The Risks of Peace: Implications for International Mediation

Analyzing the practice of international mediation raises some basic questions about what third parties can do in an intense conflict situation, under what circumstances, and to what effect. In this article, I argue that the risk management aspects of international mediation, in particular in the context of the design and implementation of negotiated peace agreements, require fuller attention. In situations of intense (i.e., violent) conflict, international mediators have a critical role to play in managing the risks of a negotiated settlement both during and after its signing. Risk is usually associated with danger or hazards involving loss. The Concise Oxford Dictionary describes risk as “a situation involving exposure to danger, the possibility that something unpleasant will happen” (Concise Oxford Dictionary, 11th ed., s.v. “risk”). Most discussions of risk in the classical economics literature provide a more precise definition of the concept: risks are, in effect, measurable, that is, the outcomes and the distribution of the probabilities associated with those outcomes are known and can be calculated (Crouhy, Mark, and Galai 2000; Marshall 2000). In many social and political contexts, however, probabilities and outcomes are not always measurable nor quantifiable with any degree of statistical precision. Thus, the term risk is often used to denote negative consequences associated with social and political actions or behaviors when there is a fairly high degree of uncertainty associated with those consequences. It is this more general conception of risk that is examined in this article.

In much of the international negotiation literature, the subject of risk management has received relatively little attention. Mediators are typically characterized as managers of a process of communication and exchange between adversaries (Bercovitch 1996; Bercovitch and Rubin 1992; Fisher 1997; Kelman 1996, 1997; Kriesberg 1986, 1992; Saunders 1996). Through their interventions, mediators can change the perceptions and attitudes of hostile or warring parties toward each other and build enough trust that the parties are prepared to negotiate, sign, and implement an agreement. In other words, mediation is a trust‐building activity that creates the conditions for reciprocity so that the parties will eventually negotiate a political settlement and lay down their arms.

Communication‐based approaches to mediation typically stress the contribution of third parties in providing a “neutral” forum where parties can explore options and develop solutions, often outside the highly charged arena of a formal negotiating structure. Mediators can also change perceptions by appealing to the overarching goals and values of the parties, playing on their aspirations for legitimacy and their desire to be part of the broader political community. The establishment of dialogue, of patterns of exchange and contact between and among official parties, other influential representatives, and even groups of civil and opinion leaders helps set the stage for a negotiated resolution to conflict. Surveys of the literature document a wide variety of communication and facilitation techniques that mediators deploy to alter parties’ relationships and promote dialogue and discussion (Bercovitch 1984; Wall and Lynn 1993).

However, in many acute conflict situations the parties never really learn to trust each other, especially if a conflict has lasted for many years (Crocker, Hampson, and Aall 2004; 2005). As Saadia Touval (1982a) has argued, mediators are essentially risk managers when parties exhibit high levels of distrust toward each other. And the persistence of high levels of mutual distrust does not necessarily preclude conflict abatement if ways can be found to effectively reduce risks and “insure” the parties against the costs of negotiation failure.

On the basis of a review of the business literature on risk management, Touval (1982a) identified a number of risk management devices that mediators employ during the negotiating process in order to reduce risk and achieve political accommodation. These include measures to transfer or shift risk such as:

• bringing a third party into the negotiation who can quietly probe and assess the intentions of the other side;

• developing deliberately ambiguous commitments that can be reinterpreted or manipulated as circumstances change (also known as hedging);

• sharing risks so that potential losses if a negotiation fails are more or less equally distributed among the parties; and

• segregating assets to limit liability, by, for example, separating issues and taking a step‐by‐step or incremental approach toward negotiations.

