The Role of Deliberative Mini-Publics in Improving the Deliberative Capacity of Multi-Stakeholder Initiatives

Abstract

Multi-stakeholder initiatives (MSIs)—private governance mechanisms involving firms, civil society organizations, and other actors deliberating to set rules, such as standards or codes of conduct, with which firms comply voluntarily—have become important tools for governing global business activities and the social and environmental consequences of these activities. Yet, this growth is paralleled with concerns about MSIs’ deliberative capacity, including the limited inclusion of some marginalized stakeholders, bias toward corporate interests, and, ultimately, ineffectiveness in their role as regulators. In this article, we conceptualize MSIs as deliberative systems to open the black box of the different elements that make up the MSI polity and better understand how their deliberative capacity hinges on problems in different elements. On the basis of this conceptualization, we examine how deliberative mini-publics—forums in which a randomly selected group of individuals from a particular population engage in learning and facilitated deliberations about a topic—can improve the deliberative capacity of MSIs.

The governance of business activities along global production networks (Levy, 2008) is crucial to dealing with the negative social and environmental side effects these activities create or reinforce (Djelic & Quack, 2018; Djelic & Sahlin-Andersson, 2006). One such governance mechanism is multi-stakeholder initiatives (MSIs) (Mena & Palazzo, 2012; Rasche, 2012). MSIs are private regulatory instruments that govern global business activities to reduce their negative societal impacts through voluntary compliance by firms to social and/or environmental standards or codes of conduct throughout their supply chains (Bartley, 2007; Scherer & Palazzo, 2007; Vogel, 2010). For example, the Forest Stewardship Council (FSC) is an MSI that governs the sustainable management of forests worldwide. ¹ In contrast with industry self-regulation (King & Lenox, 2000) or business-driven programs (Fransen, 2012), which the corporate sector sets up mostly for public relations and managing the image of their industries (Marques, 2017), MSIs involve stakeholders from two or more sectors (for-profit, public, and nonprofit) in their formal decision-making processes.

Yet, even MSIs that involve a variety of stakeholders and try to balance different interests in their procedures have been criticized when it comes to governing global business activities (e.g., Banerjee, 2018). Concerns have been raised about the excessive role MSIs give to corporate actors (Hussain & Moriarty, 2018) and their lack of impact for disadvantaged actors across global production networks, such as workers in garment factories or Indigenous communities (de Bakker, Rasche, & Ponte, 2019; MSI Integrity, 2020). MSIs’ governance role is contested in part because deliberations between stakeholders on rule setting often do not meet standards expected of regulatory actors. Yet, many political theorists view the quality of deliberations as essential for legitimate, just, and equitable governance (Habermas, 1996).

According to deliberative democratic theory, the quality of deliberations can be assessed in terms of their deliberative capacity (Dryzek, 2009). In MSIs, enhancing deliberative capacity means, for instance, the broad inclusion of different types of stakeholders affected by the regulated business activities, the structuring of fair and reciprocal deliberations, and making sure these deliberations are consequential in terms of effectively curtailing the problems created by the business activities (e.g., Arenas, Albareda, & Goodman, 2020; Mena & Palazzo, 2012; Scherer & Palazzo, 2011). The greater the deliberative capacity of MSIs is, the better they are at governing global business activities that no nation-state or company can solve alone. Yet, the deliberative capacity of MSIs has been questioned, as they only include marginalized actors in a limited fashion (Banerjee, 2018), tend to favor corporate interests (Hussain & Moriarty, 2018), and have limited regulatory impact (de Bakker et al., 2019; LeBaron & Lister, 2022).

Given these concerns, enhancing the deliberative capacity of MSIs is crucial if they are to effectively curtail the social and environmental problems created by global business activities. Prior research has already begun addressing this problem (Schormair & Gilbert, 2021). For instance, Arenas and colleagues (2020) discuss the need for some degree of dissensus in deliberations between stakeholders and explore how such fruitful contestation can be introduced in MSIs. Others have examined how the initial lack of mutual understanding between for-profits and nonprofits in MSIs can be overcome by fostering mind-sets and values geared toward finding common ground (Soundararajan, Brown, & Wicks, 2019).

In this article, we advance this research by asking the question, How can deliberative mini-publics (DMPs) improve the deliberative capacity of MSIs? To answer this question, we first adopt a deliberative systems perspective. This perspective takes a fine-grained view of polities, viewing them as systems made of different elements (such as public space or accountability) and emphasizing the division of deliberative labor across these elements (Elstub, Ercan, & Mendonça, 2019; Mansbridge et al., 2012). Deliberative capacity can thus be assessed in a more fine-grained way at the levels of the system and its elements (Curato, 2015; Dryzek & Stevenson, 2011; O’Flynn & Curato, 2015). The deliberative systems perspective is applicable to a wide range of polities, including global climate governance, transnational governance networks, and liberal democratic nation-states (Dryzek, 2016; Dryzek & Stevenson, 2011; Mansbridge et al., 2012). Because MSIs formally include different stakeholders in their governance structures, they meet the threshold of being “loosely democratic” and are thus suitable for analysis as deliberative systems (Mansbridge et al., 2012: 7–8).

DMPs are a prominent and influential democratic innovation (Elstub, 2014) that have received significant attention in research on deliberative systems given their potential to improve deliberative capacity (Curato & Böker, 2016; Felicetti, Niemeyer, & Curato, 2016). DMPs are “bodies comprised of ordinary citizens chosen through near random or stratified selection from a relevant constituency, and tasked with learning, deliberating, and issuing a judgement about a specific topic, issue, or proposal” (Warren & Gastil, 2015: 562). They advance a novel form of nonelectoral representation, “citizen representatives,” whereby participants are authorized to undertake a representative function but are not beholden to electoral accountability (Urbinati & Warren, 2008; Warren, 2008). “Standing for” (Pitkin, 1967: 60) the broader population, these representatives create presence through their descriptive representativeness (James, 2008; Zakaras, 2010), whereby they “are in their own persons and lives in some sense typical of the larger class of persons whom they represent” (Mansbridge, 1999: 629).

To build a case for how DMPs can enhance the deliberative capacity of MSIs, we first conceptualize MSIs as deliberative systems and discuss challenges MSIs often face in terms of meeting criteria of deliberative capacity. We then theorize how DMPs, when carefully translated to the context of MSIs through specific design parameters, can address deficiencies in MSIs’ deliberative capacity. In doing so, we make two major contributions to the literature on MSIs. First, we advance a new way of seeing and improving the deliberative capacity of MSIs. Indeed, by explicitly adopting a systems perspective on MSIs (which, to date, has only been alluded to in the literature, e.g., Arenas et al., 2020; Soundararajan et al., 2019), we open the “black box” of MSIs and show the variegated forms of and settings in which deliberations take place not only within MSIs, among their formal members, but also beyond, with other external actors (Fougère & Solitander, 2020). This allows for a more fine-grained analysis of deficits in deliberative capacity at different levels of the system. On the basis of this analysis, MSIs can adopt one or more of the five uses of DMPs we developed, each of which can improve the deliberative capacity of a focal element of the system and that of the system as a whole.

Second, we contribute to our understanding of democratic representation in MSIs, a topic which to date has received minimal explicit attention (for an important exception, see Reinecke & Donaghey, 2021). MSIs tend to represent the many different stakeholders affected by their operations by following a form of structural representation in their governance structures (Reinecke & Donaghey, 2021), bundling similar interests (e.g., corporate, environmental, Global South) together in constituencies and authorizing representatives (i.e., through election or appointment) to represent these constituencies. Yet, some stakeholders cannot be represented structurally and, rather, rely on representation through self-appointed representatives who make representative claims, such as when nongovernmental organizations (NGOs) represent the natural environment (Baur & Palazzo, 2011; Reinecke & Donaghey, 2021). These methods of representation have been criticized for leading to deep-seated inequalities in terms of how different actors and interests are represented in MSIs (Banerjee, 2018; Schouten, Leroy, & Glasbergen, 2012). Through our elaboration on the use of DMPs in MSIs, we advance a complementary form of nonelectoral representation that can alleviate some of these concerns and provide additional benefits to MSIs.

Our article proceeds as follows. We begin by briefly overviewing MSIs, to then conceptualize them as deliberative systems and analyze their deliberative capacity. We then introduce DMPs and their key features, which sets us up for our translation of DMPs to the context of MSIs and our theory development about five uses through which they can improve MSIs’ deliberative capacity. We conclude with a discussion of our contributions to research and directions for future research.

THE DELIBERATIVE CAPACITY OF MULTI-STAKEHOLDER INITIATIVES

Multi-Stakeholder Initiatives

MSIs are typically defined as formal “coalitions of nonstate actors [that] codify, monitor, and in some cases certify firms’ compliance with labor, environmental, human rights, or other standards of accountability” (Bartley, 2007: 298). Core aspects of MSIs thus include rule making, the monitoring of rule taking and potential sanctioning of noncompliance, and the participation of actors in a deliberative platform aimed at exchange and learning between diverse members (Palazzo & Scherer, 2010). Compliance by firms is voluntary, and MSIs thus rely on market mechanisms to ensure compliance, such as certification, product labeling, or reputational threat of activism (Vogel, 2010). A key distinctive feature of MSIs is a degree of inclusiveness of a variety of actors, including actors from two or more sectors (corporate, public, and civil society), in their decision-making and governance processes (Mena & Palazzo, 2012). This means that corporations share decision-making power, to various extents, with noncorporate stakeholders, such as governments, unions, NGOs, scientific organizations, and Indigenous communities.

