An Exploration of a Model of Social Networks and Multilateral Negotiations Open Access

Multilateral negotiations, that is, bargaining activity occurring among several parties at the same time, differ widely from the more standard variety of two‐party (bilateral) negotiations in several important ways (Druckman 1997). Economic and other incentives among the parties, the process by which agreements are crafted, as well as the views and characterizations of the outcomes can vary widely (Landau 2000). Multilateral negotiations are often international, which adds another important and complex dimension. For example, multilateral policy decisions regarding what constitutes a “global public good” such as policies concerning the Internet or general public health (Shaffer 2004) — as in the case of pharmaceutical patents — provide examples of the difficulties associated with multilateral negotiations.

Failure to reach agreement in multilateral negotiations involves more than a buyer and seller walking away from the bargaining table. The billions of dollars in cost overruns for the International Space Station (Cole 1998) and the collapse of the World Trade Organization (WTO) talks in Seattle (McClenahen 1999), which brought rioting in the streets, exemplify the challenges of international multilateral negotiation.

Despite the prevalence in the media of multilateral, cross‐national business and political negotiations, most treatment of the subject is anecdotal at best, with few studies examining and testing research propositions and hypotheses, and negotiation scholars have called for a more systematic look at how multilateral negotiations function in an international environment (Crump and Glendon 2003). In this study, we explore a previously published model of multilateral negotiations that considers social networks (Money 1998) and consider the degree to which problem‐solving activities occur to produce satisfactory or creatively beneficial outcomes.

We have used a simulation to explore constructs relevant to international negotiation, drawing on the theoretical work of Daniel Druckman (1993) — whose model included issues, background, context, structure, and immediate situation — and consistent with the work of Sarit Kraus and her colleagues (1992), who used a hostage crisis context as a backdrop. Our approach also builds upon the theoretical work of David Lax and James Sebenius (2003), who examined coalition building in multilateral negotiations. Indeed, our main focus is on the social network construct of clique formation (collaboration), which has received much attention in the management literature of late (see Shipilov, Rowley, and Aharonson 2006 for an example).

Researchers typically study international business negotiations by focusing on bilateral negotiations, that is, a dyad consisting of a buyer and seller (Graham et al. 1988) or partner‐to‐partner joint venture negotiations (Brouthers and Bamossy 1997). Multilateral negotiation is distinct from bilateral negotiation in three ways: (1) coalitions inevitably form between parties, (2) negotiators tend to vastly oversimplify the problem, and (3) various parties take on wider roles, such as leader, mediator, scapegoat, and blocker (Devine 1990). In addition, Druckman (1997) has written that multilateral negotiations are more likely to be conducted in public, involve more informal meetings between the parties, and result in fewer treaties (compared to bilateral negotiations) and deadlines. The ways in which coalitions and roles emerge in multilateral negotiations has gone largely unexplored (for an exception, see Polzer, Mannix, and Neale 1998).

The model we consider is unique to multilateral negotiations in that it primarily focuses on clique formation (coalition building) and secondarily on emergent roles (power‐oriented centrality). Furthermore, our study examines the consequences of these social network factors on the level of problem solving involved in the negotiation, as well as the outcomes as viewed by both parties (Figure One).

Social networks analysis, which considers the nature of the relationships between individuals in a social system (Granovetter 1973), has been used to study phenomenon in management (Rowley 1997; Labianca, Bass, and Grey 1998), marketing (Reingen and Kernan 1986; Iacobucci and Hopkins 1992) and economics (Berkowitz and Fitzgerald 1995). In negotiation, information from social network sources has been shown to improve the decisions negotiators make, helping them consider a wider range of outcomes (Rangaswamy et al. 1989).

Two social networks concepts are considered particularly relevant to multiparty negotiations: clique formation (the phenomenon of coalition building) and levels of centrality (the relative power among the players in a multilateral negotiation). (Both of these concepts are more fully defined below.)

Our primary focus in this research has been on coalition building. This formal network term measures to what extent ties between cohesive subgroups form among group members within the larger group of individuals (Knoke and Kuklinski 1982). Individuals who forge ties through social functions or informal conversations leading to similar positions on an issue would be considered a “clique” (Freeman 1996).

