Abstract
Facing an increasing number and variety of issues with social salience, firms must determine how to engage with issues that likely have a significant impact on them. Integrating issues management (IM) and salience theories, the authors find that firms engage with socially contested issues—where there is a high degree of societal disagreement—in a different manner from issues that have social consensus, or high agreement. Examining social issue resolutions filed by shareholders from 1997 to 2009 (3,887 total observations), the study finds that socially contested issues, as well as those issues with social consensus, are both likely to result in engagement by the firm. For social issues with consensus, a firm is more likely to opt for a low level of shareholder engagement whereas resolutions regarding contested issues lead to engaging shareholders at a higher level. These findings shed new light on the IM and issue salience literature streams that have suggested firms will react differently to these types of issues, even while they remain largely untested. Finally, firms become less engaged with perennial issues over time. rather than more, providing new guidance to researchers, shareholder activists, and firms alike. To the authors’ knowledge, such fined-grained insight into expected levels of firm engagement with social issue salience has not been put forth previously.
Researchers and practitioners alike have long been concerned with what motivates a firm to respond to certain social issues and not others. In this vein, issues management (IM) has traditionally focused on a firm’s ability to identify, evaluate, and respond to social issues that have a significant impact on it (Bartha, 1982; Dutton & Jackson, 1987; Johnson, 1983; Mahon, 1989; Wartick & Mahon, 1994), establishing some connection between societal issues and firm behavior. Although IM models have helped researchers identify issues that firms are expected to care about, most IM models remain wedded to a predictable model of the evolution of a reoccurring issue (Bundy, Shropshire, & Buchholtz, 2013; Lamertz, Martens, & Heugens, 2003) for which the firm has ample time, resources, and options to respond (Albrecht, 2000; Heath, 1997; Lamertz et al., 2003).
Recent research on firm-level issue salience—or the degree to which an issue resonates with and is prioritized by a firm’s management—suggests issue salience is a key antecedent of firm responsiveness (Bundy et al., 2013; Dutton & Jackson, 1987; Eesley & Lenox, 2006), and the iterative, perennial nature of an issue is also important to how a firm reacts (Dutton & Jackson, 1987; Lamertz et al., 2003; Proffitt & Spicer, 2006). There are also issues that are salient to society and often reflect current public opinion and special interest group attention on particular topics (Aggarwal, Erel, & Starks, 2014; Baron, 2010; Desai, 2011; Mahon, 1989). Some of these issues become salient to the firm. It is likely, then, that social issue salience can inform what issues a firm deems salient.
Therefore, issue salience is not only a function of how the firm responds to the issue (Bundy et al., 2013; Dutton & Jackson, 1987; Eesley & Lenox, 2006) and how it allocates attention and resources (Hoffman & Ocasio, 2001; Ocasio, 1997) but also of the degree of social issue salience. Although issues are at the very heart of firms’ practices and strategic planning (Arcelus & Schaefer, 1982; Arrington & Sawaya, 1984; Desai, 2011; Renfro, 1993), scholars know little about how certain types of issues affect the firm’s level of engagement given a high level of societal salience.
IM research suggests that agreement or disagreement over an issue’s meaning, definition, or label is important in determining how engaged the firm is likely to be with that issue (Bundy et al., 2013; Dutton & Duncan, 1987; Dutton & Jackson, 1987; Hoffman & Ocasio, 2001; Lamertz et al., 2003; Wartick & Mahon, 1994) and the meaning of the issue will change in important ways over time (Gerde & White, 2001; Wartick & Mahon, 1994). Researchers, however, propose contradictory ideas on how the firm will manage two specific types of issues: those issues having a high degree of social consensus and those issues that are socially contested. These types of social issues are argued to lead to different types of firm action by various researchers (Bonardi & Keim, 2005; Bundy et al., 2013; Dutton & Jackson, 1987; Hoffman & Ocasio, 2001; Jones, 1991; Lamertz et al., 2003; Wartick & Mahon, 1994). Some of this research has suggested that the agreement or disagreement about issues at the societal level also affects how the firm frames the issue and whether they then engage with it (Aggarwal et al., 2014; Bundy et al., 2013; Hoffman & Ocasio, 2001; Jones, 1991). The study extends this literature to suggest, for example, that a firm’s awareness of a social issue with consensus is more likely to result in higher levels of firm engagement because pressure to respond in acceptable ways reflects current public sentiment (Aggarwal et al., 2014). At the same time, a firm is less likely to engage with socially contested issues due to the uncertain, ambiguous nature of the issue and its impact (Lamertz et al., 2003; Wartick & Mahon, 1994).
The perennial nature of an issue and its degree of social contestation or consensus, remain important topics because they help to gain sorely needed insight into ways in which firms engage with certain social issues. And, because organizations selectively attend to some issues and not others (Dutton, 1986; Dutton & Jackson, 1987), variation in engagement is expected among firms even within the same industry (Dutton & Jackson, 1987; Hillman, Keim, & Schuler, 2004; Schuler, Rehbein, & Cramer, 2002). Driven by this literature, the study empirically tests how issues with social salience, issues that have social consensus or contestation and are reoccurring or perennial issues affect a firm’s level of engagement.
The setting for this inquiry is shareholder resolutions, a predominant and direct form of firm engagement with shareholders (O’Rourke, 2003; Proffitt & Spicer, 2006), and the primary mechanism for shareholder activists to show their dissatisfaction with a firm’s practices (Gillan & Starks, 2007; Logsdon & Van Buren, 2009). Resolutions filed by shareholders have proven especially useful in identifying important social issues and bringing them to the attention of the firm (Lawrence & Weber, 2014; Proffitt & Spicer, 2006; Reid & Toffel, 2009). In this way, shareholders have been instrumental in shaping the social issue agenda of firms by utilizing their ownership stakes and status to urge firms to pay attention to specific issues (Clark & Crawford, 2012).
