chapter  3
14 Pages

Why biased mediators bring peace institutions

The main claim of this book is that biased mediators are in a better position than unbiased ones to craft peace institutions that regulate the core issues in conflicts. In this chapter I will explain why biased mediators are important peace-brokers in civil wars. Furthermore, I will make two theoretical innovations. First, I propose a theory of biased mediation, where the effect of biased mediation is explained through four causal mechanisms. This theory systematizes and develops various insights from previous mediation theory into a comprehensive approach. Second, I utilize this theory to help to explain the variation in the contents of peace agreements. Whereas previous mediation research has discussed why biased mediators can have an effect on the parties reaching an agreement, it has paid less attention to what type of agreement is reached, and in particular, to the peace institutions crafted at the negotiating table. Hence, I will present a theory of biased mediation that can help to explain why biased mediators can bring institutional peace arrangements. The chapter also introduces and analyses the important distinction between rebel-and government-biased mediators. Lastly, the chapter addresses some possible objections to the claims of the book. In order to understand why biased mediators can bring about peace institutions in civil wars, we need to examine the logical sequence of biased mediation. There are basically four causal mechanisms at work when biased mediators seek to craft peace in internal armed conflicts. These need not play out simultaneously, but rather at different phases of the peace mediation process. Hence, one can conceive of a four-step sequence through which biased mediators influence the content of the peace agreements. First there is the motive of the biased mediator to intervene. Whereas there is always a set of different motivations for intervening, biased mediators share the basic concern of protecting their side. This gives them incentives to craft stipulations that benefit their side, the side which they have been supporting through the conflict’s armed phase. A second step in the sequence is the bargaining process between the parties, where they need to identify each other’s bottom lines (reservation points) and thereby find the exact range of possible agreements, which are mutually acceptable. The biased mediator can enhance this possibility by credibly sharing information about the parties’ capacities, their readiness to find a negotiated settlement, and their conflict cost tolerance. A third stage is the concession-making at the negotiating

table, where shifts in the bargaining positions are required for the parties to agree to peace institutions. Here the biased mediators are particularly suited for delivering their side to an agreement on how to regulate the central issues at stake through peace institutions. The last part of the mediation process sequence is related to the credibility of the parties and, in particular, how the parties making concessions can trust that the other side will not exploit these by cheating. A mediator biased in favor of the side offering the concessions can be necessary in order to be ready to take the risk of exploitation of the conciliatory offers. At the same time, accepting mediators biased against the other side is a way to costly signal peaceful intentions, and therefore a trustworthy way to show conciliatory intent. When biased mediators can offer such guarantees by counterbalancing concession cheating, the parties can trust each other enough to agree to core peace institutions. This chapter will discuss these four causal mechanisms of biased mediation. Protecting their side. The bias of the biased mediator implies that this peacemaker is not indifferent to the outcome. Biased mediators have certain preferences concerning the outcome of the conflict. This is the basic difference between unbiased and biased mediators (Kydd 2003). Biased mediators have interests that are more closely aligned with one side, rather than the other, and therefore would like to influence how those interests are secured. In other words, biased mediators engage themselves as third-party mediators at least partly because they want to ensure that the interests of their protégés are taken into account when the dispute is settled. The crucial question for biased mediators is therefore not how to get an agreement, but rather what type of agreement can be brought about through the mediation process. Biased mediators have preferences regarding the content of the agreement, and incentives to try to get agreements that have stipulations beneficial for their side. Thus, they will therefore act in order to ensure the interest of “their” side. What does this bias entail? The bias we consider here has been deduced from the actions of the third parties: they have already engaged in the conflict in order to support one side. The same reasons why they supported their side in the conflict also impel them to support them throughout peace negotiations. In order to understand the incentives of the biased third parties it is therefore important to take into account the reasons why support occurred in the first place; that is, the nature of the ties between biased mediators and their sides. These reasons can vary, but can be grouped together into three different forms: ideological, ethnic, and strategic. Bias by source can be ideologically based. Ideological affinity may tie countries together and be one of the rationales for the backing of one side by another. For a democratic state, the similarity in political culture may be a reason to side with one party in a conflict. An ideological common cause can impel countries to support one side in war and continue that support through the peacemaking efforts. The nature of the ties can also be identity-based, such as ethnic linkages, linguistic ties, or common religion. Linkages in identities may provide the motivation to support one side in the conflict, and try to get as good peace