Cliques have been studied in a variety of contexts, such as in investment banking (Rowley et al. 2005), where individual bankers build and maintain cliques based on social attraction and collaborative tasks. Cliques in corporate partnerships form subnetworks that are linked through a relatively small number of intermediaries (Baum, Shipilov, and Rowley 2003). Other research on clique formation has focused on mental health practitioners, who form networks through reciprocated referrals and case coordination (Provan and Sebastian 1998), and high school teachers, who have been shown to be active in coalition formation based on their individual orientations toward teaching (Frank 1995).

Management scholars have found coalition building among parties to be related to interest alignment (Polzer, Mannix, and Neale 1998), while negotiation scholars consider it to be a key facilitating factor in multinational negotiation (Touval 1989). Coalition building can help form “policy networks” (Raab 2002) usually based on members’ utilitarian motives or resource seeking. The transition of Eastern Germany to a market economy was an example of the development of a network clique among government officials (Raab 2002). Additionally, Caroline Freund (2000) found that smaller, regional trading blocs with tight coalitions take actions as a result of negotiations that are more effective than outcomes produced by wider multilateral negotiations. In addition, coalitions shift the balance of power, especially when smaller nations use them to counter the power of larger ones (Chasek 2005). In the absence of a cohesive coalition, multilateralism gives way to unilateralism, as it did in the case of the Iraq War (Krause 2004). Formally, network analysis clique detection has been used previously to study the subgroups among economic development organizations (Hagen, Killinger, and Streeter 1997).

Another concept from social networks theory we use in our model is that of centrality, which is basically a measure of how influential or strategically placed a member of a network is (Freeman, Roeder, and Mullholland 1980; Ronchetto, Hutt, and Reingen 1989; Rowley 1997). Traditional networking theory measures centrality using such concepts as “betweenness” of actors in the network, which is the frequency of how often an actor lies along the path of contact between other actors in a social network (e.g., a common friend between two groups of teenagers). For purposes of our multilateral model, however, negotiation centrality implies the role of emergent power in a network. Power is, of course, a key bargainer characteristic in negotiations (Lewicki, Saunders, and Barry 2006) that has been shown important in multilateral negotiations, where power differentials can shift quickly depending on options available (Landau 2000). Bilateral negotiations tends to favor the stronger power player, which may not be the case in multilateral negotiations (Turner 1988). Smaller, less powerful players may enlist the help of stronger players (the classic “north–south” multilateral division, i.e., “strong–weak”). Researchers have found that in climate negotiations, that negotiators representing business concerns have been less cohesive than those representing environmental organizations because different businesses have differing goals and various firms have emerged over time as “power players” within this coalition (Pulver 2002).

The collaborative approach in negotiation is defined as the degree to which bargainers use cooperation, collaboration, and information exchange to achieve resolution (Graham, Mintu, and Rodgers 1994), or what Roy Lewicki and his colleagues (2006), among many others, have described as “integrative” negotiation. In contrast, the absence of a problem‐solving approach would leave negotiators bargaining in a more “distributive” style (also known as “win–lose” and “zero‐sum” approaches).

We hypothesize that collaboration positively influences negotiation satisfaction. The variable of satisfaction is measured in our study in two ways: how closely the actual agreement matched the negotiator’s original position or goal and how “creatively beneficial” solutions might be in generating outcomes that were previously unconsidered but were nonetheless advantageous to the negotiator — creating a “bigger pie,” in essence.

