U.S. Fortune 500’s stakeholders engagement during the COVID-19 pandemic: Evidence for proactive approaches

Abstract

In times of a national crisis such as COVID-19, it is important for organizations to show that they are good corporate citizens. At the same time, organizations should carefully select the type of messages that resonate with stakeholders so as to reduce stakeholder skepticism. This study examines how U.S. Fortune 500 companies discussed their COVID-19 pandemic CSR actions on Facebook over 15 months and how the public responded to such messages. We identified three CSR themes: internal stakeholder proactive CSR, external stakeholder proactive CSR, and external stakeholder accommodative CSR. When publics engaged, external stakeholder proactive CSR was significantly associated with better behavioral engagement outcomes, more positive emotional engagement outcomes, and less negative emotions. However, such effects are moderated by industry type. Our findings inform public relations theory and practice and suggest that in times of major crises, organizations should prioritize proactive approaches to engage external stakeholders while being mindful of specific institutional contexts.

The COVID-19 pandemic provides public relations scholars with a front row seat to watch a crisis unfold. In real time, we can watch how organizations are responding to a global threat to their business models. We can also watch how organizations are proactively changing to survive in an uncertain environment. COVID-19 is the type of crisis that requires just about every person and every organization to change “business as usual.”

A growing body of literature has explored the implications of the COVID-19 pandemic for corporate social responsibility (Bae et al., 2021, Carroll, 2021, Crane and Matten, 2020, He and Harris, 2020). Specifically, studies have noted some unique challenges brought about by the pandemic. For example, unlike previous crises, the pandemic has no apparent end point. This requires corporations to adopt long-term strategies in response to the pandemic (Carroll, 2021). Moreover, the widespread impact of the pandemic means that corporations may not only deal with changing demands of one stakeholder group, but that of many stakeholders (e.g., consumers, employees, business partners, governments) simultaneously (Bae et al., 2021). In addition, the sharp increase in social needs may also mean that governments and publics now demand corporations to increase their donations and take more active roles in CSR (Crane & Matten, 2020). At the same time, He and Harris (2021) noted that the pandemic may allow corporations to build more genuine and authentic CSR when addressing urgent challenges.

Communication helps publics understand what organizations are doing to serve society and public engagement helps organizations anticipate public expectations (Hurst et al., 2020; Johnston, 2014). In addition, social media, with their decentralized structure and relatively cost-friendly access, offer an important channel for corporations to engage stakeholders in important conversations such as CSR efforts (Bachmann & Ingenhoff, 2016). From a research perspective, social media allow us to track how corporate messages reassure, explain and reflect a changing world. Social media also allows us to study how the public responds to those messages in real time.

CSR research shows that society expects more from corporations than returning shareholder profit (Crane and Matten, 2020, L'etang, 1994, Lim and Greenwood, 2017). However, CSR communication is not always positively received as some messages or actions are viewed as self-serving and shallow (Rim & Kim, 2016). Organizations need to ensure that their CSR communication and activities are meaningful and truly contribute to society. Otherwise, their efforts may be met with public skepticism. This study proceeds from the realities of COVID-19 as a multifaceted crisis. It is a health crisis, an information crisis and a communication crisis. As such, a central question that guides our research is: how do organizations engage their stakeholders using CSR messages related to COVID-19 during this pandemic? Answers to this question will provide insights into how organizations are adapting to the COVID-19 pandemic and may help us better understand emerging CSR practices and trends that could shape the future of CSR (Carroll, 2021).

Americans are looking to large corporations for leadership during COVID-19 (Whitaker, 2020). To gain insight into how large corporations are engaging publics, we focus on how U.S Fortune 500 companies communicated their COVID-19 related CSR strategies during the pandemic. We collected this sample because the severe nature of this crisis created considerable public demands for resource rich companies to step up their CSR efforts and contribute to society.

The first part of the paper provides a review of CSR and explores the important role that social media plays in public engagement in CSR messaging. The next section reviews foundational literature about stakeholder management and the cocreational turn to stakeholder engagement. The following sections summarize the data collection and analysis methods used to examine 16,477 CSR posts by 480 corporations during COVID-19 and report the results of discourse analysis using structural topic modeling (STM). The final section positions the findings into the CSR literature and identifies ways that corporations can better engage publics and fulfill their social license to operate. Implications for public relations theory and practice are also addressed.

1. CSR communication in trying times

1.1. CSR and crisis communication

Corporate social responsibility (CSR) can be viewed as “legal, ethical, and economic responsibilities of an organization in relationship to key stakeholders and society” (Bortree, 2014, p. 2). Research has shown that CSR may affect consumers’ perceptions of companies, leading consumers to associate CSR active companies with both positive reputation and higher economic capability (Kim, 2011). Recognizing the benefits of CSR and also recognizing that they are community members, many companies routinely incorporate CSR into their operations and communication (Lim & Greenwood, 2017).

In this study we focus on how corporations, especially highly profitable organizations such as the U.S Fortune 500 companies, communicate their COVID-19 related CSR strategies. The COVID-19 context helps us understand how companies respond to a large-scale crisis with CSR actions, and how their CSR communication changes as social expectations for businesses change during a global crisis (Crane & Matten, 2020).

Social media are important venues for corporations to provide timely updates and engage stakeholders in trying times (Abitbol & Lee, 2017). Stakeholders are any individuals or organizations that “can affect, or are affected by, the achievement of the organization’s objectives” (Freeman, 1984, p. 25). Stakeholder theory argues that corporations can be understood as a set of relationships among groups that have a stake in the activities that make up the corporation (Freeman, 1994). Further, the theory suggests that corporations should identify relevant stakeholders and they have the responsibility to operate in the best interest of all stakeholders (Freeman & Dmytriyev, 2017). Based on stakeholder theory, we expect that corporations would use social media to engage with stakeholders, understand their demands, and respond to their expectations.

Moreover, as noted by Coombs and Holladay (2015), CSR initiatives can be assets in times of crisis. Research shows that crises tend to damage organizations’ reputations and existing relationships with stakeholders (Ham & Kim, 2019). Although as an external event, COVID-19 may not damage corporate reputation, it still could adversely affect organization-stakeholder relationships. As such, corporations’ communication about CSR could be an important component of their COVID-19 crisis management (Tao & Song, 2020).

In the current study, we examine corporations’ COVID-19 related CSR communication on Facebook. We focus on Facebook because in comparison to other popular media such as Twitter, studies have found that Facebook’s user base is more diverse and reflects a representative sample of the general population. In comparison, Twitter tends to be used by “elites” for information gathering and professional purposes (Bossetta, 2018). As such, Facebook provides an ideal platform for us to assess a broad range of stakeholders’ responses to CSR engagement.

1.2. Social-mediated CSR communication

Corporations increasingly use social media to set their CSR agenda and directly interact with stakeholders (Wang & Huang, 2018). Research shows that companies routinely discuss a wide range of CSR topics on their social media accounts. For example, Cho et al. (2017) examined how corporations with good reputations communicated CSR efforts on Facebook. They found these companies’ CSR messages included a variety of topics such as community/employee involvement, philanthropic contribution, public health, environment, and education issues.

