Introduction: The Emergence of Stakeholder Media

Prior to the emergence of Internet and social media (including user forums, discussion groups and interactive websites as well as Facebook or Twitter) as global phenomena, stakeholder claims on organisations and other stakeholders were “channelled largely through the mass media” (Koopmans 2004, p. 371). Activist stakeholders were thus obliged to take account of the roles played by mainstream media (MSM) in setting public agendas (McCombs 2004, 2005). In particular, they were well advised not to appear “threatening, dangerous” (Mika 2006, p. 921) to MSM “gatekeepers” who decided what was fit to print (McCombs and Funk 2011; Shoemaker and Vos 2009).

Within only a few years, online media have begun to profoundly change this game. In an era of multiplying media channels and lowered costs in creating and diffusing content, it appears plausible, if not certain, that “traditional media’s singular, one-way power over news creation and dissemination is a past phenomenon” (Meraz 2011, p. 108). A decline of capacity at MSM, in particular for watchdog journalism (Lewis et al. 2005), appears as both a symptom and driver of this shift. Stakeholders disappointed by MSM representations of events can simply create their own channels. These new channels fit the description of “horizontal media”, which are “geared toward serving the needs of more specialized interest groups” than those served by the “vertical” MSM (Ragas and Roberts 2009, p. 45). They also resemble “media of connection” that offer members of self-interested communities a “particular array of values—an agenda—upon which those who join a group can match their own values” (Shaw et al. 1999, p. 7). However we call them, these media involve focused and purposeful content for self-selected users, allowing like-minded people to share opinions, information and objectives. We will refer to them as “stakeholder media” (Hunter et al. 2008), defined as communication instruments used by stakeholder communities to increase their resources and influence (Frooman 1999) vis-à-vis organisations and other stakeholders. We define a stakeholder community as a group “who can affect (or be affected by)” the activities of a larger entity such as a firm (Freeman et al. 2004, p. 365), and whose members are “organised around agendas” (Shaw et al. 1999, p. 7) that can affect or be affected by that entity. In other words, a stakeholder community exists because it has an agenda, and its media exist to promote that agenda, whether it involves activism, fuller enjoyment of a product or another common objective.

Agenda-setting theory seeks to describe how media impact our view of what matters, and how it matters, and ultimately, what to do about it. Our first conceptual contribution in this article is to broaden agenda-setting theory to account for the effects of stakeholder-controlled media, particularly in today’s complex online networks, on management agendas. We also contribute to stakeholder theory by providing a more detailed account of how contemporary “stakeholder agendas” (Ragas 2012, 2010) may exert influence on management. A third contribution resides in our use of system dynamics (SD) methodology to map the salience of particular issues and objectives in the stakeholder networks of three multinational firms (MNCs) involved in significant crises. We seek thus to capture the nonlinear complexity of stakeholders, their media and their effects. Though SD has been applied to analysis of the onset of corporate crises (Tsuchiya 2003), to the best of our knowledge, our article is the first effort to apply SD to agenda-setting processes. We hope to illustrate the relevance of SD to agenda-setting, within the context of stakeholder activism. Taken together, these contributions might greatly improve the ability of organisational leaders and other stakeholders to estimate which strategic initiatives might succeed or fail, and at what cost. Their understanding of which media may best be used to build their agendas, and how, might also improve. Finally, managers may gain a different perspective on how, amidst the “constant alarm and alert” of the contemporary stakeholder environment (Holmström 2005, p. 503), they can negotiate responsibility for and solutions to conflicts.

In the following passage, we review agenda-setting theory, with particular emphasis on recent developments that expand its scope to include stakeholder groups and online media. Further on, after setting out research questions and hypotheses, we describe our application of SD methodology. Next, we briefly review key events in our three contemporaneous cases of corporate crisis. Finally, we construct a preliminary model of stakeholder responses, transmitted primarily through stakeholder-controlled media, to the agendas of management and other stakeholders. In our discussion, we identify implications for agenda-setting theory, directions for further research, and practical implications for organisational leaders and stakeholders.

Literature Review: From MSM Agendas to Stakeholder Agendas

The central postulate of agenda-setting theory is that issues which are rendered “salient” through media coverage will also be considered important by the public and decision makers (McCombs and Shaw 1972). At the first level of agenda-setting, media influence the public perception of what issues matter (Carroll and McCombs 2003; McCombs 2004, 2005). A second level of agenda-setting occurs when content creators assign attributes to these objects (Ghanem 1997). The process of working through the media or a third party to reach a given public has been called “agenda-building” (Berkowitz and Adams 1990; Curtin 1999); the merger of an individual’s agenda with a group’s agenda is referred to as “agenda melding” (Shaw et al. 1999). Across these various levels of effect and process, a central path remains: “the transfer of salience from one agenda to another” (Ragas and Roberts 2009, p. 46). In essence, someone decides what matters, and what to think about it, and seeks to engage others to do something about it. Media are involved at every step of these processes, in collecting, shaping and diffusing information and ideas.

As Strömbäck and Kiousis (2010, p. 271) observe, “most of the agenda-setting research to date has focused on the correlation between issues on the [mainstream] media agenda and issues on the public agenda”. In the past decade, the focus widened to include how MSM can serve to promote or undercut corporate agendas, building on classic research into social movements that used “showmanship and display”, transmitted through MSM to influential “reference publics”, to drive change (Lipsky 1968, pp. 1145–1146). MSM reports of stakeholder protests like boycotts can impact a firm’s stock price (King 2008), or in a general sense, its reputation, defined as its perceived “capability to obtain valuable outcomes for different stakeholders” (de Castro et al. 2006, p. 362). Conversely, just as activists can use MSM to promote their causes at a firm’s expense, management press releases can influence both the choice and the tone of subjects in major newspapers (Carroll and McCombs 2003), and serve to bolster firm reputations (Kiousis et al. 2007). Here, stakeholder-controlled media appear mainly as instruments of influence on MSM, which retain power over public agendas. Similarly, Carroll (2010) found that firms’ public relations releases have zero or negative agenda-setting effects if not legitimated by the “third-party endorsement” of MSM. The salience of issues in MSM may also be driven by interest groups for whom those issues are of greater concern than for news media professionals or for a general audience (Uscinski 2009). These groups include “stakewatchers” (Fassin 2009), who monitor organisations and diffuse their findings, as well as “infomediaries”, notably corporate stakeholders, who transmit information to news media in support of their own agendas (Deephouse and Heugens 2009).