Researchers have found that small island states, through the use of coalitions in climate negotiations (Larson 2003), have proposed moreflexible alternatives, consistent with a more integrative strategy. Nancy Alder and her colleagues (1988) found that negotiators from different cultures apply different levels of integrative bargaining. Our model proposes that both clique formation (coalition building) and centrality (emergent power) are positively correlated with more frequent use of integrative bargaining and also with outcome satisfaction. Those negotiators who form coalitions and wield power would, we believe, be more motivated to use integrative approaches and generate positive outcomes, because of the difficulty of the “internal negotiation” (coalition building) in addition to the “external negotiation” with other members of the process at large (Winham 1987; Pruitt 1995). This is because the more people who become involved in the “internal” negotiation (as coalition members), the more difficult managing the negotiation becomes, thus motivating or forcing the coalition leader(s) to use a more collaborative approach to get things done. Additionally, once past the difficulty of forming a coalition and becoming central to the process, negotiators would be more likely to take pride in finishing the negotiation task (Lang 1991) by using a high level of integrative negotiation to the benefit of a wide range of parties. That is, the coalition leader would not like to seem like a “failed” negotiator to both coalition members and those he or she is negotiating with externally.

Therefore, we propose:

Finally, we suggest that problem solving might also be considered a result of network emergence in the process, a consequent construct as well as a concurrent one. In addition, studies have suggested that problem solving is more dependent on process (i.e., becomes a type of outcome) in multilateral negotiation than it is in a bilateral situation (Touval 1989). Consistent with the previously established positive relationships between more integrative approaches and mutually beneficial outcomes, we propose the following, final hypothesis:

We collected data by conducting a negotiation simulation based on actual events related to the sixteen‐nation effort to design and build the International Space Station, which was constructed between 1996 and 2004 at a cost of more than $24 billion.¹ The space station has proven to be a context‐rich, complicated, multilevel international negotiation whose many agreements have required renegotiation over the years (Cline and Gibbs 2003).

Simulation participants were MBA students from two large U.S. universities. Of the total of 375 respondents, 33 percent were women. The participants’ average age was 28.1 years old and they averaged 4.5 years of work experience.

We gave participants a brief description of the history of the space station and then randomly divided them into teams of ten people, with five subgroups on each team. Each subgroup of two people was then assigned a role in the simulation. The five roles were as follows: a representative of the National Aeronautics and Space Administration (NASA), a representative of the Boeing Corporation (the lead aircraft contractor), a representative of the European Space Agency (ESA), a representative of the Japanese space agency (NASDA), and a representative of CommBrazil, a satellite telecommunications company in Brazil. (The first four of these five roles were based on parties in the actual negotiation; the fifth party, CommBrazil, is fictitious for purposes simulating a non‐U.S. commercial interest in the station’s construction, although the country of Brazil was party to the actual space station’s development.)

The ten‐member groups were assigned to conduct a multilateral negotiation to determine four issues: which party would design and build the main module, which party would do the same for a robotic arm, what the annual funding cap would be for the entire project (what all parties would be able to spend on the station in any one year), and which party would determine and supply the payload on the space station’s first mission. (These items were loosely based on actual negotiating issues in the real project.)

We gave each subgroup a role that was manipulated to test the propositions of the study. For example, for organizational characteristics, members of the Boeing team were instructed that they were part of a large organization with much multilateral negotiation experience; Japan’s space agency, on the other hand, was described as a small organization with little experience in negotiating multilateral deals. Each two‐person team was given class time and a location for strategizing. Then we asked the ten‐person multilateral groups to convene and conduct the negotiation simulation and try to come to agreement on all four issues.

At the conclusion of the simulation, participants completed a survey that elicited both demographic information and information relevant to our research constructs. They reported that the simulation took an average of eighty‐five minutes to complete and not every group reached agreement on all four issues. Although the data were derived from self‐reported scores from survey items, this method was consistent with those of previous studies of negotiation behavior (Graham, Mintu, and Rodgers 1994; Wilkenfeld 2003).

We measured coalition building by asking participants which other party or parties to the negotiation they discussed simulation decisions with during breaks or otherwise outside formal talks, assuming that initiating and participating in such discussions indicated that the participants engaged in such conversations held similar ideas about the issues. We measured centrality by asking participants, “Who tended to emerge as the ‘power players’ in the negotiation group and why?” The number of times each party was named by another, plus who did the naming, were used to develop indicators of centrality.