Since Touval's seminal essay on the subject, other studies have explored additional strategies that mediators can use to manage perceptions of risk in situations of violent or acute conflict (Hampson 1996a; Mitusch and Strausz 2000; Princen 1991; Zartman 1989; 2001). These studies suggest that strategic behavior in ethnic or civil conflict situations is motivated by long‐standing intergroup differences that reinforce the so‐called security dilemma (a situation where even defensive actions are viewed as offensive) such that the parties view the costs of cooperation as consistently outweighing the costs of defection. In these kinds of conflicts, defection is the dominant bargaining strategy unless a mediator can be persuaded to intervene.¹ As Donald Rothchild argues, “when adversaries confront each other directly and no mediator stands between them, a shift in strategic interactions can prove difficult if not impossible . . . However, the structure for interethnic bargaining changes significantly when a third‐party mediator intercedes and attempts to influence the adversaries to alter their perceptions on the benefits of reaching an agreement” (Rothchild 1997: 1). Among the various risk‐reduction strategies identified by Rothchild and others (see Dixon 1996; Rothchild and Lake 1998; Snyder and Walter 1999; Stedman 1991, 1997) are communication, confidence‐building (through the provision of various kinds of verification mechanisms that allow each party to continuously assess the military capabilities and intentions of the other side), and “spoiler management” (devising strategies to suppress or exclude those parties who, for whatever reason, are intent on wrecking the peace process). All of these risk management strategies are designed to change the cost–benefit calculus of the parties and raise the incentives for a negotiated agreement.

In the discussion that follows, I argue that mediators must contend with not only the risks of negotiation as argued by Touval and others, but also the risks of peace when they intervene in acute conflict situations. If left unattended or otherwise ignored, these latter risks will scuttle an agreement before the ink is dry or even before final negotiations are concluded. Further, there are different kinds of risk that can bedevil a peace agreement, and only some are amenable to negotiated solutions. These risks are described in this article as covenant (i.e., settlement) risks, some of which can be completely “cured” by mediators (and by actions taken by the parties themselves) and others that cannot. Mediators must recognize that there are real limits to their ability to shape the strategic preferences of the parties and their respective understanding of who will bear the burden of risk after a negotiated settlement is concluded (Haass 1990; Pillar 1983; Touval 1996; Zartman 1989). This has important implications not only for the timing of mediated interventions, but also for mediator tactics, roles, and responsibilities in the post‐settlement phase of a peace process.

In addition to curable and incurable covenant risks, which are reflected in the terms of trade in a negotiated settlement, there are a variety of legal and systemic risks to be considered in the design of a negotiated settlement. Again, I argue that mediators must understand how to manage and anticipate these risks in their negotiated interventions.

It is widely understood that the signing of a contract, or covenant, is not a risk‐free undertaking. But what may be less well‐appreciated is that there are different kinds of risk that lurk in the terms of an agreement. Incurable covenant risks are the risks associated with the possibility that a negotiating partner will not live up to his or her specific obligations in a contractual arrangement when they fall due (or any time thereafter) and there is no judicial or political remedy for recovering those losses from the defaulting party once the contract or settlement has been enacted. Such risks are associated with the costs (current plus future) of replacing a “contract” or agreement if the other side defaults and fails to live up to its negotiated obligations. Incurable covenant risks tend to affect parties’ perceptions about the desirability of an accord not only when they enter into negotiations, but also as those negotiations play out into the endgame. (For example, it is not just the costs of settlement versus the costs of “no‐agreement” that matter, but also the anticipated future costs of replacing a failed agreement, especially if, after one side has fulfilled its own part of the bargain, the other side later reneges on its commitments.)

The concept of an incurable covenant risk has special relevance to the field of international conflict management, in particular when contractual obligations in a peace agreement have asymmetrical levels of risk exposure (e.g., the land‐for‐peace exchange in Israeli–Palestinian negotiations). The party (in this case Israel) relinquishing a nonreturnable asset (land plus sovereignty) may deem the replacement costs to be unacceptably high if the “contract”— in this case a final peace agreement — fails. This is because the exposed party (Israel) is concerned that once the transfer takes place the other side (Palestinians) will not fulfill its security obligations (which continue) under the terms of the settlement, thereby leaving it exposed and without any viable means for recouping its losses (see Kriesberg 2001; Rabinovitch 1999). (Such risks are akin to “sovereign risks” in international finance where there is no remedy if a national government defaults on its loans and the lender cannot seize the assets of the borrower.)