Shareholders bring issues they deem important directly to the attention of the firm through the annual shareholder resolution process involving a formal filing with the Securities and Exchange Commission (SEC). This process represents an important research setting to test the IM and issue salience theory literatures because a firm must publicly respond to each shareholder resolution, has only a limited time frame in which to do so, must choose among a constrained set of actions, and may need to respond to the same issue over time. Likewise, social issues brought by shareholders in these resolutions are reflective of broad public concern as they are increasingly important to both shareholders (Eesley & Lenox, 2006; Reid & Toffel, 2009) and the general public (Aggarwal et al., 2014). Almost half of all shareholder resolutions in 2014 involved social issues (Copland & O’Keefe, 2014).
Looking at shareholder resolutions enables the study to examine a variety of social issues ranging from diversity, to sustainability, to accountability standards. Much of the previous research has examined one type of social issue, environmental issues (Clark & Crawford, 2012; Eesley & Lenox, 2006; Reid & Toffel, 2009), or dealt with only generic issue types (e.g., true threat, true opportunity, frame conflict, etc.) that may or may not elicit an accommodating response from the firm (Bundy et al., 2013; Rehbein, Waddock, & Graves, 2004). Shareholder resolutions also serve as a unique context for understanding IM and firm behavior, in terms of both emerging and latent social issues, because they reveal important insight into what firms choose to do when facing public pressure, a limited response time frame and specific options for engagement.
The study’s contribution is in advancing theory by integrating issue salience and IM literatures to empirically test firm engagement with four types of issues and in doing so add to the theoretical conversation by treating shareholder activism as embedded and influenced by the salience of broader societal views as well as the consensus or disagreement among those views. In this way, we also challenge important and long-standing ideas about the effects of issue contestation and consensus on firm responsiveness to help explain how firms are engaging with social issues and which ones they engage with over time.
The article is organized as follows: first, the article presents a review of relevant literatures on IM and issue salience with an eye toward determining firm engagement with a salient social issue. Second, the article presents hypotheses regarding four types of issues—salient issues, issues with social consensus, socially contested issues, and perennial issues—and how each type of issue is likely to affect firm engagement. Finally, tests of the study’s hypotheses are followed by a discussion of its implications for activists, managers, and researchers.
Theory
Firm Engagement
According to Lawrence (2002), engagement is most likely to occur when both parties have an urgent and important goal, the motivation to participate, and the organizational capacity to engage. Furthermore, companies engage at different levels: Lower levels of engagement involve informing and explaining whereas higher levels are depicted by active attempts to involve other parties in company decision making (Carroll & Buchholtz, 2013; Hess, 2007). When a firm responds at a high level, it highlights the firm’s willingness to engage with the group sponsoring the issue in an open and meaningful way (Bundy et al., 2013; Hess, 2007). In this article, accordingly, the authors propose that engagement occurs when the firm takes action regarding an issue initiated by a proponent or a group of proponents.
A primary way in which the firm engages with shareholders is by responding to shareholder resolutions (Hess, 2007). Resolutions are a mechanism by which shareholders initiate a dialogue about firm actions and behaviors, serving as the principal means through which they show management their dissatisfaction (Gillan & Starks, 2007; Logsdon & Van Buren, 2009). Because owners of the company file resolutions, the firm is obliged to respond to all proponent filers within a specified set of actions (i.e., omit it from the proxy ballot, send it to a vote on the proxy ballot, or withdraw it prior to the creation of the proxy ballot) even though the issues of concern brought by shareholders may vary widely. 1
The next section focuses on IM and salience for insights into a firm’s response to an issue of concern.
IM
IM has been defined as the process by which the firm can identify, evaluate, and respond to social issues that have a significant impact on it (Johnson, 1983; Wartick & Mahon, 1994). A good portion of the IM literature considers the dynamic, transitory, or cumulative effect of issues over their life cycle (Baron, 2010; Bonardi & Keim, 2005). Issue life cycle literature addresses two important aspects of how issues are managed (Wartick & Mahon, 1994): the cumulative effects of a single issue over time and the cumulative effect of multiple issues affecting the same firm over time.
Mapping the trajectory of single or multiple issues yields important insights for firms (Wartick & Mahon, 1994), notably that an issue is likely to gain importance over time. For example, the importance of secondhand smoke has changed considerably over a 40-year period, with research tracing how the issues, and the responses by large U.S. tobacco firms to threats of regulation, changed significantly over several decades (Derry & Waikar, 2008; Mahon & McGowan, 1996).
Researchers examining perennial issues (Ansoff, 1980; Arcelus & Schaefer, 1982; Gerde & White, 2001) note that issues early in their life cycle are typically vaguely defined, but as time goes on information is more abundant so the costs, benefits, and assessment of alternatives can be conducted with information that becomes increasingly credible. However, others have cautioned that perennial issues may depart from the predicted linear progression of most life cycle analyses and can remain the object of public, firm, or industry attention over a sustained period or become more episodic (Hoffman & Ocasio, 2001; Lamertz et al., 2003).
Likewise, early IM literature categorized issues as threats or opportunities (Dutton & Jackson, 1987; Jackson & Dutton, 1988) whereas recent literature considers whether an issue is contested or has consensus and a consistent meaning (Bonardi & Keim, 2005; Bundy et al., 2013; Hoffman & Ocasio, 2001; Lamertz et al., 2003). For socially contested issues, there is general disagreement about the issue that might lead to uncertainty on the part of the firm regarding the impact, resolution, and desired action for the issue (Lamertz et al., 2003; Wartick & Mahon, 1994). Issues that have social consensus are those where widespread agreement exists as to whether or not the behavior is appropriate (Jones, 1991). However, scholars disagree as to whether and how contestation and consensus affect firm action (Bonardi & Keim, 2005; Hoffman & Ocasio, 2001; Lamertz et al., 2003). Comparatively, less tension exists about the salience of an issue as an antecedent of firm action (Bundy et al., 2013), even while firm engagement remains largely untested.
The next section expands on important aspects of the issue salience literature and then addresses consensus and contestation.
