan Cai; Fig. 5) increased from 951 (on 14 March) to 24,748 (on 16 March). Similarly, the Baidu search volume of “Kangshifu” (Fig. 6) increased from 4550 (on 14 March) to 137,967 (on 16 March) in 2022. Although these disclosures by CCTV are not completely random, focusing on other related firms (not the focal firm itself) can mitigate the concern of selection bias (Deng et al., 2022). Therefore, unexpected disclosures in other firms in the same industry (but not by the firm itself) represent stag- gered exogenous shocks to the firm’s exposure to stakeholder attention, allowing us to identify the causal effect of stake- holder attention on the relationship between PMC and CSR. We hand-collected data on the CCTV’s 315 Evening Party disclosures every year and use a diff-in-diff specifi- cation to assess the effect of stakeholder attention on the relationship between PMC and CSR behavior on the basis of the following regression setup: Here, Attention shock is a time-varying dummy variable that equals 1 for firms in industry j, in which at least one other firm (not the firm itself) is disclosed by CCTV’s 315 Evening Party after the disclosure, and 0 otherwise. PMC is product market competition, and X is a set of control vari- ables expected to influence CSR behavior. IMR is the inverse Mills ratio from the first-stage model. Sector*Year refers to the dummy variables used to control for broader industry sector–year fixed effects, λ connotes the dummy variables used to control for industry fixed effects, and ε is an error term. CSR scorei,t+1 = � + �1PMCj,t ∗ Attention shockj,t + �2PMCj,t + �3Attention shockj,t + �4Xi,t + �5IMRi,t + Sector∗Yearjt + �j + �i,t. Table 10   Robustness check: quasi-natural experiment to address endogeneity Robust standard errors clustered at the industry level are in parenthe- ses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 CSR score CSR score Subject to revision × After revision 2.931*** (0.524) 2.585*** (0.742) Constant 27.669*** (0.923) 9.333 (37.743) Control variables No Yes Firm fixed effects Yes Yes Year fixed effects Yes Yes Number of observations 1854 1722 Adjusted R2 0.360 0.361 Table 11   Robustness check: placebo tests for quasi-natural experi- ment Robust standard errors clustered at the industry level are in parenthe- ses; two-tailed for all variables. *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 CSR score CSR score Subject to revision × After revision (pseudo)  − 0.525 (1.696) Subject to revision (pseudo) × After revision − 1.054 (0.838) Constant 30.964 7.928 (155.592) (38.190) Control variables Yes Yes Firm fixed effects Yes Yes Year fixed effects Yes Yes Number of observations 214 1722 Adjusted R2 0.241 0.357 -4 -2 0 2 4 6 C ha ng e in re la tiv e C SR b eh av io rs 2008 2009 2011 2012 2013 2014 2015 2016 2017 Years Fig. 3   Dynamics of the unexpected revision of foreign investment guidance and CSR behavior 14  For more details, see https://​www.​163.​com/​dy/​artic​le/​G55SA​ 3MO05​19U3I5.​html. https://www.163.com/dy/article/G55SA3MO0519U3I5.html https://www.163.com/dy/article/G55SA3MO0519U3I5.html 888 Y. Wang, C. Marquis We use this exogenous shock to interact with PMC in model 2 of Table 12. The results show that Attention shock significantly strengthens ( �= 0.501,p < 0.01 ) the relationship between PMC and CSR, but the promoting effect of PMC on CSR becomes a dampening effect when the company lacks attention (the effect of PMC on CSR = −0.308 + 0.501*0). In addition, we examine another exogenous shock, the disappearance of Malaysia Airlines Flight 370, to represent an unexpected change in stakeholder attention. The Boe- ing 777 carrying 239 people from Kuala Lumpur to Beijing Fig. 4   The Baidu search volume of “Chaqi Caiye” Fig. 5   The Baidu search volume of “sauerkraut” (“Suan Cai” in Chinese) Fig. 6   The Baidu search volume of “Kangshifu” 889Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… disappeared on 8 March 2014. Although this event was unre- lated to any Chinese airlines, aircraft manufacturers, or air- craft parts manufacturers, it resulted in significant concerns about transportation safety (especially airline transportation) and spread to the transportation equipment manufacturing industry. Therefore, we also apply the diff-in-diff approach to test the contingent role of stakeholder attention. Here, the variable Crash attention equals 1 for related firms that received attention from different stakeholders (i.e., firms that belong to the transportation equipment manufac- turing industry and the transportation industry) after the disappearance of Malaysia Airlines Flight 370 and 0 for all firms before the air crash and for all firms that were never treated (i.e., firms that do not belong to the transportation equipment manufacturing industry or the transportation industry). The results of model 4 in Table 12 show that Crash attention significantly strengthens the relationship between PMC and CSR. Thus, our results are still robust. To better capture the central government’s priority in sus- tainable development, we conduct an additional test split- ting