Spoiler Problems in Peace Processes
Stephen John Stedman
Peace making in civil war is a risky business.1 The biggest source of risk comes from spoilers—leaders and parties who believe the emerging peace threatens their power, world view, and interests and who use violence to undermine attempts to achieve it.2 By signing a peace agreement, leaders put themselves at risk from adversaries who may exploit a settlement, from disgruntled followers who see peace as a betrayal of key values, and from excluded parties who seek either to alter the process or destroy it. By implementing a peace agreement, peace makers are vulnerable to attack from those who oppose their efforts. And most important, the risks of peace making increase the insecurity and uncertainty of average citizens who have the most to lose if war is renewed.
When spoilers succeed, as they did in Angola in 1992 and in Rwanda in 1994, the results are catastrophic. In both cases the casualties of failed peace were infinitely higher than the casualties of war. When Jonas Savimbi refused to accept the outcome of United Nations (UN)-monitored elections in 1992 and plunged Angola back into civil war, approximately 300,000 people died. When Hutu extremists in Rwanda rejected the Arusha Peace Accords in 1994 and launched a genocide, over 800,000 Rwandans died in less than three months.
If all spoilers succeeded, the quest for peace in civil wars would be dangerously counterproductive. But not all spoilers succeed. In Mozambique the Mozambique National Resistance (RENAMO), a party known as the “Khmer Rouge of Africa,” stalled in meeting its commitments to peace and threatened to boycott elections and return to war. In the end, however, RENAMO joined parliamentary politics, accepted losing an elec-
tion, and disarmed, thus ending a civil war that had taken 800,000 lives. In Cambodia the peace process was able to overcome the resistance from the real Khmer Rouge, the party that has provided the sobriquet for fanatic parties elsewhere.
The crucial difference between the success and failure of spoilers is the role played by international actors as custodians of peace. Where international custodians have created and implemented coherent, effective strategies for protecting peace and managing spoilers, damage has been limited and peace has triumphed. Where international custodians have failed to develop and implement such strategies, spoilers have succeeded at the cost of hundreds of thousands of lives.
This study begins to develop a typological theory of spoiler management and pursues the following research objectives: (1) to create a typology of spoilers that can help custodians choose robust strategies for implementing peace, (2) to describe various strategies that custodians have used to manage spoilers, (3) to propose which strategies will be most effective for particular spoiler types, (4) to sensitize policy makers to the complexities and uncertainties of correctly diagnosing the type of spoiler, and (5) to compare several successful and failed cases of spoiler management in order to refine and elaborate my initial propositions about strategies.
This research is a first step toward understanding spoiler problems in peace processes and evaluating the effectiveness of different strategies of spoiler management. The findings of this study are provisional. As more case studies of spoiler management emerge, as new research develops on the case studies examined below, and as more theoretical attention is trained on the problem, some of the findings will need to be reconsidered and revised. Moreover, this study addresses only the strategies and actions of external actors who oversee peace processes; the topic of spoiler management from the perspective of domestic parties committed to peace is beyond the scope of this study.
This paper argues that spoilers differ by the goals they seek and their commitment to achieving those goals. Some spoilers have limited goals; others see the world in all-or-nothing terms and pursue total power. Furthermore, some spoilers are willing to make reasoned judgments concerning the costs and benefits of their actions, whereas others show high insensitivity to costs and risks and may hold immutable preferences. Custodians have a range of strategies to deal with spoilers, from ones that rely heavily on conciliation to ones that depend greatly on the use of coercion. The case studies discussed below suggest three major findings. First, the choice of an appropriate strategy requires correct diagnosis of the type of spoiler and thoughtful consideration of constraints posed by other parties in the peace process. Second, to make good diagnoses, policy
makers must overcome organizational blinders that lead them to misread intentions and motivations. Third, implementation of a successful strategy depends on the custodian’s ability to create an external coalition for peace, the resources that the coalition brings to its responsibility, and the consensus that the coalition forms about the legitimacy (or lack thereof) of spoiler demands and behavior.
SPOILERS: A PRELIMINARY TYPOLOGY
Peace processes create spoilers. This is a statement about definition and causality. In war there are combatants, who can be identified in myriad ways—for example, rebels, bandits, pariahs, rogues, or terrorists—but not as spoilers. Spoilers exist only when there is a peace process to undermine, that is, after at least two warring parties have committed themselves publicly to a pact or have signed a comprehensive peace agreement.3 Peace creates spoilers because it is rare in civil wars for all leaders and factions to see peace as beneficial. Even if all parties come to value peace, they rarely do so simultaneously, and they often strongly disagree over the terms of an acceptable peace. A negotiated peace often has losers: leaders and factions that do not achieve their war aims. Nor can every war find a compromise solution that addresses the demands of all the warring parties. For example, the most perfectly crafted power-sharing institutions in the world are useless if one of the parties does not want to share power.4 Even the best-designed settlements may be attacked by leaders and parties who decide that the kind of peace in question is not in their interest.
Custodians of peace processes confront several different spoiler problems, which differ on the dimensions of the position of the spoiler (inside or outside an agreement); the number of spoilers; the type of spoiler (limited, greedy, or total); and the locus of the spoiler problem (leader, followers, or both).
Position of the Spoiler
Spoilers can be inside or outside a peace process. An inside spoiler signs a peace agreement and signals a willingness to implement a settlement, yet fails to fulfill key obligations to the agreement. Examples include President Juvenal Habyarimana of Rwanda, who failed to implement the Arusha Accords to end his country’s internal war; the Khmer Rouge (KR) in Cambodia, which signed the Paris Peace Accords and then refused to demobilize its soldiers and chose to boycott elections; and the Union for the Total Independence of Angola (UNITA), which signed the Bicesse Accords in 1991 but returned to war in 1992 when it lost the
election. Outside spoilers are parties who are excluded from a peace process or who exclude themselves and use violence to attack the peace process—for example, the Committee for the Defense of the Revolution (CDR) in Rwanda, which committed genocide to prevent implementation of the Arusha Accords.
Inside spoilers tend to use strategies of stealth; outside spoilers often use strategies of overt violence. Spoilers who have signed peace agreements for tactical reasons have an incentive to keep their threat hidden and to minimize the amount of violence they use; they want the peace process to continue as long as it promises to strengthen them against their adversary. Inside spoilers need to comply enough to convince others of their goodwill but not so much that it weakens their offensive military capability. Outside spoilers, on the other hand, tend to use overt violence as a strategy toward undermining peace. Favorite tactics include the assassination of moderates who stand for a negotiated peace, massacres that coincide with any progress in reaching a negotiated settlement, and the creation of alliances with conservative members in the armed forces and police to sabotage any agreement.
Number of Spoilers
The presence of more than one spoiler creates a compound challenge for custodians. Any strategy a custodian chooses to deal with one spoiler has implications for the strategy selected to deal with other spoilers. Actions taken to marginalize one spoiler may inadvertently strengthen another. To give an example from the case studies, in Rwanda the stability of the peace process was endangered because Habyarimana refused to fulfill his obligations to the peace agreement he had signed. The UN threatened to withdraw its peacekeeping operation in order to coerce Habyarimana into implementing the agreement. Yet Habyarimana was only one of two spoilers. The CDR, former members of Habyarimana’s regime, rejected the peace agreement and conspired against the peace process from outside. The UN strategy succeeded in pressuring Habyarimana but emboldened the CDR to attack the peace process.
Types of Spoilers
Recent work on civil war termination suffers from a flawed attenuated portrayal of combatants and their aims. At one extreme are analyses that posit that parties are solely motivated by insecurity and only seek party survival.5 According to this view, the only reason for parties in civil wars to fight is their fear that, if they make peace and disarm, their adversary will take advantage and eliminate them. The lack of an overarching
authority that can enforce a political settlement in civil war means that warring parties cannot credibly commit to making peace, either in the short term (through disarmament) or the long term (through a constitution). Thus, any party that violates or opposes a peace agreement does so out of fear, not some other motivation. Scholars who embrace this view believe that spoiler behavior can be addressed only by reducing the spoiler’s fears through international guarantees. Like those international relations theorists steeped in the security dilemma, these writers believe that the central theme of civil war termination “is not evil but tragedy.”6
At another extreme are those who assert that all parties in civil war seek total power.7 This, however, is too facile; all parties in civil war seek power, but not all parties seek total power. Some parties desire exclusive power and recognition of authority; some want dominant power; some seek a significant share of power; and some desire to exercise power subject to democratic controls. This should not be surprising: power is a means or resource to realize other goals. Some goals—for instance, the permanent subjugation or elimination of an ethnic group, race, or socio-economic class—need more power than the goals of creating a democratic political regime or gaining recognition of political equality among races or ethnic groups. That parties differ in their goals and commitment to total power can be seen by all of the parties that have accepted and lived with compromise solutions to civil wars (in Colombia, Zimbabwe, Namibia, El Salvador, South Africa, and Mozambique). Similarly, not every winner of a civil war creates a totalitarian regime or slaughters its opponents.8 A thought experiment makes the point: if you had to be on the losing side of a civil war, would you rather surrender to Abraham Lincoln or Hafez al-Assad, to Nelson Mandela or Mao Zedong?
Successful management of spoiler problems requires the recognition that parties in civil wars differ in their intentions, motivations, and commitment—dimensions that are crucial for understanding why some parties undermine peace agreements. Spoilers vary by type: limited, greedy, and total. These types differ primarily on the goals that the spoiler pursues and secondarily on the spoiler’s commitment to achieving its goals. At one end of the spectrum are limited spoilers, who have limited goals— for example, recognition and redress of a grievance, a share of power or the exercise of power constrained by a constitution and opposition, and basic security of followers. Limited goals do not imply limited commitment to achieving those goals, however. Limited goals can be nonnegotiable and hence subject to heavy sacrifice.
At the other end of the spectrum are total spoilers, who pursue total power and exclusive recognition of authority and hold immutable preferences: that is, their goals are not subject to change. Total spoilers are led by individuals who see the world in all-or-nothing terms, who often suf-
fer from pathological tendencies contrary to the pragmatism necessary for compromise settlements of conflict. Total spoilers often espouse radical ideologies; total power is a means for achieving the radical transformation of society.
The greedy spoiler lies between the limited spoiler and the total spoiler. The greedy spoiler holds goals that expand or contract based on calculations of cost and risk. A greedy spoiler may have limited goals that expand when faced with low costs and risks; alternatively, it may have total goals that contract when faced with high costs and risks.
