The Interaction Effects of Bargaining Power: The Interplay Between Veto Power, Asymmetric Interdependence, Reputation, and Audience Costs

For decades the probability of cooperation and the distribution of gains from cooperation have been important concepts in international relations theory and bargaining theory. Over time, scholars have developed different rationalistic theories to explain negotiators’ behavior and the outcomes of negotiations. Two of the most prominent concepts are veto power and asymmetric interdependence. Based on the veto player theory (Tsebelis 2002), scholars of institutionalism (Tsebelis 2002; see also Garrett and Tsebelis 1996) and the two‐level game (Putnam 1988; see O’Reilly 2005; Henisz and Mansfield 2006; Mansfield, Milner, and Pevehouse 2007) have analyzed how the number and preferences of veto players affect decision making in foreign policy and determine the probability of international cooperation.¹ Veto players are rational actors who have the institutional power to veto any change of the status quo. In contrast, scholars, for example of liberal intergovernmentalism (see Moravcsik 1998; Moravcsik and Vachudova 2003; Moravcsik and Schimmelfennig 2009) explain the distribution of gains from international cooperation by using the concept of asymmetric interdependence.² The concept of asymmetric interdependence is based on the idea that a state can more credibly threaten to end international cooperation when that action would hurt the other country more than itself. Veto power, on the other hand, relies on the fact that the status quo cannot be changed without the consent of all veto players.

Thus, both theoretical concepts would seem to be mutually exclusive (see Hirschman 1970; Slapin 2008, 2009). For this reason, more recent studies (Egeberg, Schaefer, and Trondal 2003; Bailer 2004, 2010; König and Slapin 2004; Schneider 2005; Slapin 2006, 2008; Tallberg 2008; Panke 2012) have tested these and other competing concepts of bargaining power (e.g., veto power, asymmetric interdependence, material resources, agenda setting, and asymmetric information) to analyze which of these concepts might better explain the outcomes of international negotiations.³ Depending on the type of international negotiations analyzed, these studies have produced different results.

But while the individual concepts of bargaining power have been well analyzed and studies have tested them against each other, the compatibility and interaction effects of the different concepts have remained unexplored. The different concepts are often seen as incompatible and competing, which is why the different theoretical approaches and empirical studies have neglected to consider how these various dynamics might interact.⁴

The nonconsideration of possible interaction effects not only diminishes the explanatory power of these separate theories, but also generates theoretical inconsistencies. Furthermore, it can lead to crucial biases in empirical studies. When the credibility of threats, for example, depends not only on the asymmetry of parties’ interdependence but also on the intensity of their potential reputational losses and audience costs, a lack of evidence that supports the role of asymmetric interdependence in that particular conflict would not necessarily contradict the underlying theoretical argument. It may just be ascribed to insufficient levels of reputational losses and audience costs. Furthermore, when reputational losses and audience costs operate in a similar way (as assumed below), it is hard to differentiate between them empirically. When researchers consider the role of bargaining power separately, the potential interaction effects that they ignore could bias their conclusions and lead to misinterpretations.

In this article, I seek to build a theoretical foundation for analyzing the interaction effects between and among the different forms of bargaining power. To achieve this, I focus here on how different forms of bargaining power affect the credibility of threats in international negotiations. I describe how asymmetric interdependence and veto power, which have been seen as mutually exclusive, can be combined and supplemented using concepts of reputational losses and audience costs to explain the credibility of threats. These fresh theoretical insights can help resolve inconsistencies that limit the power of the individual bargaining power concepts.

My analysis also raises questions about the results of empirical studies that have tested the concepts against each other but neglected possible interaction effects. Veto power and asymmetric interdependence can combine to explain bargaining behavior because domestic veto players are only “semi‐veto” players at the international level. Semiveto players can improve the status quo only through international cooperation, but they can deteriorate it unilaterally. Thus, semiveto players can use threats based on asymmetric interdependence to receive concessions vis‐à‐vis the other side.

Linking both concepts nevertheless serves to highlight a theoretical problem that analyses based on asymmetric interdependence have neglected to address. Even in the event of asymmetric interdependence, threats by the less dependent actor are noncredible if we assume the actors are rational and would thus never approve a policy that weakens their own position. Considering two additional concepts of bargaining power, negotiators’ reputation for resolve and audience costs, can help resolve this theoretical problem.