arrangements as possible for the co-ethnic, co-lingual or co-religious partners. A third type is strategic. Alliances can be formed in order to counter perceived threats from others, or in order to strategically advance interests in the realms of politics, security or economics. Supporting one side in a conflict may be due to an interest in gaining influence over a neighboring country. All these three different types of links can be reasons for why one country would support a rebel group or the government side in an armed conflict, and is also the rationale for the decision to continue to be engaged in the peace-making efforts to end the armed conflict. It is noteworthy that an external actor can transform from being the supporter of one side into a mediator between the two. Zartman (1993) is the mediation scholar who has most explicitly discussed this transformation from supporter to mediator. Zartman explores the change of rebel-supporting so-called host-countries from backing the insurgents to mediating between them and their governments. He suggests that “the most prominent candidate for the role of mediator is a third party host or even distant patron of the insurgency, under special conditions” (Zartman 1995b: 21). Those special conditions are related to the costs for the host-country, since “a host not hurting has no interest in ending support for a rebellion, because a nonburdensome rebellion is not the worst and may be the best of neighbor relations” (Zartman 1995b: 21). Change requires substantially increased costs, usually through escalation of the conflict. When the costs of supporting the rebels are perceived as higher than the benefits, a positive “triangulation” can occur and “[o]nly then can the host envisage a helpful role in negotiations – that of a mediator, through which the host-neighbor helps to find a solution” (Zartman 1995b: 21). Thus, the fact that the biased supporter of one side decides to act as a mediator, indicates that the biased mediator has an interest in seeking a negotiated settlement. Hence, although they are biased – ideologically or strategically, or in terms of identity – they have also transformed from only supporting one side to negotiating between the two sides. Thus, tied together through ideological, identity-based, or strategic reasons, biased mediators have incentives to favor their friends and it is these incentives that can help to explain why they are concerned about getting institutional arrangements in place through negotiated agreements. Peace institutions that are formed during conflict create an institutional framework that guarantees the group (be it the government or the rebel side) protection in post-war society. It is their concern for their protégé’s safety in military or political terms that gives biased mediators the reasons to make sure that protections are stipulated in the institutional arrangements designed at the negotiating table. Likewise, concerns about the possibility for their side to have influence in the post-war society compel biased mediators to craft institutions that guarantee influence and the sharing of power. Thus, since peace institutions can also help to advance the protection and influence of their side, the mediator has good reasons to try to put them in place. Two qualifications to this explanation of the effect of biased mediators are needed at this stage. First, the incentives to protect their side do not necessarily

imply an sincere motive in safeguarding their interests. In contrast, these incentives can stem from a set of instrumental reasons. Protecting their side can be carried out in order to secure other benefits. This is the motive for mediation that Zartman and Touval call “offensive,” meaning “the desire to extend and increase influence” (2007: 429). Second, the incentives to protect their side do not necessarily mean that the preferences of the biased mediators and their favored side are identical. In fact, they can have quite different perspectives on the conflict, the solution and the process of getting to it. However, it still means that there are incentives to get an agreement that safeguards the interests of their rebels or governments. Revealing information. For a peace process to be successful, the parties need to come to similar expectations about the prospects for a negotiated settlement. An essential aspect of the bargaining situation is therefore how information can be processed between the parties in a credible and reliable manner. If the parties do not share the same information, they may fail to reach agreement on peace institutions. The critical information here is about the levels of resolve and the capacity of the parties in conflict. Biased mediators can have a role to play in terms of information-sharing in peace processes. Touval (1975) suggests that biased mediators may transmit information and be valuable as communication channels, given that parties may take their bias into account when interpreting and evaluating the information that the biased mediators send. They are therefore not necessarily at a disadvantage as information channels. Kydd (2003) takes this reasoning even further. Biased mediators have, according to recent rationalist theory, a relative advantage over unbiased mediators in terms of credibly revealing information. Whereas unbiased mediators have incentives to bluff for peace – to misrepresent information in order to enhance the prospect for peace – the biased mediators have incentives to be trustworthy to their side. Given that they possess information about the other, they can therefore help to bridge the gaps in information and expectations between the parties. Thus, biased mediators would be particularly well-suited for managing socalled “information failures,” that is, bargaining situations that occur when parties have information that the other side does not possess, and at the same time, strategic reasons for not revealing this information, to cheat, conceal or exaggerate, resulting in a sub-optimal outcome, such as war (Fearon 1995; Kydd 2010; Reiter 2003). The key aspect here is the ability to supply information in a trustworthy manner. In this way, mediators can help to address and overcome the strategic reasons for manipulating information and thereby assist in resolving conflicts through peaceful means (Gilady and Russett 2002). An emerging field of rationalist conflict resolution research explores how mediators can be designed in order to most effectively overcome information failure (Beardsley 2008; Favretto 2009; Kydd 2003; Rauchhaus 2006; Savun 2008). The information-revealing capacities of biased mediators have implications for the content of the agreements. Biased mediators will, by communicating information about the “other” side’s (B) degree of resolve and capability, help their side (A) to tailor the minimum concession that will be accepted. Knowing