In addition, studies show that the topics that companies discuss on social media affect engagement outcomes. Wang and Huang (2018) found that internal CSR messages, oriented towards employees’ benefits, were more likely to elicit stakeholders’ trust, satisfaction, feelings of control mutuality and commitment towards the organization. Saxton et al. (2019) studied Fortune 500 companies’ CSR messages on Twitter and found that education and environment topics had higher public resonance (operationalized as the extent to which a tweet is retweeted). Furthermore, the company-cause fit of CSR topics1 also plays an important role because stakeholders expect congruence between a company’s mission and CSR message. Abitbol and Lee (2017) found companies that communicate CSR messages congruent with their respective industries were more likely to receive public engagement on Facebook. Similarly, Lee and Chung (2018) used an experiment to examine CSR awareness as the initial step of the outcomes of CSR communication. They found that low company-cause fit elicited more short-term memory, because violation of expectation triggers more cognitive resources, but high company-cause fit was better for long-term memory, because such aligned values were easier to remember (Lee & Chung, 2018).

As a severe crisis, COVID-19 challenges almost every aspect of organizational life. Stakeholders may have new communication expectations and thus companies may now need to discuss a wider range of topics. It is interesting, therefore, to examine what topics the most resource rich companies in the United States focused on during the COVID-19 pandemic. As such, we propose the following question:

RQ1

: Between January 1, 2020, to March 8, 2021, which COVID-19 related CSR responses did U.S. Fortune 500 companies discuss on their Facebook accounts?

The answer to this question will provide insights into which topics emerged in social media as companies engage in CSR to respond to COVID-19. However, as we further explain below, discussing CSR in times of crisis is not always a risk-free behavior. CSR skepticism may occur and lead the public to doubt companies’ intentions or view organizational actions as opportunistic or insensitive. As such, choosing the right CSR topics to discuss is a strategic decision.

1.3. CSR skepticism and risk associated with mis-communicated CSR

CSR communication does not always lead to favorable engagement outcomes and it can sometimes have counterproductive outcomes due to public skepticism. Rim and Kim (2016) explored the multidimensional construct of CSR skepticism and defined it as “publics’ inclination to question, disbelieve, and distrust an organization’s CSR motives, communication and activity, and informativeness without the consideration of predispositional traits such as cynicism toward business and management” (p. 262). Publics may hold negative attitudes toward CSR messages because they perceive that corporations engaging in such actions seek to serve themselves rather than truly caring for the community. Thus, some publics are unwilling to interact with CSR messages on social media (Cho et al., 2017). In the context of Facebook updates, CSR skepticism may be reflected as low engagement rates or negative emotional responses to company posts.

Research has shown that reducing CSR skepticism is not an easy task. When designing CSR messages, companies need to consider the expression of their motivation, the credibility of the content, and issue congruence (the consistency between companies’ long-standing positions on issues and their CSR messages) (Bachmann and Ingenhoff, 2016, Shim et al., 2017, Wei and Kim, 2021). For instance, Shim et al. (2017) compared the effects of strategic CSR, or self-interested motives for profit seeking, and normative CSR, or public-oriented motives for altruistic or social causes. They found strategic CSR led to stronger perceptions of corporate hypocrisy and negative attitude towards companies. However, companies may not always choose to communicate public-oriented motives either. In another study, Wei and Kim (2021) found that stating a firm-serving motive could reduce CSR skepticism in crisis response because it is a signal of transparent corporate actions. Such messages may help attenuate public suspicion of greenwashing.

Despite such public skepticism, Bachmann and Ingenhoff (2016) found that the disclosure of CSR actions that are proactive or exceeding normative standards could help companies gain legitimacy, even in the presence of stakeholders’ psychological reactance towards extensive CSR disclosure. In times of a national crisis such as COVID-19, it is important for companies to discuss the actions they are taking to help communities in need. At the same time, companies should carefully select the type of messages that resonate with stakeholders and reduce stakeholders’ skepticism. To understand considerations in stakeholder relationship management that may influence what messages companies should focus on, we now turn to stakeholder theory.

2. Stakeholder management evolves into stakeholder engagement

2.1. Stakeholder management

Given the definition of stakeholders (Freeman, 1984), Clarkson (1995) suggests that stakeholders can be classified as primary and secondary stakeholders. Primary stakeholders include a company’s employees, customers, investors, suppliers, government, and community with whom the corporation may have a formal, official, or contractual relationship. Among primary stakeholders, employees and investors are internal to a company whereas the rest are external to the company. Additionally, secondary stakeholders include media and special interest groups with whom the company does not have any contractual obligations.

Different stakeholders may provide different resources for companies. Primary stakeholders often have access to key resources such as labor resources (e.g., employees), raw materials (e.g., suppliers), or financial resources (e.g., consumers and investors). Secondary stakeholders may provide legitimacy and social license for companies’ operations (Clarkson, 1995). Stakeholder theory thus suggests that businesses need to understand stakeholders’ interests and build relationships with them (Freeman & Dmytriyev, 2017). At the core of stakeholder theory is the thesis that organizations should account for the interests of all their constituents, not just the interest of shareholders (Freeman, 1984, Freeman, 1994). Stakeholder theory shifts attention away from the single-valued objective of corporate profit-maximization, and seeks to address dialogic engagement between organizations and society (Laplume et al., 2008, Taylor and Kent, 2014). Effectively engaging stakeholders in such dialogic communication, according to stakeholder theory, could provide companies with competitive advantages and improve companies’ reputation and social performance (Hillman & Keim, 2001).

While the concept of dialogic communication between organizations and society may seem abstract, the process is to some degree observable on social media platforms such as Facebook. Companies across the world now utilize social media in stakeholder engagement (Manetti et al., 2017, Xu and Saxton, 2019). As such, companies’ disclosures on social media could be conceptualized as a form of early dialogic relationships between organizations and stakeholders through which corporations can share updates, observe stakeholder reactions, and adjust their communication strategies accordingly (Kent and Taylor, 2016, Taylor and Kent, 2014). Companies, especially large companies, could have a wide range of primary and secondary stakeholders. The COVID-19 pandemic could have far-reaching effects on all of these stakeholders.

2.2. Stakeholder engagement strategies

Stakeholders matter and the co-creational turn means that stakeholders co-create meanings with organizations (Botan & Taylor, 2004). Theorists have identified four stakeholder response strategies that include reactive, defensive, accommodative, and proactive (RDAP) message strategies (Clarkson, 1995, Freeman, 1984, Freeman, 1994). Both accommodative and proactive strategies would require a positive attitude from corporations to show sympathy or willingness to accommodate stakeholders. Between the two strategies, proactive strategies tend to require more substantial actions/resources to accommodate stakeholders. Accommodative strategies, on the other hand, may be more symbolic than resource based. In a reactive or defensive strategy, a company either fights against addressing stakeholder concerns or ignores them. Stakeholder theory suggests that companies take different approaches to address primary and secondary stakeholders and primarily allocate resources to address the needs of internal and external primary stakeholders (Laplume et al., 2008). In the context of the COVID-19 pandemic, since companies are not at fault for the crisis, they are less likely to take reactive or defensive strategies towards stakeholders. Therefore, in this study, we primarily focus on companies’ proactive and accommodative strategies.