Both approaches are based on the argument that backing down from a threat imposes costs ex post on the threatening actor. According to the theory of negotiator reputation for resolve, negotiators who back down will suffer a loss of the reputation for resolve and thereby their future threats will seem less credible. According to the audience costs theory, the domestic constituency will punish political leaders when their deeds are inconsistent with their promises. Potential reputational losses and audience costs are both seen to tie the hands of the threatening actor.

Thus, for threats based on asymmetric interdependence to be credible, the party receiving the threats must incur a minimum level of reputational loss or audience cost. But asymmetric interdependence still favors the less dependent semi‐veto player because he or she needs to induce comparably smaller potential reputational losses or audience costs to make his or her threats credible. Veto power, asymmetric interdependence, and reputation as well as audience costs, therefore, interact with each other to determine together the credibility of threats and demands.

My analysis suggests that the various rationalistic concepts of bargaining power interact with and affect each other in different ways, and that understanding these interactions better illuminates the individual concepts, resolves their inherent theoretical inconsistencies, and more fully explains negotiator behavior and bargaining outcomes. These four forms of bargaining power – veto power, asymmetric interdependence, reputation, and audience costs – are good candidates for analyzing interaction effects because each one provides an explicit explanation for the credibility of threats. Other rationalistic bargaining power concepts including agenda setting, asymmetric information, material resources, and domestic constraints might also be interesting in this context, but exploring them would go beyond the scope of this article and are matters for future research.

In the next section of this article, I describe each concept and its impact on the distribution of gains and the probability of cooperation. I also examine the compatibility of the different types of bargaining power and examine the role of semi‐veto players, focusing mainly on how the interaction between different forms of bargaining power affects the distribution of gains and the probability of cooperation. Finally, I illustrate these concepts with a simple bargaining situation and empirical example.

The pioneering sociologist Max Weber ([1921] 2002) described power as the probability that an actor will realize his or her own will even against the resistance of others. Similarly, a prominent definition of power comes from Robert Dahl: “A has power over B to the extent that he can get B to do something that B would not otherwise do” (Dahl 1957: 202–203). These definitions of power have in common that they are consistent with rationalism and methodological individualism. Power is based on the ability to use specific (power) resources and is differentiated from luck (Barry 1980a, 1980b). These resources are controlled and intentionally employed by an actor (Barnett and Duvall 2005: 49–50).

Power, therefore, implies an element of counterfactual reasoning because the judgment about the effects of power presupposes an implicit assessment about the unaffected state of affairs; in other words, gauging someone’s power involves assessing what would have happened if that person lacked power (Guzzini 1993). For the purposes of this article, I define the four chosen specific concepts of bargaining power – veto power, asymmetric interdependence, reputation, and audience costs –in accordance with this rationalist understanding of (bargaining) power that is also known as “the first face of power,” that is, the exercise of power by command.

With regard to the definition of veto power, especially in rational choice institutionalism and the two‐level game, scholars generally refer to “veto player theory.” Although Robert Putnam (1988) did not define veto power explicitly when he developed the theory of the two‐level game, after the work of George Tsebelis (1995, 2002) many scholars have transferred the concept of veto player to the two‐level game and international politics (Minnich 2005; O’Reilly 2005; Henisz and Mansfield 2006; Mansfield, Milner, and Pevehouse 2007; Slapin 2008). Veto players are individual or collective rational actors whose consent is necessary to change the (legislative) status quo. Negotiators usually represent a veto player. For example, U.S. trade negotiators represent the president in international negotiations. Hence, a policy change is only possible when the outcome increases the utility of each veto player compared to the status quo (Tsebelis 2002; Wiberg 2009).

The basic hypothesis states that the larger the win‐set, the more likely a policy change or cooperation will be. The win‐set is the set of all possible agreements that increase the utility of (i.e., would satisfy) each veto player. (The win‐set is also known as the zone of possible agreement or ZOPA.) The size of the win‐set depends on the number of veto players and the programmatic distance between the veto players. Because adding a veto player can decrease the win‐set or leave it unaffected but never increases it, the higher the number of veto players, the less likely is cooperation. In addition, the larger the programmatic distance between the ideal preferences of each veto player, the smaller is the win‐set and therefore the less likely is cooperation (Tsebelis 2002). It is important to note that following the veto player theory (that usually focuses on domestic politics), all veto players have equal bargaining power because they can veto any change of the status quo. Veto player theory does not consider different material capabilities or dependencies of veto players. Veto players cannot compel concessions from others by threatening to take unilateral actions. Hence, veto power can help to explain the probability of cooperation but not the unequal distribution of gains.⁵