this, the recipient (A) of the mediator’s information will find that the mediator has no incentives to exaggerate the “other” side’s (B’s) fighting spirit, military resources, or financial support. Hence, the way the biased mediator can be successful is by getting the side toward which they are biased (A) to make concessions. Biased mediators should, according to this theory, increase the likelihood of deals that disfavor their side. In fact, one of the hypotheses that can be generated from Kydd’s game theoretical model is that biased mediators will create asymmetric agreements with larger concessions for the side which the mediators are biased towards (Kydd 2003). Using Kydd’s explanation, we can therefore generate the expectation that biased mediators should bring about peace stipulations that are not necessarily beneficial for their own side. Delivering their side. Biased mediators stand in a unique position in respect of their protégés. The mediators biased toward the other side will be accepted because they hold particular possibilities to “deliver their side” (Zartman 1995b). This is especially valuable if they can use their influence over the side which needs to make the largest concessions. Hence a “biased mediator may [. . .] be the one with the greatest influence over the party that most needs to change” (Carnevale and Arad 1996: 42). Biased mediators commonly have more leverage than unbiased mediators over the parties in conflict and therefore more ability to press the parties to make concessions (Touval 1975; Touval and Zartman 2001). Indeed, leverage – an important explanation for success in the mediation literature – is often suggested to be closely linked to bias (Bercovitch 1991; Bercovitch and Houston 1993; Crocker 1992; Smith 1985; Touval and Zartman 2001). Providing support also implies the possibility of manipulating these connections in order to get one of the sides in the conflict to move in a certain direction. For instance, to provide military support to a government in conflict implies a bias, but at the same time, this means that such support can be a powerful tool of influence as the support can be decreased, paused, or terminated. Even just the threat to do so can be sufficient to get the party moving in the desired direction. Bias is commonly linked with the capability of protecting and influencing their own side: a mediator is biased because he has a history of a special relationship with one of the parties in the conflict. Biased mediators can manipulate this special relationship, for instance by withholding or increasing the resources that it gives to “its” side. According to Zartman, the so-called host-neighbor – that is, a state that has supported the rebels in another country – can have a helpful role in negotiations, to “deliver the agreement of the rebellion on which it has given support and sanctuary” (Zartman 1995b: 21). Unbiased mediators generally lack those special relationships with the parties in the conflict and are therefore less capable of using this kind of leverage. Mediators’ sources of leverage can stem from the primary parties’ need to find a way out of the conflict. A mediator can supply the services that are needed for an agreement to materialize, such as providing guarantees, offering protection and overseeing the implementation of agreements. Moreover, by exploiting the belligerents’ interests in side-payments by the mediator, a mediator can