Such risks are evident in conflicts where national governments are fearful about negotiating agreements with secessionist groups that challenge state sovereignty through violent means as in the case of Northern Ireland, at least until recently, and Sri Lanka (Jensen 1997). Parties’ perceptions that risk exposure is asymmetrical — especially when territory is an important strategic asset — may be reinforced according to prospect theory and the so‐called certainty effect, which states that parties typically tend to undervalue uncertain outcomes (e.g., improved relations with an adversary that might result from negotiation) and typically place a higher value on certain outcomes (i.e., those that can be measured and are tangible, such as control of territory) (Kahneman and Tversky 1972; Tversky and Kahneman 1992).

Asymmetric levels of risk exposure are not restricted to issues of territory and sovereignty in international relations. Many power‐sharing agreements, as well as agreements that require the parties to demobilize their military forces before elections are held, also involve asymmetric levels of risk exposure that have an “incurable covenant” element to them (Hudson 1997; Sisk 1996). This is especially true for the party that is required to demobilize (or disarm) its forces (and/or integrate them with government forces) under an agreement where participation in a political process, such as elections, is contingent on demobilization taking place first. The risk lies in the possibility that the other party, which controls the government, will not live up to its political commitment to allow the opposition to participate freely in the political process after it has relinquished its military assets (Hume 1994).

For example, the implementation of the 1992 Mozambican peace accords required both the Mozambican government and the main opposition party, the Renamo, to adhere to a strict timetable under which both sides would demobilize their troops and adhere to a cease‐fire before elections (in which Renamo would participate as a political party) were to be held. Although some government forces were to be demobilized under the accords, the agreement in a real sense shifted the burden of risk onto Renamo because once Renamo demobilized (even though some of its forces would be integrated into the Mozambican army) it would, in effect, lose its military assets and its bargaining power over the government if the government subsequently reneged on its commitments to allow Renamo to participate in elections. Renamo's leader, Alfonso Dhlakama, understood his dilemma all too well. As Aldo Ajello, the United Nations’ special representative to Mozambique, noted: “At the beginning, it was evident that Dhlakama's objective was to keep his troops in the bush as long as possible in order to preserve his bargaining power with the government. Keeping open both his military and his political options was the ideal solution for Dhlakama, but clearly impossible if the peace process was to proceed” (Ajello 1999: 632). Ajello successfully resolved the problem by reminding Dhlakama that “the presence of UN troops was his safety net, a kind of ‘life insurance’ for him in his new role as a political leader” (ibid).

Domestic political forces that are not amenable to any kind of negotiated quick fix can also increase covenant risks. The failure of a peace process or ensuing agreement can have serious repercussions at home, leading to the fall of a coalition government, an irredeemable loss of confidence in the leader who negotiated the agreement, and/or a major escalation in violence if military action is required to restore the security situation and the political or territorial status quo ante. As Robert Putnam (1988) argued, domestic as well as international political imperatives influence a leader's negotiating behavior, as does his or her ability to construct, maintain, and/or sustain coalitions among different domestic constituencies. However, it is not simply how their domestic constituents perceive the costs of “agreement versus no agreement” that influence a leader's bargaining strategies and decision‐making calculus. Anticipated risks matter also because leaders are not always able to check with their constituents at each negotiating turn; more often than not, they must only anticipate public reactions because negotiations are cloaked in secrecy.

The possible domestic political repercussions of their bargaining strategies have not been lost on the leaders of successive coalition governments in Israel, who have had to worry about how concessions they make at the negotiating table will affect the stability of their party coalitions and bases of political support in the Knesset and general electorate. For example, U.S.‐sponsored negotiations continued after the assassination of Prime Minister Yitzhak Rabin in 1995 and the subsequent defeat of the Labor party by the nationalist Likud party. But because of electoral reforms that paradoxically enhanced the power and influence of minority parties in the Knesset, the new Prime Minister, Benjamin Netanyahu, was in a relatively weak position politically vis‐à‐vis his own parliamentary coalition (Gedal 1998; Netanyahu 2000: 345). Because of the risks of defection within his own coalition, Netanyahu had limited freedom of maneuver at the negotiating table at the Wye Plantation in Maryland in 1998 when Israel agreed to further transfers of land to the Palestinian Authority. Such risks, that is the breakup of domestic political coalitions or a loss of confidence in the government, cannot be underwritten by the mediator or outside third parties. (In Netanyahu's case, his concessions at Wye polarized his own coalition. His subsequent decision to halt the implementation of the Wye accords to appease his conservative wing cost him the support of moderates and ultimately led to the dissolution of the Knesset and to new elections, in which Netanyahu lost to Labor leader Ehud Barak.)