Less attention has been given to whether and how media controlled by stakeholders can drive public agendas without the support of MSM. That is largely because prior to the online era and the advent of social media, many stakeholders lacked the means to widely deploy their own media; thus without the MSM, they could not reach their potential supporters. A second issue is the difficulty of investigating the influence and effects of individually insignificant, but networked channels “in the new, interactive media age” (Meraz 2009, p. 683). There is nonetheless evidence that stakeholder-controlled media can play a significant role in setting agendas at different levels and for different publics. Prominent stakeholder organisations like Human Rights Watch function as investigative news agencies that define agendas for large activist communities, using both traditional and new media channels (Bogert 2010). They may use, but also be used by, MSM: Political blogs influence public opinion on certain issues as much as MSM (Woodly 2008), and rather than MSM setting the agenda for blogs, the reverse may also occur (Zhou and Moy 2007). Political candidates can “permeate multiple channels” outside MSM with their agendas (Tedesco 2001, p. 2048), and the effect on other candidates’ agendas can be stronger than the MSM version of what matters (Tedesco 2005a). Moreover, candidates who release the most material to MSM may have greater influence on the latter’s agendas than their rivals (Tedesco 2005b). Like political movements, other stakeholder groups may provide MSM and their own communities with press releases, direct mail, testimony, reports or online media content. These “information subsidies” (Gandy 1982) reduce content and search costs for MSM (Berkowitz and Adams 1990; Turk 1985) as well as other stakeholders (Ragas 2010). Other stakeholder media influence consumers, notably through brand communities, in which stakeholders critique and promote the products they use (Hunter and Soberman 2010; Ragas and Roberts 2009).

At the juncture of agenda-setting, social movement and stakeholder theories, a research stream is taking shape around the effects of stakeholder agendas, transmitted through stakeholder-controlled media, on other stakeholder groups. Even before Internet attracted a wide public, certain social movements and stakeholder groups used their own media to determine hierarchies of importance (Dearing and Rogers 1996)—that is, to decide what mattered and what to do about it—whether or not MSM gatekeepers were involved. Sine and Lee (2009, p. 128) showed that in the 1970s environmentalist organisations, “by constructing and propagating the ‘problem’ of environmental degradation [and] the ‘solution’ of renewable energy” through media such as newsletters, testimony, articles and public meetings, supported the emergence of wind power entrepreneurs. Similar cheap but sophisticated media helped build the climate change and anti-GMO movements decades before the MSM noticed them (Moser 2007; Schurman and Munro 2006). Stakeholder campaign letters and news releases have been shown to be highly effective in promoting their agendas to shareholders in corporate proxy contests (Ragas 2010, 2012).

Stakeholder media appear as a structural feature of recent agenda-setting models, albeit under different names. In particular, Berger (2001, p. 114) evokes the use of “elite media”, which he distinguishes from “consumer media”, in conflicts over corporate agendas. By exposing conflicts, objectives and interests, they “provide a levelling influence on the relative power relationships of contending [stakeholders]”. Similarly, Frooman (1999) and Hendry (2005) found that “communication strategies” aimed at potential allies by environmentalist NGO stakeholders were key in pressuring managements to respond to their claims. Though these authors do not specify the media employed by stakeholders, environmentalist NGOs were innovators in the development of both traditional and online media outreach among stakeholder groups. (This is particularly true of Greenpeace; see Bakir 2006). Meraz (2009, p. 701) shows that blogs created by “citizen” public affairs journalists gain influence over time through an indirect “aggregate effect” in which the circulation of opinion gathers increasing force. It thus appears that stakeholder-controlled media can generate direct and indirect effects on other stakeholders over time. The question here is if and how this might occur in the context of conflict between management and stakeholders of a firm.

Research Questions and Methodology

Research Questions and Hypotheses

The basic research questions of this article are: How do stakeholders use their media to set and build their agendas in conflicts with management (RQ1)? How does management react to these strategies (RQ2)? And what effects are generated by stakeholder-controlled media along the way (RQ3)?

The first-level salience of a given issue can theoretically be reinforced by either opposition or approbation for stakeholders and their agendas, because “even the rejection of a demand has to reproduce that demand and thereby diffuses it further in the public sphere” (Koopmans 2004, p. 374). Further, “private issues may become more public when they become salient on the media agenda or when other actors become involved in the conflict” (Berger 2001, p. 99). The importance of media attention for adversarial stakeholders is defined by Lipsky (1968, p. 1151) unambiguously: “If protest tactics are not considered significant by the media…protest organisations will not succeed. Like a tree falling unheard in the forest, there is no protest unless protest is perceived and projected”. Stakeholder-controlled media, too, can make their owners and their moral claims more visible to other stakeholders and the public (Hunter et al. 2008). Thus, their claims become more “legitimate” and “urgent” for organisational leaders to address (Mitchell et al. 1997; Neville et al. 2011). Simultaneously, activist stakeholders typically seek to implicate other stakeholders in conflicts with management (King and Pearce 2010). For all these reasons, if stakeholders successfully use their own media to make their claims more salient, the result may be a momentum effect, in which management finds itself under scrutiny and pressure from a constantly growing number of stakeholder groups. This effect will be even stronger if MSM report the conflict; the additional media attention increases the power of adversarial stakeholders and attracts more stakeholders to engage the issue (King 2011). RQ1 asks how stakeholders use their media to set agendas. We derive the following hypotheses:

H1a

Stakeholders who focus on an issue in their own media make the issue more salient for other stakeholders and management than it would be in the absence of that attention.

H1b

As an issue becomes increasingly salient, the more other stakeholders are compelled to respond to it.

H1c

The more other stakeholders respond to the issue, the more they interact dynamically with each other, including through their media.

The difficulty for management of satisfying stakeholder demands has been observed to increase as the number and diversity of such claims increases (Livesey 2001; Marshall et al. 2007). Responses that satisfy one stakeholder group may offend another; compromises may satisfy no one (Newenham-Kahindi 2011; Reynolds et al. 2006). This is a key reason why “corporate political actors prefer… to limit the scope of issue conflict to policy agendas and policy subsystems where they maximize power relations”, notably by seeking to limit the number of stakeholders who take part in agenda setting (Berger 2001, p. 99). We therefore predict that if conflict with stakeholders continues, and a growing number of stakeholders are drawn in, management will continually be obliged to address new obstacles and competing claims. In consequence, realising management’s agenda will require growing resources to satisfy stakeholder demands. We thus predict diverse and growing costs for the firm as conflict widens. RQ2 asks how management will react to demands expressed in stakeholder media. We derive these hypotheses:

H2a

The more stakeholders interact dynamically with each other on an issue, including through their media, the greater the scope and scale of demands that will be posed to management.

H2b

The greater the scope and scale of stakeholder demands, the higher will be the costs for management of satisfying stakeholder demands.