The FSC provides a good example. It elaborates principles for sustainable forest management that inform standards with which member firms from various industries (from logging to pulp and paper to home improvement) decide to voluntarily comply and with which to get certified through an independent verification process by accredited third parties. The principles, standards, and verification procedures, along with other rules and processes, are elaborated and decided upon by various actors that are members of the FSC, including firms, social and environmental NGOs, governmental agencies, representatives of Indigenous communities, academics, and union representatives. These members are part of the General Assembly, representing equally industry, social, and environmental actors, which elects the Board of Directors that leads the initiative. The FSC’s operations are run by an Executive Team and a Secretariat, along with many country chapters.

MSIs are varied, but as the FSC illustrates, their structure typically includes 1) an assembly of members (or their delegates) who, as the highest decision-making body of the initiative, make rules for corporate conduct; 2) a board, usually made of representatives of these members and often elected by the assembly; 3) a secretariat or executive team that deals with the implementation of the rules and day-to-day operations; 4) several different working groups or committees pertaining to different aspects of the MSI (such as rule-making processes and monitoring mechanisms); and, finally, 5) some sort of oversight mechanisms regarding the implementation of the rules by firms (including monitoring, complaints, and sanction procedures).

It is important to distinguish different types of actors related to MSIs. First are members: the stakeholders that are involved in its governance and participate in the rule- and decision-making processes (point 1); they usually are the ones that are directly affected by or affect the issues governed by the MSI. Second are rule takers: the corporations that voluntarily comply with the MSI rules. Rule takers (or some of them) are often MSI members, too, involved in MSI governance, and are thus rule makers as well. Third are monitors: the actors tasked with auditing rule takers’ adoption of rules, oftentimes NGOs or consulting firms (Gereffi, Garcia-Johnson, & Sasser, 2001). They can also sometimes be members. Finally are otherwise affected stakeholders: those that are not rule takers or rule makers but are nevertheless affected by the issues addressed by the MSI (Brès, Mena, & Djelic-Salles, 2019; Martens, van der Linden, & Wörsdörfer, 2019). In the case of the FSC, these could be some consumers, Indigenous groups, features of the natural environment, or governments that are not members of the FSC.

This connects to the issue of representation of different stakeholders in MSIs, which generally takes the form of structural representation (Reinecke & Donaghey, 2021). This conception emphasizes the formal relationship between the represented and their representatives, focusing on how representatives are authorized and held accountable (Pitkin, 1967). MSIs’ member organizations usually appoint a delegate to act on their behalf at meetings like those of the general assembly, enacting the organization’s instructions and mandates (Pitkin, 1967). MSIs can have different chambers in their assembly representing particular constituencies, such as sectors, stakeholder groups, or interests (e.g., environmental stakeholders, workers’ stakeholders). This means that, indirectly, interests are bundled or aggregated by predetermined and similar interests among stakeholders. For instance, the FSC divides its assembly into equal parts between social, environmental, and economic actors. The respective members of each chamber then elect four representatives on the board for a total of twelve directors. As this example illustrates, beyond the assembly, MSIs also have a board of representatives, who serve as “trustees” (Pitkin, 1967) of a broad constituency—in this case, these “bundled” stakeholders that elect one or more representatives on the board. These representative structures mimic the “standard account” that sees representation as a principal–agent relationship whereby representatives are expected to—and, through elections, often incentivized to—advance the interests of citizens in specific constituencies (Montanaro, 2012; Urbinati & Warren, 2008). MSIs, though, differ from this standard account in important ways, including the fact that they do not govern over a clearly defined territory with citizens/residents (Bäckstrand, 2006), their population is less clearly defined and more homogeneous (Reinecke & Donaghey, 2021), and they rely on market mechanisms and potential sanctions to foster compliance (Mena & Palazzo, 2012). These peculiarities also mean that, in addition to structural representation, MSIs involve self-appointed representatives like NGOs and activists that make representative claims on behalf of some stakeholders (e.g., the natural environment, workers) but are not formally authorized to act in a representative capacity (Baur & Palazzo, 2011; Reinecke & Donaghey, 2021).

The Elements of Deliberative Systems

A major advancement offered by the deliberative systems perspective is that it focuses on how deliberation takes place across different elements of a system. As Elstub and colleagues (2019: 139) put it, this entails “an understanding of deliberation as a communicative activity that occurs in multiple, diverse yet partly overlapping spaces, and emphasizes the need for interconnection between these spaces.” Dryzek and collaborators have been particularly influential in terms of developing a more general model of a deliberative system (Dryzek, 2009, 2010; Dryzek et al., 2019; Dryzek & Stevenson, 2011; Stevenson & Dryzek, 2014). This is highly adaptable, as scholars can focus on the elements deemed most relevant to particular contexts being studied (e.g., Curato, 2015; Stevenson & Dryzek, 2014). In conceptualizing MSIs as deliberative systems, we focus on what we see as the five most relevant elements of the system: public space, empowered space, transmission mechanisms, accountability mechanisms, and meta-deliberation. In the context of the liberal democratic nation-state, the public space involves spaces like cafés, internet forums, and public hearings with few or no constraints on who can participate or what they can say. The empowered space entails spaces, such as cabinets, legislatures, and courts, that are empowered to make collective decisions. Transmission mechanisms are the means through which the public space can influence decision-making in the empowered space, including political campaigns, personal connections, and broader cultural changes. Accountability mechanisms are the means through which the empowered space provides an account to and can be held accountable by the public space. These can include public forums and the threat of electoral defeat. Finally, meta-deliberation entails the capacity of a system to self-examine (e.g., constitutional reform) and adapt. Figure 1 provides a schema of the connections among these various elements in the case of the FSC.

Figure 1: The FSC as a Deliberative System

Note. Adapted from O’Flynn and Curato (2015) and Dryzek and colleagues (2019).

We now turn to our conceptualization of MSIs as deliberative systems. Table 1 summarizes this content, with the letters in the left-hand column corresponding to Figure 1 and providing examples from the FSC. The public space of MSIs is broad and encompasses deliberations in many sites that can be open to participation to different extents. For instance, corporations—whether or not MSI members—often exchange in the context of chambers of commerce, industry associations, or think tanks, whereas NGOs concerned with the environment or trade unions often discuss in conferences or workshops. The public space includes sites that are open to more diverse participation, such as online open spaces; media of various sorts; business or economic conferences; and various types of academic, trade union, and nonprofit conferences and seminars.

Table 1: MSIs as Deliberative Systems and Common Pitfalls in Their Elements’ Deliberative Capacity

^a See https://ga2017.fsc.org/what-are-the-fsc-chambers/.

^b See https://fsc.org/en/governance-strategy.

^c See https://fsc.org/en/about-us.

^d See https://fsc.org/en/chain-of-custody-certification.

^e See https://fsc.org/en/innovation.

^f See https://fsc.org/en/unacceptable-activities.

^g See https://fsc.org/en/governance-strategy.

The empowered space of an MSI corresponds to the formal structural bodies described earlier (assembly, board, secretariat, and working groups) and is similar to the legislative, executive, and judicial organs that are involved in the collective decisions of liberal democratic nation-states.

Transmission mechanisms in MSIs typically consist of pressure from various groups from the public space targeted at actors in the empowered space, most often the board and secretariat. MSIs are usually under a lot of direct pressure from various groups, in particular, economic ones (e.g., business associations; see Fransen, 2012) or civil society representatives of various social and environmental interests (e.g., Mena & Waeger, 2014). Indeed, as mentioned previously, MSIs rely on market mechanisms, such as reputational threat, to increase compliance (Vogel, 2010). For instance, the FSC is regularly criticized by various civil society actors, and some of these critiques can be found on the FSC-Watch website. ² Pressure can be indirect, too, for instance, the result of broader societal movements (e.g., Extinction Rebellion). In addition, formal representative processes, such as elections to the board, provide another transmission mechanism for stakeholders to push their agenda on the empowered space of the MSI.

In terms of accountability, MSIs have a number of mechanisms to make the empowered space accountable to MSI members and the public more generally. Usually, MSIs and their secretariats publicly report on their activities, performance (understood as adequate compliance by participating firms), and progress. Transparency and access to information are, therefore, key to these indirect accountability mechanisms (Auld & Renckens, 2017; Bäckstrand, 2006; Hale, 2008). For instance, the FSC publishes an annual report, but also position papers, reports by certification bodies on corrective action requests, and overviews of scientific studies on the FSC’s impact. ³ To verify compliance, as described earlier, there are usually a number of monitoring procedures in MSIs to verify and ensure compliance, as well as remedies for noncompliance when found. Sanctions are also sometimes in place, such as decertification.

Finally, meta-deliberation refers to the ability of a polity to self-reflect and change itself. MSIs vary in the extent to which they self-reflect, although this element is not prominent in MSIs. Usually, mechanisms of meta-deliberation involve the ability of the assembly to change some processes and rules. For instance, the participants in the FSC Assembly can propose motions to change FSC rules before the assembly every three years.