How mediators handle the vagaries of domestic and electoral politics is a matter of supreme statecraft. But mediators should think seriously about holding back — and, if necessary, hunkering down — if the risk of agreement for one (or more) of the parties is too high and attempts by the mediator to force a settlement seem likely to throw a government into turmoil. U.S. insensitivity to these considerations has often proved counterproductive in the Arab–Israeli peace process as more than one close observer of these negotiations has noted (Bailey 1990; Kriesberg 2001; Ross 2004; Saunders 1985; Sheehan 1976; Touval 1982b).

Confronted with seemingly “incurable” covenant risks in the terms of an agreement, mediators can nonetheless do some things to help the parties manage these risks. First, mediators can help parties reduce the level of risk exposure in asymmetric situations (i.e., where sovereign risk is high) by structuring the transfer of deliverables in an agreement in increments, making sure that both parties give something up front, and by designing mutual “pay‐as‐you‐go” contracts that spread and reduce the up‐front costs of default.

In effect, this was Secretary of State Henry Kissinger's approach in the Sinai I agreements, which called for a progressive withdrawal of Israeli forces from the Sinai and the gradual assertion of Egyptian sovereignty over the territory. The immediate challenge that Kissinger confronted in his efforts to mediate a formal disengagement plan was to “reconcile Egypt's demands for sovereignty with Israel's need for security assurances” (Mandell and Tomlin 1991: 48). Under the “pay‐as‐you‐go” formula, Israel conceded some, but not all, of the Sinai and was allowed to keep some of its forces in key strategic areas.

A second way mediators can reduce incurable covenant risks is to underwrite the replacement costs of the agreement if it subsequently fails, that is, by providing “replacement” security guarantees to the aggrieved party (or parties) whose interests have been compromised. This role is generally well understood by the participants in the Camp David negotiations between Israel and Egypt, which paved the way for the complete withdrawal of Israeli forces from the Sinai. As Moshe Dayan, the former Israeli defense minister, noted in his memoirs: “the U.S. [was to] assume responsibility for there being no abrogation of the treaty we would sign for Egypt. We were concerned that Egypt, after our withdrawal from Sinai, might not honor her obligations” (Princen 1991: 61).

However, covenant risks cannot be ameliorated if the period for implementing an agreement is too long and the mediators fail to maintain pressure on the parties to act on their negotiated commitments. The incrementalist approach in the Oslo Peace Process was designed, in part, to reduce asymmetrical risk exposure by ensuring that the recognition of the Palestinian Authority took place gradually and that land (and sovereignty) were conceded by Israel in increments that were, in effect, tied to benchmarks for good behavior and an improving security situation (Makovsky 1996). As United Nations (U.N.) official Jan Egeland (1999) pointed out, the first milestone after Oslo was the Cairo Agreement of May 1994, which established Palestinian self‐rule in Gaza and Jericho. This was followed by the agreements concluded in Taba, Egypt in September 1995, which extended Palestinian self‐rule on the West Bank and established Palestinian elections in early 1996 coupled with the withdrawal of Israeli security forces from West Bank towns and villages. Subsequent agreements led to further Israeli troop redeployments and commitments by the Palestinians to fight terrorism.