the full sample into subsamples—higher versus lower— by the number of paragraphs that concern sustainable CSR scorei,t+1 = � + �1PMCj,t ∗ Crash attentioni,t + �2PMCj,t + �3Crash attentioni,t + �4Xi,t + �5IMRi,t + Sector ∗ Yearjt + �j + �i,t. development scaled by the total number of paragraphs on government objectives mentioned in the Five-Year Plans. We compare the contingent effect of Central government atten- tion across the subsamples. In the higher group, central gov- ernment attention is more CSR-related, while in the lower group, central government attention is less CSR-related. The coefficient of PMC × Central government attention is only positive and significant in model 1 of Table 13, suggesting that central government attention can amplify the benefits of CSR and exceed the cost under PMC pressure, especially when central government attention is more CSR-related. Additionally, we assume that the GDP growth priorities of local governments vary. Therefore, by utilizing the text of Chinese provincial governments’ 5-year plans,15 we capture the degree to which provincial governments prioritize GDP growth. More specifically, following Luo et al. (2017), we divide provincial governments into two groups, high GDP priority and low GDP priority, based on (1) whether a pro- vincial government’s approach toward economic develop- ment is focused on GDP growth targets or whether it also emphasizes achieving GDP growth through balanced and sustainable means; and (2) the number of paragraphs that are primarily related to GDP growth scaled by the total number Table 12   Robustness check: exogenous shocks of stakeholder attention Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 CSR score CSR score CSR score CSR score PMC 0.174*** −0.308** 0.175*** 0.174*** (0.022) (0.110) (0.022) (0.022) Attention shock −2.083 1.233 (1.595) (0.709) PMC × Attention shock 0.501*** (0.113) Crash attention 3.689* 23.035*** (1.776) (7.439) PMC × Crash attention 3.070*** (0.986) Inverse Mills ratio 2.630*** 2.629*** 2.646*** 2.639*** (0.797) (0.793) (0.812) (0.810) Constant −48.390** −54.544** −48.272** −48.353** (21.377) (21.617) (20.908) (20.857) Control variables Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Number of observations 1722 1722 1722 1722 Adjusted R2 0.282 0.284 0.282 0.282 15  Among the itemized objectives specified in the 5-year plans, the first paragraph in the list is always about overall economic develop- ment. 890 Y. Wang, C. Marquis of paragraphs on government objectives. We then compare the contingent effect of local government attention in these two groups. The results, displayed in Table 14, show that the negative contingent effect of local government attention is more profound in the subsample in which firms receive more attention from local governments with high GDP priorities. Thus, our results on the contingent effect of local govern- ment attention still hold when considering the variations in GDP priorities among local governments. Another potential concern is the overlap between tradi- tional and new media attention. Thus, we repeat our analysis using alternative measures of new media attention. We use the residual term of the regression of new media attention on traditional media attention, New media attention (residual), to capture the remaining new media attention that cannot be captured by traditional media attention. Moreover, we repeat our analysis with another measure of new media attention: the sum of Baidu search volumes of each firm in a given year (New media attention (Baidu)). Like the Google search volume, the Baidu search volume reflects the search frequency of certain keywords on Baidu, the most popular search engine in China, accounting for approximately 80% of the entire Chinese Internet search market (Kong et al., 2019). A high Baidu search volume can be considered a high level of new media attention. The coefficient of PMC × New media attention (residual) in model 2 of Table 15 and the coefficient of PMC × New media attention (Baidu) in model 4 are still positive and significant, suggesting that the role of new media attention in the relationship between PMC and CSR is robust. Finally, to rule out concerns about the correlation between central government attention and PMC, we conduct a matching approach in which we pair a firm that belongs to an industry included in or mentioned in the Chinese central government’s 5-year plans with a firm that experiences a similar PMC condition but does not belong to these indus- tries. The positive moderating effect of government attention remains robust. Because of page length requirements, the results are not included here but are available upon request. Discussion and Conclusions This study examines an important unresolved theoretical issue concerning the effect of PMC on firms’ CSR strat- egies and behaviors. Unlike prior work that has mainly focused on establishing this relationship (e.g., Duanmu et al., 2018; Flammer, 2015), our investigation focuses on how stakeholder attention affects the extent to which a firm’s CSR behavior can be recognized and