The Deliberative Capacity of Multi-Stakeholder Initiatives

A parallel area of inquiry on deliberative systems has focused on how to evaluate them and their various elements. In this article, we rely on the notion of deliberative capacity, defined as “the extent to which a political system possesses structures to host deliberation that is authentic, inclusive, and consequential” (Dryzek, 2009: 1382). Inclusiveness captures the breadth of discourses and interests circulating in a setting; authenticity captures the extent to which the deliberation “induce[s] reflection noncoercively, connect[s] claims to more general principles, and exhibit[s] reciprocity” (Dryzek, 2009: 1382); and consequentiality (sometimes equated with the notion of decisiveness, e.g., Curato, 2015) captures the direct and indirect ways in which these authentic and inclusive deliberations can impact collective decision-making and outcomes. Schouten and colleagues (2012) further usefully distinguish between output and outcome consequentiality in MSIs: output consequentiality refers to whether deliberations meaningfully affect MSIs’ decisions, whereas outcome consequentiality refers to the actual impact MSIs have on the industries they attempt to regulate.

The criteria of deliberative capacity can be used to make assessments of the system as a whole or to make a more fine-grained assessment of its elements. The former approach has been undertaken in some limited research on MSIs to date (Schouten et al., 2012; Soundararajan et al., 2019). While the deliberative capacity of different MSIs varies widely, general criticisms have been leveraged against them. In terms of their inclusiveness, MSIs are said to exclude or not include meaningfully marginalized stakeholders, such as Indigenous people or local communities (Banerjee, 2018). In terms of their authenticity, while, in general, MSIs fare fairly well in terms of reflection on preferences and reciprocity between actors (see, e.g., Moog, Spicer, & Böhm, 2015; Schouten et al., 2012), they have been criticized for lack of fairness in deliberations, mostly because of overrepresentation of corporate interests and excessive corporate power (Banerjee, 2021; Hussain & Moriarty, 2018). MSIs have also been criticized for low consequentiality (for an overview, see de Bakker et al., 2019; LeBaron & Lister, 2022). Indeed, some MSIs tend to exclude marginalized discourses—especially more radical views of sustainability—and thus have been criticized for their lack of output consequentiality, as these discourses do not meaningfully affect MSI decisions (Schouten et al., 2012). In terms of outcome consequentiality, one prominent critique is the low number of firms from a given industry that actually comply with MSI rules (e.g., Moog et al., 2015; Schouten et al., 2012). Furthermore, the lack of substantial changes in areas that MSIs regulate has also been pointed out. For instance, the FSC has been criticized for allowing human rights violations, such as the rape of women and girls in the forest industry, because of the lack of clearly assigned responsibility for this issue in the FSC’s processes (Whiteman & Cooper, 2016; see also Bartley, 2018).

Yet, there is value in a more fine-grained assessment of the deliberative capacity of the system’s distinct elements. Premised on the idea that “the deliberative system as a whole is diminished by any nondeliberative substitute for any element” (Dryzek, 2009: 1386), an analysis of a system’s elements avoids the potential pitfall of purely system-level evaluations. Indeed, there is a risk that a deliberative system could be assessed as having a high level of deliberative capacity without any meaningful deliberations occurring within it (Owen & Smith, 2015). Per Dryzek (2009; see also Curato, 2015), consequentiality is usually only examined in terms of the system as a whole, because it assesses the link between deliberations throughout the system to its decisions and outcomes. The empowered space, the public space, transmission, and accountability should be assessed in terms of their authenticity, and the public space and the empowered space should be assessed in terms of their inclusiveness and authenticity. Meta-deliberation can be assessed at the level of the system based on its authenticity and inclusiveness (Holdo, 2020; Thompson, 2008). We now examine common pitfalls in MSIs to demonstrate how this more fine-grained analysis could be undertaken. This is summarized in the right-hand columns of Table 1. These complement the system-wide pitfalls in consequentiality discussed earlier.

The empowered space of MSIs is the element that has attracted the most criticism. In terms of inclusiveness, the formal bodies of MSIs usually present high barriers to entry for marginalized stakeholders (Miller & Bush, 2015), which is linked to inadequate representation of interests (Baur & Palazzo, 2011; Reinecke & Donaghey, 2021). For instance, it is often taken for granted that environmental NGOs (such as Greenpeace or the WWF) will represent the interests of Earth and its ecosystems, yet these organizations necessarily prioritize some environmental issues and interests over others (e.g., Bendell, 2005). With regard to authenticity, the empowered spaces of MSIs are often structured in an egalitarian manner; yet, this structure has been criticized for being biased toward economic interests in practice. For example, even though the FSC structurally balances interests, it has been critiqued for favoring Northern, economic, and corporate interests (Dingwerth, 2008; Moog et al., 2015). This bias translates into a lack of fairness, as MSIs do not tend to make participation in deliberations easier for noncorporate actors, for example, in terms of language (Roussey, Balas, & Palpacuer, 2022). Beyond structural issues, the authenticity of deliberations in the empowered space is critiqued for being prone to coercion, in particular, excessive corporate power over noncorporate stakeholders (Taylor, 2005), the co-optation of minority interests (Dawkins, 2021; Maher, 2019), and the silencing of dissenting opinions (Brown & Dillard, 2013).

The public space of MSIs is also sometimes pointed out as lacking inclusiveness and authenticity. In terms of inclusiveness, the public space of MSIs is usually quite global and encompasses various sites. Yet, some actors are excluded from deliberating in those sites, such as disenfranchised groups like Indigenous communities in remote areas or workers at the bottom of supply chains (Banerjee, 2021). In terms of authenticity, these sites have been shown to be dominated by a neoliberal view that favors corporate interests over those of other stakeholders (Fougère & Solitander, 2020). Even if divergent views exist, deliberations tend to be located in particular sites and not shared more broadly to challenge mainstream views (Schouten et al., 2012; Soundararajan et al., 2019).

Prior work suggests that the authenticity of transmission mechanisms in MSIs is limited. Deliberations in the public space tend to be dominated by corporate and economic interests over those of other stakeholders, and therefore views that diverge from these interests tend not to be shared more widely in the public space. This means that when the outcomes of public space deliberations are transmitted to the empowered space of MSIs, they tend to be low in authenticity, because deliberations about what needs to be transmitted are co-opted by corporate actors and divergent views are silenced. This is reinforced by the power of industry associations and other corporate groups that pressure the empowered space through lobbying—a key transmission mechanism—more so than other actors (e.g., Marques, 2017). Although noncorporate stakeholders are able to exert some pressure on the empowered space, mostly through activism—which can be directed at the MSI or at participating firms (Mena & Waeger, 2014)—corporate lobbying usually dominates (Hussain & Moriarty, 2018).

Accountability is also criticized in MSIs in terms of its authenticity. Reporting by the secretariat (e.g., annual reporting) tends to be dictated by the dominant coalition in an MSI, which is often the corporate sector. Similarly, reports on audits by certification bodies are often one-sided and relatively superficial, as the monitoring system on which MSIs usually rely has shown its limits in terms of audit fatigue and lack of resources to investigate noncompliance in depth (Marshall, McCarthy, McGrath, & Harrigan, 2016). As mentioned earlier, symbolic adoption and mostly surface-level compliance are problematic too (Behnam & MacLean, 2011). Concerns about the lack of adequate grievance mechanisms, effective means of sanctioning, and the lack of independent sanctioning bodies have also been raised (MSI Integrity, 2020).

Finally, meta-deliberation is not usually prominent in MSIs, as they do not necessarily have the capacity to fundamentally change their structures or practices. As Barlow (2021: 16) argues, “dramatic design improvements are necessary before MSIs will fully achieve their potential.” Yet, such structural change is often brought about by the secretariat, which can be dominated by corporate actors. When such capacity exists in a more inclusive manner (such as the FSC’s potential change in statutes voted on by members at the General Assembly), these procedures are relatively cumbersome and, when used, do not change the system substantially. This means that the inclusiveness and authenticity of accountability in MSIs are usually low. For instance, when Greenpeace put a motion to the FSC General Assembly to change the requirements for certifications in primeval forests, some corporate actors threatened to leave the FSC (thus leaving the initiative much less impactful), thereby diluting the potential for the rule to protect these types of forests (Arenas et al., 2020).

DELIBERATIVE MINI-PUBLICS

Before turning to how DMPs could be used to improve the deliberative capacity of MSIs, we first introduce DMPs. They are part of a large and growing family of democratic forums intended to foster public deliberation. Pateman (2012) highlights four central features of DMPs on which we expand in the following pages: they are initiated or commissioned by a body to deliberate about a particular topic, they are selected through some form of random selection, they include many practices to foster deliberation, and their outputs are compiled and disseminated. Among these features, random selection distinguishes DMPs the most from other deliberative forums (like National Issues Forums, Study Circles, and the 21st Century Town Meetings). Other deliberative forums use one of many other selection techniques, with voluntary self-selection being the most common (Fung, 2003). Beyond these central features, DMPs are remarkably versatile in terms of other dimensions, such as size, medium, and decision rule (e.g., Fishkin, 2009; Paulis, Pilet, Panel, Vittori, & Close, 2021).

Initiators and Conveners

A variety of actors play a crucial role in the design, funding, conduct, and outcome of DMPs (Gül, 2019). First, one or more actors typically initiate (commission) the DMP and make important decisions about its terms of reference, remit, and output, including civil society and public and executive authorities (Jacquet, Talukder, Devillers, Bottin, & Vrydagh, 2020; Setälä, 2017). In some cases, third-party interest groups can be involved in these decisions (Kahane, Loptson, Herriman, & Hardy, 2013). ⁴ Although initiators often make critical strategic decisions, they in many cases delegate the actual design and operation of the DMP to a team of independent conveners, such as academics and professional consultancies (Gül, 2019; Lang, 2008). These tasks often include further specifying the focus of the mini-publics, defining the population and stratification criteria, selecting participants, planning the sessions, and executing the process (Gül, 2019). Finally, in some, more institutionalized DMPs, independent commissions can be created and resourced to oversee and implement the process (Knobloch, Gastil, & Reitman, 2015).