In contrast, the notion of asymmetric interdependence focuses more on distributive effects than on the probability of cooperation. This theoretical concept, which appears most frequently in the work of scholars of liberal intergovernmentalism (Moravcsik 1998; Moravcsik and Vachudova 2003; Moravcsik and Schimmelfennig 2009; Schimmelfennig 2015), can be found in the earlier work of Albert Hirschman (1945) and later in bargaining theory (Fisher and Ury 1981) and interdependence theory (Keohane and Nye 1977).⁶ Asymmetric interdependence assumes a situation in which an actor who profits less from cooperating is also less dependent on the cooperation partner, so she can threaten more credibly to break up the cooperation. This higher threat credibility can compel more concessions from the other side, because the less dependent party’s failure to cooperate will harm him less than it will the other party (Keohane and Nye 2001).

Although the asymmetric interdependence concept predicts the distribution of gains resulting from negotiations, it does not make clear arguments with regard to the overall probability of cooperation itself. On the one hand, many scholars (Schelling 1957; Moravcsik 1997) have argued that asymmetries of power intensify existing conflicts and thereby make it more difficult to reach a cooperation. On the other hand, Morton Deutsch (1973) noted that a clear power asymmetry helps to resolve conflicts and facilitates cooperation because the weaker actors are aware that they have to make more concessions than the stronger counterpart. William Zartman and Jeffrey Rubin (2002a) also argued that unequal power leads to more effective negotiations than equal power because small power asymmetries induce actors to pay much more attention to prevent any increase of the asymmetry. Thus, it is the perception of near‐equality that is more conducive to intensifying conflicts.

Another concept of bargaining power many international relations scholars refer to involves the negotiator’s reputation for resolve. Reputation depends on an actor’s past behavior and is used by others to anticipate future behavior. Translated into bargaining situations, this means that when a negotiator backs down from threats or demands, she suffers a reputational loss and reduces the credibility of future threats she might make. Actors will refuse to back down in order to prevent a reputational loss even when backing down would be the better option in the current situation (Satori 2002; Miller 2003).

Predicting future behavior reliably based on past behavior is only possible when the past and future situations are comparable (Mattes 2012: 684). Consequently, the more distinct the bargaining situation, the more readily actors will back down from threats, because the costs of reputational loss will be lower. Reputation usually relies on information gained not only from prior dyadic interactions but also from observation of how an actor behaves in relations to others (Crescenzi, Kathman, and Long 2007). In addition, actors do not have one single enduring reputation; they have multiple reputations that vary over time (e.g., change of governments), across issue areas (e.g., environment or security), and among various counterparts (e.g., allies or enemies) (Guzman 2008). Furthermore, an actor’s reputation comprises a range of behavioral qualities, that is, resolve, compliance, reliability, honesty, or any number of other characteristics, which can vary in compatibility (Brewster 2009; Kydd 2009).

Similarly, the audience costs theory can rationalistically explain bargaining failures that cost all parties (Fearon 1994, 1997). This theory arose originally from conflict and peace research but can also be applied generally to international negotiations in which audience costs appear to work as bargaining levers for gaining more concessions from the other side (Leventoğlu and Tarar 2005; Tarar and Leventoğlu 2009, 2013; Janusch 2015). According to James Fearon (1995), political leaders can tie their own hands by making public threats that would incur costs ex post from the domestic constituency when they should back down from their own threats. Thus, audiences punish their political leadership when their deeds are inconsistent with their words.

Audience costs can lead to a negotiation breakdown (and even war) if the threatened actor sees a threat as noncredible, but the threatener’s audience costs exceed his or her cost for going to war. According to this theory, the political leaders who can generate more audience costs accrue greater gain. First, the greater the ambiguity regarding time, place, and nature of the threatened response, the lower the audience costs, because the domestic audience will find it more difficult to pin down the government’s inconsistency (Snyder and Borghard 2011). Second, the threats and commitments made by political leaders in democracies are seen as more credible than those made by autocracies because their domestic constituency has greater opportunity for punishing them for a failure to honor such commitments, (e.g., at the ballot box) (Fearon 1994, 1999; Schultz 1998; Partell and Palmer 1999; Gelpi and Griesdorf 2001; Haynes 2012; Uzonyi, Souva, and Golder 2012).⁷ With regard to the impact on cooperation, the theory posits that audience costs in a democracy help increase the credibility of negotiator’s threats, which is why the other side is more willing to give in. Because political leaders in democracies can tie their own hands and signal their interests more credibly because their audiences costs are higher than those faced by autocrats, it is often assumed that democracies are more cooperative than autocracies.