The spoiler type poses different problems for peace processes. Total spoilers are irreconcilably opposed to any compromise peace; any commitment to peace by a total spoiler is tactical—a move to gain advantage in a struggle to the death. Limited spoilers can conceivably be included in peace processes, if their limited nonnegotiable demands can be accommodated by other parties to the conflict. Greedy spoilers can be accommodated in peace processes if their limited goals are met and they are constrained from making added demands.9
Locus of the Spoiler Problem
A key issue concerns the possibility of change in type. For example, can a total spoiler become a limited spoiler? The answer depends on the locus of spoiler behavior—that is, whether it is the leader or the followers. If the impetus for spoiler behavior comes from the leader, the parties can change type if their leadership changes. This seems particularly relevant for total spoilers because their total goals and commitment are so extreme. A change in leadership may be enough to alter a party from a total spoiler to a limited spoiler. For instance, a negotiated settlement to Zimbabwe’s civil war became possible only when Abel Muzorewa replaced Ian Smith as leader of Zimbabwe-Rhodesia. Likewise, the willingness of the South African government to implement the Namibian peace process was aided by the incapacitation and replacement of President P.W.Botha. Long-time observers of Cambodia argued that the KR would never accept a negotiated peace as long as Pol Pot was in power. Other cases, however, suggest that there are times when followers are the locus of spoiler behavior. As one of the cases below suggests, in Rwanda in 1994, Habyarimana was reluctant to fulfill his commitments to the Arusha Accords for fear that his followers would attack him.
STRATEGIES OF SPOILER MANAGEMENT
Custodians of peace processes are defined here as international actors whose task is to oversee the implementation of peace agreements. Im-
plicit in their role is the cultivation and protection of peace and the management of spoilers. International custodians can be international organizations, individual states, or formal or ad hoc groups of concerned third parties. Custodians can be tightly organized or loosely coordinated. With the exception of the implementation of the Dayton Peace Agreement on Bosnia and implementation of the South African peace settlement, the chief custodian of peace processes in the 1990s was the United Nations.
Custodians of peace processes in the 1990s pursued three major strategies to manage spoilers. In order of conciliation to coercion, the strategies were (1) inducement, or giving the spoiler what it wants; (2) socialization, or changing the behavior of the spoiler to adhere to a set of established norms; and (3) coercion, or punishing spoiler behavior or reducing the capacity of the spoiler to destroy the peace process. These strategies are general conceptual types; in practice each takes on a specific configuration likely to be more complex than the general version of it identified here. It should also be noted, as will be evident in some of the cases to be examined, that international actors can employ more than one strategy—either simultaneously (with different priority and emphasis) or in sequence.
Inducement as a strategy for managing spoilers entails taking positive measures to address grievances of factions that obstruct peace. Custodians attempt to induce the spoiler into joining a peace process or fulfilling its obligations to an existing agreement by meeting the spoiler’s demands, which can be of several types. Spoilers may insist that their behavior is based on (1) fear, and demand greater protection; (2) fairness, and demand greater benefits; or (3) justice, and demand legitimation or recognition of their position. The custodian must assess the veracity and significance of such claims.
Inducement can be rigorously applied by meeting costly demands made by spoilers, as the UN did in Mozambique in 1993–1994. Or it can be something as lax and questionable as offering a spoiler a continued role in negotiations, even when the spoiler has returned to war, as in the case of Angola in 1992, or when the spoiler has committed genocide, as in Rwanda in 1994. Indeed, the frequency of inducement attempts in peace processes suggests that it is a “default mode”—that is, a convenient strategy that is applied without adequate consideration of whether it is an appropriate strategy for the type of spoiler in question.
The strategy of socialization requires custodians to establish a set of norms for acceptable behavior by parties that commit to peace or seek to join a peace process. These norms then become the basis for judging the demands of the parties (are they legitimate or not?) and the behaviors of the parties (are they acceptable in the normative framework?). In turn this strategy relies on two components to elicit normatively acceptable behav-
ior: the material and the intellectual. The material component involves custodians carefully calibrating the supply of carrots and sticks to reward and punish the spoiler. The intellectual component emphasizes regular persuasion by custodians of the value of the desired normative behavior. Normative standards can include commitment to the rules of democratic competition and adherence to the protection of human rights. The intellectual component can be aimed at both elites (the attempt to inculcate appropriate values) and citizens (the attempt to educate the mass of citizens into norms of good governance, democratic competition, and accountability, as a means of pressuring elites).
A strategy of coercion relies on the use or threat of punishment to deter or alter unacceptable spoiler behavior or reduce a spoiler’s capability to disrupt the peace process. The strategy of coercion has several variations. The use of coercive diplomacy, or the use of threat and demand, has been used infrequently against spoilers in peace processes, the notable exception being the use of NATO air strikes against Bosnian Serbs in 1995.10 Likewise, the use of force to defeat a spoiler has been attempted infrequently—most notably, in Somalia, when the UN decided to hold Somali warlord Mohammed Farah Aideed responsible for an ambush by his forces against Pakistani peacekeepers, and in Sri Lanka, when an Indian peacekeeping force attempted to forcibly disarm Tamil rebels and capture their leader, Velupillai Prabakaran.11
Two more common variations of the coercion strategy are what I call the “departing-train” strategy and the “withdrawal” strategy. The departing-train strategy combines a judgment that the spoiler’s demands and behavior are illegitimate with the assertion that the peace process will go irrevocably forward, regardless of whether the spoiler joins. In Cambodia the strategy was linked to the holding of an election, thereby setting a deadline for joining the process and promising a change in the status quo. The departing-train metaphor implies that the peace process is like a train leaving the station at a preordained time and that, once set in motion, anyone not on board will be left behind. The departing-train strategy may require active measures to protect the parties of peace and limit the ability of the spoiler to attack the peace process.
The withdrawal variation of the coercive strategy assumes that the spoiler wants an international presence during the peace process. This strategy aims to punish the spoiler by threatening to withdraw international support and peacekeepers from the peace process. This was the dominant strategy pursued by the UN in Rwanda and the Implementation Force in Bosnia; it was also used in a tertiary manner in Mozambique. The strategy is a blunt instrument in that the punishment—withdrawal— promises to hurt parties that have fulfilled their obligations and reward any spoiler that opposes international engagement.
Matching Strategies to the Type of Spoiler
A correct diagnosis of spoiler type is crucial for the choice of an appropriate strategy of spoiler management. Total spoilers cannot be accommodated in a peace settlement; they must be defeated or so marginalized that they can do little damage. A limited spoiler can be accommodated by meeting its nonnegotiable demands. A greedy spoiler with limited goals may be accommodated, but such accommodation may whet its appetite to demand more concessions. A total spoiler, because it defines the war in all-or-nothing terms and holds immutable preferences, cannot be appeased through inducements, nor can it be socialized; moreover, both strategies risk strengthening the spoiler by rewarding it. Two versions of the coercive strategy are also dangerously counterproductive for managing total spoilers. Coercive diplomacy is unlikely to succeed given the cost insensitivity of total spoilers; they call bluffs and test wills. If custodians fail to carry through on threats or fail to establish escalation dominance, the spoiler’s position may be strengthened. By showing the inadequacy of international force, the spoiler adds to its domestic reputation for coercive strength. The withdrawal strategy also backfires against a total spoiler, who has everything to gain if custodians abandon the peace process.
Two strategies are appropriate for managing a total spoiler: the use of force to defeat the spoiler or the departing-train strategy. Because few custodians are willing to use force to defeat a total spoiler, they should strengthen the parties of peace so that they can defend themselves. The departing-train strategy can do this by legitimizing the parties of peace and delegitimizing the spoiler, by depriving the spoiler of resources— both capital and weapons—for war, and by redeploying peacekeepers to protect the parties of peace.
A limited spoiler can be included in a peace process if its demands are acceptable to the conflict’s other parties. Thus, inducement may be an appropriate strategy for managing a limited spoiler, but the strategy depends on the bargaining range established by the other parties that have already committed to peace. If the demands of the limited spoiler cannot be accommodated through inclusion, the custodian may have to choose socialization or coercion. The danger is that the threat or use of force may prompt a counterescalation of violence by the limited spoiler.
The greedy spoiler requires a long-term strategy of socialization. Because the spoiler is not total, there is at least a possibility of bringing it into the peace process. In the short term the greedy spoiler presents a serious dilemma. As inducements alone will serve to whet the appetite of the greedy spoiler, the legitimacy and illegitimacy of its demands must be clearly distinguished. Moreover, depending on the cost insensitivity and
risk taking of the spoiler, coercion may be necessary to impose costs and create a strong sense of limits to the spoiler’s demands. On the other hand, sole reliance on a coercive strategy ignores that even the greedy spoiler has legitimate security goals that can only be met through inducements.
Limitations of the Custodian
Policy makers often have concerns other than a specific conflict at hand; a strategy that may be the best from a perspective of solely managing a conflict may not be the best for a policy maker considering a range of interests. This is certainly true for U.S. policy makers when it comes to conflict resolution in small, unimportant (to U.S. national interests), and faraway countries. Even the UN considers its actions in light of its organizational interests and the need to protect the reputation and institution of peacekeeping. The optimal strategy to end a conflict and manage a spoiler may be too costly or risky for external actors.
The UN has special limitations as a peace custodian. Although it possesses formal authority, its agent on the ground (the special representative of the secretary-general) is constrained by the direction and will of its Security Council. Special representatives must borrow leverage through coalition building; their ability to induce or punish, even their ability to rule credibly on the legitimacy or illegitimacy of demands, depends on the support of member states. In some cases, such as the Salvadoran and Cambodian peace processes, the UN relied on groups of friends—formal associations of all member states that have an interest in the peace process and therefore bring their power, energy, and attention to implementing peace. In other cases the special representative has relied on ad hoc groupings of interested states, usually working with their diplomatic representatives on site.
In managing a spoiler, member states that are patrons of the spoiler are the biggest potential liability (and source of possible leverage). On the one hand, such patrons, if they are sincerely interested in making peace, may supply the special representative with assets of leverage, credibility, and trust. On the other hand, such patrons may be slow to acknowledge that their client is acting as a spoiler and may be reluctant to declare their client’s demands illegitimate. Almost every patron of a spoiler has personal networks and domestic groups that support the spoiler. Pressures from these groups, as well as prior policy commitments to the spoiler, lead the patron to continue to support the spoiler even in the face of outrageous behavior.
The Fog of Peace Making
The typology of spoiler problems described so far reflects two fundamental attributes of peace processes: complexity and uncertainty. The
typology suggests that custodians face numerous uncertainties that require skilled diagnosis, including (1) the goal of a spoiler; (2) the intent behind acts of noncooperation or aggression; (3) the degree of commitment of the spoiler; (4) the degree of leadership command and control of followers; (5) the degree of unity within the spoiler; and (6) the likely effects of custodial action on the spoiler’s willingness to continue aggression, on the other parties to the peace process, and on interested external actors.