Firm and Social Issue Salience
Generally, an issue is defined as an inconsistency between what is and what ought to be, arising from gaps in societal expectations of corporate performance that create a significant impact, present or future, on the organization (see literature review by Wartick & Mahon, 1994). Research suggests that issues become ripe for action by the firm when two factors are present: Society’s expectations are high and the issue is likely to affect the business (Heath, 1997; Lawrence & Weber, 2014). Social expectations are reflected in social norms, or the unwritten rules of appropriate conduct (Elster, 1989), and these norms affect how a firm operates given that firms can be influenced by the economic and social climate of public opinion (Aggarwal et al., 2014). Accordingly, when a firm fails to account for a gap in actual versus expected behavior, and the public takes notice of this gap, one consequence is that a social issue gains importance and the public exerts pressure on the firm to act (Wartick & Mahon, 1994). When the firm notices and identifies this social issue, as potentially relevant to its status, performance (Ashforth & Mael, 1996), or strategic outlook (Bundy et al., 2013), a firm is likely to take action (David, Bloom, & Hillman, 2007; Heath, 1997; Lawrence, 2002). Issue salience—or the degree to which an issue resonates and is prioritized by firm management—is a key driver of firm responsiveness 2 (Bundy et al., 2013), making it a fundamental stage of the IM process (Bartha, 1982; Heath, 1997; Mahon, 1989). Salience of an issue, however, depends in part upon the audience evaluating it (Hoffman & Ocasio, 2001). Public opinion and special interest groups may accelerate the salience of a social issue by repeatedly putting an issue in the headlines (Aggarwal et al., 2014; Baron, 2010; Desai, 2011; Mahon, 1989). So, firms can gauge whether a social issue is important by tracking public opinion (Aggarwal et al., 2014) and by paying attention to activists’ targeting behavior (Lawrence & Weber, 2014). Still, scholars’ knowledge about this interplay between societal issues and issues the firm engages with remains relatively limited.
Based on this discussion of firm engagement, IM, and issue salience, the following section develops and tests the study’s four hypotheses.
Hypothesis Development
Social issue salience
Conditions are conducive for a firm to engage when society’s expectations are high and the social issue is likely to affect the business (Heath, 1997; Lawrence & Weber, 2014; Wartick & Mahon, 1994). These expectations are often reflected in social norms of what is considered appropriate firm conduct (Elster, 1989), and thus affect how a firm engages given that firms are influenced by their social and economic climate (Aggarwal et al., 2014).
Therefore, a firm is likely to allocate attention and resources (Hoffman & Ocasio, 2001; Ocasio, 1997) to those issues that are likely to have a significant impact on the firm (Ansoff, 1980; Dutton & Jackson, 1987; Jackson & Dutton, 1988; Johnson, 1983). Accordingly, the authors argue that when society views a social issue as highly salient, a firm is likely to engage with it at a high level. Therefore we hypothesize the following:
Social Consensus
Previous research on social issues suggests that both agreement among societal actors on the importance of a particular issue or event (Hoffman & Ocasio, 2001), and how it should be resolved (Bonardi & Keim, 2005; Lamertz et al., 2003), are vital to determining which issue garners firm engagement. Some prior research has found support for social consensus as a predictor of a firm’s awareness of an issue (Jones, 1991; May & Pauli, 2002). Based on the work of Jones (1991), the authors define social consensus as the degree of social agreement that a proposed issue is either good or bad. The study extends this literature to suggest that a firm’s awareness of a social issue with consensus is more likely to result in higher levels of firm engagement because shareholder pressure to respond in acceptable ways reflects current public opinion on particular issues (Aggarwal et al., 2014).
Researchers have suggested that once the public accepts a standard of behavior about a certain issue, demonstrating social consensus, the opportunity window closes resulting in little choice than to comply with an issue because it has a high level of social consensus (Aggarwal et al., 2014; Bonardi & Keim, 2005; Lamertz et al., 2003; Zyglidopoulos, 2003). In this way, responding to the issue implies that “agreement exists as to whether or not the behavior is appropriate” (Jones, 1991, p. 375). Also, once issues are categorized or labeled, new information tends to confirm, not contradict, that initial label (Dutton & Jackson, 1987). Bundy and colleagues (2013) propose that issues consistent with a firm’s strategic goals will have high salience with the likelihood of higher levels of engagement. Firms facing salient issues are often pressured to conform (Bonardi & Keim, 2005) and this conformity extends to other firms making them less likely to diverge from the response norm (Bonardi & Keim 2005; Dutton & Jackson, 1987). In sum, determining the social consensus around an issue is important to the likelihood of what engagement level the firm will take. Therefore we hypothesize the following:
Social Contestation
Alternatively, some researchers suggest that socially contested issues stimulate action among firms. Social contestation deals with degree of disagreement, or contestation, on the impact and resolution regarding the issue (Lamertz et al., 2003; Wartick & Mahon, 1994). Lamertz and colleagues (2003) define issues as contestations over meaning arising from disagreements over causes and consequences. Arrington and Sawaya (1984) claim that the very heart of IM is the reconciliation of conflicting interests on pivotal issues.
The framing of the contestation is what stimulates salience for that issue and serves as a strong determinant of firm action (Dutton & Duncan, 1987). Hoffman and Ocasio (2001) focus on how public attention is situated in the business press and find that when stark contradictions exist between the interpretation of a public event by outsiders, on one hand, and the prevailing industry identity, on the other hand, the contestation among these competing interpretations will be understood as an important issue to the firm. These contestations in meaning play out in the public realm and result in sustained and high levels of public attention (Hoffman & Ocasio, 2001). Bundy and colleagues (2013) similarly acknowledge that intensified media attention to a specific issue heightens a firm’s awareness and propose that conflicting accounts of the issue will increase the likelihood of higher firm engagement. Therefore, socially contested issues may be important to a firm simply because society suggests it is an important issue to be debated, and thus, the firm should allocate attention and resources to respond appropriately. Therefore we hypothesize the following:
Perennial Issues
Another central aspect of IM is how issues evolve and reoccur. Issue life cycle literature traces the trajectory of an issue over time (Mahon, Heugens, & Lamertz, 2004) and maps the issue according to when it first created awareness, when (and if) it gained momentum, and whether awareness was sustained over a period of time. The literature on perennial issues largely suggests that issues gain importance over time (e.g., the importance of secondhand smoke, see Derry & Waikar, 2008; Mahon & McGowan, 1996). Also, issues are often vague or ill-defined making them difficult to manage in the short term (Ansoff, 1980; Arcelus & Schaefer, 1982; Gerde & White, 2001), but as time goes on, information is more abundant and the costs, benefits, and alternatives can be more easily assessed (Ansoff, 1980; Derry & Waikar, 2008).