rewarded by its external stakeholders. We propose a new framework in which CSR behavior is viewed as the result of the trade-off between the benefits and costs of CSR, and stakeholder attention is a precondition for realizing the differentia- tion effect of CSR. Our study not only examines the con- tingent effect of stakeholder attention between different stakeholder groups (e.g., the government, institutional investors, financial analysts, and media), but also within particular stakeholder groups (e.g., the central government versus local governments, long-term institutional investors versus short-term institutional investors, female analysts versus male analysts, and traditional media versus new media). By comparing the varying magnitudes of the contin- gent effects between different stakeholder groups, our study finds that, although government attention, institu- tional investor attention, financial analyst attention, and media attention all magnify the promoting effect of PMC on CSR, financial analyst and media attention have more nuanced and sensitive effects. If firms do not receive suf- ficient attention from the media or financial analysts, their CSR strategy will not be realized, recognized, or rewarded, and the promoting effect of PMC becomes a dampening effect. Furthermore, by comparing the effects of stake- holder attention within specific stakeholder groups, we find that only the attention from stakeholders who appre- ciate and prioritize CSR—i.e., the central government, long-term institutional investors, and female financial analysts—can amplify its benefits and thus lead to the dif- ferentiation effect we hypothesize. Table 13   Robustness check: subsample analysis for central govern- ment attention Robust standard errors clustered at the industry level are in parenthe- ses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Lower group Higher group PMC −0.086 −0.158 (0.326) (0.090) Central government attention 0.302 0.520 (0.535) (0.310) PMC × Central government attention 0.015 0.021*** (0.019) (0.005) Inverse Mills ratio 0.783 3.351*** (0.914) (0.615) Constant −43.555 −41.095 (25.997) (24.024) Control variables Yes Yes Industry fixed effects Yes Yes Industry category–year fixed effects Yes Yes Number of observations 506 1216 Adjusted R2 0.230 0.218 891Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… Our findings suggest that, although CSR can work as a differentiation strategy, its costs should not be ignored, and the tension between benefits and costs is not completely constant. Stakeholder attention can function as a precondi- tion or contingent condition for the trade-off between the costs and benefits of an organization’s CSR strategy. This Table 14   Robustness check: attention from local governments with different GDP priority Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 GDP priority High Low High Low PMC 0.243*** 0.434 0.416*** 0.294*** (0.051) (0.379) (0.110) (0.082) Local government attention −0.156*** −0.226** −0.403*** −0.021 (0.015) (0.103) (0.085) (0.035) PMC × Local government attention −0.009*** −0.009 −0.012** −0.002 (0.001) (0.013) (0.004) (0.004) Inverse Mills ratio 1.435 3.809*** 1.539* 3.439*** (1.171) (0.986) (0.774) (1.153) Constant 2.735 −50.195* −74.965** −37.753* (19.652) (25.692) (31.449) (21.081) Control variables Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Number of observations 670 1052 894 828 Adjusted R2 0.309 0.312 0.281 0.313 Table 15   Robustness check: overlap concerns between traditional media attention and new media attention Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Model 1 Model 2 Model 3 Model 4 CSR score CSR score CSR score CSR score PMC 0.175*** 0.157*** 0.167*** 0.032 (0.021) (0.019) (0.028) (0.060) New media attention (residual) −0.041 1.679** (0.715) (0.576) PMC × New media attention (residual) 0.213*** (0.052) New media attention (Baidu) 10.066** 14.791** (4.339) (5.825) PMC × New media attention (Baidu) 0.726* (0.344) Inverse Mills ratio 2.646*** 2.659*** 2.950*** 2.964*** (0.784) (0.760) (0.621) (0.607) Constant −48.034** −48.208** −42.691* −23.525 (19.671) (19.486) (22.028) (23.190) Control variables Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Industry category–year fixed effects Yes Yes Yes Yes Number of observations 1722 1722 1371 1371 Adjusted R2 0.282 0.285 0.228 0.229 892 Y. Wang, C. Marquis enables us to identify the boundary conditions and under- lying stakeholder mechanisms that influence how CSR can be strategically used. Moreover, we expand research on stakeholder theory and organization attention. While prior studies have explored the impact of corporate attention to stakeholders on firm decision-making (Crilly & Sloan, 2014; Flammer et al., 2019), little research has been devoted to how the attention from external stakeholders affects firms. We thus propose and examine how stakeholder effects also reflect an attentional process that then shapes the tension between the benefits and costs of CSR. In addition, our study sheds new light on multi-stake- holder perspectives. Although prior studies have recognized the