Strategic multiplexity

The ‘unitary actor’ and ‘unitary tie’ assumptions

A dominant view of an organization in a network is based on two assumptions: that of a ‘unitary actor’ and that of a ‘unitary tie’. Organizations are assumed to be monolithic entities that manage a single kind of relationship, which is chosen by a researcher for study. Before examining the strategic multiplexity perspective, it is useful to examine where these assumptions came from.

The intellectual origins of organizational networks lie in the domain of interpersonal networks (Kilduff and Shipilov, 2011a). Psychological anomalies aside, individuals usually are unitary actors when they manage different relationships. The Behavioural Theory of the Firm (BTF) provides theoretical foundations for much of the current organizational research (Cyert and March, 1963). One of this theory’s insights was the depiction of organizations as anthropomorphic entities affected by the limitations and biases of their decision-makers, and hence the unitary actor assumption of organizations also seemed justified. Yet, BTF actually doesn’t equate an organization with a ‘unitary actor’, as it discusses the origins and consequences of internal organizational heterogeneity, the presence of multiple interest groups inside the organizations as well as the multiple goals pursued by them. Despite these insights, organizational network scholars did not pay attention to what actually happened inside organizations when they managed multiple outside networks.

Studies of interpersonal networks have also made the ‘unitary tie’ assumption. Although early research on family networks collected information on different kinds of relationships which families maintained with their external environment (formal vs informal, advice vs mutual aid), researchers chose to aggregate them under a single term ‘relationship’ (Bott, 1955). This simplification was needed to develop the basic network principles, such as structural balance (Harary, 1956), structural equivalence (Lorrain and White, 1971) or tie strength (Granovetter, 1973). Yet, Harary also suggested that it was important to study the ‘mixture of relations’, such as ‘social choice, communication, formal and informal power’ because ‘the actual group of people generally has more than one relation simultaneously operating’ (Harary, 1956, in Kilduff and Shipilov, 2011b: 81). Despite the importance of this statement, it was left in the directions for future research in Harary’s paper and not much work on multiplexity happened for a long time.

Krackhardt was one of the first to explicitly model multiplexity in interpersonal networks by separating advice ties from friendship ties in order to predict an individual’s power in an organization (Krackhardt, 1990). Individuals’ centralities in the two kinds of networks were correlated, but they also had separate effects on individuals’ power. One of the first explicit recognitions of multiplexity in interorganizational networks was in Gulati and Westphal’s (1999) paper on the relationship between firms’ strategic alliances, board interlocks and the probability of future alliance formation. This paper also challenged the unitary actor view of an organization by examining the internal power dynamics between an organization’s executives and its board members.

One of the main reasons to relax the unitary actor and the unitary tie assumptions is that they are often unrealistic. Most organizations are not unitary entities and most of them have more than a single type of a relationship to manage. One exception might be small companies with highly centralized decision-making centred on a single individual (owner, CEO), but even these organizations often find themselves embedded in different kinds of relationships. Failure to capture these dynamics empirically, especially when a scholar is examining large functionally differentiated organizations, may result in models with misspecifications and omitted variables. Lack of theoretical development on this front also prevents us from getting a fuller understanding of how organizations manage complexity.

Yet, relaxing these assumptions in an article targeted at a top-level journal is clearly not effortless. It requires a researcher to collect data over time not only on a single network, but also on multiple networks. It addition, we face a theoretical problem of theorizing multilevel effects that connect ties at one level of an organization with ties at another level, plus the connection between different ties and organizational outcomes. However, as I attempt to show below, these empirical and theoretical efforts will be worthwhile not only because they will present a more realistic picture of organizations, but also because they should help us address existing puzzles in organizational network research as well as to chart new research directions.

Multiplexity and network outcomes

We know that networks act as pipes and prisms for organizations. Pipes provide access to resources and information, while prisms shape the perceptions that others have about the organization’s quality and social standing (Podolny, 2001). Strategic multiplexity allows the development of theoretical models capturing the role of multiple kinds of pipes and multiple kinds of prisms that enable organizations to achieve a network-based competitive advantage.

The view of networks as pipes has been shaped by the debate about the relative virtues and (vices) of brokerage. Positions rich in structural holes provide organizations with access to heterogeneous information, while positions rich in closure provide access to relatively homogeneous information, which nonetheless is useful because it can help gauge the extent of partners’ collaborativeness (Ahuja, 2000; McEvily and Zaheer, 1999). Moreover, brokerage enables organizations to explore new opportunities and offers flexibility in dynamic environments, while closure allows organizations to exploit existing opportunities, especially in stable environments (Rowley et al., 2000). This debate has converged on the recognition that brokerage and closure are two distinct (and not mutually exclusive) forms of social capital (Burt, 2005). Yet how organizations can simultaneously occupy network positions that can provide them with access to brokerage and closure remains an open question.