Custodians must interpret why a particular party attacks a peace process or refuses to meet its obligations to implement a peace agreement. Several interpretations are possible. A party that has signed an agreement but refuses to fulfill its obligations may be motivated by fear. It may desire an agreement but fears putting its security into the hands of its adversary. This causes it to stall on its commitments or cheat on agreements by creating a fail-safe option. A party that has signed an agreement may cheat because it is greedy and desires a better deal; it may want a negotiated settlement to succeed but also wants to increase its chances of maximizing its return in the settlement. A failure to fulfill its obligations may be a means of seeking advantage in an election that could determine partially the division of spoils and power of the settlement; alternatively, holding back from commitments may be a way to strengthen its bargaining position in the result of losing an election. Finally, a party may cheat because it has signed a peace agreement for tactical reasons; if the agreement seems as though it will bring the party to power, the party will abide by it; however, if the agreement seems like it will not bring the party to power, the party will cheat to overturn it. In such a case the spoiler is motivated by total goals and defines the stakes as all or nothing.
When external spoilers use violence to attack the peace process, a custodian must judge the intention behind the violence. Is it an attempt by the spoiler to force its way into negotiations—to alter a process so that its demands are included in a settlement? Or is it an attempt to weaken the commitment of the internal parties as a means to destroy a negotiated settlement? Again, the action must be connected to a judgment about the spoiler’s motivation. Is the spoiler motivated by limited grievances that can be incorporated into an agreement? Or is it motivated by total goals that are unalterably opposed to agreement? Custodians of peace processes must make judgments about the commitment of a spoiler to its preference. Spoilers may vary in their sensitivity to costs and risks; greedy parties may seek only limited opportunities to maximize their goals, or they may be willing to incur high costs and take large risks to improve their position.
The above interpretations assume a unified party—that the leader’s behavior reflects a group consensus about its aims. But if a warring group is
divided, a leader may sign an agreement but be reluctant to implement it for fear that an act of compromise could prompt a rebellion of hard-liners. Alternatively, a party’s aggressive behavior may or may not be evidence of a leader’s willingness to make peace; it could be the result of rogue elements that are opposed to settlement and seek to wreck an agreement.
Custodians of peace face uncertainty about the effects of actions they take toward a spoiler. Will they encourage the spoiler to desist from attacking the peace process? Or will they encourage the spoiler to continue its resistance? If divisions between hard-liners and moderates within the spoiler become evident, there will be uncertainty about how one’s actions will affect the relative strengths of the factions. A custodian’s actions likewise will have uncertain effects on the other parties to the conflict. Custodians may have to limit their use of coercion against a spoiler for fear of upsetting a fragile balance of power that might lead other internal parties to eschew a peaceful settlement. Custodians may be constrained in the use of inducements by the bargaining range permitted by the other parties to the conflict. A custodian’s failure to respond to spoiler behavior by one party may trigger a mimetic response by other parties to the conflict. If a custodian is lenient toward a spoiler, will it encourage other parties to cheat as well? If it acts aggressively toward a spoiler, might it encourage other parties to act aggressively in the belief that they have an ally that tips the balance of power against the spoiler? All of this is to say that the strategy that custodians pursue toward a specific spoiler must take into account the positions of other internal parties to the conflict and perhaps even the need to work with those parties to coordinate action.
CASE STUDIES OF SPOILER MANAGEMENT
This paper is a first step in developing a typological theory of spoiler management. The initial framework discussed above posits a typology of spoilers based on their intentions—limited, greedy, and total; it then describes a range of strategies available for managing spoilers—inducement, socialization, and coercion, with several variations of the latter strategy. It is suggested that the general appropriateness of a strategy depends on the type of spoiler: inducement for limited spoilers, socialization for greedy spoilers, and departing train for total spoilers. The framework places a heavy burden on the ability of custodians to diagnose correctly the type of spoiler they face.
This section examines five case studies of spoiler management in the 1990s as a way of refining, elaborating, and modifying the initial theory. The case studies serve the function of identifying complexities of various kinds that affect success or failure in dealing with spoiler problems that are not anticipated or explainable by the initial theory. The method of
structured, focused comparison precludes an investigation into all possible cases of spoiler management. Rather, I have chosen to present here the results of five important recent cases as an initial exploration of successful management and failed management.
The cases—Rwanda, Angola, Mozambique, and Cambodia toward the KR and the State of Cambodia—have reached stages where judgment is possible about the effectiveness of the chosen strategies. For example, even though at the time of my research, the war in Angola had not been brought to a close, I treat the period of 1991–1993 as a distinct stage in the war. Regardless of negotiations that took place after 1993, I am confident that most scholars would agree that, between the signing of the Bicesse Accords in May 1991 and the resumption of all-out civil war in 1993, the custodians of peace utterly failed to manage the spoiler problem presented by Jonas Savimbi. With respect to Cambodia, some may argue that when the research was undertaken the war continued between the KR and the coalition of the Cambodian People’s Party and FUNCINPEC (National Unified Front for an Independent, Neutral, Peaceful, and Cooperative Cambodia). While this is true, it was also possible to make an informed judgment about the relative strengths of the forces and their likely chances of victory. In contrast to some cases not studied here—for example, Burundi, Northern Ireland, and Bosnia—judgment of success and failure was premature.
The five cases examined here include variations in outcome—successful management of the spoiler (RENAMO in Mozambique and the KR in Cambodia) and failed management of the spoiler (the CDR in Rwanda, UNITA in Angola, and the State of Cambodia [SOC] in Cambodia). The judgment of successful management or failed management of the spoiler is based on whether the spoiler has been relatively weakened or strengthened vis-à-vis its opponents. The cases also vary in the mix of strategies chosen to manage the spoiler and the type of spoiler. In terms of primary strategy, Cambodia (against the KR) is an example of the departing-train version of the coercion strategy. Rwanda is an example of the withdrawal version of the coercive strategy. Angola and Cambodia (against the SOC) are examples of inducement; Mozambique is an example of a mixed inducement and socialization strategy. In terms of spoiler type, the KR and CDR are examples of total spoilers, UNITA and the SOC are greedy spoilers, and RENAMO is a limited spoiler. My judgment of spoiler type is based on evidence of intentions at the time.
Several limitations of the cases should be pointed out. First, they all took place after the end of the Cold War. Although this increases the relevance of lessons for policy makers who must grapple with the difficulties of peace making unconstrained by superpower rivalry, the findings are bounded by the historical era. Second, the primary custodian in
all of the cases is the UN. There is variation among the cases, however, in the roles played by individual states as interested actors or subcustodians who support the peace process. Third, not all combinations of strategy and spoiler type are covered in the cases, given the relatively few cases of spoiler management in the 1990s. For instance, neither coercive diplomacy nor use of force to defeat the spoiler is included. Fourth, these cases are not completely independent of one another. Strategies for managing a particular spoiler were sometimes the result of lessons derived from another case. Some UN personnel worked on more than one case. Some spoilers themselves likely drew lessons for their strategy based on evaluating the efficacy of custodians in other cases.
Following the method of structured, focused comparison of Alexander L.George,12 the case studies address the following general questions:
What was the spoiler’s behavior? What demands did the spoiler make? What was the rhetoric of the spoiler?
How did external parties interpret the empirically verifiable observations of the spoiler? What were the custodian’s judgments about the intentions and motivation of the spoiler?
What evidence existed for interpreting spoiler intentions and evaluating the organizational unity of the spoiler? What evidence did the custodian have? What evidence did it cite to support its interpretations? Was other evidence ignored or disregarded?
What strategy did the custodian choose to manage the spoiler? What was the theory behind the strategy?
Did the custodian implement the strategy effectively? What was the effect of the strategy—on the spoiler and on other parties in the conflict?
Did the custodian reevaluate the strategy during its implementation? Did the custodian reconsider its initial diagnosis of spoiler type?
Case 1. Rwanda: Threatened Withdrawal
The Arusha Peace Accords, signed in August 1993 by President Juvenal Habyarimana of Rwanda and officials from the Rwandan Patriotic Front (RPF), a rebel force composed of mostly Tutsi exiles, promised to end a three-year civil war in which approximately 10,000 people had been killed. The accords pledged an end to violent conflict between the Hutu ethnic group, which formed approximately 70 percent of the population, and the Tutsi ethnic group, which comprised nearly 30 percent.13 Simultaneously, the accords attempted to end Hutu political hegemony over the Tutsi; integrate thousands of Tutsi exiles into Rwandan life; and democratize the Rwandan government, which had been dominated for over 20 years by a
small elite group of Hutu close to Habyarimana. The accords also contained elaborate provisions for power sharing in government; integration of the two armies; a detailed plan for the return of some soldiers to civilian life; procedures for democratization of Rwandan politics; and the establishment of a coalition transition government, the Broad Based Transitional Government (BBTG). The accords were the culmination of 14 months of negotiation and mediation by the Tanzanian government, in conjunction with the Organization of African Unity and the governments of France, Belgium, and the United States. The UN was to oversee the accords’ implementation.
The mediators of the agreement apparently anticipated resistance from Hutu extremists in the army and government, who had rallied under the banner of the CDR.14 The RPF vetoed provisions that would have given the CDR a role in a new Rwandan government, arguing that it was not an independent political party and that its extreme belief in ethnic superiority was contrary to the spirit of settlement. A further point of contention concerned representation in the army: Arusha allotted 50 percent of the officer corps and 40 percent of the rest of the army to the RPF. Although the formula alleviated RPF security fears, it was contested by the CDR. Nonetheless, the Rwandan government’s negotiating team reached agreement with the RPF on those terms.
The accords contained one major flaw: they lacked a strategy to deal with the CDR. The United States and France advocated inclusion of the CDR in the peace process. Instead, the RPF and the government signatories to the accords hoped for a robust UN peacekeeping force that “would neutralize the extremists.”15 Chastened by its recent experience in Somalia, however, the UN had no intention of robust peacekeeping. Indeed, some UN diplomats foresaw implementation as a relatively easy task. The force that was deployed to Rwanda was not only less than the parties had agreed to but also less than what the UN Assistance Mission to Rwanda (UNAMIR) force commander, General Romeo Dallaire, believed necessary.
Between the signing of the peace accords in August 1993 and the full deployment of UNAMIR in February 1994, tension and uncertainty grew in Rwanda. The accords aided the growth of moderate Hutu politicians who could provide an ethnic bridge to the mostly Tutsi RPF. Habyarimana’s party seemed divided; the CDR, resolutely opposed to compromise with the RPF, emerged as a possible competitor to Habyarimana. In October 1993 a coup attempt in neighboring Burundi by Tutsi officers against its recently elected Hutu president triggered acts of genocide in that country. Between 50,000 and 100,000 people died, including Burundi’s president. In Rwanda the coup bolstered Hutu extremist resistance to the Arusha compromise, sowed suspicion among moderate Hutu politicians toward the RPF, and emboldened Hutu extremists to publicly
advocate extermination of the Tutsi as a final solution to Rwanda’s ethnic problem.16
UNAMIR’s top officials, Special Representative Jacques-Roger Booh-Booh and General Dallaire, confronted several challenges: Habyarimana’s unwillingness to fulfill his obligations to the Arusha Accords and form the BBTG, low-level political violence and ethnic killings, growing public incitement of ethnic hatred—especially by the popular radio station Radio Mille Collines—and evidence of a CDR plan to commit genocide if the BBTG were installed. On January 11, 1994, Dallaire cabled the UN Department of Peacekeeping to inform it that a high-level government defector had told UNAMIR about the formation of militias specially trained to carry out a genocide, the creation of lists of targets of Hutu moderates for assassination, a plan to kill Belgian peacekeepers in order to drive the UN out of Rwanda, and a specific threat that the BBTG would be attacked upon installation.