I begin my analysis of the compatibility of the different forms of bargaining power by examining veto power as defined in veto player theory. Although classical veto player theory often assumes complete information, in this analysis I assume incomplete information. Because veto player theory considers only domestic politics, the special features of international politics must be incorporated before it can be used for analyzing international negotiations.

The two‐level game seems appropriate for this purpose. According to Putnam (1988), the politics of international negotiations can be described as a “two‐level game.” At the international level, national governments negotiate with each other to reach a tentative agreement that satisfies domestic pressures. At the national level, the same governments negotiate with their domestic constituencies to ensure the ratification of an international agreement. Because the modification of an agreement at the national level must then inevitably be approved by the foreign government at the international level, the two levels are necessarily interconnected.

Although some studies (e.g., Kastner and Rector 2003; West and Lee 2014) have emphasized that international politics and institutions (such as the United Nations, the World Trade Organization, etc.) impose constraints on domestic politics between veto players, scholars of veto player theory, institutionalism, and the two‐level game still frequently overlook the fact that the number of veto players – and especially the function of veto power – changes when veto player theory is transferred to international politics. The number of veto players in a political system can differ depending on consideration of domestic or international politics. For example, the executive branches in parliamentary systems are often mistakenly seen as veto players in domestic politics, although the legislature can often formally pass domestic legislation without the approval of the executive branch (Wiberg 2009). In international negotiations, in contrast, the executive branch constitutes a formal veto player because it must sign international treaties in most cases before the legislature can ratify them. If the executive can approve an international agreement by decree, the legislature actually loses its veto power. Thus, the probability of policy change at a domestic level, in comparison to the probability of international cooperation, can differ depending on the institutional rules. Although this fact should be considered in empirical studies, it has no further theoretical implications.

Regarding the function of veto power, the consideration of international politics has further far‐reaching theoretical implications for veto player theory. In contrast to domestic politics, in which all veto players must approve a policy change, in international politics veto players of a single country can implement a policy without the consent of foreign veto players. Thus, it can be assumed that international negotiations will take place in bargaining situations in which an improvement of the status quo can only be reached with the consent of every veto player in every country involved in the negotiation. No need for international negotiations and cooperation would exist if veto players of a single country could improve their utility unilaterally. In international politics the status quo may still deteriorate because veto players of a single country unilaterally can change their national foreign policy without the approval of foreign countries.

Veto power bargaining is therefore different in international negotiations than in purely domestic politics, and, for this reason domestic veto players can be better described as “semi‐veto players” in international negotiations. Semi‐veto suggests that individual or collective actors can veto an improvement, but not a deterioration in the status quo. Hence, semi‐veto players can be only better off through cooperation, which requires the consent of all semi‐veto players, but they can always take unilateral action that leaves all worse off. In other words, a semi‐veto player can only worsen the opponent’s best alternative to a negotiated agreement (BATNA) (Fisher, Ury, and Patton 2011) by worsening their own BATNA to some degree. They cannot improve their own BATNA without cooperation.

Some scholars have conversely argued that states can unilaterally improve the status quo and their BATNA, respectively. For example, Lloyd Gruber (2000) described bargaining situations in which State A can unilaterally change the status quo in a way that profits that state and damages other states. Because State A can threaten to “go it alone” without hurting itself, its threat is credible, and it can use that leverage to induce an agreement in its favor (see also Voeten 2001). Heather McKibben (2015) showed that states can improve their BATNA and worsen their opponent’s by holding another issue as hostage, offering concessions on a new issue, or removing an issue from the current agenda. Roger Fisher, William Ury, and Bruce Patton (2011) emphasized that negotiators can improve their BATNA by identifying and developing alternatives in the event that negotiations fail.

Although such bargaining situations and strategies are possible and can occur in international politics, I argue that states often confront situations in which actors can veto an improvement but not a deterioration in the status quo for three reasons. First, theoretical models in which actors can “go it alone” either end with an agreement acceptable to all actors (even if the actor who had a better unilateral outside option can compel more concessions) or the actor who acts unilaterally gets the best possible deal considering the noncooperation of the other side. In both cases, the bargaining ends with a situation in which actors can no longer go it alone and improve the status quo unilaterally. Thus, from a theoretical perspective, the models described by Gruber (2000) and McKibben (2015) have an inherent tendency to end with the bargaining situations that are assumed here as the standard situation between semi‐veto players.