However, the Oslo timetable was eventually disrupted by changes in the domestic leadership of Israel, a worsening — as opposed to improving — security situation, and growing distrust and mutual recriminations between the parties as deadlines passed and negotiated commitments were broken. In addition, because the United States failed to insist on the fulfillment of the accords, each side felt more exposed to risk. Eventually, the deteriorating security situation forced a suspension of negotiations between the Israeli government and the Palestinian Authority in 2002, underscoring the fact that even a gradualist negotiating strategy and incremental peace process could not reduce the risks of reaching a comprehensive settlement to a level that was acceptable to the parties within the Oslo framework. With the collapse of the Camp David negotiations in late 2000, which was followed by an escalation in terrorist attacks and violence, the sense of hopelessness and the feeling that Israelis and Palestinians were locked in an unending struggle only grew worse. The Sharm El‐Sheikh Fact‐Finding Committee or “Mitchell Commission” (2001) offered a series of recommendations to rebuild confidence and resume negotiations, but ultimately did little to resuscitate the floundering peace process at the time.

The problem of moral hazard— the risk that a party has not entered into a contract in good faith — poses its own special challenges to third‐party interveners. In cases where moral hazard is high because the exposure to loss resulting from improper or deceptive actions by a party is considerable, some may shy away from entering into negotiated commitments with that party. One strategy for resolving this problem is for the mediator to reach out to other third parties who can exert pressure and/or impose direct costs on that party, thus leveraging the situation to reduce moral hazard, that is, in effect raising “insurance premiums” in order to discourage reckless or uncooperative behavior. But, again, it may be difficult for mediators to get others to sustain the right kind of political pressure to make this strategy work.

This dilemma plagued the efforts of Margaret Anstee, the U.N. Special Representative in Angola. During the period that followed the mediation of the 1991 Bicesse Accords in Angola, Anstee was put in the difficult position of trying, first, to implement an intricate, multitrack settlement plan and, then, when it crumbled and the parties returned to war in the wake of failed elections, to pick up the pieces. In an effort to acquire some leverage, Anstee reached out to the so‐called Troika (United States, Russia, and Portugal, who were the guarantors of the peace process) in an attempt to get the parties to return to the negotiating table. The effort was temporarily successful, and talks resumed. But the bigger problem lay in the U.N. Security Council (UNSC), which was unwilling to take decisive action and use real “carrots and sticks” to send a message to the parties — and the opposition rebel leader, Jonas Savimbi, in particular — to stop all violence and comply with UNSC resolutions. The potential bilateral leverage of individual Troika governments was not applied coherently. Consequently, any commitments that were made at successive meetings quickly unraveled and efforts to secure a cease‐fire fell on “stony ground” (Anstee 1999: 603).

Mediators can also try to reduce moral hazard by making it clear to players who have acquired a reputation for reneging on their negotiating commitment that their reputations are at risk and that the mediator and other international actors will no longer do business with them. (This is the functional equivalent of canceling an insurance policy when an individual becomes too much of a liability.) This appears to have been the strategy of President George W. Bush in the aftermath of the collapse of the cease‐fire negotiated by Central Intelligence Agency Director George Tenet in June 2001 and the terrorist attacks that were launched against Israel by various Palestinian groups. President Bush and his diplomatic emissaries sent a clear signal to Palestinian President Yassir Arafat that his political authority and credibility were exhausted and that the United States would no longer deal with him. The United States also sent a clear signal that it supported the emergence of a new leadership in the Palestinian Authority. It also conducted an intensive round of diplomacy with Arab leaders in an attempt to ensure key regional players would not give Arafat a better “political” credit rating than the United States felt he deserved.

Curable covenant risks are typically associated with a temporary failure to meet specific obligations under a set of negotiated agreements, for example, when political obligations in a settlement are not discharged according to a previously agreed timetable because of unforeseen operational difficulties or a shortage of available resources, or because the timeframe for meeting those obligations is too ambitious. In a curable covenant risk situation, a party's failure to meet its obligations is not necessarily catastrophic if the terms can be renegotiated and/or commitments can be rolled over within a reasonable period of time.