importance of moving from a single stakeholder to mul- tiple stakeholders when understanding firm strategy (Oli- ver, 1991) and explored the independent and interdepend- ent effects of political and economic stakeholders (Li et al., 2018), less attention has been paid to the different and even conflicting interests and demands within a specific stake- holder group. Focusing on the attention from stakeholders with conflicting interests within a specific stakeholder group, we provide an in-depth understanding of the multi-stake- holder perspective and provide insight about the open ques- tion in the stakeholder theory literature regarding whether investments in different stakeholder groups are equivalent. Our study also has several ethical and practical implica- tions. First, our study suggests CSR can be value-enhancing via the differentiation effect, especially under the pressure of PMC in China. Thus, for Chinese firms, as well as emerg- ing market firms operating in environments with growing awareness of CSR and business ethics, such behaviors can be used to obtain competitive advantage (Russo & Fouts, 1997; Waheed & Zhang, 2022). At the same time, we also highlight the importance of enhancing ethical awareness, which is an implicit condition for achieving more balanced and sustainable growth for firms as well as for society (Lee et al., 2019). Because not all stakeholders have the same attitude toward business ethics and sustainability (Tang & Tang, 2012) managers should comprehensively assess their stakeholders, paying more attention to the differences both between different stakeholder groups and within a specific stakeholder group to better achieve the goal of corporate sustainability. Finally, our results on the role of media (especially new media) and financial analysts also indicate the importance of information, information intermediaries, and information technology. With the development of digital technology and big data, connections between different stakeholders have been strengthened (Barnett et al., 2020; Hilbert & Lopez, 2011). Simultaneously, stakeholders’ limited attention may also be more likely to be manipulated by others, especially information intermediaries. Thus, in the digital era, both managers and stakeholders should be attentive to the role of information intermediaries in the transmission of informa- tion concerning ethical issues, such as information interme- diaries’ motivations and orientations. As an early attempt to document that the effect of PMC on CSR is contingent on different types of stakeholder atten- tion, this study is limited in a number of ways. First, we have utilized third-party data and scores based on CSR reports to depict firms’ substantive CSR performance and some related proxies. The arguments and empirical tests would be greatly strengthened if future studies could capture more directly firms’ CSR activities and behaviors. Second, we acknowledge that the limitations of archival data prevent us from identifying the detailed attentional processes whereby stakeholders allocate their limited and cognitively oriented attention, and from perfectly capturing stakeholder attention to certain firms—and their specific CSR activities in particu- lar. Multimethod research and accurate measurements will facilitate further studies. Third, we intentionally focus our multi-stakeholder approach on three kinds of well-known stakeholder groups: government, investment market, and media. However, because attention from other stakeholder groups could also affect the PMC–CSR relationship, we welcome future research that builds on our framework to consider the potential contingent effect of other specific stakeholder attention. Finally, regarding generalizability, we contextualized our analysis using China as the setting, as it is a single, large country where CSR orientation varies between different stakeholder groups and within a specific stakeholder group. Future researchers may consider testing these arguments using alternative data, such as those from developed economies. 893Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… Appendix 1: The Relationship between CSR Scores and CSR Activities Model 1 Model 2 Model 3 Model 4 Model 5 Donation 0.133*** 0.115*** (0.042) (0.035) Employment oppor- tunities 4.762*** 4.671*** (0.748) (0.655) Environmental recognition 3.351*** 2.978*** (0.497) (0.510) Sustainable strategy 3.540*** 2.851*** (0.461) (0.369) Constant 25.591*** 24.947*** 25.206*** 24.478*** 25.269*** (0.179) (0.065) (0.068) (0.105) (0.174) Industry fixed effects Yes Yes Yes Yes Yes Industry category– year fixed effects Yes Yes Yes Yes Yes Number of observa- tions 1510 1583 1583 1583 1510 Adjusted R2 0.187 0.200 0.170 0.177 0.257 Donations is the annual amount of corporate philanthropy in millions; Employment opportunities captures whether the firm drives employment in the local market in a given year; Environmental recognition is a dummy variable that equals 1 if the firm receives any environmental recogni- tion in a given year and 0 otherwise; Sustainable strategy measures whether the