Second, an actor’s ability to improve his or her own BATNA is often limited, dependent on both costs and the outcome of negotiations with third parties. Fisher and Ury (1981) offered the example of an actor who negotiates with Company X about a job offer. She can improve her BATNA by getting other job offers as backup. In the case I analyze below, the Andean countries can try to improve their BATNA in trade negotiations with the United States by diversifying their trade and negotiating trade agreements with other partners such as the European Union. But these actions cost time and resources and depend again on negotiations with others in which the BATNA is fixed.

Third, many empirical situations in international politics can be characterized as bargaining situations in which states can only veto an improvement but not a deterioration in the status quo, for example, when states threaten economic sanctions, military intervention, or the cancellation of an existing agreement. Thus, a bargaining situation as assumed here between semi‐veto players seems plausible. While the model here will not fit every international bargaining situation, it helps to explain seemingly irrational threats that put the threatening state in a worse position.

The author wants to make a theoretical contribution to explain such threats that seem to be irrational because the threating actor would put itself in a worse position. Again, the author, however, does not deny that other bargaining situations might be possible in international politics in which the model developed here does not fit.

The semi‐veto players concept not only helps illuminate the special features of international politics but can also work in conjunction with theories of asymmetric interdependence. As noted above, asymmetric interdependence means that an actor who is less dependent on cooperation can threaten more credibly to end that cooperation. Because semi‐veto players, in contrast to veto players, can detract unilaterally from the status quo, they can use asymmetric interdependence (the other party’s relative dependence) as leverage. Hence, domestic veto players with the capacity to change the status quo unilaterally but who lose less in the process than the veto players of a foreign country can more credibly threaten to stop cooperating and thereby press for more concessions.

In situations characterized by veto power and asymmetric interdependence, domestic veto players’ preferences must be coherent for threats to be credible. If the threatening nation’s veto player preferences are heterogeneous, and one veto player stands to lose more than the others, the threat becomes noncredible, because the veto player with the deviant preferences would rather block the implementation of the threat than disrupt an existing cooperation. But even when both conditions (power asymmetry between the states and coherence of domestic veto player preferences) are present, rational actor assumptions render threats based on asymmetric interdependence noncredible because rational actors would not make losing concessions. Threats to end a mutually beneficial cooperation are noncredible according to assumption of rationality.⁸

Applying theories of reputation (for resolve) and audience costs can help illuminate this problem. According to these concepts, as described above, backing down from one’s own threat or demand imposes costs ex post for the threatening actor, which is why he will refuse to back down even if it is the better option in the short term. Thus, both forms of bargaining power work in a similar way and limit the negotiator’s options. In asymmetric interdependence scenarios, the possible reputational losses or audience costs from backing down must be greater than those that would follow from the deterioration of the status quo by implementing a threat. Therefore, the parties that are less dependent on existing cooperation maintain a bargaining advantage, because credible threats on their part will entail lower reputational losses and audience costs. But threats are less credible even for less dependent actors if they suffer no costs ex post in the event of backing down, because they will suffer no penalties for backing down. Reputational losses, audience costs, or other costs imposed ex post after backing down are therefore a necessary condition for threats based on asymmetric interdependence to deteriorate the status quo to be credible.

It should be noted that, in the case of semi‐veto players, the BATNA can deviate from the status quo. If semi‐veto players make threats that would entail sufficient reputational losses or audience costs, the expected deteriorated status quo after fulfilling such a threat would replace the status quo as the new BATNA.⁹ In contrast, veto player theory assumes that the BATNA is always the status quo because each veto player can veto any change of the status quo.

The theoretical analysis suggests that the different forms of bargaining power analyzed above not only have interaction effects but are also mutually dependent, and that understanding these interaction effects can help explain some of each form’s theoretical inconsistencies. First, semi‐veto power is a necessary precondition so that an actor can make credible threats to cease cooperation should the opponent not meet specific demands. Second, the more homogenous the preferences of all domestic veto players, the higher is the credibility of their threats based on asymmetric interdependence. Third, reputational losses and audience costs or other costs that are imposed ex post if the actor backs down from a threat must be sufficiently high for threats based on asymmetric interdependence to be credible. Fourth, the hypothesis that the higher the asymmetric interdependence, the higher will be the bargaining power of the less dependent actor still holds because the less dependent that actor is, the less potential reputational losses or audience costs she or he needs for a threat to be credible.