Although Dallaire requested better equipment that would improve UNAMIR’s capacity to respond in the case of crisis, his request was ignored. He also sought permission to begin independent searches for arms caches but was told to do so only in conjunction with local authorities— some of whom were implicated in the January 11 warning. In the words of the definitive account of this period, UN headquarters was prepared to approve only “what the traffic would bear.”17
Habyarimana’s role in the plot and his motivations for stalling on implementing the Arusha Accords were uncertain; analysts disagreed on whether Habyarimana was himself an extremist who was only tactically committed to the peace process or a pragmatic peace maker whom ethnic extremists had boxed in. The faction that was implicated in the January warning to Dallaire included members of Habyarimana’s own elite troops, the Presidential Guard, several close presidential advisers, and the president’s wife. Habyarimana’s behavior could be construed as supporting either interpretation. His prevarication might have been evidence that he hoped events would strengthen the extremists for a return to war. Alternatively, it was possible that he feared for his life if he implemented the accords and therefore “buying time, without knowing exactly for what purpose, became a kind of survival reflex.”18 U.S. officials close to the situation believed that there was no split between Habyarimana and the extremists; therefore, the key to dealing with the extremists was to get Habyarimana to install the BBTG. They assumed that he would deliver his followers to the peace process.19
No coordinated unified international approach was taken toward the extremists. Booh-Booh adamantly opposed CDR demands for inclusion in the peace process, only to reverse himself and argue for their participation. France continued to have cordial relations with both Habyarimana
and officials implicated in the January warning. Arms supplies from France arrived in Rwanda in January 1994 in violation of the Arusha Accords and, according to UNAMIR officials, again in April after the beginning of the genocide. Representatives of donor nations in Kigali failed to voice clear consistent concerns about the government’s violence and human rights violations. Even the radio broadcast of genocidal threats did not yield a consensus toward the extremists. The failure of international actors to assert unified minimal standards of human rights “probably succeeded only in eroding the credibility of diplomatic suasion.”20
In late March 1994 a strategy emerged for dealing with the stalled implementation that proved completely counterproductive. The UN threatened to withdraw its peacekeeping mission unless the warring parties fulfilled their obligations to the peace plan. Pushed by the United States and endorsed by the UN, the strategy bore little connection to any of the problems in Rwanda: the presence of extremists who had pledged to attack the parties of peace, the uncertainty about whether Habyarimana was allied to the extremists, or Habyarimana’s reluctance to carry out Arusha. Although the U.S. government clearly saw the Rwandan government as the main culprit,21 the strategy threatened to punish all of the parties for its obstruction. On April 5 the UN Security Council announced that the mandate for UNAMIR would be extended but warned that its patience had worn out; if the parties did not comply with Arusha, the UN would leave.
On April 6 the strategy, combined with growing diplomatic pressure, wrested from Habyarimana a commitment to install the BBTG. In a meeting in Arusha with the regional mediators and representatives of France and the United States, he agreed to implement the accords, only to be assassinated on his return to Kigali later that night. Immediately, the Presidential Guard and CDR assassinated almost all of the Hutu moderates, and their militias began killing Tutsi throughout the country. In addition to assassinating the moderate Hutu prime minister, Agathe Uwilingiyimana, the Presidential Guard killed 10 Belgian peacekeepers who had failed to protect her.
In the ensuing two months of genocide the UN and its member states reduced the number of peacekeepers in Rwanda. Its approach became one of appeasement by inaction. For two months the UN and the United States urged the RPF and the Presidential Guard to establish a cease-fire and return to negotiations. In doing so they conveyed a clear message: committing genocide was not enough to disqualify a party in Rwanda from a legitimate place at the bargaining table.
The UN’s failure to manage the spoiler problem in Rwanda resulted from poor diagnosis, which was caused in part by organizational blinders. The poor diagnosis was threefold: incorrect assessment of the number
of spoilers, their types, and the locus of the spoiler problem. The UN targeted Habyarimana, who was a spoiler but a limited one whose hesitations stemmed from fear of his former followers. Habyarimana did not control the CDR, a total spoiler that sought to destroy the entire process. The threat of withdrawal forced Habyarimana to reveal that he was a limited spoiler who preferred peace. But the same threat had no leverage on the CDR; in fact, the threat signaled to it a basic lack of international commitment to the implementation.
The only strategy that might have avoided the cataclysm of April 1994 would have been one that combined protection for the parties of peace—the moderate Hutu parties, the RPF, and Habyarimana—through a larger, more proactive peacekeeping force; clear credible threats against the use of violence by extremists; and the defanging of extremists by reducing their capability to attack the peace process. The international community would have had to diagnose that there were two spoilers, not one; that the CDR was a total spoiler; and that the locus of the spoiler problem resided with Habyarimana’s followers. The goal then would have been to protect the coalition for peace, marginalize the Hutu extremists, and create the opportunity for Habyarimana to distance himself from the extremes to join the middle.
That such a strategy was not articulated and attempted was over-determined. Beyond the obvious intelligence failure in detecting Habyarimana’s position, myriad factors led to an absence of critical judgment. Right from the planning for implementation, the UN and its member states were only minimally committed to the peace process. The choice of the withdrawal strategy stemmed from the Clinton administration’s desire to show Congress that the UN had the discipline to say no to peacekeeping operations that seemed troubled.22 Organizational politics and the frailty of individual decision making combined to ignore the warnings of spoilers committed to genocide. Mediators could not overcome the basic contradiction in their analysis: on the one hand, Arusha excluded Hutu extremists who threatened the peace process; on the other hand, those same extremists would not really act when the peace process was implemented. The UN and its member states did not want to face up to the implications of an accurate diagnosis of the problem and failed to even create a unified disciplined message of disapproval to the extremists.
Case 2. Cambodia I: The Khmer Rouge and the Departing-Train Strategy
The Paris Peace Accords, signed on October 23, 1991, brought to a close four years of negotiations aimed at ending Cambodia’s civil war. Several factors contributed to the settlement. The war had reached a stale-
mate among the major combatants—the National United Front for an Independent, Neutral, Peaceful and Cooperative Cambodia (FUNCINPEC), the royalist party of Prince Norodom Sihanouk; the KR; the State of Cambodia (SOC); and the Khmer People’s National Liberation Front (KPNLF). The external patrons of the warring parties (China, Russia, Vietnam, and the United States) had tired of the stalemate and wished to divest themselves of their clients. The countries of the region coalesced around a framework that called for multiparty elections, demobilization and disarmament of the parties, and UN implementation of the agreement. The warring parties consented to the agreement as a result of their sponsors’ coercion and remained deeply suspicious of one another as well as distrustful of the international consortium—the Core Group, consisting of the permanent five member countries of the UN Security Council and interested regional states, including Japan and Australia— that brokered the agreement.
The party to the settlement that commanded the most scrutiny was the KR. Responsible for the deaths of nearly 2 million Cambodians during its three years in power, the KR survived because of its military prowess, support from China, and diplomatic recognition from the Association of Southeast Asian Nations (ASEAN) and the United States. Inclusion of the KR in the peace process evolved from the recognition by the United States and others that the KR could not be defeated militarily and the hope that peace would marginalize it.23
Between November 1991 and May 1992 the KR complied sporadically with the Paris settlement. Immediately after the accords were signed but before deployment of the UN Transitional Authority in Cambodia (UNTAC), a KR spokesman, Khieu Samphan, attempted to open a political office in Phnom Penh, only to be attacked by SOC-instigated rioters. Samphan retreated to Bangkok, and the KR demanded 1,000 peacekeepers to provide security in Phnom Penh. At about the same time, Sihanouk proposed a SOC-FUNCINPEC coalition government as a means of isolating the KR, which was a violation of the spirit and law of the Paris Peace Accords. As a result of these two incidents, a lobby was created that argued the KR was ready to implement the accords but was frightened into noncompliance by its antagonists. Some experts suggested there were two Khmer Rouges: a “moderate” KR that wanted peace and a “hardline” KR that wanted war. Other analysts argued that KR commitment to the accords was always tactical and the belligerence of its adversaries provided it with a convenient excuse for spoiler behavior. Evidence of KR intentions, based on interviews of its top officials and lowly foot soldiers, suggests that the party was committed to the peace process insofar as it promised to return it to power.24 The KR interpreted the Paris Peace Accords as giving UNTAC the right to dismantle the SOC’s administra-
tion and expel all ethnic Vietnamese from Cambodia. If UNTAC carried out such a program, the KR would benefit from the crisis that would ensue for the SOC.
The KR’s inconsistent behavior between November 1991 and May 1992 provided evidence for different interpretations about its intentions. The KR frequently violated the cease-fire, restricted UN mobility in its areas, boycotted joint military consultations, and attacked a UN helicopter in February 1992, wounding a peacekeeper. When UNTAC began in March 1992,25 the KR “adopted a posture of cautious cooperation, despite engaging in numerous ceasefire violations.”26 On the positive side, it allowed some UNTAC civilians into its areas. Samphan actively represented the KR on the Supreme National Council (SNC), a transitional body composed of representatives of the warring parties, and cooperated on several humanitarian initiatives with the UN.27 On the negative side, the few military observers allowed into KR territory were so limited in their activities that “at times they seemed more [like] hostages than monitors.”28
The issue of KR noncompliance surfaced publicly in May and June 1992. On May 30, 1992, a group of KR soldiers refused to allow a UN armed convoy, accompanied by the secretary-general’s special representative, Yasushi Akashi, and his force commander, General John Sanderson, to enter KR territory in western Cambodia. Instead of insisting on UNTAC’s right of passage, Akashi retreated. When cantonment, demobilization, and disarmament of all of the warring parties began in June, the KR insisted that it would not comply because Vietnamese forces were still present in Cambodia in violation of the Paris Peace Accords and because UNTAC had not established effective control over SOC. The KR insisted that it would demobilize only if UNTAC dismantled SOC administrative structures and vested the SNC with the power to run the country.