Conversely, others have recently suggested that perennial issues may depart from the predicted linear progression of most life cycle analyses and instead remain the object of public, firm, and industry attention over a sustained period (Hoffman & Ocasio, 2001; Lamertz et al., 2003), long after a recommended “solution” has been offered (Mahon, 1989; Mahon, Wartick, & Fleisher, 2001; Wartick & Mahon, 1994).
As recent research has suggested, it is likely that firms manage social issues that have sustained perennial prominence in a different manner than episodic issues (Hoffman & Ocasio, 2001; Lamertz et al., 2003). Likewise, if the IM process becomes more expensive over time, as the social issue becomes more visible to the public (Heath, 1997; Mahon, 1989; Zyglidopoulos, 2003), it may be in a firm’s best interest to detect and respond to the issue while it is still in its emergent stages (Hoffman & Ocasio, 2001; Lamertz et al., 2003; Renfro, 1993), especially given that a firm’s discretion generally recedes with time (Mahon, 1989). In all, this recent research suggests that the importance of a perennial social issue may decrease over time. Therefore we hypothesize the following:
Research Setting
Building on insights from the IM and salience literatures, we test these hypotheses in the context of shareholder resolutions. Shareholder resolutions are a common and direct means by which owners express dissatisfaction with some practice undertaken by the firm (Reid & Toffel, 2009). Shareholders “use (their) ownership position to actively influence company policy and practice” (Sjöström, 2008, p. 142) and as a mechanism that places pressure on firms to bring about change (Rao & Sivakumar, 1999). In response to a filing, firms are obligated to take action in a limited and observable time period (i.e., the first-quarter proxy season). Given the increasing attention on shareholder proposals by the public, the media, and investors, it is often not necessary for the proposal to win a majority vote to have influence on the firm (Aggarwal et al., 2014), yet filers see a high percentage of votes as a desired action taken by the firm (Logsdon & Van Buren, 2009).
Data and Measures
This study focuses on all companies listed in the Standard & Poor’s 500 Index from RiskMetrics Inc. database containing every shareholder resolution filed with U.S. firms between 1997 and 2009. The analysis includes all social issue resolutions filed during a 13-year time frame providing a broad and deep view of social issues that concern to shareholders.
Dependent variable
When a shareholder resolution is filed with the SEC, a firm has three options: seek to omit the resolution with permission from the SEC, send the resolution to a vote, or negotiate a withdrawal of the resolution with the filer. The authors refrain from suggesting that one type of a firm’s engagement is preferable to another. Based on previous research, a firm’s decision to negotiate a withdrawal represents the highest level of engagement because the firm must engage with the shareholder, usually through private negotiations, and come to a mutually agreed upon outcome (Clark & Crawford, 2012; Logsdon & Van Buren, 2009) regardless of reason for requesting a withdrawal.
The next highest level of engagement is putting the resolution on the proxy ballot for a vote by all registered shareholders because the proxy voting process is public and its results are easily tracked and analyzed. And finally, a firm may decide to try to omit the resolution from appearing on the proxy materials, by working with the SEC, the lowest level of engagement as these actions do not ultimately allow shareholders a true voice on the issue. In summary, the ordinal level of firm engagement is coded as a 2 if the firm withdraws the resolution, a 1 if the resolution is put on the ballot for a vote, or a 0 if the firm is successful in getting the resolution omitted.
Independent variables
The first variable examines the social salience of the issue brought by the shareholder through the filing of a resolution. Salience of an issue depends, in part, upon the audience evaluating it (Hoffman & Ocasio, 2001). That is, the media, public opinion, and special interest groups may accelerate the salience of an issue by continually putting the issue in the headlines (Aggarwal et al., 2014; Baron, 2010; Desai, 2011; Mahon, 1989). Firms gauge whether an issue is important by tracking public opinion (Aggarwal et al., 2014; Dutton & Duncan, 1987) and by paying attention to activists’ behavior (Proffitt & Spicer, 2006; Reid & Toffel, 2009) as a proxy for social issue salience.
Examining a broad set of social issues, the study operationalizes social issue salience as those issues included in the General Social Survey (GSS) across adjacent years in the RiskMetrics resolution data set. The GSS, conducted since 1972 by the National Opinion Research Council (NORC) is widely regarded as the single best source of data on societal trends in the United States with more than 5,416 variables captured since the inception of the GSS. 3 This large number of questions shows an overall comprehensive coverage of societal trends. Issues surveyed include tobacco, the environment (including global warming and energy), health care, sexual orientation, 4 genetically modified food, the unborn (abortion and stem-cell research), and equal employment. However, not all questions have been asked every year. Operationalizing salience of issues builds on recent research that has found that public opinion, as measured through either a public opinion survey or media coverage, at both the aggregate and firm level, influences shareholder voting (Aggarwal et al., 2014). The GSS survey encompasses both public opinion and knowledge of media accounts.
Data collected in the GSS have been used in diverse research ranging from entrepreneurship to sociology to economics and political science (see Meek, Pacheco, & York, 2010). In particular, the GSS has been used to study social norms relating to a variety of topics such as environmentalism (Meek et al., 2010), interdependence, and satisfaction (Kiecolt, 2003). The authors code salience as a “1” if the issue is included in the GSS, and “0” otherwise. For example, resolutions such as “endorse Ceres principles” and “report on greenhouse gas emissions” focus on improving the natural environment, whereas questions in the GSS ask respondents’ opinions on “improving and protecting the environment” across the years 1996-2010. Because social issues in shareholder resolutions are related to the construct behind the GSS question, they were matched with the question and codes with a “1” to indicate salience.