Firms and their managers also suffer negative impacts on their reputation from crises, especially crises in which “the organisation is deemed responsible” (Coombs 2007, p. 166). These effects are particularly pronounced when firms previously enjoyed a good reputation with the public (Dean 2004; King 2011), which applies to all three firms in our study. Moreover, firms in crisis may experience losses in key parameters of value, such as sales (Luders 2006), market share (Klein et al. 2004) or stock price (Hunter et al. 2008; King and Soule 2007; Sohn et al. 2009). A crisis may also cause damage to intangible assets like reputation and legitimacy (Formbrun and Shanley 1990; Gamson et al. 1992; Hoffman and Ocasio 2001; King 2008). These effects appear to gain strength as MSM attention to a crisis augments (King 2011). RQ3 asks what effects will result from stakeholder activism expressed through their own media. We derive these hypotheses:

H3a

As stakeholder claims and their salience increases, and their dynamic interactions multiply, including through their media, there will be a corresponding negative impact on the reputations of management and the firm.

H3b

The more stakeholders are disappointed by management’s agenda, the more the firm’s value, which can be expressed through sales, market share or share price, will decline.

Data Collection and Evaluation

Our data are drawn from three cases concerning publicly owned firms from different industries (food, energy and apparel). The cases were selected, first, because they are documented in detail by the scholarly literature (we provide our sources in the “Summary of the Danone, BP and Nike Cases” section). A second criterion is that all three cases involve sustained crises, and hence overt and explicit conflict between management and stakeholders. A further criterion is that in all three cases at least one stakeholder group was successful in obstructing management’s stated agenda, and also in hurting the firm’s reputation, sales or profits. We also chose cases in which adversarial stakeholder groups visibly interacted dynamically with other stakeholders, in particular through their proprietary media, ranging from print reports in the 1990s to websites and forums in the past decade. Though we observed proprietary media that supported the firms, we focused on adversarial stakeholder media, precisely because we are interested by their ability to counter management agendas. Moreover, recent attempts by firms to use online social media most often amount to “a marketing practice to convey messages” rather than an effective means to promote interaction with and among stakeholders (Colleoni 2013, p. 228). In sum, we sought cases in which stakeholders used their media, among other means, to exert influence on management. (We recognise that in other cases, involving stakeholders using similar media, little or no effect might be observed; we will treat this issue in our discussion.)

Two of the cases, focused on Danone SA and Nike Inc., feature union activism and consumer boycotts; the driver of the third crisis, at BP PLC, consisted of environmentalist NGOs in alliance with individual employees.

Stakeholder and management claims expressed through various media varied significantly among these cases. We forgo content analysis of those claims, and so we did not code their contents. We do so partly to avoid “implying the tautology that those who won employed the most resonant framings” (Benford and Snow 2000, p. 626). More important, we “emphasise the constraining and facilitating role of structural contexts”—the interactions among stakeholders—rather than “put agency at the centre of analysis and emphasise the purposive mobilisation of material resources and symbolic frames as the driving force” (Koopmans 2004, p. 378). We seek to understand whether a common pattern of stakeholder interactions exists across these three cases, and whether it generates common strategies and effects, regardless of specific claims or frames. We do so by mapping these interactions using SD analysis, as we explain below.

System Dynamics Methodology

Understanding stakeholder influence requires “including a structural component in stakeholder analysis” (Frooman 1999, p. 192). In parallel, management dilemmas may be usefully viewed as “arising from within a system with interdependent elements, subsystems, and networks of relationships and patterns of interaction” (Werhane 2002, p. 33). One appropriate tool for comprehending such objects is SD, introduced by Forrester (1961) as a modelling and simulation methodology for management problems. His initial goal was to capture the interaction of different stakeholders in a system, at a time of increasing complexity in organisational objectives and outcomes. SD originates from systems theory, as a method for understanding the dynamic behaviour of complex systems. SD was subsequently used to obtain insights into problems of policy resistance and change management (Coyle 1979, 1996; Sterman 2000). The methodology focuses attention to the effects of a decision on the responses of different stakeholders within a system over the long term. A particular concern in SD is why “our best efforts to solve problems often make them worse” (Sterman 2001, p. 8). Whereas optimisation applies very well “whenever the problem to be solved is one of choosing the best from among a well-defined set of alternatives… and if the system to be optimised is relatively static and free of feedback” (Sterman 1991, p. 216), SD can be usefully applied in unstable, changing environments.

The goal of the method is to identify points within the system that can most effectively be influenced by decision makers. The methodology renders more visible delayed and indirect effects of decisions, as well as of stakeholder counter strategies or resistance. SD also helps us to avoid a known risk of agenda-setting and framing studies, namely “locking [the events and discourses under study] in place, as though they were not part of a larger conversation, serving particular interests, and undergoing changes over time” (Reese 2007, p. 149). The features of SD models include time lags, information acquisition and exchange among stakeholders, and complex linkages of action and influence that render outcomes nonlinear (and hence less predictable). These phenomena appear in various forms in all of our cases—dispersion of agendas, a growing number of stakeholders and actions, and changes over time in the scope and scale of effects. For example, when management sets an altered agenda for a firm, its implementation takes time. Within this time period, the firm’s various stakeholders each adopt a stance towards the agenda, and seek to act accordingly. The effect on the firm may not immediately be evident to management. There are time lags when decisions are prepared, made and communicated, when actions are implemented, and when impacts begin to appear.

Dynamic complexity also results from the many interactions of stakeholders with one another. In SD terms, these interactions are called feedbacks. As with time lags, their effects are nonlinear, because different stakeholders are affected differently (or not at all) by specific feedbacks at different times. (For example, an investor in a firm may neither share nor act upon the outrage of a downsized worker. But if the worker’s outrage leads to a strike, and the strike impacts the firm’s stock price, the investor may well respond.) They also respond to those feedbacks in different ways, at different times.

SD seeks to capture and model feedbacks among the different components of organisational systems. As a first step, the model’s objectives (or else the research goal) must be defined—in this article, to understand the cumulative effects of stakeholder responses to change in a management agenda. Next the interactions among stakeholders, and the consequences of their diverse activities for each other and the firm, are mapped. Then, causal-loop diagrams are abstracted from these events. Causal-loop diagrams play two important roles. First, during model development they serve as preliminary sketches of causal hypotheses. Second, they describe the major feedback mechanisms of the system (Coyle 1996; Sterman 2000). In sum, the diagrams are used to construct a conceptual model of the system and the interaction of its components.

Feedback loops are components of causal-loop diagrams, and key mechanisms in the modelling process. A feedback loop is a succession of causes and effects, and it is captured by a circular arrow. A particular variable (say, a plan to increase profits) is affected by a stakeholder or event. The changed variable travels further through the system, encountering other stakeholders or events, and comes back to affect its own initial state. (For example, a profit-making plan is opposed by investors, with negative impact on the stock price, and management revises the plan.)