In the context of COVID-19, we conceptualize companies’ proactive strategies as strategies that involve substantial allocation of resources such as donations, volunteering, partnerships, and operational changes. Moreover, proactive strategies may differ between internal and external stakeholders. For example, companies may choose to provide more bonuses or protective equipment to employees (internal) while also offering donations to local communities (external). Each is a proactive strategy but may take different forms. In addition, we conceptualize companies’ accommodative strategies as strategies that do not necessarily involve allocating resources but instead demonstrate a positive attitude towards stakeholders such as providing information updates on public health behaviors (e.g., update on COVID-19 prevention knowledge). It is necessary to note that accommodative and proactive strategies reflect a continuum rather than absolute distinctions. In this study we differentiate the two mainly for the purpose of concept operationalization.

In this study, we explore how companies’ communication of proactive or accommodative internal or external stakeholder strategies may lead to different engagement outcomes. Engagement is an important goal of dialogic communication (Kent & Taylor, 1998). Although social media has abundant potential for dialogic engagement with stakeholders, not all engagement strategies lead to ideal engagement outcomes (Kent & Taylor, 2016). This is because actual dialogue requires “empathy, patience, risk, sympathy, trust, and a willingness to be changed, or admit when one is wrong” (Kent & Taylor, 2016, p. 63). To build ethical organization–public relationships on social media, Kent and Taylor (2016) argued that corporations need to offer personalized interactions on an individual basis, serve the diverse needs of multiple publics, recognize the value of others, maintain transparency and build trust. Following the dialogic engagement principle, Park and Kang (2020) showed in an experiment that in a crowdsourced CSR campaign, public-generated, rather than company-generated, CSR ideas elicited more favorable public evaluations.

When corporations post on their Facebook accounts, stakeholders may respond with both behavioral and emotional responses. On Facebook, in terms of behavioral responses, stakeholders can comment on or share a particular post. When stakeholders share messages, they transmit organizational messages to their own social circle. Stakeholders can also directly engage with organizational messages by commenting, which requires more cognitive effort (Cho et al., 2014). With regard to emotional responses, stakeholders can react through animated emojis such as like (thumbs up), love (beating heart), haha (laughing face), wow (surprised face), sad (crying face), and angry (red pouting face). Such behavioral and emotional responses are visible on posts. They are easily observable to corporate communication managers as well as viewers of the company posts. Empirical research has used these interactive features to measure engagement outcomes (Abitbol & Lee, 2017).

Previous research has generally found that proactive strategies are more likely to lead to positive responses from stakeholders (Berman et al., 1999, Mishra and Suar, 2010). This is also consistent with recommendations from the dialogic approach, because proactive strategies are more likely to show companies’ commitment to stakeholders and build trust (Kent & Taylor, 2016). Drawing from previous research, we propose:

H1

: Corporate discourse topics that indicate proactive (internal and external) stakeholder strategies would receive more engagement in terms of the number of comments and shares than accommodative strategies.

H2

: Corporate discourse topics that indicate proactive (internal and external) stakeholder strategies would receive more positive responses (such as like and love) (H2a) and less negative responses (such as angry and sad) (H2b) than accommodative strategies.

2.3. The moderating role of industry in engagement

While all corporations may be expected to take actions to help Americans survive the COVID-19 pandemic, corporations from some industries may face different expectations for pro-social CSR actions. Stakeholder theorists have long recognized the connection between stakeholder theory and institutional theory (Freeman and Dmytriyev, 2017, Verbeke and Tung, 2013). At the industry level, the institutional context may impose industry-wide norms, rules, and beliefs that define or enforce what are considered acceptable corporate behaviors (DiMaggio & Powell, 1983). Stakeholder expectations are part of the isomorphism pressures that lead organizations to adopt certain practices over others. In addition, the composition of companies’ stakeholders may differ depending on the companies’ industry and business models (Freeman & Dmytriyev, 2017), and thus lead to divergence of stakeholder expectations across industries.

During COVID-19, according to a report released by S&P Global Market Intelligence (2020), based on industries revenue lost, the most severely affected industries are transportation, leisure and tourism, restaurants, oil and gas drilling, and auto parts and equipment, whereas the least affected industry is financial services such as specialized real estate investment trusts and insurance companies. Existing studies have shown that the relation between companies’ CSR ratings and financial performance depends on industries during the pandemic (Bae et al., 2021). Combined with studies of COVID-19’s general impact on industries (Baek et al., 2020), the least affected industries such as healthcare and business equipment witnessed a significant positive effect of CSR on corporate financial performance. Those heavily affected industries such as tourism and transportation did not have this effect due to their high risk (Bae et al., 2021). One interpretation could be that corporations from severely affected industries are expected to take more actions to address stakeholder concerns, and they are also more likely to face the blame if they fail to meet stakeholder expectations. When these industries and organizations are drawn into the public spotlight, their corporate strategies and decisions are more likely to be both acknowledged by stakeholders while at the same time draw public criticisms. Furthermore, research has shown that if companies advertise nonessential products during the pandemic, they will face negative evaluative outcomes from consumers (Mundel & Yang, 2021). Thus, for those companies within industries that are strongly affected by the COVID crisis, these companies’ CSR messages may trigger different public responses than less impacted companies (Coombs, 2004). As such, we propose that regardless of these companies’ discussion topics, they are likely to receive more behavioral engagement and stronger emotional responses if they are severely impacted by the crisis:

H3

: Corporate discourse from industries that are directly affected by COVID-19 are more likely to receive higher levels of engagement in terms of number of comments and shares.

H4

: Corporate discourse from industries that are directly affected by COVID-19 would receive more positive emotional responses (such as love) (H2a) and negative responses (such as angry and sad) (H2b).

In addition, when corporations from COVID-affected industries do communicate proactive response strategies, their stakeholder responses may differ from that of other industries2 . There is no previous research on how industry type may moderate stakeholder responses, thus prompting the following research question:

RQ2

: How does industry type moderate publics’ responses to discourse topics that indicate proactive (internal and external) stakeholder strategies?

The answers to these research questions and hypotheses will provide insight into CSR social media engagement during a global crisis and also help us understand real engagement outcomes from CSR discourse. The following sections report the methods used to derive the study results.

3. Method

We collected Facebook posts by U.S. Fortune 500 companies on COVID-19 from January 1, 2020, to March 8, 2021. The 15-month data collection was done through Facebook’s internal data archive hosted by CrowdTangle3 . In total, we obtained 16,477 posts by 480 Fortune 500 companies. We first conducted a discourse analysis to explore the topics that have emerged from the corpus of corporate public communication during the pandemic. We then examined how communicating different topics associated with various types of engagement and emotional public response.

3.1. Discourse analysis: structural topic modeling

Topic modeling has been used in analyzing textual data in public relations research (Xu & Xiong, 2020). Structural topic modeling (STM) is an increasingly popular topic modeling approach based on the algorithm Latent Dirichlet allocation (LDA) (Kwon et al., 2019). Compared to other topic modeling approaches, STM has two innovations. One is that it assumes the presence of multiple topics in one single document and provides a probability score. Another advantage is that it allows covariates from metadata to affect the topic frequency, e.g., document sources (Roberts et al., 2019). In corporate discourse, incorporating message sources is essential because different industries would have entirely different priorities in public communication. Data analysis was conducted with the stm package in R (Roberts et al., 2019).

