**2. Power in MSIs**

To capture the negotiations between actors in MSIs, an understanding of power is proposed that distinguishes between three levels on which power occurs: (a) organizations, (b) individuals and (c) network-based regulations (see Table 1).

A suitable starting point for the conceptualization of (a) organization-based power is a definition from the Global Production Network (GPN) approach. GPNs capture which actors are involved in the global production, trade and consumption of goods and services and how they are related to each other [20]. In a GPN, an actor is considered powerful if it has the ability "[ . . . ] to exercise and achieve control over a particular strategic outcome in its own interests [ . . . ]" [21] (p. 66). The GPN literature distinguishes between three forms

of power in a GPN, each of which they assign to one of the actor groups: private, public and non-profit [20,22]. The authors refer to the power of companies as corporate power, meaning the ability to influence the decisions and resource allocations of other companies in their own interest due to their market power. Institutional power is exercised by states and transnational organizations over lead firms to influence their investments or other decisions. Non-governmental organizations such as trade unions have collective power and can use it against lead firms or other external actors. Empirically, to what extent these forms of power can be transferred to the actors in an MSI is an open question. GPNs are understood in the context of this paper as umbrella networks in which actors are connected to each other in various ways [20]. Actors participating in an MSI are connected through formal membership and their interactions in a separate network—the MSI—within the GPN's higher-level frame of reference. Thus, in Coe and Yeung's understanding, GPNs are a "network of networks" [23] (p. 778). Hence, the existence of power relations between actors in the MSI can be assumed. However, how interactions and negotiations take place within MSIs is currently insufficiently researched [18,24].

In an MSI, private, public and non-profit organizations are members, but they are represented by individuals. In contrast to organizations, these individuals have different possibilities to act powerfully by negotiating skillfully and making decisions. Accordingly, the proposed understanding of power in MSIs considers (b) person-based power [25,26]. Saffer et al. interpret power in MSIs as the ability of individual actors to influence communication in the network [27]. A prerequisite for this is the existence of necessary resources. Furthermore, actors in formally legitimized positions (offices) have a greater influence on communication, as this position gives additional weight to contributions [27]. This finding can be placed in theory through the analytical distinction of different effects of power, as Mossig undertakes with reference to Keohane and Nye and Bathelt and Taylor [28,29] (pp. 92–93), [30,31]: Power, through control and possession of resources, is based on formally assigned decision-making authority, e.g., due to the position as an elected board member, veto rights or the possession of relevant resources such as special financial resources ("power over" [32]). This form of power over others is related to strength and dominance and often originates in the (im-)material resource endowment outside the network. The second effect of power described by Mossig is power through relationships, which results from the power of persuasion and the ability to win over other actors to one's own ideas and conceptions ("power to" [32]) [28,29] (pp. 92–93). Although formulated more positively, this corresponds to an alternative form of influence hinted at by de Bakker et al. [18]. This reflects a relational understanding of power as a resource that understands powerful actors as those who are able to engage other actors in networks and motivate them to act together [33] (p. 65). Thus, unlike "power over", "power to" has its origin in the personality of the actors, on the one hand, and, on the other hand, this form of power is tied to the network itself. Individuals representing their respective organizations in the MSI are therefore not understood as isolated entities without external connections, but as integrated nodes in their respective individual (ego-) networks [34,35]. This means they simultaneously use their personal characteristics and capabilities, as well as the forms of power of the organization they represent. Thus, power is transferred from the conceptual forms of institutional, corporate and collective power from the GPN literature to the empirically ascertainable level of influence by individuals in an MSI.

However, conceptually linking the two levels does not resolve the open question of how power differentials are expressed in network interactions [18,24]. Thus, an unequal distribution of power can lead "less powerful groups to feel their identity threatened, hamper the ability to mobilize agreement, and result in a lack of commitment to the process" [36] (p. 246). Conversely, an unequal distribution of power among actors can also promote efficient collaboration and problem solving [31]. Well-defined hierarchies and responsibilities shorten decision-making processes [28], so that time-consuming discussions about the supposedly best solution can be reduced significantly.

In addition to organizations and individuals, (c) network-based regulations influence the cooperation of actors in MSIs. They arise from behaviors and past interactions across the actors. The resulting values and norms have a disciplining effect on the individual actors and thus reduce uncertainties [29] (p. 93). Such institutions as a form of power can be defined "[ . . . ] as ongoing and relatively stable patterns of repeated social interaction, based on mutual expectations that owe their existence to purposeful constitution or unintentional emergence" [37] (p. 123). Such established patterns of social behavior can be conceptually distinguished between formal and informal institutions. The former are based on sets of rules, such as bylaws, contracts, or regulations, while the latter emerge in the concrete practice of action and are mutually recognized and reproduced by the actors [37]. Network institutions are to be understood as interdependent with person-based power because, on the one hand, they only emerge through the actions of actors and, on the other hand, the ability of an actor to make use of the power form of network institutions corresponds to the "power to" understanding of power.

**Table 1.** Dimensions of power in the context of an MSI.


Thus, from the literature review, power can be considered in three dimensions: first, at the level of occurrence (organization, individual person, network); second, at the level of its effects ("power over", "power to", institutions); and third, closely related to this, through the forms of power (external to the network through resources and personality, internal to the network through offices and their own capabilities) (see Table 1).