In causal-loop diagrams, influence lines represent the relations among variables (such as a firm’s agenda, profits and image). The direction and sign, (+) or (−), at the end of each influence line explain the effect of the variables on each other. This effect can be balancing (−) or reinforcing (+). In a reinforcing effect, the variable at the beginning of the influence line and the variable at the end of the influence line change in the same direction. Thus, if the beginning variable increases (for example, profits grow), the ending variable also increases; if the beginning variable decreases (profits fall), the variable at the end of the line also decreases. When the effect is balancing, the variables at the start and end of the line change in opposite directions; if one increases the other decreases, and vice versa.

The same logic and language apply to feedback loops, which contain more than one influence line. A loop’s overall polarity is obtained by the algebraic product of the balancing (−) or reinforcing (+) effects that it contains. In a balancing loop (symbolised by “B”), the initial position of a variable is reversed; growth becomes loss, or vice versa. Put another way, after a disturbance, the system seeks to return to equilibrium. Conversely, in a reinforcing feedback loop (symbolised by “R”), initial effects are strengthened; growth becomes more growth, losses lead to more losses. Balancing feedback loops can neutralise such effects: For example, workers go on strike, and management offers them indemnities, which end the strike and stabilise the system.

In this article, we build causal-loop diagrams based on events from three cases, involving the publicly owned multinational firms Danone SA, BP PLC and Nike Inc., in which stakeholders advanced particular agendas. Our goal here is to build a generic model that applies well to the cases, thus demonstrating the relevance of SD for agenda-setting. We proceed by treating each successive step in the cases as a subsystem with its own inputs and feedback loops. As we advance, we combine subsystems previously diagrammed, extending their effects to successive feedback loops. (We note each step in all the feedback loops in Appendix) We also draw connections to our hypotheses. A first result is to confirm the complex, nonlinear character of interactions among stakeholder agendas. To avoid adding to that complexity, we reserve full analysis of our findings for the concluding discussion section.

The specific content of stakeholder media and actions—that is, the detail of what someone said or did—is abstracted into reinforcing or balancing stimuli in successive diagrams and the cumulative model. Every event abstracted into the diagrams is detailed in footnotes. In other words, the diagrams map the footprints of stakeholder actions and media, while their content is captured in notes. In the next section, the broad outlines of each case and sources for footnotes are presented.

Summary of the Danone, BP and Nike Cases

The Danone Boycott

In January 2001, the revelation through a leak to a major newspaper that Danone SA, one of France’s most successful and admired firms, intended to downsize biscuit operations in its home market led to what observers called an unprecedented “political boycott”. The firm refused comment, but let it be known through proxies that restructuring was necessary to improve margins and please investors; financial analysts unanimously approved. Danone’s offer of a humane and relatively generous termination program for workers did not succeed in calming public outrage that a profitable firm was shedding jobs. In April, unions launched a nationwide boycott. MSM generally predicted the boycott’s failure. However, activist online media promoted it, diffusing lists of Danone brands. Rumours that the boycott was succeeding began circulating in May, panicking some investors, but were dismissed by the firm’s CEO at a Shareholder General Assembly. Management’s denials were accepted by the news media, which largely abstained from further coverage of the boycott and labour protests, and the stock briefly recovered. The following autumn, financial analysts reported that the boycott had indeed hurt Danone’s sales and market share, and that continuing labour action was further impacting operations, sales and growth across Danone’s three main product lines. An investors’ flight was accelerated by the events of 11 September 2001. The firm’s market share in the biscuit division and stock price remained depressed for several years. This summary and page references for footnotes concerning Danone are drawn from Hunter et al. (2008).

The Beyond Petroleum Campaign

The announcement that BP PLC, the world’s second-largest oil firm, was branding itself as “Beyond Petroleum” in 2000 generated widespread scepticism in MSM. However, MSM soon accepted BP’s claim that it was contributing to environmental progress. In contrast, Greenpeace’s website redefined BP as “burning the planet”. Its members occupied one of BP’s arctic drilling operations, and sought to organise BP’s activist shareholders against management. The main threat to BP’s operations, however, came from within, as employees at the firm’s Prudhoe Bay field in Alaska seized on the high ethical standards proclaimed by “Beyond Petroleum” to attract attention for their claims that BP was failing to meet its own safety and environmental benchmarks. A series of accidents at Prudhoe Bay, publicised first by activists operating their own media online, and then by news media, drew the attention of regulators, who ultimately posed a serious threat to BP’s operations in Alaska. Early in 2005, a catastrophic accident at BP’s refinery in Texas City and an oil spill in Alaska validated the charges of BP’s stakeholder critics. This summary and page references for footnotes concerning BP are drawn from Hunter et al. (2011).

The Nike Anti-sweatshop Movement

Nike Inc. is well-known not only for its shoes, but as one of the most controversial and crisis-ridden corporations (McHale et al. 2007). Beginning in the 1970s, Nike outsourced nearly all its production to Asian suppliers who employed low-skilled, low-wage labour (Lim and Phillips 2008). This strategy proved highly profitable, and Nike became a model for its industry. However, throughout the 1990s labour activists and NGOs reported wage law violations, child labour, excessive overtime, physical abuse of workers and unsafe working conditions among Nike’s suppliers. In June 1996, Life magazine showed a 12-year-old boy surrounded by the pieces of a Nike soccer ball that he would spend most of a day stitching together for the grand sum of 60 cents. In October 1996, the CBS news program “48 Hours” reported mistreatment of workers in developing countries by Nike (Basin 1996). Both MSM (Cherin 1999; Gevirtz 1996; Ikram 1996; Klein 1997; Manning 1996; Melville 1997; Read 1997) and activist groups and their media (including Global Exchange, UNICEF, Canadian Catholic Organization for Development, Sweatshop Watch, Citizen Action and Peace and Free The Children) supported union and NGO protests with further revelations. Student groups supported by university websites, newspapers and radios joined in the campaign in the late 1990s (Duncan 1997; Saunders 1997; Stancill 1997; Wilson 1997). Their calls for a boycott of Nike generated response from consumers and regulators (Collins et al. 2004; McHale et al. 2007; Palazzo and Basu 2007). Nike moved from denying the charges to approving routine independent inspection of its subcontractors (Holt 2002), and establishing a code of conduct on labour and environmental practices (Cropanzano et al. 2004; Graham and Woods 2006; O’Rourke 2006). Adversarial stakeholders continue to monitor the firm and promote protest actions through online media at this writing. Sources concerning Nike are cited in the footnotes and included in references.

Modelling Management and Stakeholder Agenda-Setting Interactions

Our analysis has identified multiple feedback loops underlying agenda-setting events in the Danone, BP and Nike cases. Of course, not all such loops can be of equal importance for organisational leaders or their adversaries; we will address that issue later. However, first we must understand the structure of the loops. Readers who find this process fastidious may wish to skip to Fig. 5.