Participant Selection

A first step in selecting participants in a DMP is to define the population. Often this is done based on the population of a particular jurisdiction (e.g., the residents of a particular state), though it can also be more precisely bounded based on the specific focus of the DMP, as in the case of the Citizens’ Jury on waste incineration in Dublin, where the population was defined as those who would be serviced by the focal proposed treatment plant (French & Laver, 2009). Interest groups, in particular, are often excluded from serving as participants to avoid exerting too much influence over the process but are sometimes granted other opportunities to play a role in the DMP, such as providing information and serving on advisory boards (Carson & Martin, 2002; Kahane et al., 2013; MacLean & Burgess, 2010).

Once a target population has been set, participants are selected through some form of random selection. Participants in DMPs can be seen as a particular form of nonelectoral representatives, sometimes termed “citizen representatives,” which are distinct from self-appointed representatives (Urbinati & Warren, 2008). They gain much of their legitimacy through their descriptive representativeness (James, 2008; Warren, 2008). Descriptive representation can emphasize correspondence in visible characteristics but also in shared experiences, such as one’s professional background (Mansbridge, 1999). Regardless of represented characteristics, the initial focus is placed on who the representative is in terms of their correspondence on those characteristics as opposed to what their specific interests or desires are (Brown, 2006). In this sense, participants in DMPs do not represent specific constituencies and are not held to account through elections (Brown, 2006; Urbinati & Warren, 2008).

Descriptive representation has many merits. Pitkin (1967) highlights its utility as a means of providing information about those being represented. In her seminal work, Mansbridge (1999) argued that descriptive representation could help increase the substantive representation of disadvantaged groups in two contexts: when the group’s interests are uncrystalized and when there is impaired communication and distrust between dominant and marginalized groups. Goodin (2008: 248) discusses how aiming for complete mirror representation may be unrealistic but notes how representing “the sheer fact of diversity” could have important benefits, including making decision makers more aware of and interested in soliciting diverse perspectives. At the same time, efforts to foster descriptive representation in DMPs have received some important criticism. First, unlike their formally elected peers, participants in DMPs are neither authorized nor accountable to a particular constituency (Brown, 2006). Second, they are also, overall, likely to be less skilled, knowledgeable, and experienced (Mansbridge, 1999). Third, what and how groups ought to be represented can be complex and subject to the discretion of organizers (Gül, 2019; Mansbridge, 1999). The first two challenges are an important reason why many proponents of DMPs see descriptive representation as a complement and not a wholesale substitute for other forms of representation (Mansbridge, 1999; Setälä, 2017, 2021). The third challenge points to the importance of reflecting carefully on who ought to have the power to decide on what descriptive characteristics to emphasize.

When it comes to executing the random selection process, there are differences in the specific method of random selection that will have implications for MSIs. Some designs, notably deliberative polls, use a combination of statistical random sampling and a large sample size to select a microcosm of the population in terms of demographic and attitudinal perspectives (Fishkin, 2018). Statistical random samples are said to give everyone an equal chance of being selected (Brown, 2006). Yet, statistical random samples can also be prone to distortions brought on by potential bias in who declines the invitation to participate (O’Flynn & Sood, 2014) or the use of small sample sizes (Bächtiger, Setälä, & Grönlund, 2014). Other designs use stratified random sampling—often involving oversampling certain groups—to increase the representation of underrepresented groups in the population (Bächtiger et al., 2014). The use of stratified random sampling helps generate descriptive representativeness across characteristics identified as important for a particular topic (Lubensky & Carson, 2013) and thus helps ensure that diverse perspectives are brought into DMPs (Brown, 2006). It is important to select stratification criteria deliberately using heuristics like which identity characteristics are most relevant within a particular society and most clearly related to the focus of the DMP (James, 2008).

Deliberation

When it comes to practices to foster robust, well-informed, and inclusive deliberation in DMPs, they commonly include trained and impartial facilitators and moderators, mixing small-group discussions and plenaries (Goodin & Niemeyer, 2003; Harris, 2019; Landwehr, 2014; Leydet, 2019). New techniques have been developed to accommodate groups of participants that speak less than others, including encouraging different types of expression and communication and regularly rotating facilitators across different small groups (Curato, Dryzek, Ercan, Hendriks, & Niemeyer, 2017; Harris, 2019). Overall, these practices result in DMPs performing well when it comes to their internal deliberations (Setälä & Smith, 2018). For instance, effective facilitation can help prevent domination by some members, encourage brainstorming, boost interactivity, and track ideas (Carson & Martin, 2002).

DMPs also provide balanced and accessible information on the focal topic, including briefing materials (O’Flynn & Sood, 2014), expert testimony and presentations (Brown, 2006), and information from third parties, such as relevant interest groups, that complements expert information (MacLean & Burgess, 2010). This helps participants transition from having a “raw” to a “refined” opinion on the topic, one that “has been tested by the consideration of competing arguments and information conscientiously offered by others who hold contrasting views” (Fishkin, 2009: 14). It also helps contribute to the inclusion of a broader array of relevant discourses in tandem with the descriptive representation of participants (Felicetti et al., 2016). The use of random selection enables participants to engage with this information in a more impartial and open-minded manner than if they were selected through other selection methods, such as elections (e.g., having to fulfill electoral promises), contributing to higher-quality decision-making (Gastil & Wright, 2018; Vandamme & Verret-Hamelin, 2017). In the case of randomly selected deliberators in a citizens’ jury focused on container deposit legislation, they found that “the citizens in this case study showed no sign of susceptibility to outside pressure and they displayed no obvious biases or preconceptions that inhibited rational deliberation” (Carson & Martin, 2002: 13).

Outputs

Following the deliberation phase, DMPs enter a decision-making phase to consolidate conclusions ahead of their dissemination. DMPs vary in terms of how decisions are made. Some strive for a form of consensus, whereas others use voting and individual surveys (Escobar & Elstub, 2017; Fishkin, 2009). These conclusions are then compiled (e.g., in a position report) and diffused to the broader public (Fournier, van der Kolk, Blais, & Rose, 2011). Initiators have a significant amount of power in deciding what the outputs of DMPs will be and how they will be used (Setälä, 2017). For DMPs focused on informing policy, the outputs of DMPs are often consultative. However, in rare cases, initiators can have some more direct influence and authority (Setälä & Smith, 2018), such as when political actors commit to implementing a DMP’s suggestions (Fishkin, He, Luskin, & Siu, 2010).

Deliberative Mini-Publics in Deliberative Systems

In light of these unique characteristics, DMPs have attracted significant attention in research on deliberative systems (e.g., Curato & Böker, 2016; Felicetti et al., 2016; Hendriks, 2016; Niemeyer & Jennstål, 2018). Scholars have developed frameworks focused on the tasks and responsibilities DMPs could undertake in various polities (e.g., Goodin & Dryzek, 2006). Some of this work examines how DMPs could be used in specific parts of the deliberative system (see, e.g., Dryzek, 2009, for their use in the public space; Felicetti, 2014, for their use as a means of transmission). In a related area of research, scholars have explored, directly and indirectly, how the use of DMPs can contribute to the deliberative capacity of deliberative systems. For instance, Lafont (2017, 2020) outlines various ways DMPs could revitalize public deliberation in a more participatory manner. Finally, in another stream of research, scholars have adopted a goal-oriented perspective on the uses of different types of DMPs, whereby they may be more or less suitable in different contexts and to solve different types of problems in particular systems (Curato, Vrydagh, & Bächtiger, 2020).

DELIBERATIVE MINI-PUBLICS IN MULTI-STAKEHOLDER INITIATIVES

We now theorize how DMPs can be translated to the context of MSIs by elaborating on four key design parameters. We then develop theory about how five different uses of DMPs can contribute to improving MSIs’ deliberative capacity.

Design Parameters

As elaborated on earlier, DMPs were developed for, and have almost exclusively been used in, the context of polities like nation-states or municipalities. However, MSIs differ from these polities substantially in terms of membership, processes, structure, and decision-making institutions (i.e., empowered space). Moreover, MSIs also vary extensively among themselves. These features necessitate careful reflection about how DMPs are to be structured and implemented in MSIs. While some DMPs’ characteristics, structures, and processes can be readily translated to MSIs (e.g., using trained moderators and facilitators), other aspects need more careful and specific consideration (e.g., defining the population from which to draw participants). We conceptualize four design parameters we see as particularly relevant to translating DMPs to MSIs.

Initiating and Convening Deliberative Mini-Publics in Multi-Stakeholder Initiatives

In terms of the initial impetus to adopt DMPs in the first place, there are two possible scenarios. In the first case, it could happen at the time of the founding and elaboration of an MSI, as national governments and the nonprofit sector (Bartley, 2007; Marques & Eberlein, 2020) can serve as key instigators. In the second case, it could happen after the founding of the MSI, as would be the case for the large variety of existing MSIs, for instance, as a result of a motion by one or more MSI members.