create strong incentives for the warring parties to reach an agreement (Smith 1985: 367; Touval and Zartman 2001; Zartman 1995a, 1995b). A particular type of leverage stems from the ability to engage in balance-of-power relationships. Touval (1975) suggests that one important source of leverage for the biased mediators lies in their potential threat to disengage from their support of one side, or even to threaten to shift side and to join a coalition with the other side. The biased mediators can be accepted precisely because the disadvantaged side can see the possibility of creating a split between the mediators and their protégé. Biased mediators can strategically utilize this potential in order to press their side to make concessions. Hence, the biased mediator can bargain with their relationship towards the parties at stake, making their side agree to stipulations that benefit the other side. The leverage that biased mediators possess helps to explain why they can be particularly effective in bringing about concessions related to the basic contentious issues at stake. Thus, biased mediators can utilize their ties, and manipulate their side’s dependence on them as a way to press them to agree to stipulations that are not necessarily attractive or beneficial for the side they have been supporting. Giving up long-held positions regarding the basic issues and agreeing to make concessions will not come easily for the conflicting parties. The crafting of peace institutions is therefore likely to be done only after a certain degree of pressure. Biased mediators are accepted into the process in the anticipation that they can help to create this shift on the side of their friends. Strategically utilizing their leverage, the biased mediators can therefore contribute to creating central peace institutions. Counterbalancing concession cheating. The difference in biased and unbiased mediators’ incentives is also likely to influence the conditions under which belligerents will accept different types of mediators. Touval (1975) suggests that biased mediators may be accepted because they represent the best out of a set of alternatives. That is a logical starting point: it is the expectations of the different alternatives, and the costs and benefits involved in these, that can help to account for why biased mediators are accepted. Here it is important to recall the definition of mediation as a voluntary process, implying that the parties in conflict have a degree of power in determining the choice of mediator. It is not the fact that biased mediators can be accepted that is of primary interest here, but rather the implications that their acceptance can have for the content of the agreements between the parties: the possibility to achieve jointly crafted peace institutions, which regulate the core issues at stake. The primary parties in the conflict should be aware of the mediators’ bias and the incentive they have to secure the interests of those they have been supporting. Since mediators that are biased toward one side have incentives to protect their interests, they will be attractive as mediators to that side. The belligerents will therefore accept biased mediators for different reasons, depending on the direction of the bias. Hence, rebel groups will have incentives to get mediators involved who have supported the insurgency. Similarly, governments will have incentives to try to get their supporters as mediators. In fact, governments have

good reasons to request government-biased mediators in anticipation that they will seek to protect their side. Biased mediators can more easily understand their point of view. Using the bargaining terminology, they know the parties’ reservation lines. As mediators they will therefore not take the risk of proposing peace institutional arrangements that are far beyond what is acceptable. This is particularly important when the parties (both or one side) anticipate that there is a need for substantial concessions on core issues. The party who expects that it will have to make concessions (usually a status quo power that needs to give up at least some of its positions) has incentives to request a third party biased toward their side as a way of protecting against exploitation of the conciliatory moves they foresee making. Yet, biased mediators should, for precisely the same reasons, be considered unattractive to the other side. How should we understand the seemingly counterintuitive situation of accepting a mediator biased against oneself? Put differently, why would any side agree to accept someone as a mediator who has supported the other side? The explanation for this can found in the bargaining situation in which the parties are situated. A seemingly disadvantaged side can accept biased mediators as a way to help them to overcome commitment problems arising from the asymmetrical setting in which internal armed conflicts are fought. Commitment problems occur when well-meaning parties fail to reach agreements that are mutually beneficial because (at least) one side will have future incentives to renege or exploit an agreement. A party that is increasing in power in the future will therefore have difficulty in making its commitments credible in the present. In this kind of bargaining situation of rational mistrust, biased mediators can play an important role. The acceptance of a mediator biased toward the other side is a costly signal of peaceful intent, which can mitigate commitment problems that can otherwise hinder effective bargaining. Civil wars are asymmetrical, but that asymmetry is tilted differently depending on the dimensions and phase of the peace process. Governments are recognized, established, and generally more powerful in terms of capabilities. If the relative balance of power between the parties is expected to move from the rebels to the government in peace time, this presents the government with a problem of trustworthiness. For instance, the government side can have a credibility problem in terms of disincentives for implementing the agreements on various reforms once the rebels have demobilized (Walter 2002). Rebels, on the other hand, also have commitment problems. They may exploit the lull in the fighting in order to rebuild their strength and military capacity. They may utilize their international recognition as part of the mediation process, in order to advance their agenda. And they may exploit the possibility of obtaining access to territory or political structures in order to achieve a situation where they can negotiate from a stronger position, and subsequently increase their demands (Svensson 2007). Consequently, in situations where peace institutional arrangements are about to be formed which imply concessions from the parties, both governments and rebels have incentives to seek mediators who are either government-biased or