In this section, we model the feedback loops in our cases. Two different types of lines are included: communication lines (in which stakeholders transmit information to others), and lines that show the deployment of resources. The information can concern the altering of an agenda, news about movements like boycotts or strikes, opinions or demands, and it can be transmitted through websites, reports or other media. The resources can include monetary or other capacities, and can be tracked through sources such as the stock value, investor briefings or annual reports. Moreover, different stakeholders are included (management, investors, workers, media, society and regulators), to capture their interactions. When a line starts from one stakeholder but ends with a different one, then the actions or the decisions of the initiating stakeholder affect the stakeholder where the line ends. While lines have starting points, feedback loops do not necessarily begin or end with a specific variable. (For example, a firm’s management may alter an agenda because its value has decreased, or because management expects a future decrease. In the first case, the loop begins with a decrease in value; in the second, it begins with an altered agenda.) In the text, variables appear in italics to emphasise their significance; regular typefaces are used in diagrams.

Firm’s Value and Interaction of an Altered Management Agenda with Workers: A General Case

Figure 1 depicts the following general dynamic: The vast majority of firms aim at improving their value (Firm’s Value), measured principally in profits and stock price. For example, Danone’s margins and stock price were at risk because of overcapacity in the biscuit sector,Footnote 1 BP’s management aimed to increase its salesFootnote 2 and Nike’s management wanted to reduce its cost by producing overseas.Footnote 3 The firm’s Profits increase through sales or market share, which depend in part on the Firm’s Image. These lines capture the deployment of resources, especially financial resources.

Fig. 1
figure 1

Firm’s value and interaction of an altered management agenda with workers: a general case

Full size image

In order to improve the Firm’s Value, management announces an Altered Agenda that involves changes in the firm’s extant operations (such as improving the firm’s processes, capacity or quality of output), for the benefit of various stakeholders. This move appears in the diagram as a communication line from the Firm’s Value to the Altered Agenda. The Altered Agenda may take the form of restructuring, as in the Danone case, or of perceived alterations in corporate priorities or operating standards, both of which affected BP, or of contracting with suppliers that employ low-wage labour, as in the Nike case. In all these cases, management perception of a potential decrease in the Firm’s Value drives the Altered Agenda. Through the Altered Agenda, the Firm’s Value is expected to increase. In SD terms, an initial decrease in the Firm’s Value eventually leads to an increasing effect of the Altered Agenda on the same variable, identifying a balancing feedback loop (B1). This loop is evident in all three cases.

However, the announcement of changes (communication line) may decrease Worker Trust in management’s decisions. A further decrease then occurs in the Firm’s Image, which is impacted both internally and externally by the protests of workers. The initial decrease in the Firm’s Value will be reinforced by declines in Worker Trust and in the Firm’s Image, as in loop R1. This feedback loop appeared at Danone when workers learned of the planned closing of factories and went on strike.Footnote 4 In BP it appeared when workers grew concerned about safety at the Prudhoe Bay oil fields.Footnote 5 At Nike, workers at subcontractors made contacts with unions to defend their rights.Footnote 6

In all three cases, management seeks to demonstrate responsible behaviour and reduce damage to the Firm’s Image by announcing the provision of Indemnities (such as benefits for displaced workers or safety inspections at production sites) to workers.Footnote 7 This creates the B2 (balancing) feedback loop, which neutralises loop R1.

Media Enter the Game

Distrustful workers can leak or openly volunteer information and counter-agendas to stakeholders who control their own media (referred to in Fig. 2 as SHM, for “Stakeholder Media”). In our cases, these media included tracts, online articles, financial analyst reports, blogs and pro-worker websites.Footnote 8 The support of Adversarial SHM Reports reinforces the decrease in Worker Trust (see loop R2, which appears in all cases).

Fig. 2
figure 2

Media enter the game

Full size image

Through Adversarial MSM Reports, news media cite or echo Adversarial SHM Reports. Likewise, Adversarial SHM Reports cite Adversarial MSM Reports in support of the stakeholders’ cause. The result is an increase in the overall number of reports, and hence in the salience of the pertinent issues (as predicted in H1a and H1b). This reinforcing feedback loop, R3, appears in all three cases.

As Worker Trust further declines, workers provide increasingly damaging information on the firm to MSM, leading to more Adversarial MSM Reports (in loop R4, which appears in all cases).Footnote 9 These reports negatively affect the Firm’s Image. Footnote 10 Thus, are created two reinforcing loops (R5, including Adversarial MSM Reports, Profits, Firm’s Value, Altered Agenda and Worker Trust, and R6, including Adversarial SHM Reports influenced by workers as well as the variables of R5). Because workers may receive indemnities at some point, two balancing loops (respectively, B3, involving Adversarial MSM Reports, Profits, Firm’s Value, Altered Agenda, Indemnities and Worker Trust, and B4, involving the same variables plus Adversarial SHM Report) are also created. R5 is neutralised by B3, and R6 by B4. All of these loops appear in the BP and Nike cases.

Subsequently, Adversarial MSM Reports may moderate management’s determination to execute the Altered Agenda as originally planned.Footnote 11 This creates two reinforcing loops that incorporate previous reports, management’s agenda, indemnities, worker trust and further reports (loops R7, including Adversarial MSM Reports, Indemnities and Worker Trust, and R8, involving Adversarial SHM Reports as well as the variables of R7). In the eventual absence of indemnities, two balancing loops emerge: B5, through Worker Trust, Adversarial MSM Reports and Altered Agenda, neutralises R7; B6, through the same elements plus Adversarial SHM Reports, neutralises R8). These loops appear in all three cases.

Interaction with Society

New stakeholders appear in response to the attention afforded to the conflict by media. Adversarial SHM Reports increase the Mobilisation of adversaries, which can lead to a Social Movement (in the Danone and Nike cases, this dynamic preceded MSM coverage).Footnote 12 These changes decrease the Firm’s Image,Footnote 13 Profits and the Firm’s Value,Footnote 14 which supports hypotheses H3a and H3b.

The introduction of Mobilisation and a Social Movement creates two new reinforcing loops, respectively, R9 (involving Adversarial MSM Reports through Firm’s Image, Profits, Firm’s Value, Altered Agenda, Worker Trust and Adversarial SHM Reports) and R10 (which contains the same elements, minus MSM reports). The dynamic interaction of workers, MSM and SHM across these loops, with each feeding and reinforcing the other, supports hypothesis H1c.