We first processed corporate messages following standard natural language processing (NLP) procedures. We removed non-English posts and removed duplicate messages. When tokenizing texts, we removed punctuation, numbers, stop words and conducted stemming. After preprocessing, 9977 posts from 469 companies were left for further analysis. To select the optimal number of topics, the combination of computation and human judgment is essential (Chen et al., 2020). The built-in selection function showed that 19 and 20 topics had relatively high held-out log likelihood, low residuals and high semantic coherence ( Fig. 1). Two researchers independently checked the most frequent words associated with each topic and the most probable posts assigned to each topic. Both researchers agreed that 20 topics made the best fitting model. The two researchers also checked every topic to see if they are indeed about CSR. Assisted by visual representation of inter-topic relations ( Fig. 2), we classified these topics into three semantic themes: community information update, organizational crisis response and organizational contribution to the society.

Fig. 1.

Optimal topic number detection.

Fig. 2.

Visual representation of the correlation of 20 topics including information update (red), crisis response (blue), organizational contribution (green).

3.2. Independent variables

3.2.1. CSR themes

Based on the probability of topic membership assignment and clustering of topics under three broad themes, we calculated the probability that each post belonged to the three CSR themes. We then assigned each post to the CSR theme with the highest probability. To be cautious of disputable cases, we calculated the ratio of the highest probability to the second highest probability for each post (M = 5.73, SD = 9.17). The majority of posts had a prominent CSR theme membership with high confidence. About 2 percent of the posts had a small difference (below 0.1), and these posts were manually checked to ensure accuracy. ( Fig. 3)

Fig. 3.

Boxplots of engagement measures by CSR themes.

3.2.2. Corporate industry

We followed the Standard Industrial Classification (SIC) to code corporations into different industries. Since some companies didn’t have the number of followers as the control variable, the final dataset with complete observations included 9514 posts by 438 corporations. We chose to remove posts without complete variables after topic modeling because their messages made an integral part of the Fortune 500 public discourse. The manufacturing (n = 75), transportation (n = 46), retail (n = 89), finance (n = 77) and service (n = 102) industries had the most companies in the sample. Wholesale (n = 37), mining (n = 8) and construction (n = 4) have fewer companies, and thus they were merged into the category “others” (n = 49).

3.3. Control variables

3.3.1. Number of followers

The number of followers at the time a corporation made the post. This variable measures corporations’ social media status (M = 1,571,029, SD = 5,330,893)

3.3.2. Corporate revenue

We collected corporate revenue from the most recent year from Fortune.com (M = 80.84, SD = 315.42). According to stakeholder theory, the larger the financial revenue, the higher the expectation from stakeholders. It is likely that stakeholders expect those most resourceful companies to take actions during this crisis, and these companies are more likely to receive stronger engagement.

3.4. Dependent variables

3.4.1. Behavioral engagement outcome

On Facebook, users can create a comment on a post by clicking the comment button beneath the post. Users may also share a post by pushing the post to their friends and followers. The number of comments (M = 109.19, SD = 1738.00) and shares (M = 111.64, SD = 1108.50) provide the measure of the engagement outcomes.

3.4.2. Emotional engagement outcome

Facebook users can also react to posts by communicating a range of emotional responses. Users can express their emotional reactions through clicking these icons. Positive emotions include the number of likes (M = 451.49, SD = 4266.58) and loves (M = 64.51, SD = 674.29) that posts individually receive. Negative emotions include the number of sad (M = 2.98, SD = 69.80) and angry (M = 7.51, SD = 273.35) that posts receive. Figure 3 shows how engagement outcomes are different for CSR themes.

Descriptive statistics of original continuous variables are shown in Table 2. Social media engagement variables were all log transformed to meet the normality assumption. We adopted mixed effects models to test our hypotheses. This modeling technique is necessary because post-level observations were clustered within corporations and such dependency should be taken into account.

Table 2.

Descriptive statistics for continuous variables.

	Mean	SD	1	2	3	4	5	6	7
1. Comments	1.35	1.94
2. Shares	2.02	1.89	.72
3. Likes	3.78	1.71	.74	.82
4. Love	1.32	1.77	.77	.80	.86
5. Sad	.10	.55	.45	.38	.40	.38
6. Angry	.23	.75	.70	.48	.49	.49	.58
7. Followers	11.64	2.34	.53	.35	.56	.44	.20	.34
8. Revenue	80.84	315.42	.23	.14	.12	.15	.03	.14	.10

Note: 1. Variables from Facebook (1-7) are log transformed. Raw numbers are indicated in the text. 2. Corporate revenue is in billion USD.

4. Results

RQ1 asks which COVID-19 related CSR responses did U.S. Fortune 500 companies discuss on their Facebook accounts? The breakdown of each topic and associated top keywords and top sample posts are shown in Table 1. Among the three CSR themes, the community information update theme is mainly about general information on COVID-19 including testing, vaccines, use of technology and research. It also involves issues that concern ordinary people’s daily life in the pandemic, such as social distancing, sanitation, mental health and awareness of financial scams. Since these CSR updates rarely suggest actual corporate actions or donations, but rather an accommodative attitude, we term this theme as “external stakeholder accommodative CSR.”

Table 1.

Topics and associated top key words and top posts.