To illustrate the interconnected concepts of bargaining power, a simple bargaining situation between three veto players from two different countries serves here as an analytical example. These players are the president (P), congress (C), and a foreign government (F). For simplification, the foreign country will be seen as a unitary actor and the preferences of P and C₁ as nearly similar. (C₁ and C₂ represent the same actor with different ideal points in two bargaining situations.)

In Figure One, each veto player is represented by an ideal point at which his or her utility is maximized in a two‐dimensional space. Each circle represents an indifference curve of one player, that is, all points on the curve have equal value for that player so one is as good as another. The status quo (SQ) determines the indifference curve for each actor and therefore the domestic win‐sets – that is, the sets of all possible agreements that increases the utility to each veto player – which, for the sake of better illustration, are in this case identical with the complete indifference curves. The overlapping area of both domestic win‐sets is the set of all possible international agreements that raise the utility of the domestic veto players and would be approved by them, or the ZOPA (see, e.g., Sebenius 2002).

Assuming rational actors, each veto player will try to approximate the new SQ as closely as possible to its ideal point, but will reject any point that is further away than the current SQ. According to Putnam’s hypothesis of the two‐level game, the larger each win‐set, the more likely international cooperation will become, because the more likely it will be if everything else stays the same that the win‐sets will overlap. Conversely, the smaller the win‐sets, the greater the probability that negotiations will break down. At the same time, however, a negotiator prefers a smaller domestic win‐set because it can be a bargaining advantage at the international level. The smaller the win‐set, the more credibly a government can refer to their “tied hands” at the domestic level in order to receive more concessions vis‐à‐vis the foreign counterpart (Putnam 1988; see also Tsebelis 2002).

Let us now consider a threat from semi‐veto players based on asymmetric interdependence. For example, if P and C₁ can change the status quo unilaterally to point x by ceasing cooperation, as illustrated in Figure One, they will weaken their own outcome, but less than F’s, so a threat on their part will be more credible. Assuming rational actors, however, this threat is credible only if it is clear that P and C₁ will face sufficient reputational loss and audience costs ex post from backing down. The reputational loss and audience costs must reach at least the value indicated by the distance between the indifference curve at point SQ and the indifference curve that would be at point x for both actors. This is indicated in Figure One by the dashed line. In this case, point x replaces the status quo as the BATNA. If the reputational costs, at least for one of the semi‐veto players, and audience costs are less than the distance between the indifference curves, backing down will be the better option and maintaining the status quo remains as the negotiator’s BATNA.

As previously discussed, threats become less credible as the preferences of domestic semi‐veto players become more heterogeneous. For example, if the ideal point of C is at C₂ instead of C₁, C will lose a great deal more than in the earlier instance from implementing a threat that results in point x. Therefore, the reputational loss must be higher for C₂ to be resolute rather than back down, because the relation between C₂ and F is less asymmetrical than between P and F.¹⁰

If a threat is perceived as credible, the size of the intersection between the win‐sets will change. At point x, for example, the intersection of win‐sets will increase by the medium‐gray area in favor of P and C₁. If, on the other hand, threatening actors expect that their reputational losses and audience costs from backing down would be higher than the losses they would incur by implementing the threat, the intersection of the win‐sets will comprise only the medium‐gray and dark‐gray areas. The light‐gray area will, in those circumstances, form no part of the set of possible international agreements, because P and C₁ would rather implement their threat at point x than accept an agreement in this area. Thus, the threat increases the domestic win‐set of F in such a way that the bargaining space at the international level includes more outcomes favorable for P and C_1.

Figure One illustrates that a set of agreements originally lying outside the intersection of the win‐sets becomes possible in the event of a threat based on asymmetric interdependence. Threats, in other words, can create an intersection of win‐sets that did not exist before. Assuming Putnam’s hypothesis, cooperation becomes more likely in the event of such a credible threat because of the larger intersection of the win‐sets.¹¹ However, arrangements in the medium‐gray area can be seen in terms of coercive rather than negotiated cooperation (Young 1989).

Although asymmetric interdependence is usually used to explain threats to revoke existing cooperation, it could also be applied to threats to abandon current negotiations for future cooperation (see also Moravcsik 1993). Thus, an actor who stands to gain less than his or her opponent from a putative future agreement can threaten more credibly to cancel current negotiations. This form of asymmetric interdependence adds a time dimension to the theoretical argument. Again, the threat is only credible if the reputational loss or audience costs incurred from backing down ex post are higher than the gains from a possible agreement. If this is the case, asymmetric interdependence will evidently tend to support any future cooperation that creates more nearly equal gains for all sides; because when threatening Actor A stands to gain more than Opponent B from a future cooperation based on a currently reached agreement, A immediately becomes more dependent and B gains bargaining power.