Depending on how an actual agreement is structured, however, the nature of the risk differs. Under a negative covenant, a party agrees not to do certain things over the lifetime of an agreement. In an affirmative covenant, the party agrees to fulfill a particular set of specified obligations. A covenant can be tripped by a negative or positive (affirmative) default, that is, a party does something it agreed not to do or it fails to do something it previously agreed to do. The situation is deemed curable because there is a good possibility that the defaulting party will be able to settle or make good on its negotiated commitments later on and/or take remedial action to correct the default. (Curable covenant risks are somewhat akin to liquidity risks in a financial agreement, i.e., the risk that a party will not settle for full value at the due date but might be able to do so later.) Curable covenant risks reflect the short‐term costs of adjustment and differ from contract replacement risks, which represent the current and future costs of replacing a failed agreement.

As discussed in the investment literature, netting and novation are two kinds of correctives to a curable default.² The concepts are instructive for international mediation. Netting is a means for allowing positive and negative values in an agreement — which, in a conflict setting, are usually related to the incompatible political objectives of the warring parties — to cancel each other out by bundling them together into a single package of commitments and mutual obligations. When properly “netted,” the different contractual elements are discharged in such a way that the parties cannot cherry‐pick the agreement apart during its implementation.

Novation is a contract replacement strategy whereby an existing (and typically unmet) obligation or commitment in a formal agreement is discharged later on by replacing it with a new set of obligations that substitute for the original — that is, outstanding obligations are effectively rolled over into new ones, but in such a way that does not diminish the political intent of the original agreement. When used in tandem, netting and novation can help reduce both replacement and shortfall implementation risks in a negotiated agreement.

The successful negotiation of the Dayton Peace Accords, which ended the war in Bosnia, was in part the result of a netting strategy in which the United States wired together a series of separate agreements into a comprehensive package of mutually binding commitments. As Saadia Touval, a close observer of these negotiations, noted, the key elements of the Dayton Accords were not new. The concept of a single Bosnian state was an element of the Cutileiro Plan of March 1992. “It was revived in 1993 and revised to give the Serbs contiguous territory. It was a central element in the Union of Three Republics (Owen–Stoltenberg) Plan, its revised version labeled ‘European Action Plan’ discussed in the latter half of 1993, and the 1994 Contact Group Plan . . . All these plans also proposed that Bosnia be recognized as a single state under international law” (Touval 1996: 559). However, “the formula represented by the preliminary agreements . . . was flawed because it lent itself to contradictory interpretations, binding the parties to unity, and legitimizing their separation. While the Muslims hoped that the constitutional provisions could be turned into an instrument for creating a unified state, the Serbs hoped that it would facilitate the secession of Republika Srpska from Bosnia” (Touval 1996: 564). By negotiating a series of bilateral agreements between the parties, the United States was able to reconcile these conflicting objectives and firmly knit together the key elements of the tripartite settlement at Dayton. The key element in these bilateral undertakings was the Bosnian Federation accords, which ended the war between Croats and Muslims and provided a temporary solution to part of the conflict until the Dayton peace process could take hold (Serwer 1999).

Mediator novation has also rescued a peace process when it threatened to go off the rails because timetables were unrealistic or because the implementing party did not have adequate resources. For example, in the Salvadoran peace accords, which were signed at Chapultepec Castle in Mexico City on January 16, 1992, the parties committed themselves to a series of security‐related obligations and reforms, which had to be completed according to a preset timetable (de Soto 1999; Cañas and Dada 1999). Difficulties soon arose when key security obligations were not met. Under a schedule determined by the Ad Hoc Commission on the Purification of the Armed Forces, which was a part of the accords, two public security bodies — the Treasury Policy and the National Guard — were supposed to be abolished by the government by March 1, 1992 and their members incorporated into the army. However, the government failed to carry out the disbanding of these two bodies. For several weeks after their incorporation into the army, the former members of these two bodies (some 3,500 personnel in all) remained in their original barracks. The Frente Farabundo Marti para la Liberacion Nacional (FMLN) denounced this as a violation of the peace agreement and refused to complete the redeployment of its own forces until the problem was resolved. Although the cease‐fire between the two sides continued to hold during the spring and summer, by the fall of 1992, it was quite apparent that both parties would not be able to comply with the October 31 date for ending the conflict.