rebel-biased, respectively. It follows then, that the acceptance of biased mediators is associated with negotiation situations where one or both sides are about to make concessions. Anticipating that they are about to give up their maximalist positions in order to agree to peace institutions, the parties can have good reasons for wanting mediators biased toward their side. Only with the protection of a mediator they can trust, can a concession-making side be ready to agree to stipulations that regulate the core issues in a way that is not attractive to them. Consequently, the parties’ acceptance of a mediator biased against them can be the signal that helps to overcome rational mistrust and to create a bargaining situation where concessions on core issues can be made in a way that paves the way for the creation of peace institutions. Guarantees can be more credible when third parties biased toward the weakening side in an asymmetrical commitment problem make them (Schmidt 2005), or when they are made by mediators biased toward the government side in a peace process where the rebels are empowered through a settlement (Svensson 2007). On a similar line of thought, Walter (2002: 164) suggests that biased mediators can deliver the weak side and are more equipped to play a constructive role in the guarantee process than unbiased mediators.1 Hence, there are four reasons why biased mediators should be associated with peace institutions in civil wars. Biased mediators can intervene in order to protect their side. Biased mediators can also credibly reveal information between the parties. They may be accepted because of their capability in persuading their side to make costly concessions, since they possess unique access to their side and are therefore capable of using their leverage on the belligerents. Accepting biased mediators is a way for one side who will make concessions to protect against attempts by the other side to exploit such moves, and for the other side it can be a way of credibly revealing itself as a trustworthy actor. Taken together, these mechanisms suggest processes that are helpful in explaining why biased mediators will be associated with peace institutions. Up to now, we have treated biased mediators as one unified category of third party mediator. In this section, we introduce the distinction between government-and rebel-biased mediators. The basic relationship between parties in civil wars is that of asymmetry, in both material (resources, external support) and immaterial (recognition, commitment) terms. Hence, the two sides in civil wars are inherently different types of entities, with one side being a recognized and legal representative and the other a non-state entity, challenging the basic legitimacy of the government side. This implies that, theoretically, the effects of biased mediators can be quite different depending on whether the mediator is biased toward the government side or the rebel side. Opening this up into a more disaggregated analysis of biased mediators also enables us to theorize about the causal relationships between the different types of biased mediators on the one hand, and the various peace institutional arrangements on the other. This section will describe the basic expected effects of government-biased and rebel-biased mediators on different forms of peace institutions. In particular, we will focus on our three dimensions of peace institutions in civil wars – power-sharing, security

guarantees, and justice arrangements – and analyze why and how these peace institutions are more likely to be formed under the auspices of biased mediators, either supporting the government side or the rebel side. Power-sharing arrangements. As we have discussed above (in Chapter 1), there are three basic aspects of power-sharing. It can relate to the horizontal dimension of the state, regulating the formation, boundaries, and degree of sovereignty and power in different territories. This is territorial power-sharing. The other basic form of power-sharing is political power-sharing. Here the focus is on the regulation of central power, the division and sharing arrangements of the central authority of the state. These two types of power-sharing arrangements are clearly related to the types of conflict, so we could expect territorial powersharing to occur in territorial conflicts. Yet, the relationship is not clear-cut: power-sharing pacts can be made to address territorial conflicts, and vice versa. The third form of power-sharing has to do with security aspects in conflict: military power-sharing, through which the parties in conflict regulate some type of joint control over the armed forces. A causal process through which biased mediation affects power-sharing has to do with the demand for mediation. A party in conflict turns to a biased mediator when it expects to make considerable concessions regarding political and/or territorial power. To agree to power-sharing is costly, in particular for the government side. It implies formally recognizing entities that challenge the government’s legitimacy. It can hand over resources of state apparatus to a side that may have incentives to use them to increase their aspirations. In particular, if the agreement on regulating political or territorial power may put one or both sides in a vulnerable position, then a potential concession-maker may seek mediators biased toward their own side in order to have protections and guarantees. Unbiased mediators have, just because of their impartiality, fewer possibilities to counter-balance these risks for exploitation. Another way in which biased mediations may be causally related to political power-sharing has to do with the resources that biased mediators commonly possess and that they can bring to bear in a peace process. These resources give the biased third parties leverage over the parties by affecting their analysis of their costs and benefits. Unbiased mediators are either less resourceful, or can alternatively bring less resources because their interests and commitment to the peace process are generally at a much lower level. Thus, in general, biased mediators, rather than unbiased mediators, are expected to increase the likelihood of provisions for political powersharing. The effect of bias may be different, however, dependent on the direction of the bias. Political power-sharing is generally a stipulation beneficial for the insurgency, since it guarantees access and influence over political decision making in the aftermath of an armed conflict. In internal armed conflicts, rebels generally strive to achieve changes in the distribution of political power. If biased mediators have preferences close to the side towards which they are biased, then we would consequently expect rebel-biased mediators to try to get provisions of power-sharing into an agreement. Thus, rebel-biased mediators,