The recognition of strategic multiplexity can provide a relatively simple solution to this puzzle. That is, if organizations are connected by multiple interdependent relationships, then the structure of one kind of relationship might provide them with access to brokerage while the structure of the other kind of relationship can provide them with access to closure. For example, a company might have a strategic alliance network with a high degree of closure and the information circulating in this network can help its decision-makers to exploit existing opportunities with these partners. At the same time, this company might obtain diverse information from a network of board interlocks that is rich in structural holes (i.e. if individuals who sit on the board of the focal company don’t sit on the same boards outside this company), and this information can enable the organization to explore new opportunities.

Strategic multiplexity can also inform discussion about the benefits of networks as prisms. Work on status has shown that the structure of external relationships shapes perceptions of the organization’s quality. If an organization is connected to other well-connected organizations, then it is perceived as having a higher quality as compared with organizations that are less well connected (Podolny, 2001). Status then becomes a key element of an organizational identity that has positive performance consequences. Identity, however, is a double-edged sword. On one hand, maintaining a consistent identity (especially as a high status firm) helps an organization to project a consistent image to the external audience, making it easier for the audience to categorize and ‘typecast’ the organization, which ultimately leads to better quality evaluations (Hsu, 2006). On the other hand, the need for maintaining a consistent identity constrains an organization to what it can do, how it should behave (Padgett and Ansell, 1993) and even the types of partners with which it needs to affiliate itself (Benjamin and Podolny, 1999). This prevents organizations from engaging in ‘robust action’ based on multivocal identities in that they could potentially be associated with multiple roles or groups, retain flexibility in responding to interactions and avoid restricted lines of action (Zuckerman et al., 2003). While partnering with high status organizations can be advantageous for the development and retention of one’s high status, high status firms might not have access to the novel ideas and disruptive innovations that typically come from the industry’s periphery.

How can a firm maintain high status, but still be connected to the innovations from the periphery? Strategic multiplexity can provide a solution to this dilemma. Organizations can employ multiplexity to build a ‘multiplex status’, i.e. different projections of status derived from different kinds of relationships. For example, organizations can use the most visible kinds of relationships to build connections to high status organizations (e.g. by forming strategic alliances or board interlocks with them), while less visible relationships (e.g. career mobility linkages, market exchange or membership in industry standard setting committees) might connect the organizations to lower status entities.

Multiplexity and network dynamics

Path dependence is the dominant model of network dynamics, suggesting that networks reproduce over time and firms engage in local searches for future partners within the pool of their prior partners. Yet, a big puzzle is what makes organizations engage in non-local partnering – forming relationships with completely new partners. Such relationships provide access to new information for the organizations and build the bridging ties comprising structural holes. Existing research suggests a variety of explanations for this risky behaviour, ranging from firms experiencing performance shortfalls (Baum et al., 2005) to the nature of uncertainty facing firms (Beckman et al., 2004) and their desire to access non-network resources (Mitsuhashi and Greve, 2009).

A strategic multiplexity perspective can suggest an alternative explanation for non-local partnering. If we assume the interdependence of different kinds of relationships, then relationships that appear new in one kind of network might actually be existing relationships with respect to another network. For example, if coaches of professional sports clubs A and B have developed a friendship relationship when these coaches worked for different clubs X and Y, then A and B will engage in player exchange under these coaches’ leadership even if these clubs did not exchange players in the past (Barden and Mitchell, 2007). In a different example, strategic alliance between two firms might be preceded by a long period of market exchange within which one firm was buying another firm’s products and the strategic alliance formalizes an existing relationship. In the latter case, organizations gradually reduce uncertainty about non-local partners and escalate their partnership intensity.

Furthermore, one kind of relationships that firms maintain with a common partner might affect the formation of their direct relationship of a different kind. Two producers might have no prior direct relationship with each other in a network of strategic alliances. Yet, these producers might actually share a common partner in a different network, such as a common customer with whom they had individual buyer–supplier relationships. Because a common customer can act as an indirect channel of information exchange between two producers, it can bring them together into a ‘multiplex triad’, even though neither this customer nor the relationships that this customer maintains to producers are observable if we only look at the relationships between producers themselves (Shipilov and Li, 2010).

Many companies have already recognized the importance of multiplex relationships. Investment banks, for example, are embedded in at least two types of networks – those formed as a result of their syndication of initial public offerings (Baum et al., 2003) and those formed as a result of their participation in merger and acquisition advisory teams (Shipilov, 2009). These two networks are managed by two different departments: syndication networks are managed by bankers in equity and capital markets (ECM) departments, while M&A networks are managed by bankers in corporate finance departments. When bankers in these departments exchange information, public offering deals can be orchestrated based on the information from the M&A department, while M&A deals can be done using the information received from the ECM department (Li and Shipilov, 2012).