In contrast to threats to impair the status quo, threats to hinder future gains affect the distribution of gains but have no impact on the size of the win‐sets and their intersections if they cause no reputational losses or audience costs. This argument helps illuminate the negotiation situation in Figure One, in which P and C₁ can credibly threaten with point x, and helps explain why agreements with a high unequal distribution such as at point y are unlikely, because, at point y, P and C₁ would gain much more from agreement, which lowers their bargaining power vis‐à‐vis F. Thus an agreement that lies in the middle of the overlapping area of the veto players’ indifference curves (which cross at point x) is more likely.¹² The transfer of asymmetric interdependence from threats to cancel existing cooperation with actual gains to threats to break up current negotiations over future gains provides a rational answer to the classical conundrum of why weaker actors still achieve sizable – and often better than expected – results in negotiations with a stronger party (Zartman and Rubin 2002b).

Trade negotiations between the Andean countries and the United States provide a good example of a case in which governments function as semi‐veto players in international politics. The United States began negotiations over a regional free trade agreement (FTA) with the Andean countries of Bolivia, Colombia, Ecuador, and Peru in May 2004. In these negotiations, the United States used a possible nonextension of the Andean Trade Promotion and Drug Eradication Act (ATPDEA) of 2002 and cancelation of the FTA negotiations to pressure the Andean countries for concessions, especially regarding investment disputes and market access for agricultural products. Under the ATPDEA, the United States had temporarily granted Ecuador, Colombia, and Peru unilateral preferential market access (Villarreal 2011). A cancelation of the ATPDEA would have damaged the Andean countries economically much more than the United States, and the political costs for the United States in the fight against coca production and drug trafficking are arguably much less than the economic costs the Andean countries could have faced if the deal fell through.¹³ Therefore, an asymmetric interdependence characterized U.S.‐Andean trade relations.

This bargaining situation illustrates that the U.S. administration and Congress, whose approval was required for any extension of the ATPDEA, were semi‐veto players. Although an improvement of the status quo was only possible with the consent of the Andean countries, the U.S. administration and Congress could unilaterally deteriorate the status quo by abrogating the ATPDEA and not replacing it with an FTA. Because the administration of President George W. Bush and Congress supported a tough stance, the preferences of all domestic semi‐veto players were relatively similar, which is necessary for a threat to be perceived as credible. Furthermore, the U.S. administration expected to face relatively high reputational costs for backing down, because it was negotiating several FTAs at this time and expected comparable future investment disputes with the Andean countries and other trading partners.

Thus, the threat increased the intersection of the win‐sets in favor of the United States. Because of the high asymmetry, the coherent preferences of the U.S. administration and Congress, and the expected high reputational loss for the U.S. side, the threat did not miss its target. Before the official beginning of the FTA negotiations, the United States was able to reach settlement of most investment disputes and the Ecuadorian government agreed to U.S. labor standard demands (Inside U. S. Trade2004a, 2004b).

During the official FTA negotiations, the United States also made decisive progress with the Peruvians, who were willing to make more concessions, but the negotiations with Ecuador and Colombia were still at a deadlock in November 2005. Despite the asymmetric interdependence and coherent position of the Bush administration and Congress, their former threats showed no effect vis‐à‐vis Ecuador and Colombia. The upshot was that the Bush administration canceled the regional FTA negotiations with the Andean countries (Inside U.S. Trade2005; Villarreal 2006; Anonymous Interviewee Number One 2012a). This course of action boosted the U.S. reputation for resolve and hence the credibility of the threats to deteriorate the status quo. As a result, the Colombian government quickly signaled its willingness to make concessions in the bilateral negotiations (Inside U.S. Trade2006c; Anonymous Interviewee Number Two 2012a). The Colombian government finally gave in on many issues that it had been unwilling to accept in November (Gómez and Gamboa 2010; Inside U.S. Trade2006c, 2006d; Villarreal 2011). The Bush administration concluded the FTA negotiations with the Peruvian government in December 2005 and with the Colombian government in February 2006 (Inside U.S. Trade2006a, 2006b; Anonymous Interviewee Number One 2012a). In the case of Ecuador, the Ecuadorian government was also ready to make more concessions in spring 2006 to achieve an FTA (Reuters 2006). The FTA negotiations finally failed, however, because of massive protests in Ecuador and the resulting termination of a contract with the Occidental Petroleum Corporation by the Ecuadorian government, which was seen by the U.S. government as a violation of the existing bilateral investment treaty (Associated Press 2006; Crutsinger 2006; Palmer 2006; Anonymous Interviewee Number Three 2012; Janusch 2015).