rather than unbiased mediators, are expected to increase the likelihood of provisions for political power-sharing. A second dimension of power-sharing in civil war peace agreements concerns the control of military power. Protecting their side, we can expect biased mediators to seek to incorporate military power-sharing arrangements into agreements. On the other hand, unbiased mediators will seek to enhance the prospects of a settlement, without going through the hurdle of negotiating military powersharing. Thus, in general, biased mediators, rather than unbiased mediators, are more likely to increase the likelihood of provisions for military power-sharing. Here again we can expect the difference between the effects of bias to vary according to the direction of the bias: for the government or for the rebel side. Military power-sharing will provide some sort of insurance to the minority, which fears future exploitation, that the government will not use force against them (Hoddie and Hartzell 2005). Hence, we could expect rebel-biased mediators to try to get the parties to reach agreements through military pacts. Moreover, rebels commonly fear that the government will not abide by the terms of the settlement and that laying down or handing in their weapons will put them in a vulnerable position vis-à-vis the government. Thus, previous literature has suggested that rebels commonly fear disarmament processes since these may leave them vulnerable to future attacks by the government side (Walter 1997). Similarly, rebels are commonly, on good grounds, suspicious that the government will fail to abide by joint agreements and neglect to implement costly reforms and therefore have disincentives to give up their control of military means. Mediators that are biased toward the insurgents will therefore have incentives to negotiate stipulations including guarantees into peace agreements. Thus, we can expect that, taking into consideration the side of the bias, that rebel-biased mediators, rather than unbiased mediators, will probably increase the likelihood of provisions for military power-sharing. A third dimension of power-sharing regards the decentralization of territorial power. We can expect biased mediators to look after the interests of their protégés, resulting in biased mediators being more likely than unbiased mediators to be associated with these kinds of arrangements. Thus, in general, biased mediators, rather than unbiased mediators, will probably increase the likelihood of provisions for territorial power-sharing. Territorial power-sharing is different from political or military power-sharing, since it gives the rebel groups certain benefits that are very costly for the government side and which can be used against it later on in the process. Unlike political and military power-sharing, territorial power-sharing gives the rebels a basis for demanding separation by creating the possibility of creating and maintaining regional identities. Moreover, territorial decentralization might enhance the organizational capacity of the rebels and allow for regional leadership to emerge (Cornell 2002; Lake and Rothchild 2005). Lake and Rothchild (2005) argue that territorial power-sharing can provide a costly signal of intent, indicating to the insurgency group that the government side consists of moderates willing to share power. Hence, territorial pacts present a high-risk institutional arrangement for