These delays and the reactions of each party to them were clearly leading the peace process into a cul de sac as each party held the other responsible for the delays while insisting on its own interpretation of key clauses in the accords. In order to break the impasse, U.N. Secretary General Boutros Boutros‐Ghali sent Marrack Goulding and Alvaro de Soto to San Salvador to mediate a solution. De Soto conducted extensive separate discussions with both the government and the FMLN. The result was an adjustment of the Chapultepec timetable and an exchange of letters stipulating that compliance with specific undertakings by one side would be contingent upon compliance with specific undertakings by the other side. In this case, an affirmative covenant was rescued during its implementation by novation techniques, which rolled over and substituted a new agreement for a series of obligations that had not yet been met.

Legal and systemic risks are two other kinds of risk that can influence the trajectory of an agreement during its implementation (Crouhy, Mark, and Galai 2000). Legal risks typically fall within the curable category of risk. However, they are different from covenant risks because the alleged violation does not involve willful intent (as in the case of moral hazard), but arises from the unintended (or unanticipated) consequences of deficiencies within the political or legal framework of the agreement itself. A legal risk is the risk that a transaction is unenforceable because there is no sound political or legal framework for ensuring that negotiated obligations are properly fulfilled. Ambiguous or vague language in a settlement — which Touval (1982a) explains may be the direct consequence of “hedging,” a risk‐avoidance strategy during negotiations — can come back to haunt an agreement during its implementation.

Suffice to say, most peace agreements contain provisions with ambiguous or inadequate legal and constitutional mechanisms (and/or procedures for implementing them). This has adversely affected the prospects for peace, although through the provision of mediated interventions these risks are sometimes successfully managed. The land tenure provisions in the Salvadoran peace accords, which were major sources of disagreement and conflict between the parties, are a case in point. The peace agreements themselves did not sanction an overall land redistribution program (of the sort that many post‐revolutionary regimes implement after a civil war). Rather, the peace accords specified a land transfer program as “the main venue . . . through which ex‐combatants and supporters of the FMLN would be reintegrated into the productive life of the economy” (de Soto and del Castillo 1994: 11–12). Land tenure questions were especially sensitive issues, given the importance of agriculture to the economy and the fact that arable land was in short supply and unevenly distributed. Ownership of land also made available other potential benefits, like housing credits and assistance for agricultural production. Additionally, because the peace accords themselves only reflected broad principles, the actual details of land transfer had to be worked out during the course of the implementation of the peace accords and with the assistance of a third party, the U.N. Observer Mission in El Salvador.

The peace accords stipulated that, pending agreement on various issues, the land ownership situation would be respected in former conflict zones and current landholding occupants would not be evicted (Hampson 1996b). They also assigned the task of verifying implementation of these provisions to a special commission that reported to the Comision Nacional para la Consolidacion de la Paz (COPAZ) — a body responsible for overseeing implementation of all political agreements reached by the parties. One of the difficulties the special commission faced derived from the peace agreement's failure to define the “conflict zones.” February and March 1992 saw tensions rise in the countryside after various peasant groups seized properties, only to be evicted by security forces. These actions were also of concern to FMLN combatants, who were waiting to move into designated concentration areas. When conditions failed to improve, the U.N. representatives met with the parties, who agreed to suspend land seizures and evictions in order to facilitate the processing of cases submitted to COPAZ's special commission. The U.N. also convened a special outside group of experts who worked on the land transfer issue and submitted a set of recommendations, which were eventually accepted by both sides.

Systemic risks derive from adverse political developments in the general regional and/or international environment. Unlike the typical “spoiler” (Stedman 1997), who is usually a direct party to the conflict, the threat in this case is an indirect one, resulting from untoward (or possibly unforeseeable) actions by regional and international actors (or, in some cases, transnational, nonstate actors) that jeopardize an agreement's implementation. Some of the recent literature on “greed and grievance” also draws attention to the political economy of violent conflict and the role that natural resource and illicit commodity markets (which represent another kind of systemic risk) play in undermining nascent peace processes (Berdal and Malone 2001).