We measured how integrative the negotiators’ approaches were using eight items adapted from John Graham and his colleagues (1988). We asked each member of the ten‐person group to rate their own organization and all the other organizations (i.e., NASA was asked to rate Boeing) for their level of interest in problem solving and also these organization’s perceptions of bargaining strategies (i.e., exploitive or accommodating, honest or deceptive, and biased or unbiased). Thus, this construct captured a general perception of the entire group’s negotiation approaches as perceived by each participant.

We derived our measures of outcome satisfaction, our second consequent construct, from six statements that described how negotiators perceived the outcomes of the negotiation, including assessments of satisfaction. Descriptive statistics of raw measures and reliability scores for composite measures for the use of the integrative approach and outcome satisfaction are reported in Table One. An exploratory factor analysis (EFA) of the raw measures revealed two constructs with eigenvalues greater than 2.0. Eigenvalues greater than 1.0 are generally accepted indicators of construct discriminant validity, that is, the different variables are not measuring the exact same thing and are appropriately included in the study. The Alpha reliability scores of each construct also exceeded .70, which is the acceptable threshold for convergent validity; that is, the variables, while not measuring the exact same concept, are sufficiently related to each other to give a more complete picture of what is happening in the negotiation.

Table Two (below) shows the number of times individuals within a simulation subgroup (e.g., NASA) indicated that the target negotiation involved discussions with another subgroup (e.g., ESA) outside the formal rounds of talks with the entire ten‐member group.

Table Two indicates that NASA and Boeing were involved in the greatest number of discussions as they developed coalitions with other organizations (NASA at sixty‐eight discussions and Boeing at sixty‐nine). ESA had the fewest at forty‐eight. Interestingly, those who engaged most in coalition building were also those who attempted to form coalitions with each other. For instance, NASA mentioned Boeing thirty‐one times and Boeing mentioned NASA thirty‐three times. Totals along the right side of the matrix show that, overall, organizations attempted to form coalitions most with Japan’s space agency (seventy‐seven) and Boeing (sixty‐two), least with NASA (fifty‐three) and ESA (forty‐five). Perhaps this is because the former were viewed as those who could help sway the opinions of others in the group, or that the coalition targets were most likely to agree with the initiators’ positions.

We also created a proximity matrix from the counts in Table Two.² A lower proximity indicates greater similarity between organizations relative to the focal measure. A Multi‐Dimensional Scaling (MDS) map based on the derived proximity matrix is shown in Figure Two. Organizations that are more likely to form cliques or coalitions with each other appear closer on the map; thus, Boeing and NASA were more likely to form a clique with each other than with other organizations, as was the Japanese space agency with the Brazilian communications company. ESA acted independent of these two cliques.

Table Three indicates the number of times a respondent identified another organization as one that became a “power player” in the simulation.

NASA is shown to be the dominant power player, being mentioned most often as a powerful member of the negotiation group (189 mentions), followed by Boeing (137 mentions). The Brazilian communications company was viewed least often as a “power player,” with fourteen mentions. Also, the Brazilian organization, more than any other, recognized other organizations as powerful (127 mentions), perhaps reflecting its own perceived lack of power. We created a second MDS map to explore similarities in centrality across groups, as shown in Figure Three. Maps transformed through multidimensional scaling often include interpretive dimensions. In this case a dimension has been added to indicate greater levels of emergent power.

The MDS map shows that, again, NASA and Boeing were both viewed as most powerful. Figure Three also shows that the ESA was also viewed by other parties as relatively powerful, although not as powerful as NASA and Boeing.

Hypotheses One (a and b) predicted that the level of the collaboration (integrative) approach perceived by the organizations would be positively correlated with levels of coalition building and emergent power. The mean values for how frequently the integrative approach was used for each organization as derived from survey responses are shown in Table Four. Results show that NASA (3.22) ranked highest in use of integrative approaches and Boeing (2.82) ranked the lowest.³ That is, NASA and its partners were perceived as being the most concerned with solving the problems of the group while Boeing and its partners were perceived as being most concerned with solving their own problems.