Specific decisions about a DMP’s initiation, whether one-off or ongoing, can be made by a dedicated, independent commission based on predefined parameters (Knobloch et al., 2015). In the case of the FSC, a related MSI secretariat, such as the Marine Stewardship Council (dealing with sustainable fishing), could serve as such a body. This independent commission could include MSI members in an advisory capacity to seek their input on the remit and output to legitimize the DMP, increase buy-in, and leverage members’ insights in the process (French & Laver, 2009; Kahane et al., 2013).

In terms of convening DMPs, MSIs may be susceptible to subtle co-optation or domination from powerful actors who may, for instance, advocate for preferred venues or times for deliberations (Maher, 2022). Best practice from DMPs indicates that such co-optation can be reduced by appointing neutral conveners (e.g., with no conflict of interest with the issue[s] at hand) to design and implement the DMPs (Beauvais & Warren, 2019). The independent commission described before could also serve the role of convener.

Defining the Population

DMPs have been developed for the most part in the context of a clearly delimited sovereign territory (although there are a few exceptions, such as those focused on governing the use of the internet; see Fishkin et al., 2018). Even then, it can be difficult to delineate clearly who is sufficiently affected by a particular decision to warrant inclusion (O’Flynn & Sood, 2014). Some citizens may be legally bound by a decision, others may be directly affected, and still others might be indirectly affected.

MSIs face an even greater issue regarding which actors to include in the population given that they span national boundaries and their main constituents are not citizens but mostly organizations, such as corporations or NGOs (Martens et al., 2019). In terms of the overall population of DMPs when used in the context of MSIs, drawing on Martens and colleagues’ (2019) categorization of different actors involved in MSIs, we take as the focal population of actors (individuals and organizations) the combined group of MSI members, rule takers (i.e., firms), and all otherwise affected actors. Importantly, as we discussed earlier, consideration of the scope of a DMP is crucial when defining the population (French & Laver, 2009). For instance, in the case of the FSC, a DMP addressing a more specific issue related to sustainable forest management, such as FSC’s Principle 3, “Indigenous Peoples’ Rights,” would have a more narrow population. This is because not all actors from the population of the FSC will be affected by issues related to Indigenous people (such as firms exploiting forests with no Indigenous communities). Moreover, Indigenous notions of sovereignty and self-determination (Banerjee, 2021) will also be important in determining the population of such an issue-focused DMP.

Following Bader (2018), decisions around the population of a DMP in an MSI are best not based on an initial drawing board but made deliberatively by conveners and, when necessary, expert advisors given the issue and techniques like stakeholder mapping. It is important to recognize that defining who is affected by a particular topic is context-dependent and may be challenging and, in the case of MSIs, may be the result of manipulation and power dynamics (Brown & Dillard, 2013; Dawkins, 2021). We later discuss how DMPs could be used to support this process when discussing meta-deliberation in MSIs.

Delineating and Selecting Participants

Once the population is defined, a second decision pertains to which actors will be invited and how they will be selected. The first question pertains to who should be excluded, which is particularly important given the differentials of power between actors in MSIs (Dawkins, 2021; Taylor, 2005). While a full examination of exclusion criteria is beyond the scope of our article and, in any case, will depend on each DMP, we discuss the important consideration of official representatives of organizational members of MSIs (e.g., the CEO of IKEA, the president of the WWF). We think it is important to exclude them from participating in DMPs as participants to avoid biasing the process, but argue that they should be granted roles commonly given to interest groups in DMPs, such as the opportunity to testify and the opportunity to help inform the information that goes out to participants (e.g., Carson & Martin, 2002). This approach will have the benefits of gaining access to their knowledge and perspectives and increasing their buy-in while helping to reduce the risk that they will dominate or derail the actual deliberations (Kahane et al., 2013).

Second, in terms of which actors can be invited as participants of DMPs, MSIs are unique in that some actors of the population will be formal organizations (e.g., corporate rule takers) made up of individuals (e.g., employees), whereas others will be individuals or informally organized individuals (e.g., local community members). In this light, we envision three main options. First, at one logical extreme, only individuals would be involved in the DMP, with all organizations being disaggregated so that the individuals working for them become individuals of the population (for such a use in the case of a DMP for the governance of the internet, see, e.g., Fishkin et al., 2018). In the context of the FSC, this would mean that FSC organizational members, like IKEA or the WWF, would see each of their employees (from a senior marketing manager to the janitor in a store to a research associate) be potential participants in the same vein as local community or Indigenous individuals. Yet, this approach carries the risk that large organizations (in particular, corporations) would represent a larger piece of the pie of the composition of the DMP. Second, at the other logical extreme, individuals and organizations could both be treated as single actors. This would increase the likelihood that individuals would be selected and, in so doing, decrease the relative weight of organizations (see, e.g., Gleckman, 2018, who discusses such a possibility applied to MSI governance). In the case of the FSC, organizations like IKEA and unions would each have one randomly selected member eligible as a potential participant, and each of the Indigenous people living in FSC-related forests would count as one potential participant. Third, a hybrid approach could be used, whereby a multiplier would be used to grant organizations a number of potential participants based on their size. In this way, individuals would be the only actors who can serve as participants of the DMP, but there would be comparatively more participants from organizations due to the multiplier. Home Depot (one of the largest firms participating in the FSC) could, for instance, be granted twenty employees as potential participants. Although each approach has its unique pros and cons, all of them would help rebalance the power from large organizations toward dispersed and unorganized but numerous actors—which is often the case of marginalized stakeholders (e.g., Banerjee, 2018).

As discussed earlier, participants do not serve as delegates of any constituency; rather, grounded in the principles of descriptive representation, they “stand for” the broader population as themselves. For instance, in the case of internet governance, participants represented the broader population of users of the internet, not the organizations with which they were affiliated (Fishkin et al., 2018). Participants deliberated “as netizens, changing their views based on substance rather than simply taking instruction from their home institutions” (Fishkin et al., 2018: 12–13). Despite prior evidence, some participants may be inadvertently influenced by their organizational affiliation, making the aforementioned practices to foster robust deliberation all the more important. Additional techniques to overcome this include anonymizing proceedings to reduce the perceived risk of retribution (Carson & Martin, 2002) and using secret votes (if voting is used) to reduce the risk of retribution (Gastil & Wright, 2018). Even if some participants cannot overcome their organizational affiliations, random selection can help ensure they would be a minority and would pose less disruption to deliberations (Leydet, 2019). For these reasons, while recognizing these potential risks, they are surmountable and unlikely to significantly detract from the main benefits offered by DMPs in MSIs.

Third, we illustrate what the sampling process to select participants could look like. We apply the first option from earlier, whereby all organizations would be disaggregated into their respective individuals. The pool of potential participants would thus comprise all individuals and employees of organizations that are members of the MSI, rule takers, or otherwise affected (excluding any individuals ultimately subject to exclusionary criteria). As described earlier, two main methods could then be used to select participants in the DMP from this population: statistical random sampling and stratified random sampling. The methods and stratification criteria should be made with the support of impartial, expert conveners based on the broader principles of representative characteristics within the population and the goal of the DMP (James, 2008). For example, for the purpose of illustration, stratification criteria in the case of a DMP tasked with input on the FSC rules could be sector (public, for-profit, and nonprofit organizations; general public), position in the supply chain (upstream vs. downstream), and views about the place of humankind in nature (e.g., anthropocentric vs. ecocentric; Purser, Park, & Montuori, 1995).

Efforts to Increase Willingness to Participate and Support

It is essential to ensure that all actors invited to participate in the DMP are willing and able to do so. Given the unique nature of MSIs, different approaches are likely to be necessary for different types of actors, such as for-profits and related actors (e.g., industry associations), nonprofits, and loosely or unorganized actors (e.g., local communities, natural environment). In terms of the latter two, though research suggests that nonparticipation is complex and multifaceted (Jacquet, 2017), some suggestions have been developed to increase participation, such as honoraria or covering participation-related expenses (e.g., accommodation) (Harris, 2019). Given that an important reason for not participating is a lack of perceived impact of the DMP, conveners can clearly spell out how the outputs of the DMP will be used within the MSI deliberative system (Jacquet, 2017). French and Laver (2009) found that the possibility of having input into decision-making was perceived by participants as an important incentive. In the context of MSIs, however, and as heavily emphasized in the research critical of them (e.g., Banerjee, 2018; MSI Integrity, 2020), actors other than for-profits are often sidelined in decision-making processes. Some actors may also refuse altogether to participate in such deliberations in the first place because of intractable differences, such as some Indigenous communities’ emphasis on self-determination (Banerjee, 2021). Yet, apart from intractable differences, being excluded from MSIs should, in general, be sufficient motivation for marginalized groups to participate in DMPs that would be able to influence MSI processes and decisions and make them less marginalizing. Some marginalized groups may require additional support to be able to participate that DMPs can provide, such as supplementary learning materials (O’Flynn & Sood, 2014). Yet, as we describe later, DMPs can also help build the skills and capacities of participants and nonparticipants (Niemeyer, 2014) so that they could address how and why these groups have been marginalized in the first place in the MSI. In the spirit of reciprocity core to authentic deliberations, DMPs allow participants to learn from each other despite differing worldviews (Caluwaerts & Kavadias, 2014).