governments, more so than political or military power-sharing, in which governments still keep some control over joint institutions. Government-biased mediators are therefore needed to convince the government side to accept these types of costly concession. Hence, it is reasonable to examine the effect of government-biased mediators in this regard. Thus, government-biased mediators, rather than unbiased mediators, will probably increase the likelihood of provisions for territorial power-sharing. Security guarantees arrangements. Another peace-agreement stipulation that has been identified as important in previous research is third-party security guarantees. We can expect biased mediators to look after the interests of their protégés, meaning that biased mediators will be more likely than unbiased mediators to be associated with these kinds of arrangements. Biased mediators can affect the parties’ decision to regulate military relationships through our four causal mechanisms. Biased mediators have incentives to seek structural arrangements that serve to protect their side. The parties’ security considerations are vital in civil wars, and they have strong interests to protect in this regard. We therefore have good reason to expect that biased mediators will seek to incorporate military power-sharing arrangements into agreements. On the other hand, unbiased mediators have incentives to seek a quick settlement, without going through the hurdle of negotiating military power-sharing. A similar logic would apply to the decision to request or accept third-party security guarantees – biased mediators would have stronger incentives than unbiased mediators to get thirdparty security guarantees built into a peace agreement, so that the interests of their side are protected. The effect of biased mediators is related to their ability to gain concessions for the parties in the conflict. Biased mediators are more likely to deliver their side to an agreement on security arrangements. Getting parties to agree to give up autonomy over security arrangements will be immensely difficult, in light of the considerable perceived and real risks involved. For those making concessions on security, which basically means giving away some degree of power over their own ability to defend themselves and advance their interests through military means, there are good reasons to expect them to be in a vulnerable position. Therefore, if their concessions are counter-balanced by the presence of a mediator biased toward their own side, this can be a powerful argument to agree to security arrangements. Biased mediators commonly have essential resources that they can put on the table. In relation to security arrangements, there are two important resources. First, the biased mediators can bring other types of resources (military support, economic investments, political policy-changes) that will increase the relative merit of accepting security arrangement, in the form of military power-sharing, or third-party security guarantees. Second, the biased mediators can themselves supply the third-party security guarantees. This is an important resource. By guaranteeing agreements, mediators can also enhance their influence (Touval and Zartman 2001). Considering the substantial risks and costs involved in acting as a security guarantor, the incentives for unbiased mediators are to get

“cheaper” deals, without the peace institutions for security. Biased mediators, on the other hand, in light of the fact that they have already provided support to one side, will in general have the resources to be guarantors of agreements. Thus, the mechanisms of protecting one’s side, delivering one’s side, and possessing resources lead to expectations regarding the chances of the parties agreeing on military power-sharing arrangements and to the supply of third-party security guarantees. In general, biased mediators, rather than unbiased mediators, increase the likelihood of provisions for third-party security guarantees. In particular, rebel-biased mediators, rather than unbiased mediators, increase the likelihood of provisions for third-party security guarantees. Justice arrangements. An amnesty is one stipulation that very clearly benefits the side that is able to secure it. If biased mediators achieve biased outcomes, we would expect biased mediators, rather than unbiased ones, to be associated with the stipulation of amnesties in peace deals. In particular, mediators biased in favor of either the government or the insurgents should increase the likelihood of amnesties for government representatives or rebel representatives, respectively. This reasoning leads to the expectation that, in general, biased mediators, rather than unbiased mediators, would probably increase the likelihood of provisions for amnesties. More specifically, we would expect that rebel-biased mediators, rather than unbiased mediators, would increase the likelihood of provisions for rebel-sided amnesties, and that government-biased mediators, rather than unbiased mediators, would increase the likelihood of provisions for governmentsided amnesties. Restorative justice involves repair and reparation for the damage done to victims (Shriver 2003). Hence, it involves measures to restore the society and the social fabric by compensating the victims of atrocities. This study uses provisions for the repatriation of civilians in peace agreements as an indicator of restorative justice. Repatriation of civilians is another peace-agreement stipulation that benefits the belligerents and their followers. The peaceful return of civilians affected by the armed conflict should be a concern for mediators that have particular links to the parties. If, as my argument suggests, biased mediators commonly intervene to secure the interests of their preferred side, and unbiased mediators may be predominately motivated to mediate to rapidly end the conflict, we would expect to see that biased mediators rather than unbiased mediators try to get provisions for the return of refugees and internally displaced persons into an agreement. Thus, in general, biased mediators, rather than unbiased mediators, increase the likelihood of repatriation stipulations in peace agreements.