The variables social contestation and social consensus suggest that there are differing levels of agreement about a social issue in society. They attempt to capture both ends of a continuum that includes high and low levels of agreement where low levels of disagreement are the measure of consensus and high levels of disagreement are the measure of contestation—which cannot exist at the same time. The study operationalized social consensus as a subset of salient issues that had strong public sentiments and defined strong public sentiments by examining the responses to the relevant GSS questions. Strong sentiments may be composed of agreements around strongly positive or strongly negative views (i.e., consensus). Whereas when there is a mix of strongly positive and strongly negative views on a social issue, there is contestation. Specifically, questions asked by the GSS regarding public opinions were matched to resolution topics (e.g., the environment, sexual orientation, etc.). 5 All questions asked by the GSS were examined and a subset of GSS questions were selected as potential measures if they were (a) consistently fielded during the years of resolution data and (b) contain measures of attitude toward a topic as opposed to facts such as demographics. Next, the general idea behind each question was summarized, often through the variable name given in the GSS. When multiple questions regarding the same topic were asked, the more general attitude question was selected. For example, the GSS question (variable name, homosex) “about sexual relations between two adults of the same sex” with responses options ranging from “always wrong” to “not wrong at all” was matched to a resolution submitted to Berkshire Hathaway asking to “adopt sexual orientation anti-bias policy.” 6
Responses in the GSS were used to create a variable measuring the social consensus/contestation around a resolution. So, contestation was measured as the percentage point (%) difference from the mode response using a representative question for each category since 1996. To account for greater potential levels of contestation arising from varying number of response categories across GSS questions, this measure of disagreement was scaled by the ratio of total response categories divided by the total response categories minus 1. That is, contestation is calculated with the following expression:
Without scaling by the number of categories, an issue that is fully contested would appear to have greater disagreement as an artifact of the number of response options in a question. As a result, all measures of disagreement are continuous and bounded by the same theoretical minimum (0: maximum consensus) and maximum (1: maximum contestation). Analyzing this continuum using a squared term (disagreement2) introduces the potential of a “U” shaped relationship, which can be used to test the hypotheses that assert consensus and contestation are associated with an increase in the level of firm engagement.
A perennial issue examines those social issues that persist over time. The study operationalized perennial issues as previously voted upon resolutions that are resubmitted, given that they were not previously passed, and are eligible to be resubmitted. To be eligible for resubmission, the resolution must have reached a certain threshold of votes based on the number of times it has been submitted in the past 5 years. If the resolution has been submitted 1, 2, or 3 times in the past 5 years, it must have received at least 3%, 6%, or 10%, respectively, of the total votes in the last submission. Operationalizing perennial issues excludes all omitted resolutions because to be a perennial issue the resolution must be, by definition, previously voted upon (and not omitted). Because operationalizing perennial issues requires considering the previous 5 years of voting history, the models of perennial issues are truncated to the time period from 2002 to 2009.
Consistent with previous research on social issues, firm data by year was matched to the Compustat database to control for organizational size by examining sales, number of employees, assets, and earnings before interest and taxes (EBIT; McWilliams & Siegel, 2000; Waddock & Graves, 1997). The analysis also included a dummy variable indicating whether or not the firm is involved in a manufacturing sub-sector using the North American Industry Classification System (NAICS)
3-digit indicator because previous research has noted that an industry sector may or may not experience significant social problems (Waddock & Graves, 1997). Finally, the analysis controlled for the general type of resolution sponsor, that is, who proposed the resolution (i.e., religious, individual, special interest, or other), as has been suggested by previous research (Gillan & Starks, 2007).
Findings
The findings begin with a descriptive overview of the data, which contains 3,887 total social issue resolutions (see Table 1). Of these observations, 709 (or 19%) of the resolutions were omitted with 1,902 resolutions voted upon representing 63% of the non-omitted resolutions.
Social Issue Resolutions.
A preliminary analysis of the resolutions shows the top five social issues from 1997 to 2009 were (a) energy/environmental issues, (b) human rights, (c) domestic worker issues, (d) political and government relations, and (e) animal welfare (see Table 1). Overall, the total number of resolutions ebbs and flows year over year but has increased over time (Figure 1).
Resolutions by year.
Hypotheses Testing
Models displayed in Tables 2 to 6 are based on the three levels of engagement available to a firm regarding shareholder resolutions: omit, vote, or withdraw. To test Hypothesis 1, the analysis consisted of an examination of the ranked order of levels of engagement regarding the firm’s actions: omissions as the lowest level of engagement (coded “0”), those sent for a vote (coded “1”) to the shareholders, or withdrawal (coded “2”) as the highest level of engagement. We examined the likelihood of each action occurring using a generalized logit transformation, which simultaneously allows for more than two discrete categories in the modeled variable. Just as in binary logistic regression, coefficients in this model are in terms of log odds ratios. Vote, the most frequent outcome, is used as the baseline reference category. This means that negative coefficients in the “omit” (vs. vote) model indicate an increase in engagement, whereas positive coefficients in the “withdraw” (vs. vote) model indicate an increase in engagement.
Hypothesis 1—Social Issue Salience (Generalized Logistic Regression Parameter Estimates).
Note. DV = Dependent Variable.
Social Consensus/Contestation (Generalized Logistic Regression Parameter Estimates).
Hypothesis 2—Social Consensus (Logistic Regression Parameter Estimates).
Hypothesis 3—Social Contestation (Logistic Regression Parameter Estimates).
Perennial Issues (Logistic Regression Parameter Estimates).
Although these ordered relationships are often examined using ordinal regression with cumulative link models, such relationships require a proportional relationship between responses (Kutner, Nachtsheim, & Neter, 2004). In this case, the relationship between omit versus vote as described by the parameters would be assumed to be proportional to that of the relationship between vote versus withdraw. This proportional odds assumption may be untenable.
Rather than force this assumption on the categorical dependent variables, the study tested the hypotheses using a generalized logistic regression model. The process began by fitting a nominal response model to examine the effect of social issue salience on firm engagement while controlling for firm characteristics. First, a model comprised of the control variables was fit and compared with an intercept-only model. The likelihood ratio indicates the baseline model comprised of the control variables has a better fit relative to an intercept-only model (χ2 = 373.74, df = 16, p < .01).