Also included in Table Four are the mention counts of clique formation (coalition building) and centrality (emergent power) reported in Table Two and Table Three, respectively. Totals in the right‐most column of Table Two represent clique counts in which Others Mention You (OMY) as a partner, while totals along the bottom row of the table represent clique counts where You Mention Others (YMO) as a partner. For the analysis, clique counts were divided by the relative number of respondents to reduce bias. For example, NASA clique (YMO) used in the regression is 68 mentions among 73 respondents, or 0.93 mentions per respondent. One would generally expect clique‐OMY and clique‐YMO to be highly correlated, but our data indicate that these two measures were unique and statistically uncorrelated and thus are included as independent predictors in the estimated model.⁴

Totals in the right‐most column of Table Three represent centrality counts in which Others Mention You (OMY) as a power player, and totals along the bottom row of the table represent centrality counts in which You Mention Others (YMO) as a power player. These counts are also included in Table Four, along with a measure where You Mention Self (YMS) as a power player. Power (YMS) is the unreported diagonal of the matrix in Table Three. For the analysis, we converted these counts to average mentions per respondent to reduce bias. We found a negative and highly significant correlation measurements of OMY and YMO.⁵ Similarly, we found a highly significant and positive correlation between the OMY and YMS measures.⁶ Power (OMY), power (YMO), and power (YMS) are essentially indicators of the same construct; as such, only power (OMY) is used in the regression model.

Hypothesis Two (a) predicted a positive correlation between the antecedent constructs and the collaborative (integrative) negotiation approaches. We constructed and evaluated alternative regression models, controlling for bias and testing for nonlinear effects. The final model showed clique formation and centrality to have nonlinear effects on integrative approaches, after controlling for negotiation duration and gender. These results are shown in Table Five.

Low variance inflation factors (VIF) serve to confirm the independence of the modeled effects. Our results support the proposition that centrality has a positive effect on the frequency with which negotiators used integrative approaches.

Interestingly, the effects of clique formation (coalition building) are mixed. The frequency of the use of integrative approaches increased when Others Mention You⁷ yet decreased when You Mention Others.⁸ If these perceptions were consistent (i.e., Others Mention You=You Mention Others) the net effect is positive (see standardized betas). But coalition building had a negative effect on the use of integrative approaches when You Mention Others more than Others Mention You or, rather, when an organization’s self‐perceived contribution was not recognized by its partners.

Hypothesis Two (b) predicted a positive correlation between the concurrent constructs and satisfaction. Tests of alternative models show power (OMY) and clique (OMY) to have nonlinear effects and clique (YMO) to have a linear effect on the use of integrative approaches. Estimates reported in Table Six support the proposition that centrality has an overall positive, albeit nonlinear, effect on satisfaction (see standardized betas). Again, clique (OMY) squared and clique (YMO) have similar, but offsetting effects. Finally, consistent with our third hypothesis, we found a positive correlation between the use of more integrative approaches and participants’ overall satisfaction with the negotiation.⁹

Our results reveal a surprising relationship between the constructs of our study. The effect of centrality (which we also call emergent power) was positively correlated with greater use of more integrative negotiation approaches and with greater satisfaction with negotiation outcomes, but the effect of coalition building was dependent on participants’ relative perspectives. When organizations reported coalition building efforts that were not acknowledged by their negotiating partners, they believed that the negotiations were governed by greater self‐interest, exploitation, and deception (i.e., fewer efforts at integrative negotiating) and, as a consequence, were less satisfied with negotiation outcomes. This is an interesting finding for negotiators who consider building coalitions, particularly as they consider whether or not to acknowledge the coalition‐building efforts of others. Otherwise, coalition building has a positive overall effect on both integrative efforts and satisfaction. That is, those participants who reported they engaged in more efforts at coalition building reported greater satisfaction with the negotiation outcome.

One explanation of this phenomenon is that dominant organizations may exaggerate the value of their own contributions, perceptions not shared by their weaker partners. Table Four indicates that the perceived contributions of the “power players” were often not acknowledged by their partners. For example, NASA and Boeing mentioned others less than others mentioned them [clique (YMO) > clique (OMY)] while representatives of the Japanese and Brazilian entities mentioned others more than others mentioned them [clique (OMY) >clique (YMO)].