Themes	Topics	Top keywords and sample posts with highest probability
Community Information Update (external stakeholder accommodative CSR)	Topic 6: Sanitation and social distancing	Keywords: help, time, make, work, get, way, home, stay, need, people, keep, import, together, differ, like, find, know. Sample post: Celanese Corporation: While we are experiencing the effects of COVID-19 and many of us are working from home, here are a few helpful tips on how to be effective even when you're not in the office. 1. Maintain regular work hours. Set your alarm and get up at your regular time. 2. Create a morning routine. Take a shower, make your coffee and get dressed! Staying in pjs all day can make you feel sluggish and less productive! 3. Try to set boundaries with people in your space. Let your kids/spouse know that when you are at your computer, you are working and cannot be disturbed.
	Topic 9: Technology in healthcare	Keywords: health, care, patient, healthcare, hospital, provide, pandemic, help, center, system, learn, need, cerner, medical, access, support, read, community, use Sample post: Cerner Corporation: Using a collection of de-identified, HIPAA-compliant health data, leaders at OSU Center For Health Systems Innovation created two models of potential mortality risk in an effort to diagnose, stage and monitor disease progression and save lives. Read how they're analyzing anonymous patient data to improve care.
	Topic 10: Personal financial scam	Keywords: finance, call, bank, online, payment, use, video, due, tip, account, question, protect, credit, learn, pay, take Sample post: TransUnion: Our Financial Hardship Studies have found COVID-related digital fraud is on the rise. If you think you are a victim of fraud, it may be a good idea to add a fraud alert to your credit report. A fraud alert tells lenders to take extra steps to confirm your identity before extending credit.
	Topic 16: COVID testing and vaccine	Keywords: vaccine, test, visit, learn, inform, health, medical, please, member, cvs, call, avail, office, today, clinic, symptom, distribute, flu Sample post: CVS Health: No patient will be charged for COVID-19 vaccines once available, which will be provided by the federal government. For the vast majority of individuals, administration fees will be covered by their insurer or a program for the uninsured. Hear more from Dr. Sree Chaguturu, Chief Medical Officer at CVS Caremark, on what to expect from the COVID-19 vaccine experience.
	Topic 17: COVID research	Keywords: test, learn, use, lab, develop, research, author, new, solution, fight, kit, detect, antibody, diagnostic, infect, read, emerge, virus Sample post: PerkinElmer: The U.S. Food and Drug Administration (FDA) has provided Emergency Use Authorization (EUA) for Euroimmun USs Anti-SARS-CoV-2 ELISA (IgG) serology test. Clinical laboratories certified to perform high complexity tests under Clinical Laboratory Improvement Amendments (CLIA) can immediately begin using this ELISA for the detection of antibodies of the immunoglobulin class G.
	Topic 20: Mental health during pandemic	Keywords: health, share, pandem, feel, read, stress, may, mental, children, resource, social, distance, tip, anxiety, live, care, physical Sample post: Magellan Health: Anyone who has gone through a traumatic event can develop symptoms of stress disorders. Although the COVID-19 Pandemic in general wouldn't meet the criteria for causing a stress disorder, many of the same issues can result from such a widespread and overwhelming event.
Organizational Crisis Response (internal stakeholder proactive CSR)	Topic 3: Future of work	Keywords: world, change, year, new, company, challenge, crisis, pandemic, global, look, future, sustain, people, woman, read, lead, today, like Sample post: Philip Morris: Now is the time to reset, reposition, and rethink what we want our future to look like and rediscover our ability as humans to work together for better. Our COO Jacek Olczak wrote this editorial for the Diplomatic Courier on the challenges of 2020, COVID-19 and the UNs Sustainable Development Goals during this years UNGA.
	Topic 5: Data security for remote work	Keywords: remote, secure, learn, work, data, organization, solution, business, communicate, new, use, crisis, manage, technology, enable, plan, quick Sample post: Unisys Corp: Given the findings from the 2020 Unisys Security Index, it is evident that securing personal and corporate data and systems from malicious attacks is the need of the hour. What can individuals and organizations do to achieve this? Unisys Leaders Sumed Marwaha and Seshadri P.S. share recommendations on.
	Topic 8: Manage remote work	Keywords: post, digital, business, virtual, invest, strategy, work, new, world, technology, say, space, transform, experience, ready, learn Sample post: Cognizant: orgs can address post- new normal by integrating virtual channels into legacy business models.
	Topic 13: Assistance to business	Keywords: business, small, help, learn, resource, pandemic, new, impact, finance, need, owner, program, employee, client, relief, insure, loan, inform, navigate Sample post: Paychex: The Paycheck Protection Program (PPP) went into effect under the CARES Act on April 3, providing relief for small businesses impacted by COVID-19. To learn more about the program and get detailed instructions on how to apply, visit our dedicated PPP page here:
	Topic 14: Market impact	Keywords: market, impact, expert, latest, webinar, insight, discuss, report, coronavirus, register, global, episode, economy, join, watch, share Sample post: CME Group: Equity volatility has returned amid concerns over a resurgence in COVID-19 cases. Dan Gramza discusses bullish and bearish scenarios for the September E-mini Dow futures contract.
	Topic 15: Leadership view on industry impact	Keywords: ceo, discuss, president, pandemic, read, office, blog, join, industry, share, recent, chief, hear, global, director, execute, chairman, impact, article Sample post: Echo Global Logistics: In an interview with Logistics Management, Echos CEO Doug Waggoner discusses the impact of COVID-19 on the freight industry, peak season prospects, and more.
Organizational Contribution for Society (external stakeholder proactive CSR)	Topic 1: Donation and sponsorship	Keywords: community, support, donate, food, organization, foundation, fund, local, help, relief, provide, need, partner, million, effort, response, family, world, proud Sample post: Gucci: Gucci donated 1 million to the United Nations Foundations COVID-19 Solidarity Response Fund in support of the World Health Organization though Facebooks US $10 million Matching Fundraiser. Help us reach the $10 million target by donating now.
	Topic 2: Enhance manufacturing and production	Keywords: supply, product, mask, medical, help, face, use, manufacture, hand, protect, equipment, chain, hospital, produce, air, sanitation, need, fight, critic Sample post: Cummins Inc: To help respond to the pandemic we've announced additional efforts to help protect the needs of health care professionals. In collaboration with 3 M, Cummins will manufacture high efficiency particulate filters for use in Powered Air Purifying Respirators (PAPRs). The 3 M designed filters will be made in Cummins Neillsville, Wisconsin, plant.
	Topic 4: Support young people and education	Keywords: learn, student, school, education, new, virtual, year, support, event, program, hit, impact, due, affect, nasdaq, job, initial Sample post: Regions Bank: When our students and communities succeed, so do we. Even in the midst of the unknown, Regions partnered with Jobs for America's Graduates (JAG) to ensure education and job training for at-risk youth and young adults.
	Topic 7: Transportation industry’s response	Keywords: travel, flight, mask, fly, wear, test, hand, guest, wash, free, new, family, visit, airline, cover, well, offer Sample post: Frontier airlines: Frontier and Miracle Flights Caring for Some Extra Special Individuals! Miracle Flights, headquartered in Las Vegas, is the nations leading medical flight charity and typically arranges more than 600 flights a month to medical facilities across the country. Meet Levi, a Miracle Flight recipient who has traveled with us:
	Topic 11: Express gratitude to frontline workers	Keywords: team, work, thank, employee, proud, pandemic, member, deliver, around, support, colleague, essential, keep, across, dedication, service, continue, hero Sample post: UGI Utilities: Who would you recognize as being a Hero in the fight against COVID-19? Heroes do not necessarily wear capes and fly around the skies. Today we'd like to recognize delivery and postal service workers - all of these every day heroes that have been bringing essential items to our doorsteps - from food to PPE to mail. THANK YOU! Want to recognize your delivery or postal service worker? Nominate them by entering our Heroes of Hope Sweepstakes! Just visit and complete our short entry form before August 25, 2020. We’ll be randomly selecting three heroes to each win a $500 Visa Gift Card. And, if your hero is randomly selected as a winner, you will win a $100 Amazon eGift card!
	Topic 12: Energy industry’s response	Keywords: custom, energy, ugi, service, utility, project, electricity, power, gas, pandemic, program, bill, company, assist, nature, time, construct Sample post: Dominion Energy: In addition to suspending service disconnections for customers unable to pay their bills, we will be reconnecting residential customers who were previously disconnected for nonpayment. Please give us a call during business hours to make arrangements for service restoration.
	Topic 18: Express gratitude to health workers	Keywords: worker, healthcare, frontline, thank, line, family, day, front, first, nurse, fight, hospital, hero, story, live, today, show, care Sample post: Dunkin': While our hometown would've been today, we're still raising a cup, and cheering all the brave Bostonians on in this new marathon. You can cheer on the Boston hero in your life, too. Send a small thanks with an e-gift card, at Dunkincoffeebreak. Well donate $1 for every e-gift card purchased to our Joy in Childhood emergency COVID-19 relief funds.
	Topic 19: Commitment to customers	Keywords: customer, continue, employee, safety, safe, health, ensure, community, keep, take, committee, spread, protect, update, priority, service, inform, measure Sample post: Madewell: Madewell’s top priority is the health and safety of our associates and our customers. Learn what measures we are taking to safeguard our community from the spread of COVID-19 here.