UNTAC officials debated the use of force to gain KR compliance. French General Michel Loridon, UNTAC’s deputy force commander, believed that a show of strength would compel the KR to meet its obligations and would establish a reputation among the other factions that the UN would enforce compliance. If the UN did not act, Loridon maintained, it would lose credibility with all of the parties. He asserted that UNTAC had the legal authority to enforce compliance and did not need to seek a Chapter 7 mandate to do so.29 Human rights organizations and nongovernmental organizations in Cambodia supported Loridon’s call for toughness against the KR.30
Akashi and Sanderson opposed the use or threat of force against the KR for six reasons. First, Sanderson drew no distinction between threatening the KR with force to gain compliance and going to war with the KR.31 This dovetailed with Akashi’s assessment that the troop contributors to the mission, as well as the Core Group, would oppose fighting a
war. Second, Sanderson thought that UNTAC was not organized for offensive operations and would thus prove ineffective. Third, both Akashi and Sanderson argued that any attempt to use force would destroy the Core Group’s consensus; it was unlikely that the group would immediately agree to condemn KR behavior and condone the use or threat of force. Fourth, they believed that, if the Core Group disintegrated, the operation would collapse. Fifth, they thought that any use of force would undermine the attempt to negotiate compliance, which was Akashi’s preferred approach and which meshed with Sanderson’s belief in the “good Khmer Rouge-bad Khmer Rouge” line; as long as there was a good KR, there was a possibility of earning its voluntary compliance. Sixth, Akashi feared that using force against the KR would upset the balance of the peace process. FUNCINPEC and KPNLF derived some of their power from the KR counterbalance to the SOC; to weaken the KR might tempt the SOC to seek an outright victory.
Akashi sought to discuss with the KR its noncompliance. He quickly realized that the KR interpreted two key components of the Paris Peace Accords in ways counter to their spirit. First, the KR believed that an injunction for the removal of all “foreign forces” meant all foreigners, regardless of their status as combatants. At stake was the political status of non-Khmer Cambodians, especially ethnic Vietnamese who lived in Cambodia. Second, the KR insisted that the accords required the complete destruction of SOC administrative structures. Akashi realized that meeting the first demand would violate human rights and pander to ethnic extremism and that meeting the second demand was impossible: UNTAC did not have the administrative personnel or know-how to replace the SOC. The likely result would be chaos, which, although satisfying KR aspirations, would destroy the peace process.32 Akashi chose not to appease KR demands and began to build a strategy for managing the party’s spoiler behavior. He met with local representatives of KR patrons—Thailand and China—to create a unified approach to the problem, privately condemned KR noncompliance at SNC meetings, and warned then-UN Secretary-General Boutros Boutros-Ghali that pressure would likely have to be applied in the future.
In diagnosing the problem in a letter to Boutros-Ghali in July 1992, Akashi described the KR rejection of UNTAC’s efforts to address its concerns and stated that KR behavior had demonstrated that it was not sincerely committed to the peace process. He attributed KR noncompliance to its attempt “to gain what it could not get either in the battlefield or in the Paris negotiations, that is, to improve its political and military power to such an extent that the other parties will be placed at a distinct disadvantage when UNTAC leaves.”33 Akashi then chronicled acts of bad faith
by the KR and asserted that Samphan “is little more than a glorified mouthpiece of…Pol Pot,” who “seems to be dedicated to the doctrine of simultaneously ‘talk and fight.’”34 He argued that the circumstances dictated a strategy of “patient persuasion” and “sustained pressure.” He insisted that UNTAC “adhere to an impartial stand, while criticizing any acts in violation of the Paris Agreement.” Although doubting KR good faith, Akashi asserted that keeping an open door to its participation would prevent turning KR followers into a “permanent disgruntled minority.”35
Akashi pointed out that leverage over the KR could come from Thailand and China. He expressed skepticism about the former because of the unwillingness of the Thai government to control several army generals who collaborated with the KR in illegal timber and gem trading across the Thai border. Akashi also argued that China’s influence over the KR waned after the signing of the peace accords and its cessation of assistance to the KR. He requested that, if KR noncompliance continued, Boutros-Ghali should mobilize economic pressure against the KR: “This should not however involve any spectacular action, but rather a steady strengthening of our border checkpoints adjacent to the DK [KR] zones, in order to control the inflow of arms and petroleum and the outflow of gems and logs, a major source of DK’s [KR’s] income.”36
By leaving open the door to the KR to rejoin the peace process, Akashi hoped to contain its dispute with UNTAC and to limit its hostility to the peace process. An aggressive stance toward the KR would make targets of all UNTAC’s personnel. If KR opposition could be contained, UNTAC could redeploy its peacekeepers to protect the election, which would go forward without the KR. The Australian foreign ministry promoted Akashi’s strategy in a September 1992 policy paper that became the basis of the international response to the KR.37 It sought a concerted response from the Core Group, based on the judgment that KR “demands are not in strict accord with the actual terms of the Paris Agreements.”38
Although Sanderson was loathe to try to enforce KR compliance with the peace process, he reconfigured the peacekeepers to protect the strategic objective of holding the elections. He redeployed his battalions in ways that could contain KR attacks and reinterpreted the traditional peacekeeping doctrine of neutrality and impartiality, going so far as to use military units of all of the factions (save the KR) to assist UNTAC in providing security during the election. Sanderson justified using soldiers from the other parties by insisting that these armies were not deployed against the KR per se but rather against any force determined to disrupt the election. In a rather ingenious formulation, he described the changed military mission as “an interposition strategy, but not between opposing
forces. Rather, it was between a highly moral act sanctioned under international law and supported by international consensus, and any person or group which might threaten it.”39
To help establish an atmosphere where civilians would feel secure participating in an election, UNTAC created a radio station in December 1992. Although its purpose was to convince voters of ballot secrecy and to explain UNTAC’s mission and activities, it also aimed to neutralize KR propaganda. The establishment of Radio UNTAC overcame objections of the UN Secretariat, among others, that an independent media outlet would endanger UNTAC’s perceived neutrality.
From September 1992 to May 1993, the scheduled month for elections, UNTAC held firm in its strategy. The Core Group clearly signaled that the peace process would go forward without KR participation. China and Thailand acceded to a nonbinding UN Security Council resolution to impose economic sanctions on the KR. And as the election date drew near, both China and Thailand explicitly supported the elections. Although the KR increased attacks against UNTAC during March and April 1993, it did not unleash a military offensive against the elections, which were held as planned.
UNTAC’s strategy for dealing with the KR was imaginative and effective and serves as the prototype of the departing-train strategy for managing spoilers. When faced with KR attempts to undermine peace, UNTAC emphasized that the peace process would not exclude the KR nor would it be held hostage by it. UNTAC tried first to address the KR’s specific demands, but when such demands threatened the core agreement of the peace process, UNTAC sought international consensus to delegitimize KR demands and to approve the strategy of continuing the peace process in its absence. UNTAC reconfigured itself militarily to protect the electoral process from KR attack but left open the door to the KR if it wanted to reengage the peace process.
Case 3. Cambodia II: The SOC and Inducement
The motivation of the KR was difficult to gauge because its grievance toward UNTAC’s lack of control over the SOC had some merit. UNTAC faced an ongoing problem of SOC obstruction of its mandate. It never established control of SOC’s administrative structures, given the lack of qualified personnel in the numbers that were needed, SOC’s tendency to ignore and sabotage UNTAC directives, and UNTAC’s unwillingness to assert its administrative prerogatives as outlined in the Paris Peace Accords. An equally pressing matter of control concerned the SOC’s police and security forces. Throughout UNTAC’s life span, SOC police intimidated civil society organizations and physically assaulted and assassi-
nated members of the opposition. While UNTAC reconfigured its military mission to protect the election from KR attack, SOC security personnel waged a low-level reign of terror against its party’s electoral competition.
Despite warnings and protest by some UNTAC human rights officials, Akashi did not perceive the SOC as a potential spoiler and was unprepared when it attempted to undermine the peace process immediately after the May 1993 election. The election results devastated the SOC. Despite the SOC’s use of intimidation, assassination, and fraud, FUNCINPEC beat SOC’s political party, the Cambodian People’s Party (CPP), which disputed the election results and attempted to grab power. The president of the SOC, Hun Sen, charged that the election was fraudulent; some CPP officials declared secession for one region of the country; and SOC-orchestrated riots throughout Cambodia threatened UNTAC officials. At the very moment that UNTAC had seemingly achieved success, the SOC threatened to tear down the whole edifice of peace in Cambodia.
From the beginning of implementation of the Paris Peace Accords, Akashi, the UN, and the Core Group focused mostly on the KR as a threat to peace and ignored the potential for the SOC to be a spoiler. Moreover, KR spoiler behavior created incentives and excuses for the SOC to undermine the peace process. Unless both spoilers carried out their commitments to the peace process, each could claim that its behavior was a function of the other. Witnessing violence or lack of commitment by their opponent, they asserted that playing by the rules would leave them vulnerable. This rationale can become a barrier for peace makers who seek to determine the real motivation behind acts of violence and subversion. In cases of mimetic spoilers, peace makers tend to accept a situational explanation for spoiler behavior—an explanation that may be correct and will seem reasonable—and overlook the possibility that such an explanation may also be a facade for a party already committed to tactical subversion of the peace process.
The departing-train strategy toward the KR increased Akashi’s perception that UNTAC was dependent on the SOC and thus limited in its options for controlling it. That UNTAC was dependent on the SOC is not in doubt; if the SOC left the peace process, there would be no elections and the parties would return to war. But Akashi failed to comprehend the SOC’s dependence on UNTAC. UNTAC had greatly strengthened the SOC, which had a stake in holding the election and gaining international legitimacy and support. If the SOC had to wage another war against the KR, it would do so from a stronger position with international support and FUNCINPEC’s abandonment of its former coalition partner. Akashi also misread command-and-control relations within the SOC. He told
aides of his fear that Hun Sen had only tenuous control over hard-liners, who if pushed too hard would rebel against Hun Sen and return to war. Akashi’s staff, however, believed Hun Sen to be firmly in control of his followers.
Although the accords contained numerous references to administrative control, the UN interpreted UNTAC’s mandate in a limited way. UNTAC was “urged to rely on ‘codes of conduct and guidelines for management’” and to avoid issuing binding directives.40 Akashi, under the advice of Boutros-Ghali, envisioned UNTAC exerting control by monitoring and supervising existing administrative structures.41 Moreover, Akashi worried about UNTAC’s lack of domestic legitimacy; he envisioned the SNC as a governing body that could make hard decisions, referee the peace process, and therefore provide domestic legitimacy for actions against spoilers. Although the SNC did assist the strategy against the KR by ruling that its demands were illegitimate, it proved much less effective toward the SOC.