Furthermore, mean satisfaction scores were generally lower for NASA and Boeing. This would seem to contradict findings that power is positively correlated with satisfaction. It appears that while power is desirable in complex multinational negotiations, asymmetric perceptions of relative contribution can be either detrimental or perhaps indicative of serious problems. This finding is consistent with recent negotiation literature (Lax and Sebenius 2003) that suggests that in conducting negotiations, even from a position of dominant power, an attitude of amicability and mutual benefit tends to produce superior results in terms of satisfaction and prospects for doing business again in the future. Fortunately, it is possible for negotiators to express collegiality and their awareness of other parties’ contributions without relinquishing their centrality or power.

While the results of the study are revealing, some limitations should be noted, which future research, in addition to furthering knowledge in this area, could address. First, data were collected by way of a classroom simulation. Although consistent with prior international negotiation research (Graham, Mintu, and Rodgers 1994), this approach is subject to the weaknesses of contrived circumstance. The complexities of multilateral negotiation necessitated that the variable measurements be taken in a controlled environment, consistent with the difficulties and trade‐offs of attempting to design research that is at the same time real, accurate, and generalizable (McGrath 1982). An omniscient point of view would be required to analyze and test the myriad of the dynamics in a multilateral negotiation, actual or contrived. Nevertheless, future studies might address this limitation by emphasizing the “field” aspects of data collection, perhaps with meaningful in‐depth interviews of negotiators who have participated in multilateral negotiations, for example (McCracken 1988). Another interesting finding that clearly deserves further study is the question of why the average satisfaction levels in our study were lower among more powerful players than they were among the weaker players.

Notable is the relatively low R², which indicates that only 10 percent of the variation in the dependent variable is captured by the constructs included in the model. This suggests that integrative approaches and outcome satisfaction are also affected by forces other than coalition building and centrality, such as loyalty. The low model fit does not negate the significance of our reported findings but rather suggests a need for future research of the phenomena.

In sum, we found interesting associations as we explored the hypotheses herein, dealing with relationships between coalition building, centrality/power, collaboration/integration, and satisfaction with the negotiation outcome. We found that the aspects of multilateral negotiations affected by the relationships between these variables bring their own set of unique challenges and influences on the outcomes, an important contribution of our study.

Our results indicate that coalition building can have a mixed effect on the level of problem‐solving collaboration that negotiators bring to the table, as well as their satisfaction with the outcome. Centrality or power, on the other hand, was found to have a positive influence on both problem solving and satisfaction, although the relative levels of satisfaction were actually lower among the powerful players when compared with the weaker players. Furthermore, when parties mutually acknowledged coalition‐building efforts, these effects were positive, but the frequency of integrative efforts and negotiator satisfaction both declined when negotiators’ coalition‐building activities were not acknowledged by their counterparts. This finding is important because it is something that negotiators can do without a great deal of effort (acknowledging coalition building) that could improve satisfaction with outcomes for both parties. Finally, we found, as expected, that higher levels of integrative activity in negotiations increases satisfaction with the outcomes.

Our study has important implications for both negotiation academics and managers. For negotiation researchers, we empirically explored combined constructs from business, social network theory, and the negotiation literature to examine and illuminate the complexities of multilateral negotiations (Pruitt 1994). This study therefore adds to the body of knowledge in a wide range of areas and lays groundwork for future research in multilateral negotiations.

For business managers and public policy officials making substantive decisions, our study underscores the complexity of multilateral negotiations (Tarar 2005). The social networks aspect of negotiations, which has previously been given little attention in research, can significantly affect how multilateral negotiations operate and how satisfied the parties are with the outcomes. In particular, our research points to the fact that building coalitions may not be as satisfying as wielding power, but those who are more powerful seem to be less satisfied than those with less power. This finding may be particularly important to managers and policy officials who must consider the most effective ways to accomplish their negotiation objectives, particularly when groups of constituencies are involved. Compared to traditional two‐party negotiations, such complexities of multilateral negotiations are quite varied and expansive, worthy of significant further theoretical exploration and sustained empirical inquiry.