In the event of Indemnities, two balancing loops are created. B7, which includes Firm’s Image, Profits, Firm’s Value, Altered Agenda, Indemnities, Worker Trust, Adversarial MSM Reports, Adversarial SHM Reports, Mobilisation and Social Movement, neutralises R9; and B8, which contains the same variables except for MSM reports, neutralises R10. These loops appear in the Danone and Nike cases. In another loop, the Social Movement openly supports workersFootnote 15; Worker Trust in management further decreases. This conjunction of forces creates two reinforcing feedback loops (R11, including Worker Trust, Adversarial MSM Reports, Adversarial SHM Reports, Mobilisation and Social Movement, and R12, which combines the same elements minus MSM reports). They are evident in all three cases (Fig. 3).

Fig. 3
figure 3

Interaction with society

Full size image

Interaction with Regulators

Mobilisation can lead to judicial retaliation by the firm’s management against stakeholder activists,Footnote 16 or conversely, to regulatory intervention or lawsuits on behalf of stakeholders who believe their interests are affected by the Altered Agenda. Such Judicial/Regulatory Action can further hurt the Firm’s Image. Footnote 17 We thus see two new reinforcing loops: R13, which includes Firm’s Image, Profits, Firm’s Value, Altered Agenda, Worker Trust, Adversarial MSM Reports and Adversarial SHM Reports, and R14, containing the same variables except for MSM reports. They are driven by the effect of Mobilisation on Judicial/Regulatory Action. They are neutralised when management provides Indemnities Footnote 18 in the balancing loops B9 and B10. These four loops are evident in all the cases, and support hypotheses H2a and H2b, which predict rising costs of satisfying stakeholder demands as more of them join the fight.

Different drivers create different loops: Judicial/Regulatory Action may also support the agenda of the Social Movement. Footnote 19 This leads to four new reinforcing loops: R15 and R16 both include Worker Trust, Adversarial SHM Reports, Mobilisation, Judicial/Regulatory Action and Social Movement, and the former also includes Adversarial MSM Reports. R17 and R18 contain Firm’s Image, Profits, Firm’s Value, Altered Agenda, Worker Trust, Adversarial SHM Reports, Mobilisation, Judicial/Regulatory Action and Social Movement, with the addition of Adversarial MSM Reports in R17. We also see two new balancing loops in the event that Indemnities are offered; B11 neutralises R17, while B12 neutralises R18. All six new loops appear in the Danone and BP cases.

Judicial/Regulatory Action also informs and intensifies coverage in Adversarial SHM Reports Footnote 20 (reinforcing loop R19, including Mobilisation,) and in Adversarial MSM Reports Footnote 21 (reinforcing loops R20, containing Adversarial SHM Reports, Mobilisation and Judicial/Regulatory Action, and R21, which also includes Altered Agenda, Indemnities, and Worker Trust). The absence of Indemnities will create balancing loop B13, which neutralises R21 through Adversarial MSM Reports, Altered Agenda, Worker Trust, Adversarial SHM Reports, Mobilisation and Judicial/Regulatory Action. These four loops are evident in all our cases (Fig. 4).

Interaction with Investors

A decline in the Firm’s Value impacts its Investors’ Trust, decreasing the Stock Price. After a time delay, falling investor trust likewise decreases the Firm’s Value (Fig. 5). These variables create the reinforcing feedback loop R22, which appears in all the cases. Support for hypotheses H3a and H3b is confirmed. (In fact, investors often react to rumours or reports of a decline in profits before that news is officially announced by a company.) Criticism of management by financial analysts or other investors (Investors’ Trust) attracts Adversarial MSM Reports, further decreasing Investors’ Trust,Footnote 22 and creating reinforcing loops R23, including Firm’s Value and Altered Agenda, and R24, involving Firm’s Image, Profits and Firm’s Value. These loops appear in the BP case.

The loss of Investors’ Trust also generates Adversarial SHM Reports,Footnote 23 which further influence the Investors’ Trust. Thus, all our cases show the reinforcing loops R25 (including Firm’s Value, Investors’ Trust, Adversarial SHM Reports, Adversarial MSM Reports and Altered Agenda) and R26 (including Firm’s Image and Profits along with the variables of R25 but excluding Altered Agenda). The reinforcing loop R27, involving Investors’ Trust and Adversarial SHM Reports, also appears in all cases.

Management may seek to forestall further loss of Investors’ Trust due to a decline in the Firm’s Value and Stock Price by denying losses in sales, market share or quality resulting from protests, as occurred in the Danone case.Footnote 24 In SD terms, this equates to trying to balance the system; hence we now see balancing feedback loop B14, including Firm’s Value, Rumours’ Denial, Investors’ Trust and Stock Price. The Rumours’ Denial appears in SHM such as analyst reports as well as MSM. This creates balancing loops B15 (Firm’s Value, Rumours’ Denial, Investors’ Trust, Adversarial SHM Reports, Adversarial MSM Reports and Altered Agenda) and B16 (similar to B15, with Firm’s Image and Profits in the place of an Altered Agenda). The denial also surfaces in MSM that quote only management. The result is balancing loops, B17 and B18, which resemble the previous pair but do not include SHM reports.

However, even if management denials initially succeed, eventually a public firm must acknowledge a genuine decline in sales.Footnote 25 The admission will negatively impact Investors’ Trust, captured in reinforcing loop R28, which includes Firm’s Value, Rumours’ Denial, Correct Information, Investors’ Trust and Stock Price. The Correct Information also appears in analysts’ SHM; the result is reinforcing loops R29 (Firm’s Value, Rumours’ Denial, Correct Information, Investors’ Trust, Adversarial SHM Reports, Adversarial MSM Reports and Altered Agenda), and R30, in which Firm’s Image and Profits take the place of Altered Agenda.

Finally, we note that exogenous events occurring in parallel with the Altered Agenda can trigger Adversarial MSM Reports as well as Adversarial SHM Reports that impact the firm. Such events may thereby obscure or override management attempts to resolve conflicts that affect the firm’s agenda. An example occurred in the Danone case, when the brutal closing of a major retailer in Paris overshadowed the firm’s attempts to appear responsible.Footnote 26 Exogenous events create impacts but not loops in our cases, because there is no way for stakeholders to influence the source of such events.

From the above analysis, we observe that in all our cases, management’s announcement of changes compelled different stakeholders (workers, stakeholder and news media, society, regulators and investors) to advance alternate agendas. Management and stakeholder agendas are shaped in a total of 34 lines (counting the lines only the first time that they appear). Of this total, six lines show the deployment of resources and 28 are communication lines in which stakeholders transmit information. Concerning the latter, we note that SHM reports appear in eleven loops in the absence of MSM reports (while MSM reports appear in eight loops without SHM reports). In other words, stakeholders are not necessarily dependent on MSM in order to inform or influence each other. This insight reinforces hypotheses H1a and H1b.