Finally, it is also important to consider how to encourage corporate members of MSIs to support the adoption of DMPs. Research on the involvement of interest groups in DMPs suggests various techniques like conveying that DMPs may be just one part of a broader decision-making process and that their interests will also be formally represented in other ways in MSIs (e.g., being a member of the assembly, participating in the election of the MSI board), offering them the chance to provide feedback on the DMP after its completion, and emphasizing the organizational learning benefits of participating (Kahane et al., 2013). Additionally, prior work on MSIs points to several factors that could similarly increase corporate support for DMPs, including market rewards like certifications (Potoski & Prakash, 2009); managerial or supply chain advantages, such as improved traceability (Zadek, 2004); lock-in effects of participation therein (e.g., Bartley, 2007; Fransen, 2012); reputational sanctions stemming from pressure from activist groups (Mena & Waeger, 2014); and high-publicity events. For instance, Huber and Schormair (2021) forcefully show how the 2013 Rana Plaza collapse coerced some firms to participate in the Accord on Fire and Building Safety, an MSI in Bangladesh focused on improving working conditions in the textile industry. This did not prevent some conservative firms from maintaining their reluctant stance toward MSI regulation and collaboration with stakeholders, but some showed a willingness to engage over time (Huber & Schormair, 2021).

Improving Multi-Stakeholder Initiatives’ Deliberative Capacity through Deliberative Mini-Publics

We now develop a framework of how DMPs can improve the deliberative capacity of MSIs. In line with recent goal-oriented perspectives on the use of DMPs (Curato et al., 2020), MSIs should make decisions about which uses are more relevant based on undertaking a comprehensive evaluation of their deliberative capacity. We conceptualize five uses of DMPs corresponding to each element of the MSI system: public space, empowered space, transmission, accountability, and meta-deliberation. Our conceptualization details how each use can improve the deliberative capacity for that element and the MSI system as a whole.

While we conceptualize how each use of DMPs uniquely improves deliberative capacity depending on the characteristics of the system’s element, we also identify effects that span all our uses. In terms of their effects on the deliberative capacity of the system’s elements, DMPs can improve inclusiveness and authenticity. When it comes to inclusiveness, the use of a DMP can overcome deficiencies in the breadth of interests and discourses spreading in a particular element of the system. As discussed earlier, the use of random selection helps bring together a descriptively representative body to ensure that a broader array of perspectives and interests are brought to bear on deliberations in the element, and the careful provision of balanced and comprehensive information that covers relevant arguments and discourses would help ensure that relevant discourses are included in these deliberations (Carson & Martin, 2002; Dryzek, 2010; Felicetti et al., 2016). The use of DMPs can also improve authenticity in a particular element of the system through “deliberation-making” (Niemeyer, 2014: 179). DMPs’ myriad practices to foster deliberation equip and enable participants to carefully reflect on, parse, and synthesize various arguments, discourses, and sources of information, such as carefully scrutinizing potentially manipulative discourses (Curato & Böker, 2016; Niemeyer, 2014; Niemeyer & Jennstål, 2018). This can improve the quality of deliberations outside the DMP if the takeaways are synthesized in a manner that is understandable to the broader MSI population, as defined earlier (Niemeyer, 2014). We tailor these two effects that span our uses to each use in what follows.

One effect spans all our uses of DMPs to improve the deliberative capacity of the system as a whole. DMPs can improve system-wide inclusiveness and authenticity through fostering deliberative skill and capacity building (Niemeyer, 2014), by which they “contribute to building the capacity of a polity to host inclusive and authentic deliberation” (Curato & Böker, 2016: 178, emphasis original). Those who participate directly in DMPs have the opportunity to develop numerous civic capacities and habits (Knobloch & Gastil, 2013, 2015). Yet, even those who do not participate directly can see their skills and capacities develop through the spillover of deliberative norms and broader cultural change brought about through well-executed DMPs (Boswell, Hendriks, & Ercan, 2016; Niemeyer, 2014). In MSIs, by building their deliberative skills and capacities, DMPs can increase the likelihood marginalized, uninterested, or unaware actors from the population participate in MSI deliberations (in any element), thereby improving inclusiveness. They can also improve authenticity by increasing actors’ ability to enact deliberative norms and expectations and challenge acts of coercion, unfairness, and imbalance. We now turn to an overview of each of our five proposed uses of DMPs. We summarize our main arguments in Table 2.

Table 2: Uses of DMPs in MSIs and Their Effects on Deliberative Capacity

Our first use posits DMPs as a means of fostering deliberation in the public space of the MSI. Such a DMP could focus on any topic deemed relevant to the population of the MSI and would disseminate its outputs and underlying arguments broadly throughout various sites of the MSI’s public space. The Citizens’ Initiative Review is a useful example given its focus on distributing its outputs pertaining to ballot measures to the broader public to inform their decision-making (Knobloch et al., 2015). This use is well suited to overcoming the aforementioned deficiencies in inclusiveness and authenticity that sometimes afflict an MSI’s public space. By bringing together a descriptively representative group of actors from the population of the MSI, it would improve the inclusiveness of the public space of an MSI by surfacing additional voices and opinions. In the case of the FSC, such a DMP would include various stakeholders from which participants can be drawn, some of which are not included in the General Assembly of the FSC, such as some social or environmental stakeholders from some African or Southeast Asian countries ⁵ who may not be members due to a lack of knowledge about the FSC or the inability to pay membership fees. It would also bring alternative discourses to the table, which is often dominated by powerful actors like corporations, by providing balanced information on relevant arguments—the “contestatory” use of DMPs that Lafont (2017, 2020) outlines. This use would also improve the authenticity of deliberations in the public space by formalizing them and providing a forum in which coercion may be less present than in more natural deliberative settings. Indeed, fairer deliberations stemming from using a DMP could allow less powerful, often noncorporate actors to push against the hegemonic view(s) in the public space. In the case of the FSC, this hegemonic view is that self- and private regulation can limit deforestation through market means with limited state intervention (Moog et al., 2015). Because not all conversations in the public space of the FSC (see, e.g., FSC Watch website) follow this hegemonic view, a DMP could help bring alternative views to the fore and confront the hegemonic view(s) by, for instance, putting these views under scrutiny.

This use can also contribute to improving the deliberative capacity of MSIs at the system level. The effects on deliberative capacity through skill and capacity development could be particularly powerful in this use, given the emphasis on broad-based dissemination throughout the public space. In terms of output consequentiality, this use could help mobilize various sites in the public space to more directly scrutinize and engage with the empowered space on perceived issues and concerns with the MSI. This corresponds to Lafont’s (2017, 2020) vigilant and anticipatory roles of DMPs, in which the MSI’s decisions and processes do not match views in the public space and when actors of the public space have not yet formed opinions about these issues, respectively. By making the gap between the public and empowered spaces evident and making explicit some issues with the MSI for various sites in the public space, this use would strengthen the link between deliberations in the public space and decisions in the MSI, ultimately improving system-wide output consequentiality.

Our second use posits DMPs as a means of making decisions in the empowered space of the MSI, with or without binding decision-making authority. In the case of the former, this use would be particularly useful for topics with which the existing empowered space is ill suited to deal, especially when there are conflicts of interest or a lack of motivation to address important issues (Bouricius, 2018; Kuyper & Wolkenstein, 2019). For example, the FSC’s empowered space was conflicted around protected forests (“intact forest landscapes”), and the General Assembly eventually decided to push this issue aside to move on to other problems and decisions (Arenas et al., 2020). In such a case, a DMP could take over authority from the general assembly to help move forward on this value-laden, contested issue. A variant of this proposed use with less decision-making authority could build off proposals for permanent chambers comprising randomly selected representatives (e.g., Gastil & Wright, 2018). For example, one such chamber could have the “power of initiative, consultation, and amendment” (Vandamme & Verret-Hamelin, 2017: 20). In the case of the FSC, the Board of Directors established a Policy and Standards Committee ⁶ that fulfills this role. A DMP could replace or complement this committee, which provides recommendations to the Board of Directors on value-laden decisions.

This use of a DMP could improve authenticity and inclusiveness in MSIs’ empowered space. DMPs are well suited to overcoming deficits to authenticity, such as bargaining, horse trading, and politicking, that can influence decisions in empowered spaces (Kuyper & Wolkenstein, 2019). The relatively fairly structured general assembly of the FSC could benefit from such use, as some interests have nevertheless been favored in decisions (Dingwerth, 2008; Moog et al., 2015). Moog and colleagues detail, for instance, the strategic struggle within the FSC in its early years between proponents of radical, high ecological integrity and those favoring a market approach based on consumer recognition and a large market share for FSC-certified products, which won the battle. This use of a DMP could help rebalance interests by increasing the fairness and reciprocity of, as well as limiting coercion in, deliberations through their contributions to deliberation making (e.g., participants becoming more aware of manipulative discourses). In terms of inclusiveness, barriers to entry in the empowered space could be reduced by this use. Through the use of random selection, stakeholders from the population that are not currently formally involved in the empowered space could now be included. In the same vein, this would solve some problems with representation through self-appointed third parties in MSIs (Reinecke & Donaghey, 2021) by representing stakeholders directly and in a descriptively representative manner. Once these stakeholders are included, this use would also foster a broader inclusion of discourses by stakeholders through balanced and comprehensive information.