firm conducts a sustainable development strategy in a given year. Robust standard errors clustered at the industry level are in parentheses; two-tailed for all variables *p < 0.10; **p < 0.05; ***p < 0.01 Acknowledgements  We thank Tilly Feng, Cathy Lu, Kunyuan Qiao and Eric Zhao for their comments and suggestions on this research and the University of Electronic Science and Technology of China (UESTC) School of Management and Economics for its support. Yichen Wang acknowledges the funding support from the Scientific Research Foundation of the Chengdu University of Information Tech- nology (KYTZ202448). Data availability  The data utilized in this study are sourced from pub- licly available records or obtained through commercial databases, as described in the Methods section. Declarations  Conflict of interest  We declare there are no potential conflicts of inter- est. Open Access  This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. References Allen, F., & Gale, D. (2000). Financial contagion. Journal of Political Economy, 108(1), 1–33. Awaysheh, A., Heron, R. A., Perry, T., & Wilson, J. I. (2020). On the relation between corporate social responsibility and financial performance. Strategic Management Journal, 41(6), 965–987. Barber, B. M., & Odean, T. (2001). Boys will be boys: Gender, over- confidence, and common stock investment. Quarterly Journal of Economics, 116(1), 261–292. Barnett, M. L. (2014). Why stakeholders ignore firm misconduct: A cognitive view. Journal of Management, 40(3), 676–702. Barnett, M. L., Henriques, I., & Husted, B. W. (2020). The rise and stall of stakeholder influence: How the digital age limits social control. Academy of Management Perspectives, 34(1), 48–64. Bear, S., Rahman, N., & Post, C. (2010). The impact of board diversity and gender composition on corporate social responsibility and firm reputation. Journal of Business Ethics, 97(2), 207–221. Borghesi, R., Houston, J. F., & Naranjo, A. (2014). Corporate socially responsible investments: CEO altruism, reputation, and share- holder interests. Journal of Corporate Finance, 26, 164–181. Bouquet, C., & Birkinshaw, J. (2008). Weight versus voice: How for- eign subsidiaries gain attention from corporate headquarters. Academy of Management Journal, 51(3), 577–601. Branstetter, L. G., & Feenstra, R. C. (2002). Trade and foreign direct investment in China: A political economy approach. Journal of International Economics, 58(2), 335–358. http://creativecommons.org/licenses/by/4.0/ 894 Y. Wang, C. Marquis Campbell, J. L. (2007). Why would corporations behave in socially responsible ways? An institutional theory of corporate social responsibility. Academy of Management Review, 32(3), 946–967. Capriotti, P. (2011). Communicating corporate social responsibility through the internet and social media. In Ø. Ihlen, J. L. Barlett, & S. May (Eds.), The handbook of communication and corpo- rate social responsibility (pp. 358–378). Wiley. Carroll, G. R., & Hannan, M. T. (2000). Why corporate demography matters: Policy implications of organizational diversity. Cali- fornia Management Review, 42(3), 148–163. Chan, M. C., Watson, J., & Woodliff, D. (2014). Corporate govern- ance quality and CSR disclosures. Journal of Business Ethics, 125(1), 59–73. Chen, J., Su, X., Tian, X., & Xu, B. (2022). Does customer-base structure influence managerial risk-taking incentives? Journal of Financial Economics, 143(1), 462–483. Chen, X., Harford, J., & Li, K. (2007). Monitoring: Which institu- tions matter? Journal of Financial Economics, 86(2), 279–305. Chen, X., Liang, X., & Wu, H. (2023). Cross-border mergers and acquisitions and CSR performance: Evidence from China. Journal of Business Ethics, 183(1), 255–288. Cho, T. S., & Hambrick, D. C. (2006). Attention as the mediator between top management team characteristics and strategic change: The case of airline deregulation. Organization Science, 17(4), 453–469. Chung, R., Firth, M., & Kim, J. B. (2002). Institutional monitoring and opportunistic earnings management. Journal of Corporate Finance, 8(1), 29–48. Cox, P., Brammer, S., & Millington, A. (2004). An empirical exami- nation of institutional investor preferences for corporate social performance. Journal of Business Ethics, 52(1), 27–43. Crilly, D., & Sloan, P. (2014). Autonomy or control? Organizational architecture and corporate attention to stakeholders. Organiza- tion Science, 25(2), 339–355. Declerck, M. D., & M’Zali, B. (2012). Product market competition and corporate social responsibility. Working paper, University Lille Nord de France, Lille, France. Deng, Y., Liao, L., Yu, J., & Zhang, Y. (2022). Capital spillover, house prices, and consumer spending: Quasi-experimental evi- dence from house purchase restrictions. Review of Financial Studies, 35(6), 3060–3099. Duanmu, J. L., Bu, M., & Pittman, R. (2018). Does market com- petition dampen environmental performance? Evidence from China. Strategic Management Journal, 39(11), 3006–3030. Dupire, M., & M’Zali, B. (2018). CSR strategies in response to competitive pressures. Journal of Business Ethics, 148(3), 603–623. Economist. (2020). The newspaper industry is taking a battering. https://​www.