Next, the dummy variable indicating inclusion in the GSS is added to the baseline model and a deviance test is conducted. This test shows adding the variable GSS significantly improves the model fit (χ2 = 43.64, df = 2, p < .01). Parameter estimates of this model are reported in Table 2 and are used to formally test Hypothesis 1 by examining the effect of salient issues, represented by the resolutions’ topic being included in the GSS. The authors find socially salient issues increase the level of firm engagement by increasing the likelihood of a vote versus omit (βGSS_Omit = −.24, Wald χ2 = 4.32, p = .04) and increasing the likelihood of a withdrawal versus vote (βGSS_Withdraw = .51, Wald χ2 = 30.82, p < .01) in support of Hypothesis 1.
Hypotheses 2 and 3 involve one independent variable measuring the degree of consensus to contestation around an issue. Taken together, these hypotheses would indicate a “U” shaped relationship, as high levels of both consensus and contestation are hypothesized to increase engagement. To test Hypotheses 2 and 3, a nominal regression model was used similar to the examination of Hypothesis 1. However, instead of using a dummy variable indicating inclusion in the GSS, the amount of disagreement was used as well as the square of this variable. The inclusion of these two variables, Disagreement and Disagreement2 significantly increase the model fit (χ2 = 29.49, df = 4, p < .01) and a Type 3 analysis of effects shows the squared term to be significant (χ2 = 6.67, df = 2, p = .04). Examining the components individually (Table 3), the authors find that both squared terms are individually significant (βDisagreement2_Omit = .002, Wald χ2 = 5.14, p = .02; βDisagreement2_Withdraw = .002, Wald χ2 = 5.73, p = .02), indicating a parabolic relationship across firm responses.
However, the significance of the squared term does not necessarily imply a “U” shape across the valid range of disagreement observed in the data set. Examining the disagreement measure, the authors find results as low as .345 and as high as .900 with an overall mean of .536. Viewing the expected probabilities derived from the model across the disagreement measure (at the average values for all the other variables), the authors find the parabolic parameter estimates are not centered at the mean of the observations but rather shifted to the left (Figure 2). This possibly indicates the relationship between disagreement and firm engagement is not “U” shaped. For example, examining the modeled probability a resolution will be omitted (Figure 2, solid black line), the effect appears to be closer to a “hockey stick” shape as the relationship appears flat at lower levels of disagreement. As a result, the authors conducted further analysis to formally test Hypotheses 2 and 3 by allowing for a non-symmetric relationship between disagreement and level of firm engagement.
Effect of social contestation/consensus—Generalized logit model.
To allow for this more flexible/non-symmetric relationship, the analysis used multivariate adaptive regression splines to fit a piecewise model as proposed by Friedman (1991). This type of analysis is important to the present study because it does not presuppose any structure on the relationship between variables. Multiple adaptive regression splines starts with a dependent variable related to an intercept term and then adds independent variable relationships over some or all of their range. If the independent variable is only found to be related to the dependent variable over a portion of the valid range (e.g., higher levels of disagreement increase the probability of increased engagement, whereas no effect is found at lower levels of disagreement), the model is said to have a knot, or elbow. These relationships are continually added to the model until the decrease in the overall lack of fit statistic reaches a minimum level of change (Kuhfeld & Cai, 2013). After these terms are added, the model is pruned using a backward elimination technique to create a parsimonious model and guard against over fitting. As a result, this modeling technique is well suited for uncovering the relationship between disagreement and level of firm engagement without imposing any a priori structure, such as a parabola.
This technique can be used with a binary variable using a logit transformation; the analysis modeled firm responses with two different models as a result of the three levels of firm response to mirror the processes that firms are likely to follow when deciding how to engage. Consequently, the first step is to separate the resolutions that are omitted from those that are not omitted and then examine those resolutions that are not omitted: either withdrawn or sent for a vote to the shareholders. If the firm does not omit the resolution, the firm can opt for negotiating an agreement with the proponent to withdraw. Or, the firm can put the resolution on the proxy ballot for a vote by all registered shareholders.
As such, the analysis modeled the probability that a resolution is not omitted: P(Omit = “no”). Then, the authors examined the probability that a resolution is withdrawn, given that it was not omitted: P(Withdraw = “yes” | Omit = “no”). If a resolution is not omitted, the firm has only two responses available—vote or withdraw—meaning that the opposite sign of the parameters in the second model is the probability that a resolution was voted on.
Multivariate adaptive regression splines is applied to these two variables using Proc Adaptivereg in SAS (Kuhfeld & Cai, 2013). The measure of disagreement and all control variables from Table 3 were eligible, but not forced, to enter the model. The relationship between disagreement and level of firm engagement that emerged has only one elbow for each dependent variable (Figure 3, Panels A and B). These results show that as social consensus increases (i.e., disagreement decreases), a firm is less likely to omit the resolution (Figure 3, Panel A), supporting Hypothesis 2. Figure 3 (Panel B) also shows that as social contestation increases (i.e., disagreement increases), a firm is more likely to negotiate a withdrawal of a resolution, given that it was not omitted.
Adaptive regression—Disagreement component.
These “hockey stick” style relationships are not the “U” shapes originally hypothesized, but rather may support a more nuanced level for firm engagement with respect to how the firm responds. To compare the results from this multivariate adaptive regression analysis with the previous generalized logit regression analysis (Figure 2), the expected probabilities, after a logit transformation, resulting from the adaptive model across the disagreement measure (at the average values for all of the other variables) are also charted (Figure 4). Interestingly, at lower levels of disagreement, the slope of the regression lines appear flat, indicating very little associated change. By letting the structure of the curves emerge through this modeling technique, it appears as if firms are sensitive to higher levels of social contestation.
Multivariate adaptive regression splines—Predicted probabilities.
To formally test Hypotheses 2 and 3, the authors created a dummy variable to dichotomize the disagreement measure. The cut points of the dummy variable correspond to the knot, or the elbow in the disagreement measure “hockey stick,” in both the “not omit” (vs. omit) and the “withdrawal” (vs. vote) models. Using a logistic regression analysis, the authors tested whether high levels of consensus increase the likelihood a resolution will not be omitted. Similarly, the authors tested whether high levels of contestation increase the likelihood a resolution will be withdrawn, given that it was not omitted.