This use can also contribute to improving the deliberative capacity of MSIs at the system level. As with other uses, it can contribute to participants’ and nonparticipants’ skill and capacity development, particularly in the case of members of the empowered space who can gain new perspectives about how their work could be approached. This use would also improve the output consequentiality of the MSI system by guiding decision-making in the empowered space to be more in line with the full array of deliberations in the public space. Indeed, the empowered space of MSIs usually consults stakeholders other than only their members when deciding on or changing their processes. For instance, the FSC regularly puts its policies, standards, and procedures to external consultation when developed or revised. ⁷ A DMP would help bring a broader set of views from the public space in such consultative processes and ultimately link the empowered space’s decisions to some suggestions made by stakeholders.

Our third use posits DMPs as a means of transmitting outputs of deliberations from the public space to the empowered space of the MSI. Such a DMP can provide input to the assembly, board, and other bodies of the empowered space of the MSI about what issues ought to be pursued and how they could be pursued. This input could include what items to put on the agenda, feedback on proposed rules, and feedback on which members should be included in the MSI in the first place. In this way, they would function as “mediating institutions” (Parkinson, 2012: 162) that could overcome deficiencies in the transmission of informed opinions and ideas from the public space to the decision-making bodies in the empowered space (Boswell et al., 2016; Hendriks, 2016).

This use is particularly well suited to improving authenticity in MSIs’ transmission mechanisms. As discussed earlier, a major challenge pertaining to MSIs’ deliberative capacity is their imperfect method of transmission of the large variety of voices, discourses, and interests that go beyond formally involved actors (e.g., in the assembly as members), such as the otherwise affected stakeholders (Martens et al., 2019). Although transmission through market means (e.g., NGO activism, pressure on firms and MSIs) has been shown to generate some results (Mena & Waeger, 2014), it nevertheless does not transmit the wide array of views in the population of an MSI (Moog et al., 2015; Schouten et al., 2012). This use would help balance the dominant corporate interests with divergent views that tend to be silenced in MSI transmission. This would be the case thanks to DMPs’ ability to bring various different discourses to deliberations and to allow participants to see through manipulative discourses. In particular, rather than relying only on social and market pressures to attract the attention of the MSI’s empowered space, a DMP would offer a formal setting in which deliberations are less confrontational, more reciprocal, and thus more authentic.

In terms of effects on the deliberative capacity of the system as a whole, in addition to improving authenticity and inclusiveness through its contributions to skill and capacity development described earlier, this use can improve output consequentiality by increasing transparency. It would increase the transparency of particular MSI processes by bringing views from the public space that may have otherwise been co-opted, coerced, or silenced. For example, grievances that had been previously unknown or downplayed could be brought to the attention of the empowered space (Schouten et al., 2012). This would serve to link deliberations throughout the system, particularly those from the public space, to decisions made by the empowered space of the MSI.

Our fourth use posits DMPs as a means of holding the empowered space accountable. In essence, this would be a form of “back-loaded” (Fung, 2003: 346) DMP that would be tasked with critically reviewing the decision-making and implementation processes of the empowered space, particularly those related to reporting and monitoring. It would engage in what Mansbridge (2009; see also Dryzek & Stevenson, 2011) termed deliberative accountability, which, in contrast with narrative accountability, entails a two-way dialogue between a principal and agent about the agent’s explanations.

This use could help improve authenticity in MSI accountability. It could help increase the transparency of MSI operations and reporting thereof. In particular, it would help improve the authenticity of accountability through reporting by ensuring that deliberations about what is reported and how are less biased toward corporate interests, thanks to the DMPs’ ability to counteract coercive behaviors by providing the opportunity to reflect on and parse out various discourses. In the same vein, it would help improve fairness and reciprocity in deliberations related to the monitoring system, such as those relating to audits, noncompliance, and sanctions—another key aspect of MSI accountability. Here, again, DMPs’ ability to allow participants to balance information about discourses, as well as counter manipulative actions, would work toward this effect. Finally, this use can help bring more independence to the monitoring system from the empowered space and powerful actors therein, in particular corporations. Indeed, through their capacity to foster fairer and noncoercive deliberations, notably through allowing participants to arrive at more balanced and synthesized information, the DMP outputs may substantially differ from the accounts on monitoring compliance provided by the empowered space.

In terms of effects on the deliberative capacity of the system as a whole, in addition to improving authenticity and inclusiveness through skill and capacity development, this use could improve the authenticity of the system in another way. The institutions of the empowered space may have made decisions that do not fit some views and perspectives in the public space. Indeed, this use would provide the opportunity to discuss the decision-making processes behind, for example, a particular accreditation decision, which would help foster greater reciprocity and learning. Additionally, this use can help improve output consequentiality by inducing the empowered space to act in line with the public space’s interests, as the former would anticipate having to justify its decisions through dialogue (Setälä, 2017). It can also help improve outcome consequentiality as it pertains to compliance by member firms. Such a use would allow greater transparency around compliance because a DMP would be less risky as a way to complain about corporate behavior. It could also surface issues with compliance and the monitoring system that would not have appeared otherwise. And finally, this use reduces chances of symbolic adoption, as it would allow for wider and more authentic scrutiny of compliance. In the case of the FSC, there are various reporting mechanisms by participating firms, monitoring mechanisms by auditors (and associated reporting procedures), and complaint mechanisms. Submitting these accounts of verification and compliance to a DMP, and thus to a wider array of actors that can examine these accounts through reciprocal and fair deliberation, rather than by a single accredited third party, is likely to improve the identification of sources of noncompliance and, ultimately, outcome consequentiality.

Finally, our fifth use posits DMPs as a means of engaging in meta-deliberation about the system, focusing on identifying potential gaps and opportunities pertaining to inclusiveness, authenticity, and consequentiality. Although meta-deliberation can occur anywhere within the system (Stevenson & Dryzek, 2014), DMPs have sometimes been used to foster meta-deliberation as one of their objectives (e.g., Thompson, 2008).

This use can make meta-deliberation more inclusive by counteracting the usually dominant perspectives and voices in the empowered space (e.g., secretariat, general assembly to some extent). It would do so by bringing in a more diverse and broader set of opinions about what needs to be changed and how, by virtue of descriptive representativeness and balanced information on various discourses. This use would also help improve the authenticity of meta-deliberation. Deliberation-making practices will particularly help in informing stakeholders in a more comprehensive manner and help them deliberate with more reciprocity on what needs changing. It can reduce coercion in meta-deliberations and thus help counterbalance corporate threats of leaving if the MSI changes.

This use can also improve the deliberative capacity of MSIs at the system level. As with other uses, this use could improve inclusiveness and authenticity through skill and capacity development, therefore increasing the use of meta-deliberation by reducing its complexity for stakeholders. Beyond this effect, this use could unveil more systematic gaps in inclusiveness and authenticity. It could, for instance, help identify some affected stakeholders that should be included in the population. For example, a currently nonparticipating home improvement company may decide to obtain FSC certification for its wood products. If not already supplying other FSC-participating firms, this company’s suppliers will become part of the FSC’s population (as we defined previously). As a ripple effect, a number of other stakeholders (e.g., local communities that the firm’s supply chains affect, workers and unions) will also become part of the population of the FSC. This use of a DMP could focus on including these “new” stakeholders; bring their views, opinions, and discourses into deliberations with other existing stakeholders; and provide feedback to the initiators and/or conveners of the DMP and to the empowered space of the MSI about whether and how to incorporate these new actors and views. In the same vein, it may also identify systemic gaps in authenticity, such as the general bias in many elements of the MSI system (beyond meta-deliberation) toward corporate and economic interests, underlying their overwhelming power and potential for coercion. Such a use can first help identify these gaps and then allow participants to deliberate on how to tackle them. In terms of output consequentiality, such a use can more closely link the deliberations between affected actors to the decisions taken by the empowered space of the MSI, by virtue of examining when these deliberations and decisions are not aligned. For instance, by including in a fairer manner Indigenous communities living in forests in which FSC-certified firms are operating, this use of a DMP can help bring to the fore issues that are not dealt with adequately at the moment in terms of preserving livelihoods. Finally, it can improve MSIs’ outcome consequentiality, such as how much of an industry is covered by the rules of the MSI. Indeed, by aligning its rules more closely to the actual stakeholders affected by or affecting the industry, it increases the likelihood of dealing more adequately with problems arising from this industry.

DISCUSSION

MSIs have received significant criticism for their regulatory role in global production networks because they often struggle to adequately represent and incorporate the views of marginalized actors (Banerjee, 2018; Fougère & Solitander, 2020) and are prone to co-optation by economic interests (Dawkins, 2021; Moog et al., 2015). From a deliberative democratic perspective, these problems refer to a lack of deliberative capacity in MSIs—that is, they do not host deliberations that are authentic, inclusive, and consequential. We have applied a deliberative systems perspective to MSIs to understand how DMPs can improve these systems’ deliberative capacity. In doing so, we respond to recent calls to elaborate on normative-conceptual notions of deliberative democracy in the context of MSIs to develop phenomenological and pragmatist solutions to issues faced by real-world MSIs (e.g., Arenas et al., 2020; Barlow, 2021; Fougère & Solitander, 2020; Sabadoz & Singer, 2017). We developed theory about how DMPs, when translated to the MSI context through specific design parameters, can improve MSIs’ inclusiveness, authenticity, and consequentiality. We now turn to discussing our two main contributions and resultant implications for future research, followed by a discussion of limitations and practical considerations.