In international relations theory and bargaining theory, scholars emphasize different rationalistic concepts of bargaining power to explain the distribution of gains and the probability of cooperation. Although the individual concepts are well analyzed and empirical studies have tested the competing theories, the compatibility and interaction effects of the different forms of bargaining power have remained unexplored. Two of the most prominent concepts tested in these studies, veto power and asymmetric interdependence, are emphasized by different theoretical approaches and seem to be incompatible. The power to veto a change in the status quo contradicts the argument of asymmetric interdependence whereby less dependent negotiators can threaten credibly to impair the status quo if their demands are not met.

This inconsistency, however, can be resolved by considering the special features of international politics. In international negotiations, domestic veto players constitute only semi‐veto players. Semi–veto players can veto any improvement of the status quo for which international cooperation is necessary but they cannot veto a deterioration that arises from the unilateral policy of a foreign country. Assuming the presence of semi‐veto players, asymmetric interdependence becomes compatible with veto power at the international level. Furthermore, it emphasizes that the preferences of all domestic veto players must be homogenous for a threat based on asymmetric interdependence to be credible.

Threats to abrogate existing cooperations nonetheless create a theoretical problem according to the assumption of rationality. Because rational actors would never implement a threat that decreased their own utility, all threats to cease cooperating would seem to be noncredible. Two additional concepts of bargaining power, negotiators’ reputation for resolve and audience costs, suggest a theoretical solution. Both concepts are based on the argument that negotiators face costs ex post if they should back down from their own threats, which damages their reputation and, therefore, lowers the credibility of future threats or risks generating costs to be imposed by the domestic constituency.

Reputational losses and audience costs can explain why negotiators do not back down from their own threats even if this would be the better option. But actors less dependent on agreement will still have a bargaining advantage because they require less potential reputational losses or audience costs to make credible threats. Thus, threats become more credible when asymmetric interdependence is high, the preferences of the domestic veto players in the threatening state are relatively coherent, and the threatening semi‐veto player would suffer a critically high level of reputational loss or audience costs for backing down.

Understanding the interaction effects between different forms of bargaining power helps analysts evaluate the credibility of threats in international negotiations. For example, when the administration of President Donald Trump threatens to leave the North American Free Trade Agreement (NAFTA), asymmetric interdependence obviously makes the threat more credible. First, Mexico’s economy is much more dependent on trade with the United States than vice versa. Second, the trade deficit gives the United States a bargaining advantage. It is easier to find new sellers for imports than new buyers for exports.

The theoretical insights emphasize, however, that all domestic semiveto players must support a threat for it to be perceived credible. Although the president can withdraw from a trade agreement, it is unclear (from a legal standpoint) whether this action requires Congressional approval. Because Congress passes an implementing bill for any trade agreement, it is likely that Congress is an additional semi‐veto player. In this case, a threat by the Trump administration loses credibility because the Republican‐dominated Congress supports free trade. Furthermore, although U.S.‐Mexican trade relations are characterized by an asymmetric interdependence, a withdrawal from NAFTA and the resulting increase in trade barriers could also damage the U.S. economy. Thus, the Trump administration must suffer sufficiently high reputational losses or audience costs. Otherwise, the threat is again not seen as credible. If the Trump administration expects no reputational losses or audience costs for backing down – for example, because the United States is not negotiating any comparable trade agreements at the same time and public opinion does not show that voters would punish a backing down – the threats remain noncredible despite an asymmetric interdependence.

Illuminating how veto power, asymmetric interdependence, reputation, and audience costs interact helps to better explain the distribution of gains and the probability of cooperation in international politics and helps resolve theoretical inconsistencies of the individual concepts. In addition, it also highlights that the concepts are theoretically mutually dependent to some extent. This analysis also raises questions about empirical studies that have analyzed a single specific form of bargaining power or that have tested the different concepts of bargaining power as competing theories. If the different forms of bargaining power are dependent on and can affect each other, empirical studies should consider this circumstance. The insights expressed in this article are useful not only for the light they shed on the compatibility of different forms of power and their interaction effects, but also because they help to bridge competing theories based on different concepts of bargaining power in international negotiations.