The organizational crisis response theme is mainly about corporation management during this crisis. It involves market and industry impact, business assistance for organizational survival through hard times, and effective management of remote work and virtual teams. Discussions in this theme mainly focus on internal stakeholders such as employees and investors and how companies take actions to solve their own problems, and we thus term this theme as “internal stakeholder proactive CSR.”

The organizational contribution theme is about corporations’ actual actions to serve the broader society. It includes donations and community relief actions, and company actions to protect customers. It also includes actions specific to different industries, such as increasing production in the manufacturing sector and delivery by the transportation industry. Discussions in this theme mainly focus on external stakeholders and how companies take actions to help communities in need, and we thus term this theme as “external stakeholder proactive CSR.”

H1 hypothesizes that the corporate CSR theme of proactive corporate actions would receive more comments and shares. As shown in Table 3, compared to community information update (external stakeholder accommodative CSR), the organizational contributions theme (external stakeholder proactive CSR) associated with higher likelihood of receiving comments, B =.10, SE =.03, p <.01. It is also positively associated with receiving more shares, B =.10, SE =.04, p <.05. However, the organizational crisis response theme (internal stakeholder proactive CSR) associated with a lower chance of receiving comments and shares. H1 is partially supported.

Table 3.

Mixed effects models predicting Comments and Shares.

	Comments		Comments		Shares		Shares
	B	SE	B	SE	B	SE	B	SE
Fixed effects
(Intercept)	-3.71***	.35	-3.78***	.38	-.63	.34	-.55	.38
Crisis response	-.13***	.04	-.13	.19	-.34***	.05	-.75**	.23
Contribution	.10**	.03	.21	.17	.10*	.04	.09	.21
Manufacturing	-.20	.25	-.10	.29	-.29	.24	-.31	.31
Transportation	.94***	.27	1.12***	.33	-.11	.26	-.18	.34
Retail	.79**	.25	.84**	.31	.15	.25	.13	.33
Finance	.40	.24	.47	.29	-.43	.23	-.51	.30
Service	-.17	.23	-.14	.28	-.68**	.22	-.86**	.29
Crisis response x Manufacturing			-.14	.21			.17	.25
Contribution x Manufacturing			-.09	.19			.00	.23
Crisis response x Transportation			-.11	.23			.16	.28
Contribution x Transportation			-.22	.21			.04	.25
Crisis response x Retail			-.19	.23			.23	.28
Contribution x Retail			-.05	.21			-.04	.25
Crisis response x Finance			.08	.20			.57*	.24
Contribution x Finance			-.15	.19			-.12	.23
Crisis response x Service			.06	.20			.56*	.24
Contribution x Service			-.06	.19			.11	.23
Followers	.45***	.03	.45***	.03	.28***	.03	.28***	.03
Corporate revenue	.67***	.18	.66***	.18	.38*	.18	.38*	.18
Random effects
Corporation SD	1.26		1.26		1.17		1.17
Observations	9514		9514		9514		9514
Corporations	438		438		438		438
AIC	28161		28194		31787		31792
BIC	28247		28351		31873		31949
Loglikelihood	-14069		-14075		-15881		-15874

Note: 1. CSR themes are compared to “community information update”. 2. Industries are compared to the “other” category. 3.

^*p <.05.

^**p <.01.

^***p <.001.

H2a hypothesizes that the CSR theme of proactive actions would receive more positive emotional responses. As shown in Table 4, compared to community information update (external stakeholder accommodative CSR), organizational contributions (external stakeholder proactive CSR) associated with a higher likelihood of receiving likes (B =.37, SE =.03, p <.001) and loves (B =.57, SE =.04, p <.001). However, the organizational crisis response theme (internal stakeholder proactive CSR) associated with a lower chance of receiving likes and love responses. H2 is partially supported.

Table 4.

Mixed effects models predicting Likes and love responses.

	Likes		Likes		Love		Love
	B	SE	B	SE	B	SE	B	SE
Fixed effects
(Intercept)	.27	.29	.35	.32	-2.41***	.31	-2.22***	.35
Crisis response	-.12***	.03	-.23	.17	-.16***	.04	-.50*	.19
Contribution	.37***	.03	.29	.16	.57***	.04	.41*	.18
Manufacturing	-.41*	.21	-.41	.25	-.48*	.22	-.49	.27
Transportation	-.29	.23	-.40	.28	-.04	.24	-.15	.30
Retail	-.08	.21	-.10	.27	.12	.23	-.10	.29
Finance	-.74***	.21	-.86***	.25	-.74***	.22	-1.07***	.27
Service	-.82***	.20	-.93***	.24	-.74***	.21	-.90***	.26
Crisis response x Manufacturing			.01	.19			.22	.21
Contribution x Manufacturing			-.02	.17			-.06	.19
Crisis response x Transportation			.13	.21			.25	.24
Contribution x Transportation			.12	.19			.09	.21
Crisis response x Retail			.00	.21			.12	.24
Contribution x Retail			.02	.19			.25	.21
Crisis response x Finance			.21	.18			.56**	.21
Contribution x Finance			.12	.17			.38*	.19
Crisis response x Service			.13	.18			.33	.20
Contribution x Service			.13	.17			.12	.19
Followers	.35***	.02	.35***	.02	.36***	.02	.36***	.02
Corporate revenue	.31*	.15	.31*	.15	.27	.16	.27	.16
Random effects
Corporation SD	1.05		1.06		1.10		1.10
Observations	9514		9514		9514		9514
Corporations	438		438		438		438
AIC	26151		26191		28592		28607
BIC	26236		26348		28678		28764
Loglikelihood	-13063		-13073		-14284		-14281

Note: 1. CSR themes are compared to “community information update”. 2. Industries are compared to the “other” category. 3.

^*p <.05.

^**p <.01.

^***p <.001.

H2b hypothesizes that the corporate CSR theme of proactive actions would receive fewer negative emotional responses. As shown in Table 5, compared to community information update (external stakeholder accommodative CSR), the organizational contribution theme (external stakeholder proactive CSR) had no effect on the chance of receiving sad responses, but it negatively associated with the likelihood of receiving anger responses (B = -.05, SE =.02, p <.05). H2b is partially supported.

Table 5.

Mixed effects models predicting Sad and Angry responses.