Contributions and Implications for Future Research

Our first contribution is to provide a novel way of understanding and improving the deliberative capacity of MSIs by leveraging the deliberative systems perspective and DMPs. We thus advance a comprehensive understanding of MSIs as deliberative systems, which contributes to opening the “black box” of the many different forms of deliberations and the locations in which they take place in and around MSIs. Indeed, our translation of the systems perspective to the unique context of MSIs helps us provide a more fine-grained analysis than current research allows, considering different locations and aspects of deliberation in MSIs. Consequently, it also allows for a more precise diagnosis of problems and elaboration of applicable solutions for MSIs. For example, Soundararajan and colleagues (2019) examine how to foster a collective stakeholder orientation in global supply chain actors so that they can come together in an MSI to collaborate and govern their supply chain in a way that is highly authentic, inclusive, and consequential. As one example, they propose that “more neutral and accommodative spaces” (405) are needed in MSIs to bring out authentic deliberations. Our conceptualization of MSIs as deliberative systems allows for a more fine-grained approach to such an insight: these spaces may not necessarily be required in the empowered space but could relate, for instance, to a more effective transmission between existing, accommodative sites of the public space and the empowered space of the MSI.

While we have focused on DMPs as solutions to problems in MSIs, our endeavor opens up future research leveraging the deliberative systems perspective to study MSIs in all of their deliberative complexity, rather than being treated monolithically. This echoes the call of Fougère and Solitander (2020) to examine interactions between stakeholders not only within MSIs (e.g., Zimmermann, Albers, & Kenter, 2022) but also beyond. The systems perspective also allows for comparing different MSIs with each other (O’Flynn & Curato, 2015) in a systematic and more fine-grained manner (see, e.g., Schouten et al., 2012, who compare two MSIs’ deliberative capacity without considering them as systems). Another avenue for future research opened up by this perspective is to view MSIs as but one part of a broader deliberative system of transnational governance (de Bakker et al., 2019). As explored in the case of global climate governance (Stevenson & Dryzek, 2014), MSIs could be seen as a part of the empowered space of global business governance (of a specific industry or not), along with other regional, national, multilateral, and transnational institutions. In this vein, it would be interesting to explore how MSIs—particularly those that adopt some form of DMP—contribute to the deliberative capacity of the transnational business governance system.

A corollary contribution is to show how DMPs can enhance the deliberative capacity of various elements of the MSI system. Yet, we also conceptualize how DMPs can ultimately enhance the deliberative capacity of the system as a whole by, for instance, facilitating the creation and convening of the spaces to which Soundararajan and colleagues (2019) point. In a similar vein, our conceptualization of the functions of DMPs in an MSI system could be particularly helpful in addressing how to put in practice some of the normative recommendations of the deliberative literature on MSIs. For instance, Dentoni, Bitzer, and Schouten (2018) elaborate on how structuring MSIs in relation to the nature of, or harnessing, the wicked problem(s) the MSI addresses can help enhance their quality. DMPs can be useful in better grasping the nature of these wicked problems in the first place, from a more plural and inclusive perspective, and thus incorporating solutions to these problems more effectively within MSI structures. Our conceptualization also contributes to an improved understanding of the possibilities of MSIs in light of critiques they have received, such as the need for local and regional governance in dealing with global business activities (e.g., Banerjee, 2018). Indeed, DMPs open the possibility for “democratization processes ‘from below’ that are more participatory” (Banerjee, 2021: 13) despite the global scale of MSIs.

In this respect, our endeavor paves the way for future research that could explore what types of DMPs would be more or less suitable to fulfilling different uses. Future work could explore in greater depth how different design choices like sequencing (Beauvais & Warren, 2019), decision rules for DMP participants (i.e., whether consensus-based or aggregation-based; Fishkin, 2009), or what medium is used (e.g., online or face-to-face; Paulis et al., 2021) affect the suitability of different DMPs for different uses. In a similar vein, our framework is but a first step toward understanding how deliberative capacity can be improved through DMPs. Future research should investigate additional pathways by which DMPs can do so. In particular, there is room for investigation of the system-wide effects of specific uses on authenticity and inclusiveness. Another interesting avenue for research would be the reverse process: not just how DMPs can help improve MSIs’ structures and processes but also how the latter can in turn shape and potentially help improve the dynamics of DMPs in MSIs.

Our second main contribution is to advance alternative notions of democratic representation to MSIs beyond structural and self-appointed representation. Reinecke and Donaghey (2021) highlight how both these forms of representation vary across the dimensions of creating presence, authorization, and accountability and identify areas of complementarity and competition across these dimensions. Our engagement with research on DMPs helps us advance another form of nonelectoral representation beyond self-appointed representatives: citizen representatives, of which DMPs are an exemplar (Urbinati & Warren, 2008; Warren, 2008). In contrast with self-appointed representatives, they create presence through their descriptive representativeness (James, 2008; Zakaras, 2010) and are formally authorized to take on a representative function (Urbinati & Warren, 2008; Warren, 2008). Finally, although they are not bound by electoral accountability, they can perform well on other criteria of accountability, perhaps most notably discursive accountability (Warren, 2008). A major reason for the appeal of this form of representation is that they can give presence to perspectives that may not be captured by either structural or self-appointed representatives (Urbinati & Warren, 2008). Indeed, one of the main problems with structural and self-appointed representations used in MSIs is that the representations are prone to inequalities in representing diverse perspectives and voices (Baur & Palazzo, 2011; Reinecke & Donaghey, 2021). Citizen representatives help rebalance represented interests in MSIs. By virtue of random selection, the power of large organizational actors will be more balanced because any employee could be selected to participate in a DMP. In this respect, a particularly promising direction for future research would be to investigate potential complementarities and tensions stemming from the introduction of citizen representatives within systems of representation that contain both structural and self-appointed representation.

We also advance the broader notion of descriptive representation, which has a rich history and many applications beyond its pivotal role in DMPs (Mansbridge, 1999; Pitkin, 1967). Descriptive representation has many additional benefits for MSIs, facilitating the adoption of cognitive and cultural characteristics that foster higher-quality deliberations. Indeed, deliberative MSI research has emphasized some cognitive and cultural characteristics or features important for high-quality deliberations, such as embracing contestation (Arenas et al., 2020) or a collective stakeholder orientation (Soundararajan et al., 2019). Descriptive representation can facilitate the adoption of such cultural-cognitive stances by stakeholders (e.g., mind-sets, orientations, values) by exposing participants to a larger variety of such stances in the first place. As discussed earlier, descriptive representation has many merits and demerits, and future research could explore how other further applications, such as the use of different types of quotas, would need to be translated to the MSI context.

Limitations and Practical Implications

Although DMPs have many merits, they have also faced important criticisms. Their grounding in descriptive representation comes with complications, including a lack of formal authorization and accountability, potential skill and experience deficiencies when compared to their elected peers, and complexities related to what groups to represent and how (Brown, 2006; Mansbridge, 1999). As Lafont (2017, 2020) forcefully points out, an uncritical embrace of DMPs may come at the expense of making broader, system-level improvements. Moreover, DMPs can be susceptible to strategic action on the part of initiators, who may use them for purposes other than improving the deliberative capacity of the system (e.g., Wakeford, Pimbert, & Walcon, 2015). We have sought to address these criticisms head-on throughout the article when considering the MSI context, such as emphasizing the use of independent commissions and expert conveners in our design parameters and emphasizing the benefits of viewing DMPs as complements to existing MSI structures and processes. Nevertheless, we undoubtedly leave room for future research on how the limitations of DMPs could be overcome, particularly in terms of how they could be combined with other efforts to foster deliberation in MSIs and how nefarious efforts can be prevented.

Additionally, given their relative novelty, the implementation of DMPs may be challenging for some MSIs, particularly in terms of introducing them in the first place. Yet, MSIs could start off with what could be seen as lower-risk DMPs to reduce potential resistance from some members. Those that deliberate about less contentious issues and have more of an advisory function, such as many that fall under our first suggested use of fostering deliberation in the public space, would likely draw the least resistance, paving the way for the subsequent adoption of other uses. Moreover, as addressed earlier, MSIs often consult various actors on some revisions of their processes or standards. DMPs could be relatively easily introduced to formalize such consultations that already make an attempt to bring the opinions of different stakeholders to bear on some MSI decisions. DMPs also come with operational complexity, such as bringing diverse participants together and organizing costs. Yet, existing DMPs like an individual Citizens’ Initiative Review have proven relatively affordable, costing approximately US$100,000 (Knobloch et al., 2015). As DMPs would be part of the regular operations of MSIs, the specific source of the budget (e.g., participating firms’ fees) should not pose problems to the use of DMPs given the arguments we advanced, such as appointing neutral conveners and creating a dedicated commission. Moreover, operational complexity could be reduced given the recent improvement in technologies for large online gatherings and conferences, as long as existing inequalities are taken into account.

Conclusion

We have endeavored to explore how to enhance the regulatory role of MSIs for the many negative externalities of global business activities. Ours is an attempt to bring pathbreaking research on deliberative democracy to our current understanding of deliberations in and around MSIs to improve their governance capabilities. Yet, MSIs, because they involve corporate actors, will always be critiqued in their regulatory role (Hussain & Moriarty, 2018). Even though our conceptualization can help alleviate some of these concerns, for instance, around bringing more local and “on-the-ground” solutions for governing global business activities (Banerjee, 2021), the intractability of some underlying worldviews will always result in dissensus and contestation around MSIs’ role (Fougère & Solitander, 2020). Yet, working to improve the parts, such as MSIs, of the overall transnational governance system, as we have done, will, we hope, help deal with the social and environmental problems generated by transnational business activities.