For the most part, UNTAC refrained from attempting to enforce compliance with its administrative directives. Akashi did not want to use the prerogative of replacing or repositioning SOC bureaucrats. His restraint in the face of SOC obstruction prompted the head of UNTAC administration, Gerald Porcell, to resign in February 1993. At the time Porcell lamented that as long as UNTAC did not “have the political will to apply the peace accords, its control cannot but be ineffective.”42
UNTAC was also lax in holding the SOC accountable for human rights violations. Although UNTAC’s mandate for creating an environment conducive to human rights was the most ambitious ever for a UN peacekeeping operation, its enforcement of violations was “dilatory, sporadic, and improvised.”43 Akashi interpreted UNTAC’s human rights mandate narrowly. He believed that rigorous action to enforce human rights would endanger UNTAC’s neutrality; moreover, he felt that a broad interpretation of human rights “seemed to be based on unrealistically high standards in the context of Cambodia’s reality.”44
As violence increased and the political climate deteriorated at the end of 1992, Akashi acceded to the establishment of a special prosecutor’s office. But as William Shawcross notes, “the office languished as Akashi, Sanderson, and other UNTAC officials began to fear that prosecutorial zeal might destroy the entire mission’s fragile links with the Phnom Penh regime.”45 Akashi was indirectly supported in this in February 1993 when ASEAN and China exerted pressure to limit the human rights component of UNTAC to education and training.46
Between May 1992 and May 1993, UNTAC pursued a de facto policy of inducement against the SOC. It usually did not act against SOC violations; when it did, it sought to deter SOC obstruction through private
persuasion. Akashi asserted that the parties need not adhere strictly to all of their commitments. As he later wrote, “too rigid, legalistic interpretations of the agreements would have hindered my work.”47 He believed that, although the accords were “based on the concepts of Western democracy, Asian methods and procedures should be used in the negotiations.”48 Such methods found public reprimand (or acknowledgment of violation of agreements) distasteful.
In May 1993 UNTAC’s tacit strategy of inducement became explicit in its response to SOC’s attacks immediately after the election. UNTAC’s support of the election results wavered. Within the first 48 hours, Akashi attempted to console Hun Sen and promised to investigate fully his charges of electoral fraud. He also sought out the leader of FUNCINPEC, Prince Norodom Ranariddh, to urge him to be conciliatory toward the CPP.
In the two weeks after the election the CPP resorted to violence “to blackmail both FUNCINPEC and UNTAC in an attempt to reverse the election results.”49 As Shawcross writes, the CPP attempt at blackmail was largely successful.50 Fearing a return to open civil war, the UN acceded to a power-sharing arrangement mediated by Sihanouk that provided the SOC with more power and cabinet positions than its electoral performance deserved. Akashi acknowledged that the deal was “unorthodox by universal democratic principles,” but he defended it on the basis of the “practical wisdom” of combining FUNCINPEC’s political appeal with the administrative experience and power of the CPP.51
Akashi believed that compliance on most of the dimensions of the peace process, including demobilization and disarmament, and human rights protection, was secondary to compliance with holding an election. The election became a Holy Grail for UNTAC; Akashi defined the mission’s success solely on the basis of achieving it, and the myriad goals of UNTAC’s mandate—promotion and protection of human rights, disarmament and demobilization, and administrative control during the transition—were made subservient to this quest. In the end this even included rejecting a “strict adherence” to the results of the election; Akashi and UNTAC did not insist that the political outcome of the election accurately reflect the electoral outcome for fear that it would undermine the triumph of holding the election.
In hindsight it is possible to hazard a judgment about the effectiveness of UNTAC’s strategies for managing the spoiler problems in the Cambodian peace process. The strategy it chose to deal with the KR has been vindicated. The KR’s power declined; in the summer of 1996 a severe factional split decimated the party. In June 1997 an internal rebellion by soldiers who wanted to end the war deposed Pol Pot. Time has not been so kind in evaluating UNTAC’s strategy toward the SOC. For instance, the
SOC’s steadily increasing grip on power since 1993 caused several Cambodian experts in 1996 to warn of a “creeping coup.”52 Such warning turned prophetic in July 1997 when Hun Sen and the SOC attacked FUNCINPEC, sent Prince Ranariddh into exile, and assassinated FUNCINPEC officials and prodemocracy advocates. As the United States, ASEAN, and the UN engaged in collective hand wringing, the SOC skillfully manipulated an internationally negotiated and implemented peace process to triumph in a war that it could not win on the battlefield.
Case 4. Angola: UNITA and the Failure of Inducement
In May 1991 the two main antagonists in the Angolan civil war, the government of Angola and UNITA, signed a peace agreement in Bicesse, Portugal. The agreement, mediated by Portugal, the United States, and the Soviet Union, called for an 18-month transition period during which each party would canton its troops, demobilize some of them, and then join the remainder in a unified Angolan army. At the end of this period, elections would determine the presidency and composition of a national assembly. The agreement contained no provisions for power sharing, nor was there a provision for the election’s loser to receive a share of ministerial portfolios or provincial governorships in the highly centralized state structure. Although each party had been urged to consider various power-sharing options, both vetoed them in the belief that they would win the elections.
The government of Angola reluctantly accepted a role for the UN to monitor and assist implementation of the peace agreement. The role and size of the UN presence were the result of hard bargaining between UNITA, which wanted a large UN force with an active mandate to implement the agreement, and the government of Angola, which perceived a large UN presence as an infringement on its sovereignty.
The UN operation in Angola was done on a small budget with little independent latitude. The parties themselves were left to carry out the demobilization, and they failed. By May 1992 about 70 percent of the estimated 160,000 soldiers had been processed at assembly points, but only 6,000 had been demobilized. By the elections on September 29–30, substantially more government troops had demobilized than had UNITA soldiers; nonetheless, both sides had intact armies. Between January and September 1992 there were numerous violations of the cease-fire, with UN monitors estimating that at least 16 skirmishes could have escalated into major combat between the parties.53 That they did not was attributed to the parties’ determination to see the process through to elections and their command and control over their armed forces.
Until late summer 1992 the United States and the United Nations
were more apprehensive about the Angolan government’s willingness to abide by the peace process than they were about UNITA’s. The biggest worry for U.S. policy makers regarding Angola was that the government might not accept an electoral defeat and would throw the peace process into crisis.54 Nonetheless, UNITA’s president, Jonas Savimbi, provided signs that he might be the obstacle to ending the war. Under cover of the agreement, UNITA stationed soldiers throughout the country—especially in areas it had previously not engaged—and cached arms for quick access. In addition, rumors suggested that UNITA was holding back armed battalions across the Zaire border. A top Savimbi aide defected and informed the United States of a plan for a quick strike offensive to take the country by force.
The transition period had upset the balance of power between the two militaries. The cantonment process had worked decisively in UNITA’s favor; UNITA’s army maintained its discipline and remained a unified force that could be mobilized quickly for fighting purposes. The government’s army, on the other hand, suffered from poor morale; desertion and drunkenness were rife. Savimbi’s generals informed him that Angola could be taken by a military surprise attack—a sentiment that Savimbi relayed to officials in Washington in August, two months before the elections.55
As elections drew near, U.S. policy makers grew alarmed about Savimbi’s potential as a spoiler. Reports circulated that his standing among Angolan voters was declining, raising the possibility that Savimbi would lose the election. Then-Assistant Secretary of State for Africa Herman Cohen and his Portuguese counterpart traveled to Angola at the beginning of September to persuade Savimbi and President Eduardo dos Santos to share power after the elections. While Savimbi responded enthusiastically to the plan, dos Santos privately expressed interest but felt he could not publicly commit to such a deal. In the end no contingency plan was formed in case Savimbi lost the election.
The elections were peaceful. The UN took multiple intricate precautions to prevent fraud: representatives of the competing parties were present at the 5,800 polling stations and at every municipal, provincial, and national electoral center. Party representatives signed off on results sheets at each level, a process that added days to the vote counting. There were numerous logistical foul-ups, but international observers judged the elections to be free from intimidation and fraud. The results jibed with the UN electoral unit’s quick count, thus confirming that fraud did not occur.56 In the legislative elections the ruling Movement for the Popular Liberation of Angola (MPLA) outpolled UNITA by a margin of five to three. Dos Santos received about 49.5 percent to Savimbi’s 40 percent for the presidency; as no candidate received 50 percent of the vote, a runoff election would have to be held in 30 days.
Problems emerged as soon as the preliminary results were reported. The first results came from Luanda, an MPLA stronghold; not surprisingly, they showed the MPLA and dos Santos with a large majority. Savimbi and UNITA immediately cried foul and issued a bombastic five-page memo (in English) declaring that the MPLA was engaged in massive fraud and warning in veiled terms that UNITA would resume the war if it lost the election. When returns from the central provinces (UNITA’s stronghold) did not offset MPLA’s early lead, UNITA withdrew its generals from the joint command of the newly established united Angolan army. Savimbi retreated to a redoubt, refused to meet any foreign officials or take a phone call from Cohen, denounced the UN, and repeatedly ignored international calls for reasonableness. His army then launched attacks throughout the Angolan countryside, quickly seizing large amounts of territory and destroying government arms depots.
The international response to Savimbi was scattershot. The UN at first insisted on the primacy of the elections and condemned Savimbi for obstructing them. The United States initially urged Savimbi to use the established mechanisms for investigating his electoral grievances, making its appeal over Voice of America because Savimbi would not speak with U.S. officials. As the UN attempted to press Savimbi to honor the election results and participate in a presidential runoff election, South Africa’s foreign minister, Pik Botha, visited Savimbi and unilaterally put forward a plan to shelve new elections and to provide for a government of national unity. Policy makers in Washington quickly chose a similar option: to press the MPLA into a power-sharing agreement to appease Savimbi.
The U.S. response established the international strategy toward Savimbi. By initially equivocating, the United States failed to challenge him; then by interpreting his actions as understandable and reasonable, it chose a strategy of inducement, which served only to encourage further aggression. Savimbi continued his attempts to defeat the Angolan government and to avoid serious negotiations. A year later, in November 1993, Savimbi returned to negotiations, but only after the rearmed Angolan military had rolled back UNITA’s gains, the United States had granted diplomatic recognition to the Angolan government, the UN had imposed sanctions against UNITA, and 300,000 Angolans had died.
Washington’s strategy of inducement toward Savimbi was disputed by U.S. officials in Angola, including Ambassador-designate Edmund De Jarnette. From the beginning of the crisis he and others stationed in Luanda diagnosed the problem as stemming from Savimbi’s personality and ambitions. They believed he was motivated by a desire to win complete power in Angola. These officials counseled giving Savimbi an ultimatum to return to the peace process, backed by a threat to use U.S.
military force. From the onset of the crisis through 1993, De Jarnette argued that an inducement strategy toward Savimbi would fail.
U.S. officials in Washington, however, believed that Savimbi did not want to overturn the peace process but simply wanted a better deal. Long-standing ties between Savimbi and the Defense Department and intelligence agencies were part of the problem; individuals who knew Savimbi and had been romanced by him could not bring themselves to find him at fault. Likewise, negotiators who had worked hard to get an agreement could not believe that one of the signatories was rejecting a compromise solution outright. And those in Washington who were swayed by the analysis of the U.S. diplomats in Luanda were stymied by the tight connections between Savimbi’s Washington lobbyists and the Bush administration.