	Sad		Sad		Angry		Angry
	B	SE	B	SE	B	SE	B	SE
Fixed effects
(Intercept)	-.52***	.14	-.56***	.15	-1.09***	.16	-1.14***	.18
Crisis response	-.02	.02	.01	.08	-.03	.02	.00	.10
Contribution	-.01	.02	.05	.08	-.05**	.02	.02	.09
Manufacturing	-.17	.10	-.16	.12	-.07	.11	-.07	.14
Transportation	.00	.11	.21	.13	.43***	.13	.67***	.16
Retail	.24*	.10	.30*	.13	.18	.12	.23	.15
Finance	-.16	.09	-.15	.12	.18	.11	.25	.14
Service	-.12	.09	-.05	.11	.04	.11	.08	.14
Crisis response x Manufacturing			-.01	.09			-.05	.11
Contribution x Manufacturing			.00	.08			.03	.10
Crisis response x Transportation			-.18	.11			-.17	.12
Contribution x Transportation			-.25**	.09			-.29**	.11
Crisis response x Retail			-.15	.10			-.04	.12
Contribution x Retail			-.07	.09			-.05	.11
Crisis response x Finance			.04	.09			-.01	.10
Contribution x Finance			-.02	.08			-.11	.10
Crisis response x Service			-.04	.09			-.01	.10
Contribution x Service			-.09	.08			-.05	.10
Followers	.07***	.01	.07***	.01	.11***	.01	.11***	.01
Corporate revenue	-.03	.07	-.04	.07	.29***	.08	.29***	.08
Random effects
Corporation SD	.48		.46		.58		.58
Observations	9514		9514		9514		9514
Corporations	438		438		438		438
AIC	12923		12936		15465		15493
BIC	13009		13058		15551		15650
Loglikelihood	-6450		-6451		-7721		-7724

Note: 1. CSR themes are compared to “community information update”. 2. Industries are compared to the “other” category. 3.

^*p <.05.

^**p <.01.

^***p <.001.

H3 proposes that corporations from industries that are directly affected by COVID-19 are more likely to receive higher levels of engagement in terms of comments and shares. Research on the financial impact of COVID-19 of different industries shows that industries including energy, tourism, retail, transportation industries were most affected during the initial period of the pandemic (Baek et al., 2020). Table 3 shows the transportation and retail industries received more comments, and service industries received fewer shares. Since the transportation and retail industries are among the worst hit industries during COVID (Baek et al., 2020), H3 is partially supported.

H4a proposes that CSR messages from industries that are directly affected by COVID-19 would receive more emotional responses. Table 4 shows that manufacturing, finance and service industries were less likely to receive “like” and “love” responses compared to the baseline category. Since the baseline category “others” mainly included the wholesale industry, which was also severely affected, H4a received partial support. Table 5 shows that the retail industry was more likely to receive sad responses, and the transportation industry was more likely to receive angry responses. H4b is partially supported.

Taken together, the findings show that severely affected industries were more likely to receive behavioral engagement and emotionally intense responses. For example, Southwest Airline shared a post: “This photo was taken earlier this week by Robert, one of our Flight Attendants. There was one Customer onboard the flight, a pulmonary critical care doctor flying to the west coast to volunteer with coronavirus relief. We’re flying fewer passengers than normal these days, but we will continue to fly. Our CEO and Chairman Gary Kelly put it this way: We need to be there for those who still have to travel for essential work. To all of our healthcare workers, thank you.” This post was shared 4530 times, generated 2920 comments, and received 37,518 likes and 789 sad responses.

RQ2 inquiries into the moderation effects of industry type on publics’ responses to their discourse topics. In Table 3, no interaction effect was observed for comments as the dependent variable. Service and finance industries positively moderated the negative effect of organizational crisis response (internal stakeholder proactive CSR) and shares. In other words, although the service and finance industries had a tendency to receive fewer shares, communicating organizational crisis responses might help those engage social media users. In Table 4, no interaction effect was observed for “likes” as the dependent variable. The finance industry was less likely to receive love responses. It positively moderated the effects of both crisis response (internal stakeholder proactive CSR) and organizational contribution (external stakeholder proactive CSR) on love responses. In Table 5, the transportation industry negatively moderated the negative effect of organizational contribution (external stakeholder proactive CSR) on angry responses.

For the control variables, as expected, the number of followers consistently correlated with all engagement types. Corporate revenue was a stronger predictor for comments and angry responses than others. Consistent with the previous argument on the relationship between companies’ resourcefulness and stakeholder engagement, and because commenting is a type of high-level engagement (Cho et al., 2014), stakeholders are more likely to engage with more resourceful companies and when they fail stakeholder expectations, they are more likely to get angry responses.

5. Discussion

Our study examined how companies should discuss their CSR activities during the COVID-19 pandemic and identified which CSR strategies are most effective in terms of stakeholder engagement outcomes. Our analysis reveals a number of important findings that we discuss below.

5.1. Which CSR topics to talk about during a global crisis?

This examination of U.S. Fortune 500 companies’ COVID-related CSR communication during the first fifteen months of the COVID-19 pandemic extends previous research that suggests that CSR topics matter a great deal in eliciting public engagement and shape public perceptions about corporations (Cho et al., 2017, Saxton et al., 2019). In the current study, instead of using predetermined CSR topics as specified in previous studies, we adopted an inductive approach to let CSR topics emerge from the corporate discourse. This is because the COVID-19 pandemic is a novel crisis with large-scale societal risks. We argue that CSR communication, which has been viewed as one tool for managing individual corporate risk, should be reconceptualized as a set of tools for collectives of companies to manage social risk (Crane & Matten, 2020). During a social crisis, companies are expected to play a responsible role in addressing the crisis. This idea suggests that companies in the same industry or related industries recognize shared social risks and stakeholders’ expectations (e.g., COVID-19, climate change), and develop a range of CSR response strategies to address emerging social problems and changing stakeholder concerns. As shown in our findings, although certain CSR strategies in general were not perceived positively by stakeholders, certain industries may benefit more from such strategies than others, as different industries are expected to play different roles in addressing the social crisis.

The novel nature of this crisis requires an inductive research method that explores and reflects new territories of research. Our computational discourse analysis contributes to such an updated conceptualization of CSR. Specifically, we found that three CSR thematic clusters emerged from the corporate discourse on COVID-19: community information update (external stakeholder accommodative CSR), organizational crisis response (internal stakeholder proactive CSR), and organizational contribution to the society (external stakeholder proactive CSR). It is worthwhile to pay special attention to the specific CSR topics within the thematic cluster of external stakeholder proactive CSR. Apart from external CSR topics that have been extensively studied, e.g., donations, sponsorships, education, commitment to customers, there are also topics that directly reflect the crisis scope and changes in the compositions of primary stakeholders. The additional efforts of businesses to enhance the manufacturing of essential goods and services should be considered as showing responsibilities to the society, because the crisis has highlighted corporate capability and willingness for modified or increased production to address societal needs (Crane & Matten, 2020). These topics are rarely a routine component of CSR communication, but they became important during the COVID-19 pandemic. Topics about how companies assist frontline industries in healthcare and transportation are also proactive strategies because these industries or sectors are key to keeping the society functioning during the crisis. Increasingly, many of these companies’ primary stakeholders are expecting them to contribute to society during COVID-19 (Whitaker, 2020).

5.2. CSR themes and public engagement

Among the three CSR themes, we found that proactive strategies targeted at external stakeholders elicited more behavioral public engagement and positive emotion than accommodative strategies, whereas proactive strategies targeted towards internal stakeholders led to lower levels of engagement outcomes. It is likely that as many external stakeholders are in considerable hardship due to the pandemic, they expect help and respond well to companies that make substantial efforts to contribute to the community. These results confirmed our hypothesis that a CSR response strategy that exceeds stakeholder expectations would receive more public attention and that those organizations’ CSR would be more likely to be positively evaluated. This finding is consistent with findings on disclosure of CSR, where corporations could gain legitimacy by communicating extensive CSR efforts that exceed normative standards (Bachmann & Ingenhoff, 2016).