​econo​mist.​com/​brita​in/​2020/​04/​18/​the-​newsp​aper-​ indus​try-​is-​taking-​a-​batte​ring Fernández-Kranz, D., & Santaló, J. (2010). When necessity becomes a virtue: The effect of product market competition on corporate social responsibility. Journal of Economics & Management Strat- egy, 19(2), 453–487. Fisher, G., Kotha, S., & Lahiri, A. (2016). Changing with the times: An integrated view of identity, legitimacy, and new venture life cycles. Academy of Management Review, 41(3), 383–409. Fisman, R., Heal, G., & Nair, V. (2006). A model of corporate philan- thropy. Working Paper, Columbia University New York, NY. Flammer, C. (2015). Does product market competition foster corporate social responsibility? Evidence from trade liberalization. Strate- gic Management Journal, 36(10), 1469–1485. Flammer, C., Hong, B., & Minor, D. (2019). Corporate governance and the rise of integrating corporate social responsibility criteria in executive compensation: Effectiveness and implications for firm outcomes. Strategic Management Journal, 40(7), 1097–1122. Gaspar, J. M., Massa, M., & Matos, P. (2005). Shareholder investment horizons and the market for corporate control. Journal of Finan- cial Economics, 76(1), 135–165. Gentry, R. J., & Shen, W. (2013). The impacts of performance relative to analyst forecasts and analyst coverage on firm R&D intensity. Strategic Management Journal, 34(1), 121–130. Godfrey, P. C., Merrill, C. B., & Hansen, J. M. (2009). The relationship between corporate social responsibility and shareholder value: An empirical test of the risk management hypothesis. Strategic Management Journal, 30(4), 425–445. Goldeng, E., Grünfeld, L. A., & Benito, G. R. (2008). The performance differential between private and state-owned enterprises: The roles of ownership, management and market structure. Journal of Management Studies, 45(7), 1244–1273. Gong, G., Huang, X., Wu, S., Tian, H., & Li, W. (2021). Punishment by securities regulators, corporate social responsibility and the cost of debt. Journal of Business Ethics, 171(2), 337–356. Graves, S. B., & Waddock, S. A. (1994). Institutional owners and cor- porate social performance. Academy of Management Journal, 37(4), 1034–1046. Harrison, J. S., Bosse, D. A., & Phillips, R. A. (2010). Managing for stakeholders, stakeholder utility functions, and competitive advantage. Strategic Management Journal, 31(1), 58–74. Hartzmark, S. M., & Sussman, A. B. (2019). Do investors value sus- tainability? A natural experiment examining ranking and fund flows. Journal of Finance, 74(6), 2789–2837. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161. Hilbert, M., & Lopez, P. (2011). The world’s technological capacity to store, communicate, and compute information. Science, 332, 60–65. Hirshleifer, D., & Teoh, S. H. (2003). Limited attention, information disclosure, and financial reporting. Journal of Accounting and Economics, 36(1–3), 337–386. Hoffman, A. J., & Ocasio, W. (2001). Not all events are attended equally: Toward a middle-range theory of industry attention to external events. Organization Science, 12(4), 414–434. Hu, L., & Wang, J. (2020). Product market competition and corpo- rate innovation efficiency—A research based on one quasi natural experiment. Modern Economic Research, 9, 98–106. (In Chinese). Ioannou, I., & Serafeim, G. (2015). The impact of corporate social responsibility on investment recommendations: Analysts’ per- ceptions and shifting institutional logics. Strategic Management Journal, 36(7), 1053–1081. Jacquemin, A. P., & Berry, C. H. (1979). Entropy measure of diversi- fication and corporate growth. Journal of Industrial Economics, 27, 359–369. Jia, N., Huang, K. G., & Zhang, M. C. (2019). Public governance, corporate governance, and firm innovation: An examination of state-owned enterprises. Academy of Management Journal, 62(1), 220–247. Jiang, L., & Lu, Y. (2018). Whether market competition increasing boosts the leverage of SOEs. China Industrial Economics, 11, 155–173. (In Chinese). Jin, H., Qian, Y., & Weingast, B. R. (2005). Regional decentraliza- tion and fiscal incentives: Federalism, Chinese style. Journal of Public Economics, 89(9–10), 1719–1742. Jones, T. M., Harrison, J. S., & Felps, W. (2018). How applying instru- mental stakeholder theory can provide sustainable competitive advantage. Academy of Management Review, 43(3), 371–391. Kahneman, D. (1973). Attention and effort (Vol. 1063, pp. 218–226). Prentice-Hall. Kim, H. D., Kim, T., Kim, Y., & Park, K. (2019). Do long-term insti- tutional investors promote corporate social responsibility activi- ties? Journal of Banking & Finance, 101, 256–269. https://www.economist.com/britain/2020/04/18/the-newspaper-industry-is-taking-a-battering https://www.economist.com/britain/2020/04/18/the-newspaper-industry-is-taking-a-battering 895Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility… Kim, K. H., Kim, M., & Qian, C. (2018). Effects of corporate social responsibility on corporate financial performance: A com- petitive-action