A top U.S. policy maker admitted in retrospect that the strategy of inducement toward Savimbi was “clutching at straws” but insisted that a tougher policy was out of the question.57 The Bush administration was preoccupied with the November 1992 presidential election and then with the humanitarian crisis and intervention in Somalia in December. Moreover, Savimbi’s Washington connections precluded any use of coercion. Inducement failed in Angola because U.S. policy makers erroneously believed that limited incentives would satisfy Savimbi. Furthermore, they were not prepared when inducement emboldened Savimbi to continue his spoiler behavior. In part, Savimbi’s personality defined the conflict in all-or-nothing terms; a combination of racism, paranoia, and megalomania led him to believe that the MPLA had stolen the election from him and that he had the right to rule all of Angola. In October 1992, at the time that he rejected the elections, anything less than an absolute firm stand against his pursuit of war and a credible threat of force and sanctions against UNITA had little chance of persuading Savimbi to return to the peace process. The initial choice of inducement convinced Savimbi that the international community would likely defer to his return to war. Inducement proved that the U.S. and South African governments saw Savimbi’s demands and actions as legitimate. There was little international support for rallying behind the sanctity of the Bicesse peace process.
The change in power positions of the MPLA and UNITA also worked against the inducement strategy. By October 1992 the peace process had greatly strengthened UNITA; Savimbi was confident that UNITA could win the war, and he underestimated the resolve and capability of the MPLA to fight. If Savimbi had been militarily weak in October 1992 and the international community held a monopoly on rewards, inducement might have brought him back into the fold. Savimbi, however, continued to have uninterrupted sources of revenue through UNITA’s control of diamond mines and support from some of his neighbors—Mobutu in
Zaire and rogue elements in the South African defense forces—who would continue to supply him with arms, ammunition, and fuel to fight the war.
Case 5. Mozambique: RENAMO and Successful Inducement
In October 1992 the government of Mozambique signed a peace agreement with RENAMO, a South African-trained and -assisted guerrilla movement. Although RENAMO was unable to defeat the Mozambican government, it had rendered most of the country ungovernable. RENAMO’s use of terror, indiscriminate killing of civilians, press-ganging of child soldiers, and destructive capacity had earned it the appellation the “Khmer Rouge of Africa.” Upon assuming the task of implementing the peace accords, the UN, having witnessed Savimbi’s return to war in Angola and being concerned about the character of RENAMO and its leader, Afonso Dhlakama, doubted the sincerity of RENAMO’s commitment to the negotiated settlement.58
The tardy deployment of UN peacekeepers and establishment of an administrative capacity to oversee cantonment and demobilization of troops provided RENAMO and the government with an excuse to stall on their obligations under the peace treaty.59 But as the necessary UN units and personnel arrived in Mozambique and eliminated the excuse, UN Special Representative Aldo Ajello found himself stymied by the warring parties’ continuing noncooperation—especially RENAMO, which embarked on a three-month boycott of the implementation process.60
To bring RENAMO back into the fold, Ajello pursued two policies. First, the peace accord’s mediators stressed that Dhlakama, in addition to searching for security assurances, placed a high priority on the issue of legitimacy—that his movement had fought for a just cause, had domestic roots, and was not simply a puppet of South Africa—and hungered for recognition as a Mozambican nationalist who had fought for democracy. Much of RENAMO’s behavior belied such a self-image, and Dhlakama’s vision of democracy was not multiparty pluralism where parties do not return to war if they lose an election or eliminate their adversary if they win. A key task for Ajello therefore was to socialize RENAMO into the rules of democratic competition and to make its legitimacy contingent on fulfilling its commitment to peace. The subcustodians of the peace process—the representatives of the countries supporting the UN Observation Mission in Mozambique—assisted Ajello by continuously reinforcing Dhlakama’s desire for legitimacy. Leaders of neighboring countries overcame their distaste for RENAMO, met with Dhlakama, and treated him as a legitimate national leader.
Second, Ajello understood that it was crucial to wean RENAMO from its military raison d’être. To do so he fulfilled a promise that mediators
made to Dhlakama to provide financial assistance to RENAMO to transform it into a democratic party. Italy, Ajello’s home country, gave him a $15 million fund to assist RENAMO’s renovation. This fund, combined with the flexibility to use it in conjunction with gaining incremental compliance to the accords, gave Ajello enormous leverage with RENAMO. This leverage was amplified because of the unique context of the Mozambican case. Unlike Angola, where Savimbi bankrolled his spoiler behavior through the illegal diamond trade, or Cambodia, where the Khmer Rouge replenished its weapons and ammunition through illicit gem and timber deals, Mozambique’s paucity of accessible valuable commodities deprived RENAMO of resources if it chose to reject peace.
Although Ajello’s use of inducements kept RENAMO engaged in the peace process, it also encouraged Dhlakama to continue voicing grievances and making demands on the UN. Ajello resolved several incidents involving cantonment and demobilization by acceding to RENAMO demands that were not contained in the peace accords.61 Eventually, Ajello grew frustrated with RENAMO’s tactics and threatened withdrawal. Ajello used a visit by Secretary-General Boutros-Ghali to Mozambique in October 1993 to convince RENAMO that continuation of the peace process was conditional on RENAMO meeting its obligations. The withdrawal threat yielded short-term results; RENAMO pledged anew its commitment to peace, and UN officials reported progress on implementing key provisions of the agreement.
Ajello discovered, however, that the closer the peace process came to the election in October 1994, the more the UN’s threat of withdrawal declined in credibility. Having spent $800 million on the peace process, the UN could not convince RENAMO of its pledge to walk away without at least holding elections.62 Ajello therefore relied increasingly on inducement, even though the risk of whetting Dhlakama’s appetite grew more dangerous as the elections drew near. What would happen, for example, if RENAMO lost the elections and demanded that the results be annulled in the hope of being rewarded for its obstructionist behavior?
The UN had sought to make the actions of a postelection spoiler irrelevant by fully demobilizing both armies and creating a new unified army. Demobilization, however, was incomplete; both RENAMO and the government had armies at the time of the elections in October 1994. Neither side, however, had anything close to the amount of troops wielded by the parties in Angola in October 1992; RENAMO and the government had several thousand troops held in reserve.63 If the loser decided to defect from the peace process, it would still possess a destructive capacity but not a force capable of winning the civil war in a short period.
Given the UN and U.S. experience of Savimbi’s overturn of the elections in Angola in 1992, as well as the use of violence by the SOC to
blackmail UNTAC after the Cambodian elections, it is remarkable that as late as June 1994, four months before the Mozambican elections, neither the UN nor the U.S. embassy in Maputo had contingency plans in case the loser of the election rejected the results.64 Instead of thinking strategically about the possibility, the United States unsuccessfully urged the parties to consider a power-sharing pact that would establish a South African-style government of national unity after the election. On the other hand, the governments of states in the region, very much chastened by the experience of Savimbi in 1992, began consultations on responses if Dhlakama rejected the results.
RENAMO did attempt to boycott the elections and reject the results. Because the United Nations had not set a limit on how far inducement would go, diplomats from the neighboring countries of Zimbabwe and South Africa had to meet with Dhlakama and warn in no uncertain terms that they would not accept any obstruction of the elections. The UN and United States followed the warning with both an appeal to Dhlakama’s desire to be a legitimate national actor and democrat and a promise to investigate any alleged electoral fraud. RENAMO rejoined the elections, withdrew its charges of fraud, and took its seats in the newly elected Mozambican parliament.
SPOILER MANAGEMENT: EVALUATING SUCCESS AND FAILURE
The case studies discussed here suggest several general findings about managing spoilers in peace processes. First, they illustrate that spoiler type, number of spoilers, and locus of the spoiler problem are key variables that affect the robustness of strategies for managing spoilers. Second, the cases underscore the need for custodians to diagnose correctly the spoiler problem, a task made difficult by the fog of peace making. Third, the cases show that quite often the very rules, beliefs, and frames that custodians use to cope with the complexity of peace processes defeat them. Fourth, the cases of successful spoiler management demonstrate that spoilers need not destroy peace—if custodians create robust strategies, seek international consensus behind the strategy, and recognize the much maligned force of normative power.
Diagnosis and Treatment
The case studies in this paper demonstrate that the creation of an effective strategy of spoiler management rests first on correct diagnosis of the spoiler problem and the selection of an appropriate strategy to treat the problem. Choice of strategy depends on judgment about the inten-
tions of the spoiler and awareness of constraints posed by other parties to the conflict.
The failure to manage spoilers in Rwanda, Angola, and Cambodia (SOC) stemmed from poor diagnosis and the choice of a strategy that was inadequate to the problem at hand. In contrast, common to the successful cases—Cambodia (KR) and Mozambique—was early identification of a potential problem, a good diagnosis, and the selection of an appropriate strategy to treat the problem. By examining the specific strategies—withdrawal, departing train, and inducement—we can begin to understand the conditions that favor or mitigate against their effectiveness.
Custodians used the threat of withdrawal as a primary strategy in Rwanda and as a secondary strategy in Mozambique. In Rwanda the strategy pressured Habyarimana, a limited spoiler, into implementing the Arusha Accords but provided a green light for the CDR, a total spoiler, to commit genocide to stop implementation. In Mozambique the strategy created a short-term sense of urgency among the warring parties to fulfill their obligations to peace before the opportunity to end the war disappeared.
These two cases suggest that the withdrawal strategy’s effectiveness depends in part on the type of spoiler at which it is aimed, the number of spoilers in the peace process, and the credibility of the threat. The withdrawal strategy is a coercive strategy that aims to punish recalcitrance by taking away the opportunity for peace. For the strategy to have its intended effect, the target must view the withdrawal of international engagement as a punishment. Moreover, for the threat to be credible, the custodian must seem indifferent to ending the war. Against a total spoiler like the CDR in Rwanda, the strategy cannot work because the spoiler does not view withdrawal as punishment. Indeed, the strategy can backfire if it signals to the total spoiler that the custodian lacks a commitment to peace.
The withdrawal strategy is a blunt instrument of spoiler management. The strategy can work only if all parties to an agreement are equally culpable in failing their obligations and all spoilers are limited spoilers who want a settlement to succeed. The threat of withdrawal can then provide a test of motivation: if the parties are really committed to the peace process, the possibility of losing international support might provoke them into beginning the implementation process. Where not all parties are spoilers, as in Rwanda, the strategy punishes even those parties committed to peace. Where one party is a total spoiler, as in Rwanda, the withdrawal strategy gives it a veto on the peace process.
The Departing Train
The departing-train strategy succeeded in Cambodia against the KR, a total spoiler. Success derived from the ability of the custodians to learn the spoiler’s intentions, the strong degree of international consensus and unity behind a judgment that the spoiler’s demands were illegitimate, the deployment of force to protect the peace process, and the custodian’s willingness to stand firm in implementing the peace process.