However, there is no common sequence of events across the three cases. For example, in the Danone and Nike cases, workers were the first to oppose the firm’s Altered Agenda, while for BP the first opposition was expressed by MSM and environmentalist SHM. Therefore, the order in which stakeholders and feedback loops are presented above does not define a temporal order. It indicates that certain feedback loops will emerge, but not exactly when. In practice, feedback loops take time to actualise; specifically, feedback loops containing more stakeholders and interactions require more time to actualise than “shorter” ones. One implication is that the more actors enter the scene, the less likely it is that major outcomes of their actions will be immediately obvious. We will return to this point in our discussion.

Comparison of the Model with Full Cases

Our generalised model illustrates the complex nature of the agenda-setting effects generated by the multiple stakeholders who act in each case and the different goals that they pursue. Though built using cases from three different industries and involving different contents, it exposes common structural characteristics, such as the similarity of stakeholder strategies, amplification of feedback among different actors, and the inability of management to address complexity. However, in each case different effects are activated (or not). In the Danone case (Hunter et al. 2008) two effects sketched in the generic feedback loop diagram do not appear entirely valid. Specifically, Adversarial MSM Reports did not affect the Firm’s Image. Moreover, the decline in Investors’ Trust was not fully reported in Adversarial MSM Reports. In particular, MSM did not make a connection between management’s delayed and partial acknowledgement of the impacts created by protestors and workers, and the subsequent flight of investors from Danone’s stock. In contrast, financial analysts at this point repeatedly warned that Danone management information was not sufficiently “visible”. Therefore, the Danone case contains 42 feedback loops compared to the 48 feedback loops found in the generalised model.

Likewise, in the BP case (Hunter et al. 2011) some impacts sketched in the generic feedback loop diagram are not reflected in events, so far as we can discern from firm announcements, news and stakeholder-controlled media. Specifically, we note that the Social Movement did not affect the Firm’s Image. The difference here from the Danone and Nike cases may be due to the fact that MSM coverage of worker protests in Alaska was largely restricted to the business press, and in particular The Financial Times. Thus, stakeholders who do not follow business news had little opportunity of being exposed to worker agendas, unless they followed certain environmentalist and left-wing political websites. Hence, in the BP case there are 36 feedback loops.

Finally, in the Nike case (see “The Nike Anti-sweatshop Movement” section) we also find that some impacts sketched in the generic feedback loop diagram are not reflected in events. Specifically, we note that unlike Danone, management at Nike did not find itself obliged to deny and then confirm “rumours” that cast the firm in a bad light. Nor did Judicial/Regulatory Action support the agenda of the Social Movement. Finally, as in the Danone case, news media did not correlate decreases in Nike’s stock price with stakeholder protest mobilisations. In fact, news media eventually gave Nike credit for a good faith effort to end abuses in its suppliers’ plants, at a time when other leading apparel manufacturers likewise moved their production to Asia and faced similar accusations. Therefore, the Nike case contains 32 feedback loops.

We note that specific loops cannot always be designated as responsible for a given outcome, because certain influence lines appear in multiple loops. Thus, for example, any or all loops containing impacts on sales or stock price may affect a rise or fall in those variables. This makes it more difficult to control outcomes, a point we will return to below.

We conclude that the model can be used as a dynamic foundation that one can customise by adding pertinent factors. For example, Tsuchiya (2003) found that explicit ethical concerns and whistle blowers play a larger role than in our model, reflecting her case studies and their Japanese cultural context. We thus confirm that SD can help to weigh what can and should be done in the face of complex issues (Pruyt and Kwakkel 2007), including stakeholder agenda-setting.

Discussion

How Stakeholder Agendas Prevail

We began with three questions: How do stakeholders use their media to set agendas? How does management react? And what effects result? Our main objective was to widen agenda-setting theory to a dynamic and nonlinear networked stakeholder context, in which stakeholder-controlled media assume part of the role previously ascribed to MSM. In effect, our diagrams describe a generic process through which stakeholder groups intervene to counter agendas, set their own and compel management to negotiate responsibility for conflict.

The process begins with a management agenda that aims at making the firm more profitable for its investors. However, certain stakeholders resist a change that they view as contrary to their own interests. They engage in protest actions, and use their own media to publicise those actions. Thus, H1a is validated. Whether or not MSM take note immediately, the stakeholder resistance is noticed and amplified by at least some other stakeholders. Our model captures that, as the social movement literature shows, stakeholder claims gather the most attention from media and other stakeholders when they are accompanied by action such as strikes, boycotts, etc. (Amenta et al. 2009). As H1b predicted, new stakeholders appear in the conflict. They include most of those identified by Ragas (2010) as actors in corporate proxy fights: management, shareholders, analysts, unions, workers, suppliers, government, customers, MSM, courts and regulators. All of these actors are structural elements of the business environment, and the noise from other stakeholders alerts them to situations requiring their attention. They also begin to pay attention to each other, as H1c predicts.

H2a predicted that once stakeholders get involved, they formulate their own demands, and Figs. 4 and 5 capture the resulting thicket of dissatisfaction that enfolds management. H2b predicted that in trying to satisfy those multiplying demands, management would also face rising costs. Our model shows two specific mechanisms that raise them: indemnities and regulatory sanctions. The indemnities required to satisfy workers and regulators may be very disappointing to investors. That is one reason H3a, which predicted a negative effect on management reputation, likewise appears supported: Rising costs make investors worry that management cannot succeed. There is also a hit on the firm’s value, as predicted in H3b: The circle of resisting stakeholders eventually includes customers whose purchases support the firm’s profits. The consequent fall in stock price further alarms investors and confirms H3a.

Fig. 4
figure 4

Interaction with regulators

Full size image
Fig. 5
figure 5

Interaction with investors

Full size image

The result is that harassing management becomes a common agenda for stakeholders as diverse as investors, regulators and protestors. Management is surrounded by a web of watchdogs, who closely monitor its moves and rapidly sanction any future misstep.

Watchdogs are hardly unique to the stakeholder environment; they are also a prominent feature of high-quality MSM news operations. In analysing six cases of investigative journalism reports that drove positive change, Protess et al. (1992) discovered two explanatory models of their effects. The first, “mobilisation”, occurs when public response to journalistic revelations is so outraged that policy makers leap to placate it. The famous Watergate affair is cited as an example: Reporters revealed successive truths until a public backed by honest officials forced the President to resign. The second model, “coalition”, takes place when journalists assemble a diverse stakeholder constituency to promote their revelations to policy makers. A common element in both models is that to maintain pressure for change, watchdogs must constantly widen their circles of influence to new allies and observers, using their cumulative responses to drive a first-level agenda (“this affair is important”).