In comparison, organizational crisis responses targeted at internal stakeholders received less behavioral engagement and fewer positive public responses. The lukewarm responses associated with internal stakeholder proactive CSR may partly be due to the fact that Facebook is a public-facing medium where external stakeholders probably far outnumber internal stakeholders. In addition, although businesses themselves have also encountered severe hardships during the pandemic, it seems stakeholders are not very interested in learning about the market impact and remote working policies of companies. This finding is inconsistent with previous research that CSR messages targeted at internal stakeholders (i.e., employees) could also increase stakeholders’ positive attitude towards a company (Wang and Huang, 2018, Whitaker, 2020). A possible explanation is a mismatch of what publics expect and what these messages provide in times of a great societal crisis. Although publics understand these corporations, like them, are also negatively affected, publics may still perceive the corporations as more resourceful than themselves and thus expect the corporations to care more about the communities around them (Whitaker, 2020).

It is interesting to note that our finding is consistent with other studies about how CSR influences corporation values during the pandemic. Qiu et al. (2021) found that community-related COVID-19 CSR messaging could significantly influence stock return more than employee-related CSR messaging. Together, our studies demonstrate that in times of a major societal crisis, external audiences as well as investors in the stock market would respond positively to CSR efforts that aim at the betterment of society rather than the betterment of individual corporations.

Overall, for public relations practitioners that manage CSR communication during and post pandemic, our study suggests that they should focus on the needs of external primary stakeholders, and proactively demonstrate how their companies are helping to address the public’s needs. For example, for a crisis such as the COVID-19 pandemic, its impact may be long lasting, especially on vulnerable communities. As the pandemic moves into new phases, practitioners need to carefully consider stakeholders’ changing needs and challenges and adjust accordingly. Our analysis shows that not all CSR strategies would be perceived positively. Only timely and relevant strategies are likely to engage publics and co create positive meanings about the organization’s response to COVID-19.

5.3. Industry influences in stakeholder expectation

We also found that industry type influenced public engagement outcomes because stakeholders’ expectations vary towards different industries. In general, we found that industries severely affected by the pandemic (e.g. retail, transportation, and wholesale) were more likely to receive behavioral engagement and emotional public responses from publics. It is likely that the public sees that these companies have lost millions due to the pandemic but are still taking actions to address social concerns. Such actions may have exceeded public expectations and thus lead to better engagement outcomes. Another possibility is that because these companies have endured considerable economic hardship and thus stayed in the public and media spotlight during the crisis because of their hardship. Their actions appear more likely to receive heightened attention due to already concentrated media and public attention.

Overall, our results revealed some interesting findings regarding the role of industry in terms of moderating the relationship between CSR themes and engagement outcomes. For the finance industry, they received fewer negative responses when they discussed their organizational crisis responses targeted at internal stakeholders. This is probably because market impact on the finance industry would be contagious to other industries. In addition, since COVID-19 has pushed many people into financial hardship, many publics may be dependent on the finance industry for insurance, loans, and debt relief. As the public received help from the industry, they might be less critical towards these companies. We also observed that the transportation industry was more likely to receive angry responses. This is likely due to the many travel restrictions and new rules (e.g., wearing a face covering during an entire flight trip, social distancing at the airport, cancelled or reduced options) that transport companies imposed during the pandemic have caused confusion and backlash. Only when transportation companies communicated now they had actively contributed to society (e.g., delivering essential medical devices) were there fewer angry responses. This again shows the disparate stakeholder expectations for different industries in the COVID-19 crisis. Public relations practitioners need to carefully consider these special conditions and modify their messages based on societal expectations for their industry.

5.4. Implications for public relations theory

Our findings also have implications for public relations theory. Co creational theory argues that publics are partners in the meaning making process (Botan & Taylor, 2004). Dialogic engagement argues that communication and action need to be interactive in that organizations respond to public expectations (Taylor & Kent, 2016). The social media data tell us that publics respond positively to organizational messages that show community-based CSR contributions. Of the three CSR stakeholder engagement strategies: internal stakeholder proactive CSR, external stakeholder proactive CSR, and external stakeholder accommodative CSR, Facebook publics engaged more with the external-focused messages. These findings add new dimensions to thinking about co creation theory in public relations. Social media responses provide organizations with insights into stakeholder values and priorities. Engagement responses are part of the dialogic process. Over the 15 months of the study, our research team watched as organizations communicated their contributions to society. Social media users “judged” the value of these messages through their engagement and sent both explicit and implicit co created messages. A future study should study the longitudinal changes in organizational response and public response to look for trends in the data that can further extend co creational theory.

Additionally, our findings suggest a more nuanced approach to CSR theory. CSR theory has viewed communication as one tool for managing individual corporate risk. But, the data suggest that it should be reconceptualized as a set of tools for collectives of companies to manage social risk. Publics have clear expectations about industries and CSR communication is really a set of strategies (proactive, reactive and accommodative) that speak to publics’ interests and priorities. Fortune 500 companies now need to listen and adapt to meet societal expectations. Consumers and scholars also need to be vigilant to recognize organizations that are acting irresponsibly during the pandemic as well.

5.5. Limitations and future research

This study does have several limitations. First, we only focus on the largest American corporations’ communication on Facebook. Although Facebook is one of the most popular social media platforms for organizations to engage stakeholders, future research could adopt a cross-platform perspective to compare CSR efforts across different arenas (e.g., Twitter, Instagram) and different corporate populations. A cross-platform, cross national comparative approach would be useful because the platform ecology and target audiences will be different. Future research could compare how different CSR themes emerge on different social media platforms and how they lead to different public engagement outcomes. Future research could also delve into the ways in which organizations operating out of other nations coordinate their CSR efforts across platforms and if such concerted strategies are more effective.

Second, since our study mainly focuses on observational data, we did not collect stakeholders’ self-reports. In other words, we did not assess stakeholders’ expectations. Future studies may examine if stakeholders with different demographics or stakeholders who are affected by COVID-19 pandemic in different ways may have varying expectations for corporations.

Third, to our best knowledge, this study is one of the first to examine corporate response strategies during this crisis. We used an inductive approach to identify discourse topics. However, due to the lack of previous research, we do not know if these topics may persist after the crisis or resurface in another major crisis. Future research could continue to explore how corporations conduct CSR practices and communicate CSR in a major social crisis and what strategies are most effective. Future research could further examine how corporations manage social risks and prioritize different stakeholders given the changing social needs in a major crisis.

In conclusion, our findings demonstrate that corporations attempted to engage stakeholders on social media. Their CSR efforts actively supported their employees, consumers, and communities during the pandemic. COVID-19 was a time for corporations to show their local community and the world that they are valued corporate citizens and community members. Our analysis found that different strategies evoked different emotional and behavioral engagement responses. As we move toward more public centered communication, this study provides evidence that organizations that demonstrate they are responsible members of society will be viewed favorably. Future research should look at long term outcomes of COVID-19 engagement to see if these CSR efforts have transformed these companies’ stakeholder engagement approaches for the long-run.

Conflict of interest

There is no conflict of interest.

Data availability

The data that has been used is confidential.