perspective. Journal of Management, 44(3), 1097–1118. Kish-Gephart, J. J., Harrison, D. A., & Treviño, L. K. (2010). Bad apples, bad cases, and bad barrels: Meta-analytic evidence about sources of unethical decisions at work. Journal of Applied Psychology, 95(1), 1–31. Kölbel, J. F., Busch, T., & Jancso, L. M. (2017). How media cover- age of corporate social irresponsibility increases financial risk. Strategic Management Journal, 38(11), 2266–2284. Kong, D., Lin, C., & Liu, S. (2019). Does information acquisition alleviate market anomalies? categorization bias in stock splits. Review of Finance, 23(1), 245–277. Kong, D., Zhu, L., & Yang, Z. (2020). Effects of foreign investors on energy firms’ innovation: Evidence from a natural experiment in China. Energy Economics, 92, 105011. Lau, C., Lu, Y., & Liang, Q. (2016). Corporate social responsibility in China: A corporate governance approach. Journal of Busi- ness Ethics, 136(1), 73–87. Lee, J. H., Byun, H. S., & Park, K. S. (2018). Product market com- petition and corporate social responsibility activities: Per- spectives from an emerging economy. Pacific-Basin Finance Journal, 49, 60–80. Lee, S. Y., Zhang, W., & Abitbol, A. (2019). What makes CSR com- munication lead to CSR participation? Testing the mediating effects of CSR associations, CSR credibility, and organization– public relationships. Journal of Business Ethics, 157, 413–429. Lennox, C. S., Francis, J. R., & Wang, Z. (2012). Selection models in accounting research. Accounting Review, 87(2), 589–616. Li, H., & Zhou, L. A. (2005). Political turnover and economic per- formance: The incentive role of personnel control in China. Journal of Public Economics, 89(9–10), 1743–1762. Li, J., Xia, J., & Zajac, E. J. (2018). On the duality of political and economic stakeholder influence on firm innovation performance: Theory and evidence from Chinese firms. Strategic Management Journal, 39(1), 193–216. Li, S., & Lu, J. W. (2020). A dual-agency model of firm CSR in response to institutional pressure: Evidence from Chinese publicly listed firms. Academy of Management Journal, 63(6), 2004–2032. Lin, N. (2011). Capitalism in China: A centrally managed capitalism (CMC) and its future. Management and Organization Review, 7(1), 63–96. Luo, X., Wang, H., Raithel, S., & Zheng, Q. (2015). Corporate social performance, analyst stock recommendations, and firm future returns. Strategic Management Journal, 36(1), 123–136. Luo, X. R., Wang, D., & Zhang, J. (2017). Whose call to answer: Institutional complexity and firms’ CSR reporting. Academy of Management Journal, 60(1), 321–344. Luo, X. R., Zhang, J., & Marquis, C. (2016). Mobilization in the internet age: Internet activism and corporate response. Acad- emy of Management Journal, 59(6), 2045–2068. Luo, Y., Xue, Q., & Han, B. (2010). How emerging market govern- ments promote outward FDI: Experience from China. Journal of World Business, 45(1), 68–79. Madsen, P. M., & Rodgers, Z. J. (2015). Looking good by doing good: The antecedents and consequences of stakeholder atten- tion to corporate disaster relief. Strategic Management Jour- nal, 36(5), 776–794. Margolis, J. D., & Walsh, J. P. (2003). Misery loves companies: Rethinking social initiatives by business. Administrative Sci- ence Quarterly, 48(2), 268–305. Marquis, C., & Lee, M. (2013). Who is governing whom? Execu- tives, governance, and the structure of generosity in large US firms. Strategic Management Journal, 34(4), 483–497. Marquis, C., & Qian, C. (2014). Corporate social responsibility reporting in China: Symbol or substance? Organization Sci- ence, 25(1), 127–148. Marquis, C., & Qiao, K. (2022). Mao and markets: The communist roots of Chinese enterprise. Yale University Press. Marquis, C., & Raynard, M. (2015). Institutional strategies in emerg- ing markets. Academy of Management Annals, 9(1), 291–335. Marquis, C., Toffel, M. W., & Zhou, Y. (2016). Scrutiny, norms, and selective disclosure: A global study of greenwashing. Organi- zation Science, 27(2), 483–504. McDonnell, M. H., & King, B. (2013). Keeping up appearances: Reputational threat and impression management after social movement boycotts. Administrative Science Quarterly, 58(3), 387–419. McWilliams, A., & Siegel, D. (2001). Corporate social responsibil- ity: A theory of the firm perspective. Academy of Management Review, 26(1), 117–127. Ocasio, W. (1997). Towards an attention-based view of the firm. Strategic Management Journal, 18(S1), 187–206. Ocasio, W. (2011). Attention to attention. Organization Science, 22(5), 1286–1296. Ocasio, W., Laamanen, T., & Vaara, E. (2018). Communication and attention dynamics: An attention-based view of strategic change. Strategic Management Journal, 39(1), 155–167. Oliver, C. (1991). Strategic responses to institutional processes. Academy of Management Review, 16(1), 145–179. Orlitzky, M., Schmidt, F. L., & Rynes, S. L. (2003). Corporate social and financial performance: A meta-analysis. Organization Studies, 24(3), 403–441. Phillips, L. W., Chang, D. R., & Buzzell, R. D. (1983). Product qual- ity, cost position and business performance: A test of