Consulting companies like McKinsey and BCG maintain relationships with their former employees who move to work elsewhere. From these relationships, the companies get insight into what is happening in other companies and industries. If an alumna of a consulting company maintains her/his friendships at the former place of employment and also builds friendships with new colleagues at the new employer, then this person acts as an informal communication bridge between two companies. The consulting company will encourage its current employees to maintain relationships with former colleagues so that the company can learn about work opportunities in the other company and the former employee can eventually vouch for the former employer’s quality and skill. Thus, an informal tie between two companies formed as a result of employee mobility can eventually lead to the creation of a buyer–supplier relationship between the consulting company and the potential client (Somaya et al., 2008).

Reciprocal cross-level effects and aggregating mechanisms

It can easily be argued that most individuals embedded in multiplex relationships are aware of these relationships and their action in one network is shaped by relationships in another. The situation with organizations is very different, because multiple relationships can happen at different levels, i.e. between individual employees, between work groups, between business units or between organizations. To argue that relationships at one level influence relationships at another, or to propose that an organization can benefit from the structure of multiple kinds of ties, research on strategic multiplexity will need to take into account two related issues: reciprocal cross-level effects and aggregating mechanisms.

The prevailing logic of management research has been that the upper-level context within which lower-level variables are nested has a stronger downward influence than the lower-level variables on the upper-level ones (Hitt et al., 2007). With multiplex relationships, however, just as lower-level ties (e.g. interpersonal relationships between companies) are impacted by higher-level ties (e.g. strategic alliances), the lower-level ties can also impact higher-level ones. For example, informal relationships between scientists working for different pharmaceutical companies might initiate information exchange between them, which can lead to the subsequent formation of strategic alliances between these companies. At the same time, informal relationships between scientists working for different pharmaceutical companies will also be affected by strategic alliances between these companies. Ultimately, theoretical models within the domain of strategic multiplexity have to simultaneously examine reciprocal effects of relationships at different levels.

Econometrically, it seems that the reciprocal cross-level effects should be modelled as simultaneous regressions. In such regressions, variables capturing the structure of both networks should be allowed to influence each other. Ideally, one also needs to find instrumental variables that are correlated with the structure of one network but not with the structure of another in order to properly estimate such simultaneous equations.

To theoretically argue for any cross-level effects (including but not limited to the reciprocal ones) the researchers also need to theorize about the aggregating mechanisms. These can be defined as enablers of information and action flows across different levels of relationships in a collectivity (such as an organization). In the example above, how is information that the scientists collect by dealing with their counterparts in a different organization used by individuals in charge of strategic alliances (and the other way around)? This is where BTF with its focus on decision-making in organizational contexts (Cyert and March, 1963), and especially the attention-based view (Ocasio, 1997), can be very useful. To develop convincing theoretical arguments about the aggregating mechanisms one needs to theorize about specific players in an organization who are involved with using information from one network to build relationships in another. Who are these players and what are their roles and structural positions? It would also be important to examine formal and informal principles of action, interaction and interpretation that guide and constrain these players in building different kinds of networks. In addition, one needs to examine what resources these players have to allocate across networks and what technological infrastructure can facilitate aggregation of information across networks. What are the interests of these players, how are these interests related to their identities, what are the issues that these players have and how does managing relationships across different networks answer these issues? While this list inspired by the attention-based view is clearly not exhaustive, it hopefully presents a good starting point to think about aggregating mechanisms, as without them a paper on strategic multiplexity will not be complete.

Multiple methods to study multiplexity

In the introduction to a special issue on multilevel research, Hitt et al. (2007) emphasize the importance of multiple research methods. This could not be truer for research on strategic multiplexity. Given the importance of the insight into aggregation mechanisms, combining qualitative fieldwork with archival or survey data collection is vitally important for any scholar doing research within the strategic multiplexity perspective. Just like in institutional theory (Greenwood et al., 2011) and process research (Langley, 2007), the context within which one examines dynamics and outcomes of multiplex networks will also help researchers answer the questions about which kinds of relationships to include in any specific study.

Conclusions

In this essay, I attempted to sketch the research agenda for a strategic multiplexity perspective. This perspective has a rich theoretical and empirical basis within existing network research and it can be enhanced to examine how organizations manage multiplex relationships. Because of the inherent complexity of studying multiple relationships, this perspective calls for doing multilevel and multimethod research that will help us better understand the dynamics and consequences of networks affecting individuals, groups and organizations.

The strategic multiplexity perspective can have a broad impact on social science. As the world around us is becoming more interconnected, reciprocal cross-level effects become more pronounced in our economic and social lives. For example, established formal relationships of exchange and power in many societies have influenced informal communication for centuries, until they were ripped apart by the grassroots communication revolutions enabled by social media. New technologies increasingly play the role of aggregating mechanisms in collectivities, such as societies or organizations. The downward influence of macro- to the micro-structures is increasingly giving way to reciprocal influences in which both micro- and macro-structures co-evolve. By forcing ourselves to better understand strategic multiplexity, the associated aggregating mechanisms and reciprocal cross-level effects, we should be able to better understand the momentous changes happening in the social systems around us as well.