First, a model composed of the control variables was constructed to examine the likelihood a resolution would not be omitted. This model was then compared with an intercept-only model. The likelihood ratio indicates the baseline model composed of the control variables has a better fit relative to an intercept-only model (χ2 = 81.85, df = 8, p < .01). Next, the authors added the dummy variable corresponding with a higher level of consensus and find an improved fit over the control-only model (χ2 = 5.95, df = 1, p = .01). Examining the parameter estimates (Table 4), the study finds that a higher level of consensus increases firm engagement by increasing the likelihood a resolution will not be omitted (βConsensus_NotOmit = .99, Wald χ2 = 6.31, p = .01). These results support Hypothesis 2.
When examining the effect of contestation on the likelihood a resolution will be withdrawn, given that it was not omitted, the study first finds the control variables increase the overall model fit compared with an intercept-only model (χ2 = 61.45, df = 8, p < .01). Adding a dummy variable to account for a higher level of contestation increases the model fit compared with the control-only model (χ2 = 15.05, df = 1, p < .01). Table 5 shows the parameter estimate associated with high contestation. In support of Hypothesis 3, the study finds a higher level of social contestation is associated with an increased likelihood a resolution will be withdrawn (vs. voted on) given that the resolution was not previously omitted (βContestation_Withdraw = .72, Wald χ2 = 14.73, p = .01).
Taken together the results for Hypotheses 2 and 3 suggest that social consensus may lead to higher engagement through a decreased likelihood the resolution will be omitted. At the same time, social contestation is associated with an increased likelihood of negotiating a withdrawal.
Finally, to examine the effect of perennial issues on level of firm engagement, perennial issues were operationalized as a social issue that must be voted upon previously and is not eligible to be omitted. Thus, firm engagement with perennial issues can only be examined through two responses: withdraw and vote. A logistic regression model was constructed to examine the likelihood a resolution would be withdrawn versus voted on. The analysis finds the control variables significantly increase the model fit relative to an intercept-only model (χ2 = 82.83, df = 8, p < .01). In addition, the authors find segmenting perennial issues from non-perennial issues increases the overall model fit (χ2 = 106.30, df = 1, p < .01). Examining the parameter estimates (Table 6), the study finds perennial issues decrease firm engagement by decreasing the likelihood a resolution will be withdrawn versus voted upon (βPerennial = −1.25, Wald χ2 = 97.49, p < .01). These results support Hypothesis 4. A summary of the findings can be found in Table 7.
Summary of Results.
Note. +/− significantly higher/lower likelihood versus reference category. NA = test not applicable; H = Hypothesis.
Discussion
This research sought to provide insight into the level of firm engagement for different types of issues fundamental to IM literature: salient, contested, consensus, and perennial. The study finds evidence to support increased firm engagement is associated with social issue salience, contestation, and consensus. Perennial issues were associated with lower levels of engagement.
High social issue salience (Hypothesis 1) is more likely to result in a higher level of firm engagement, empirically confirming prior theory (Bonardi & Keim, 2005; Bundy et al., 2013; Dutton & Jackson, 1987). The support for Hypothesis 1 is consistent with the predictions of Logsdon and Van Buren (2009) by finding that salient issues are more likely to be withdrawn. That is, the firm appears to be more willing to engage at a high level and directly with shareholders to negotiate a withdrawal of a resolution due to the issue’s perceived salience. Using shareholder resolutions as the study context also provides a unique setting to analyze what social issues firms are likely to engage with. By providing empirical testing of social issue salience, these findings support the assertion that firms do pay attention to issues that have a significant impact on them (Ansoff, 1980; Bundy et al., 2013; Johnson, 1983). By using this context, the study shows that firms gauge whether an issue is important by tracking public opinion (Aggarwal et al., 2014; Dutton & Duncan, 1987) and by paying attention to activists’ targeting behavior (Proffitt & Spicer, 2006; Reid & Toffel, 2009) to determine the degree to which it should respond to a request (Eesley & Lenox, 2006). This context then links firm action on social issues to agreement or disagreement on those social issues by society. In all, this examination strongly suggests that shareholder activism is embedded and influenced by the broader societal views, as well as the consensus or disagreement among those views.
The study results suggest social consensus (Hypothesis 2), may lead to higher engagement but only through a decreased likelihood that the resolution will be omitted. Social contestation (Hypothesis 3), or a high degree of disagreement over an issue, is also likely to produce higher levels of engagement (i.e., a withdrawal) if it passes the omission hurdle confirming some prior research (Bundy et al., 2013; Hoffman & Ocasio, 2001; Lamertz et al., 2003).
In finding support for both social contestation and consensus leading to higher levels of firm engagement, this study sheds new light on the IM and salience literature streams. Taking these results together indicates the omit versus not omit decision is different than a vote versus withdraw decision in that consensus is associated with increased engagement in the former whereas contestation is associated with increased engagement in the latter. With contestation, it may be the high uncertainty surrounding contested issues that make these issues compelling to the firm because it gives it a chance to shape the discussion of the issue directly with the proponent. Thus, contested issues, significant enough to pass the omission hurdle, may require a firm to take this type of nuanced approach to properly deal with the specific filer’s issue. Such private negotiations might indicate the firm’s desire to maintain maximum discretion (Mahon, 1989) with an opportunity to provide language and action showing that they have an inclination to advance mutual interests over time (Freeman, Wicks, & Parmar, 2004). When the firm engages with an issue of consensus, it may reflect the idea that once consensus takes hold of an issue, conformity extends to other firms making them less likely to diverge from the response norm (Bonardi & Keim, 2005; Dutton & Jackson, 1987). Likewise, the bulk of the shareholder activism literature has been focused on social issues in a single group, such as environmental (Goranova & Ryan, 2014) versus the variety of social issues captured by the present study’s constructs. Therefore, analyzing each of these decisions uniquely is important for IM, salience, and corporate governance researchers.