Our feedback loops show another way in which contemporary stakeholder watchdogs drive outcomes. As before, stakeholder groups (in our cases, workers) initiate opposition to management, and other groups are affected, who affect still others. However, unlike the coalition or mobilisation models, in our diagrams stakeholders gain influence not only by building coalitions or alerting potential allies, but also by compelling stakeholders who are not in the same camp to press their own claims on management. The web of watchdogs eventually widens to include not only adversarial stakeholders, but also management’s putative supporters. In the worst case (as occurred to Danone and BP), management is isolated, its credibility is weak, and its initial agenda is effectively moribund. Though the media employed to this end may be “social”, the general concern of stakeholders is hardly to be “liked” by anyone else in the network. Investors rejected the goals of the “alters” whose websites drove the Danone boycott, but the former punished management very effectively for denying that the boycott worked. Agenda-setting strategies are evolving in this new, complex era by embracing stakeholders whose motivations and objectives may vary hugely.

What drives this dynamic is that initiating stakeholders diffuse information that even non-allied stakeholders must take into account. These information subsidies may be appropriated by MSM, as in the BP case, or not, as in the Danone case (where MSM never reported that analysts sanctioned management). However, in our three cases, stakeholder media were quite sufficient to disturb the system through new information.

The first key providers of that information to stakeholder media, in every case, were workers. This phenomenon is intuitively logical: Workers may enjoy privileged access to information about a company, they may have strong reasons for revealing it, and they may not enjoy privileged access to MSM, at least in comparison to management. Our model captures how these factors can combine with stakeholder media to shift some power from MSM and management to stakeholders. Specifically, it shows that even stakeholders with limited resources can drive feedback loops that raise management costs to unsustainable levels. What is an appropriate management response to this new reality?

Management Responsibility in an Age of Stakeholder Media

In each of our cases management seeks to act responsibly first by proposing a strategy to increase the firm’s value. When opposition arises, management displays responsibility by offering indemnities to key stakeholders. In all three cases, the tactic fails to resolve the issues. One possible explanation is that the stakeholders who were directly affected by the strategy considered the indemnities as merely symbolic gestures, woefully disproportionate to the real needs created by management’s agenda. They responded by seeking to disrupt the agendas, notably by pulling other stakeholders into the conflict. In that case, what is management’s further responsibility?

In every case, management behaved as though it were vital to maintain their initial agenda, regardless of growing costs. Nike did not shift production to higher-cost environments. Danone management proceeded with restructuring. BP continued with environmentally and physically risky operations. Stakeholders changed the circumstances, but management did not take those changes into account quickly, or at all. It could be argued that these managers believed their plans were vital to the future of their firms. However, in two out of three cases, rising costs overwhelmed potential gains of the strategy. It could thus also be argued that by failing to heed the watchdogs, managers failed to protect value for all their stakeholders, and in particular investors. Even from a shareholder value perspective (Friedman 1970), this makes little sense.

Leaders, especially, are supposed to prevail; yet they must also accept the facts when their strategies cannot succeed. In all of our cases, to a greater or lesser extent, management instead displayed a typical response in crisis situations: denial (Coombs and Holladay 1996; D’Aveni and MacMillan 1990). The problem was someone else’s, or under control, or did not exist except in the minds of certain stakeholders. At an extreme, firms (like Danone) may deny that activist protests impact financial performance, even if they do. Certainly, sometimes this management stance is justified. Stakeholder activism does not always target urgent issues, or materially impact a firm’s operations or sales. (In fact, this was the case at BP, until successive safety disasters severely damaged the firm’s image with regulators and the public.) Nor do stakeholders always attract support from others: Shareholder activists’ proposals to management are more often than not defeated.Footnote 27

The underlying and key issues are whether and how management must respond to stakeholder claims. We can add two criteria that may help managers decide when those claims are legitimate and urgent. The first is the density and depth of feedback loops around an issue—that is, the number of other stakeholders who respond to one group’s initial claims. When response appears among stakeholders who are not allied to protestors, it is a sign that feedback is spreading into an “ambiguous, legitimizing environment” (Holmström 2005, p. 503) where management’s influence will continually decline.

The second criterion is the quality of information that circulates in these feedback loops, and particularly in stakeholder media, which often surpass MSM in the early stages of a conflict. Other stakeholders cannot always judge for themselves if a particular issue raised by protestors (such as inadequate safety or environmental precautions at an installation) is valid. Management, however, either has the means to verify such claims, or is failing to run this aspect of the firm. In the latter case, denial may lead to disaster. In the former, indemnities can be targeted and scaled to effective solutions for stakeholders, and the feedback loops that carry information and opinions damaging to the firm can thereby be transformed into balancing forces. Alternatively, management can identify and address other players in the same loops—one way to do so is by monitoring stakeholder media—to find an acceptable solution (Henisz et al. 2011). In any or all of these cases, management is responsible for taking into account the information made available through stakeholder media.

Managers who persist with an agenda in the face of widening and diversifying stakeholder feedback loops are engaged in a high-risk strategy. In confrontations with expanding stakeholder networks, where the underlying issues pose a “dichotomy of risk versus danger, of decision taker versus victim” (Holmström 2007), time is not on management’s side. Adversaries do not need to score a decisive victory; they need only continue to raise the marginal costs for management of pursuing its agenda, in order to change the agendas of other stakeholders towards the firm. Even if one is not interested in SD or modelling, this key message should be clear. Leaders in an age of stakeholders networked through their own media are only one node in a complex system they cannot totally control, in which actions and reactions can come back like a boomerang. Whatever the final outcome, management will not determine or impose it alone.

Further Directions for Research

We acknowledge that relying on a relatively small number of case studies, while appropriate for opening a new line of research, limits the general applicability of our findings. For example, family owned firms may display very different stakeholder dynamics from publicly owned ones. It would also be instructive to apply our model to cases where activist stakeholders do not prevail in whole or part—for example, where the costs to management created by opposition are trivial. Our proposed models must be further validated through case studies in circumstances that differ substantively from the three cases shown here.

The following foundation steps must be undertaken: first, formulation of a simulation model with estimations of parameters and initial conditions, which can be fine-tuned by adding more cases and data, including theoretical and empirical material from the literature. An integrative approach is required. SD can be used to identify the most interesting kinds of loops, and then to guide more quantitative research. Actual cases could be followed in real time, to predict what is likely to happen next and propose more options for action (Sterman 2000). The evolving model should be tested by extreme condition cases, to explore its limits and verify its quality. Finally, the effects of different policies or actions on the system should be evaluated by conducting simulations. Probability sampling and quantitative longitudinal data should be used to validate or falsify resultant models. Eventually, robust SD models may enable us not only to better dissect the paths of stakeholder feedback, but also to foresee and resolve the underlying issues behind it.