some key hypotheses. Journal of Marketing, 47(2), 26–43. Porter, M. E. (1980). Corporate strategy. Free Press. Porter, M. E. (1985). Corporate advantage. Free Press. Qian, C., Lu, L. Y., & Yu, Y. (2019). Financial analyst coverage and corporate social performance: Evidence from natural experi- ments. Strategic Management Journal, 40(13), 2271–2286. Russo, M. V., & Fouts, P. A. (1997). A resource-based perspective on corporate environmental performance and profitability. Acad- emy of Management Journal, 40(3), 534–559. Shleifer, A. (2004). Does competition destroy ethical behavior? American Economic Review, 94(2), 414–418. Siegel, D. S., & Vitaliano, D. F. (2007). An empirical analysis of the strategic use of corporate social responsibility. Journal of Economics & Management Strategy, 16(3), 773–792. Su, W., & Tsang, E. W. (2015). Product diversification and financial performance: The moderating role of secondary stakeholders. Academy of Management Journal, 58(4), 1128–1148. Tang, Z., & Tang, J. (2012). Stakeholder–firm power difference, stakeholders’ CSR orientation, and SMEs’ environmental performance in China. Journal of Business Venturing, 27(4), 436–455. Tian, Z., Wang, R., & Yang, W. (2011). Consumer responses to cor- porate social responsibility (CSR) in China. Journal of Busi- ness Ethics, 101, 197–212. Toussaint, L., & Webb, J. R. (2005). Gender differences in the rela- tionship between empathy and forgiveness. Journal of Social Psychology, 145(6), 673–685. Tuggle, C. S., Schnatterly, K., & Johnson, R. A. (2010). Attention patterns in the boardroom: How board composition and pro- cesses affect discussion of entrepreneurial issues. Academy of Management Journal, 53(3), 550–571. Waheed, A., & Zhang, Q. (2022). Effect of CSR and ethical practices on sustainable competitive performance: A case of emerging markets from stakeholder theory perspective. Journal of Busi- ness Ethics, 175(4), 837–855. 896 Y. Wang, C. Marquis Walker, L. J. (2006). Gender and morality. In Handbook of moral development (pp. 93–115). Lawrence Erlbaum Associates. Wang, D., & Luo, X. R. (2019). Retire in peace: Officials’ political incentives and corporate diversification in China. Administrative Science Quarterly, 64(4), 773–809. Wang, H., Gibson, C., & Zander, U. (2020). Editors’ comments: Is research on corporate social responsibility undertheorized? Academy of Management Review, 45(1), 1–6. Wang, H., & Qian, C. (2011). Corporate philanthropy and corpo- rate financial performance: The roles of stakeholder response and political access. Academy of Management Journal, 54(6), 1159–1181. Wang, J., & Zhang, H. (2019). An empirical research on the relation- ship between product market competition and investment effi- ciency. Journal of Finance and Economics, 45(10), 125–137. (In Chinese). Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Xue, F., Chen, Q., Chan, K. C., & Yi, Z. (2022). Is corporate social responsibility value relevant? Evidence from a quasi-natural experiment of anti-corruption campaign. Journal of Business Research, 140, 520–532. Yan, X., & Zhang, Z. (2009). Institutional investors and equity returns: Are short-term institutions better informed? Review of Financial Studies, 22(2), 893–924. Yin, J., & Zhang, Y. (2012). Institutional dynamics and corporate social responsibility (CSR) in an emerging country context: Evidence from China. Journal of Business Ethics, 111, 301–316. Yu, T., Sengul, M., & Lester, R. H. (2008). Misery loves company: The spread of negative impacts resulting from an organizational crisis. Academy of Management Review, 33(2), 452–472. Yuan, X., & Zuo, J. (2011). Transition to low carbon energy policies in China—from the Five-Year Plan perspective. Energy Policy, 39(6), 3855–3859. Zhang, J., Marquis, C., & Qiao, K. (2016). Do political connections buffer firms from or bind firms to the government? A study of corporate charitable donations of Chinese firms. Organization Science, 27(5), 1307–1324. Zhao, E. Y., Fisher, G., Lounsbury, M., & Miller, D. (2017). Optimal distinctiveness: Broadening the interface between institutional theory and strategic management. Strategic Management Jour- nal, 38(1), 93–113. Zhu, R., & Lin, B. (2021). Energy and carbon performance improve- ment in China’s mining Industry: Evidence from the 11th and 12th five-year plan. Energy Policy, 154, 112312. Zyglidopoulos, S. C., Georgiadis, A. P., Carroll, C. E., & Siegel, D. S. (2012). Does media attention drive corporate social responsibil- ity? Journal of Business Research, 65(11), 1622–2162. Publisher's Note  Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Does Product Market Competition Promote or Reduce Firms’ Corporate Social Responsibility Behavior? How Stakeholder Attention Shapes Responsiveness to Stakeholders Abstract Introduction Stakeholder Attention and the Effect of Product Market Competition Government Attention Investment Market Attention Media Attention Methods Estimation Method Results Robustness Checks Discussion and Conclusions Appendix 1: The Relationship between CSR Scores and CSR Activities Acknowledgements References