For social issues that are perennial (Hypothesis 4), the firm is more likely to put the resolution up for a vote rather than negotiating a withdrawal, representing a lower level of engagement as hypothesized. Traditionally, the literature on perennial issues has suggested that issues gain importance over time (Derry & Waikar, 2008; Mahon & McGowan, 1996). This study’s findings challenge this line of thinking. Instead, the authors find evidence to suggest that when an issue is reoccurring the firm is less likely to engage with a perennial issue indicating firms choose to respond while issues are still in their emergent stages (Hoffman & Ocasio, 2001; Lamertz et al., 2003; Renfro, 1993), in contrast to previous findings suggesting that an issue gains importance over time (Derry & Waikar, 2008; Mahon & McGowan, 1996). This action is likely to be in recognition that a firm’s discretion recedes with time (Heath, 1997; Mahon, 1989; Zyglidopoulos, 2003) or, more specifically, that a firm sees a repeatedly filed resolution as a signal of sorts that the issue is losing, rather than gaining, its potential impact. Prior shareholder resolution research has not considered the effect of such perennial issues.
Limitations and Future Research
Although this study used the shareholder resolution context as an insightful proxy for the way in which firms engage with issues that have differing characteristics, namely social salience, consensus, contestation, and perennial, it has not tested the possibility that managers can actively manipulate the meaning of an issue making it more or less salient (Dutton & Jackson, 1987), even while it has accounted for the closing of a window in that effort by looking at perennial issues (Zyglidopoulos, 2003). Further research is encouraged into what additional drivers there might be for determining firm-level issue salience (Bonardi & Keim, 2005; Bundy et al., 2013; Dutton & Jackson, 1987) and how it may or may not be connected to symbolic management through decoupling tactics in the short term (Bundy et al., 2013; Westphal & Zajac, 1998) or to longer term social movement perspectives (Proffitt & Spicer, 2006).
The authors also recognize that there may be other ways to determine social issue salience than shareholder resolutions and the GSS and that it may not be a perfect match, even if there is significant overlap in the underlying concepts, because the GSS is not primarily designed to be matched with a data set of resolutions. Although the GSS is designed to study social trends by holding samples and measures constant over time (Marsden, 2012), future work could attempt to triangulate the influence of social issues on firm engagement. At the same time, the authors offer these as appropriate measures for the study because it enables testing of issues that actually get attention from the firm.
The ordering of firm responses is based on previous research indicating a withdrawn resolution represents the highest level of engagement (Clark & Crawford, 2012; Logsdon & Van Buren, 2009) and an omitted resolution represents the lowest level of firm engagement. However, the results show there may be different processes associated with taking a resolution beyond the base level of engagement (i.e., omit) versus negotiating a resolution withdrawal. For example, it is conceivable that a firm may not be able to omit a resolution due to certain SEC rules, and therefore, can only put the resolution up for a vote or negotiate to have the resolution withdrawn. The data set used here does not indicate whether the firm attempted to have the resolution omitted or attempted to negotiate a withdrawal.
Future research could also focus on additional nuances of firm engagement with shareholder coalitions by considering intra-group differences such as shareholders having multiple roles (e.g., a shareholder, activist, and employee) in much the same way that Mitchell, Agle, and Wood (1997) identified various stakeholder attributes to provide managers a way to prioritize them. By focusing on levels of engagement, the authors leave to future research if a firm’s decision to opt for a low level of engagement, such as putting the resolution up for a vote rather than a high-engagement withdrawal, is prudent or effective. Similarly, the authors leave to future research the examination of the conditions under which a firm will use an omission strategy, which remains under-examined.
Another promising area might consider what the shareholder is specifically asking the firm to do (e.g., label, report, review, change operations, adopt principles) with a given issue because a shareholder resolution about the same issue could have a different “ask” attached to it (Clark, Griffin, & Bryant, 2013). Additional nuances about the level and nature of issue salience have been underexplored (Bundy et al., 2013). In a similar vein, a resolution is only one activity that the firm engages with a shareholder about (Goranova & Ryan, 2014); other activism tactics could be considered and could apply equally to the construct of levels of engagement.
Last, to broaden the discussion of shareholder-initiated issues to the ongoing work in corporate governance, the analysis provides additional future research directions. Because it has not specifically tested how similar firms, industries, or boards respond differently to the same issues, the authors suggest this line of inquiry could be fruitful to better understand the filtering process of the firm (see Dutton & Jackson, 1987; Eesley & Lenox, 2006; Rehbein & Schuler, 1999) at the board level. And, given that the study has not considered the effect of issue salience on the shareholder voice research stream (Bebchuk, 2005), or on the current and ongoing debate about shareholder primacy versus director primacy (Bainbridge, 2012), the authors suggest that considering whether and how resolutions shift the power away from boards is an important future direction. Yet, because this research focuses on the firm–shareholder relationship, it does suggest a bridge can be formed between advocates of both the firm and the shareholder perspective. The authors welcome future work that attempts to puzzle out these questions.
Conclusion
This research represents a new and important setting to test the IM and salience literature because a firm must publicly respond to each shareholder resolution, has only a limited time frame in which to do so, must choose among a constrained set of actions, and may need to respond to the same issue repeatedly. This research setting helps to illustrate that scholars need to consider IM from a more holistic perspective, not just from the firm’s perspective, by acknowledging society as more than a single, monolithic variable.
To the authors’ knowledge, such fined-grained insight into expected levels of firm engagement with issue types has not been put forth previously. Yet, firms and shareholders exist in an increasingly complex relationship, much of which is playing out publicly through activism. As a result, a firm’s response to social issues brought by shareholders has never been more important. It is through nuanced investigations such as this initial foray into issue types and the levels of firm engagement that the authors hope to make progress in understanding if, when, and how firms are more likely to meaningfully engage with shareholders on the vital issues of our time.
Footnotes
Acknowledgements
The authors express their sincere thanks to Richard Priem and Jeffrey Harrison and the session participants at the 2013 Strategic Management Society conference where an earlier version of this article received the Best Paper Proposal Award. They also thank Jill Brown, Kathy Rehbein, Doug Schuler, Andrew Millington, and Johanne Grosvold, participants at the University of Bath 2014 Corporate Governance Workshop along with John F. Mahon and Anu Phene for comments on earlier versions of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.