Systemic risks are not necessarily curable, but they are sometimes manageable or controllable. Multilateralizing a peace process by bringing in key, affected regional or international actors as “friends of the peace process” is one way to manage these risks. Another is to forge an explicit political entente among great powers to attenuate the adverse effects of political competition. Much has been written about the conditions that led to the successful negotiation of the Cambodian peace accords, which culminated in international supervised elections in 1993. But as U.S. negotiator Richard Solomon has argued, the success of the accords depended critically on the ability of two great powers, Russia and China, whose geostrategic interests were changing, to manage the risks of withdrawing their forces from Indochina in a way that would not be exploited by the other side as well as other great powers and regional actors (Solomon 1999, 2000). The irony was that the United States, which had withdrawn from the region after the Vietnam War, was the catalyst for a negotiating framework that allowed the parties to manage these risks (Solomon 1999). In the negotiated (and largely U.S.‐mediated) geopolitical entente leading up to the Cambodian peace accords, the United States helped the parties manage the risks of exit from Indochina by vigorously promoting the idea that a comprehensive political settlement with significant U.N. involvement would not only bring the full weight of the United Nations Security Council and its five permanent members to bear on all of the parties, including Pol Pot's Khmer Rouge, but also “internationalize” the diplomacy of a settlement. Within the framework of the U.N.‐sponsored peace plan, the key players were able to reduce their regional rivalries, exit from military commitments that were increasingly costly, and bring about a withdrawal of Vietnamese troops from Cambodia.

Great power mediation, coupled with multilateral approaches that engage the interests of critical regional and great powers, is one way to control systemic risk. Another is to promote issue‐party linkages so that key systemic risks offset each other in a way that facilitates a negotiated settlement. In the negotiation of the Angola–Namibia peace accords in the 1980s, the articulation of the linkage between the issue of Cuban troop withdrawals from Angola, which addressed South Africa's fundamental security concerns, and the negotiated withdrawal of South African troops from Namibia to secure Namibia's independence became the cornerstone of the U.S. policy of constructive engagement and a key element of the subsequent peace agreement. At the same time, these strategic goals were tied to a broader U.S. interest in promoting peaceful democratic change in South Africa, which would see the dismantling of apartheid where engaging — as opposed to isolating — South Africa was seen as key (Crocker 1992, 1999).

Finally, it is worth noting that mediators can mobilize political support for targeted sanctions and certification schemes, such as the U.N.‐instigated “Kimberley Process” (which is directed at combating international trade in blood diamonds in countries such as Angola, the Democratic Republic of Congo, and Sierra Leone) in order to try to control illicit commercial transactions that fuel conflict processes and undermine a peace settlement (Kimberley Process 2005).

In this article, I have argued for a conception of international mediation that explicitly recognizes the importance of different strategies of risk management and control in third‐party mediated peace settlements. I suggest that mediators have to contend with a wide variety of risks that involve not simply the risks of negotiation but also the risks of agreement.

I also argue that there is more than one kind of risk in a peace agreement and that the type of risk will, to some extent, determine the kinds of negotiating tactics and resolving formula that mediators bring to an inherently risk‐laden situation. Mediators can underwrite the risks in a peace agreement in a variety of ways. They can provide political and security guarantees; design pay‐as‐you‐go delivery schedules on commitments that reduce risk exposure; leverage political assets in the regional and international environment to reduce moral hazard; wire agreements together through netting tactics to stop the parties from cherry‐picking an agreement apart; and substitute new obligations and timetables for those that the parties have been unable to fulfill via a process of novation. (Table One summarizes mediation response strategies that are part of comprehensive risk control strategy.) However, there are real limits to what mediators can do as risk managers because the risks of peace are not completely irreducible or curable.

What mediators must also understand is that they are required to remain “on call” after the agreement is signed in order to help the parties manage the many risks that threaten its implementation. When the guns fall silent, there is no fast or easy exit from a conflict zone.