The departing-train strategy was developed by UNTAC in Cambodia to deal with the KR. Evidence suggests that the KR was a total spoiler and that a coercive strategy was necessary, but UNTAC judged that it was constrained from using either force to defeat the KR or coercive diplomacy to threaten it. The use of force against the KR ran the risk of escalation, which could have endangered the overall UNTAC mission. Moreover, use of force against the KR could have weakened FUNCINPEC, which used the existence of the KR as leverage against the SOC. The departing-train strategy held out the possibility of continuing the peace process while marginalizing the KR. UNTAC ruled that KR demands were illegitimate, continued the peace process without the KR, and established a deadline—the election—for it to join the process. UNTAC then redeployed its force to protect the election.
The departing-train strategy requires that external actors take a stand on the spoiler’s demands: Are they legitimate and important enough to halt the progress at compromise that the other parties have made? For the strategy to work, external actors (including current or former patrons of the spoiler) must concur that, if the spoiler’s demands are met, peace may be unattainable. Having made a decision that the peace process will go on without the spoiler, external actors must find ways to protect the parties of peace. Attempts can be made through the spoiler’s patron to warn the spoiler of dire implications if it escalates its attacks, and the custodian can deploy military forces to protect people and processes such as elections.
The departing-train strategy also depends on the custodian’s ability to convince all parties that the peace process will proceed without everyone on board. International consensus is crucial for sustaining such credibility. In Cambodia the commitment of China and Thailand to the election timetable clearly conveyed to the KR that the peace train was departing.
The Cambodian case suggests two limitations of the strategy. First, the strategy is a gamble that the parties to the peace process will be strengthened sufficiently to deal with the spoiler on their own after international disengagement from the process. The strategy’s long-term success rests on the ability of the parties favoring peace to form a strong enough front to withstand future challenges. For example, when UNTAC
departed in 1993, the KR had not been eliminated, but it had been marginalized. Subsequent developments have vindicated the strategy: a factional split in the KR all but destroyed it in 1997. Second, the strategy can make the custodian dependent on the parties that get on the train. In Cambodia, for example, the strategy against the KR required that the SOC be part of the peace process. UNTAC’s sense of dependence on the SOC constrained it from opposing the SOC’s behavior as a greedy inside spoiler.
Inducement was attempted against the SOC in Cambodia, UNITA in Angola, and RENAMO in Mozambique. Although the strategy failed in the first two cases, inducement toward RENAMO largely succeeded, albeit in conjunction with a heavy emphasis on socialization of RENAMO and an implied threat by regional actors that established the limits of inducement.
The three cases of inducement suggest that this strategy can only succeed with limited spoilers. Toward the SOC, a greedy spoiler, the inducement strategy backfired by strengthening the SOC and weakening FUNCINPEC. The only strategy that might have managed the SOC would have been a strong socialization effort that would have established strong norms of human rights, democracy, and good governance, as well as systematic use of carrots and sticks to gain compliance. In the case of UNITA, another greedy spoiler, inducement whetted its appetite for power. Thus, even if UN and U.S. policy makers were correct in 1992 that Savimbi was not a total spoiler but simply greedy, the only strategy that would have managed him would have been a socialization strategy with a heavy dose of coercion. This relates to the second aspect of spoiler type: commitment to achieving goals. Savimbi’s insensitivity to risk and cost meant that he would fulfill his obligations to the peace process only if the price of not doing so was extremely high.
The Angolan and Mozambican cases suggest that leadership personality is a key variable behind successful inducement. Unlike Savimbi, Dhlakama did not characterize his conflict in all-or-nothing terms. UN and U.S. policy makers came to believe that Dhlakama’s goal was not total power in Mozambique but rather to gain legitimacy for his movement, for the war he waged, and for the people and region that supported him.65 When faced with Dhlakama’s continuous demands for money and recognition of various grievances, UN Special Representative Ajello believed that financial resources would keep Dhlakama and RENAMO in the peace process. Moreover, the neighboring countries (especially Zimbabwe and South Africa), the United States, and the UN continued to
focus on the legitimacy theme—to try to socialize Dhlakama and RENAMO into playing by the rules and transforming themselves into democratic politicians. The message was clear: Dhlakama’s newly recognized legitimacy depended on his willingness to meet his obligations. When the moment of truth came and Dhlakama’s commitment wavered, the neighboring states, the UN, and the United States delivered a strong unambiguous signal: if he returned to war, his legitimacy would be lost and there would be a high cost to pay.
The Mozambican case suggests several requirements for a successful inducement strategy. First, the spoiler must be a limited type. Diagnosing Dhlakama’s type was not easy, however; the view that he sought limited goals was disputed by veteran Mozambique watchers and was at odds with RENAMO’s past brutal behavior. Second, external actors must be unified in establishing the legitimacy and illegitimacy of spoiler demands and behavior. Again, this seems deceptively simple. At the time many in the human rights community disputed the legitimacy of RENAMO as an actor. The United States, the UN, and neighboring countries all had to reverse their characterization of Dhlakama both as a puppet of the former apartheid regime in South Africa and as a leader with no domestic constituency.66 Third, inducement is aided if the custodian, by acting in coordination with others, is the sole source of rewards to the spoiler. Unlike Savimbi in Angola, Dhlakama had no independent source of capital if he chose to return to war. Fourth, in civil wars where the goal is the establishment of sustained peace, inducement is most likely best carried out in conjunction with a concerted international effort to socialize the spoiler into accepting the basic rules of good governance and democracy. Fifth, inducement must be accompanied, if necessary, by a credible threat to establish its limits and break any cycle of grievance, reward, new grievance, reward, new grievance. Even in the case of a limited spoiler like RENAMO, inducement can encourage increased obstructionist behavior in the hope of getting more rewards. If a limited spoiler continues to undermine peace, it will run the risk of prompting its opponent to view it as a total spoiler for whom no concession will gain its commitment.
In addition to pointing to the crucial role of good diagnosis and choice of appropriate treatment, the case studies provide a vivid reminder of the uncertainties, complexity, and ambiguity of peace processes—what I refer to earlier as the “fog of peace making.” But the case studies also provide evidence that several organizational rules, beliefs, and frames that custodians use to cope with uncertainty can contribute to poor diagnosis
and in extreme cases to avoiding the spoiler problem. Such organizational blinders include (1) prior commitments of the organization to the spoiler, (2) organizational doctrine, (3) organizational Holy Grails, (4) organizational interests, and (5) organizational roles.
In several of the cases, prior commitments between individual states and spoilers blocked a correct interpretation of the intention and behavior of the spoiler. In the case of Angola the U.S. government had long been a patron of Savimbi, had supplied him with arms and supplies, and had cultivated his friendship. When Savimbi rejected the election results and returned to war in late 1992, the initial tendency of the United States was to believe Savimbi’s versions of events, to discount interpretations that Savimbi intended to win complete power, and to emphasize that even-handedness was still called for because the MPLA was no better than UNITA. Savimbi’s impressive network of friends, supporters, and representatives in Washington buttressed the psychological commitment to him. While this commitment led U.S. policy makers to benign interpretations of Savimbi’s actions, or at least to assertions that there were no good guys in the conflict, the personal networks served to constrain the influence of those who sought a more aggressive policy toward Savimbi.
The Angolan case is similar to France’s relationship to the CDR and Presidential Guard in Rwanda. First, the French military’s prior organizational commitment to the Presidential Guard led it to demonize the RPF— to see it as the all-or-nothing party, a “Khmer Noir,” an image that reinforced and legitimized the Hutu extremist version of the conflict. Second, the prior relationship caused the French military to ignore clear signs that the CDR was planning genocide. Third, the personal network of relationships extended high into the Mitterand government, thus paralyzing conflicting policy tendencies in the French foreign ministry.
Beyond prior ties to the spoiler, the ability of organizations to interpret the intentions and behavior of spoilers and to fashion effective responses is constrained by their doctrines. The UN, for example, approaches its custodial role with several assumptions. First, its representatives assume that the parties are acting in good faith when they sign a peace agreement. This leads UN personnel to downplay violations by signatories to agreements. Second, its representatives tend to be slavish in their devotion to the troika of traditional peacekeeping values: neutrality, impartiality, and consent. Such values, when followed blindly, constrain attempts to challenge spoiler behavior, as custodians fear being seen as partial to the victim. Even when
spoiler behavior is recognized as such, the doctrine insists on reestablishing consent and not forcefully challenging violations.
In the successful cases of spoiler management, custodians took a stand on the merits of issues under dispute. In Cambodia, UNTAC rejected KR grievances as illegitimate. UNTAC’s success stemmed from its ability to garner international consensus against KR interpretations of the Paris Peace Accords. In the case of Mozambique the regional custodians of the peace process confronted RENAMO when it attempted to withdraw from the election at the last moment and threatened to return to war.
The successful cases of spoiler management, however, do not diminish the more significant point that doctrine poses constraints; they simply show that custodians had to reinterpret their actions so as not to appear to be in conflict with their principles and had to triumph in intraorganizational battles over the doctrinal implications of their actions. For instance, in Cambodia General Sanderson succeeded in employing soldiers from the warring factions to protect the election against the KR by redefining the meanings of impartiality and neutrality. Radio UNTAC, which earned universal praise for its role in combating KR propaganda, was established only over strenuous objections from UN headquarters that such a radio station would imperil UNTAC’s neutrality.
A third organizational blinder is the tendency of custodians when faced with complexity and uncertainty to redefine their goals and standard of success to one overriding accomplishment, a Holy Grail. When implementing peace, the UN often drops its commitments to various components of the peace agreement in order to focus on holding an election. When spoilers plunge their countries into war, the UN focuses on obtaining a cease-fire.
For example, when faced with KR intransigence in Cambodia, UNTAC narrowed its mission to holding an election rather than fulfilling complete implementation of the Paris Peace Accords. This redefinition of mission was an appropriate response to a spoiler’s attempt to veto the Paris agreements. Yet UNTAC became so focused on attaining the election that it ignored SOC violence and obstruction during the electoral campaign. The need to reach an election took precedence over how the parties conducted themselves during the election. When an election finally took place, UNTAC acquiesced to SOC blackmail and encouraged its quest for a coalition deal that was disproportionate to its electoral result. UNTAC feared that the SOC’s threat would nullify the achievement of the election, so it compromised the quality of the election to appease the SOC.
When spoilers in Rwanda and Angola attacked and plunged their countries back into civil war, the UN responded by insisting on a ceasefire and return to negotiations. In both cases, spoilers were willing to kill hundreds of thousands of people to demonstrate that they did not want a negotiated settlement, yet the UN responded by pleading with them to return to negotiations.
An overriding sense of organizational interest can also prevent custodians from recognizing and effectively managing spoilers. In Rwanda, like Angola before it, the UN’s interpretation of the conflict and its consideration of an appropriate response were based heavily on “what the traffic would bear.” Faced with information that requires costly and risky action, the UN and many of its member states chose to ignore the information.
The conceptions that mediators and UN special representatives have of their roles can lead them to misinterpret evidence of spoiler intention. Both mediators and special representatives invest enormous time and energy into negotiating and implementing peace; therefore, when faced with spoiler behavior, they assume spoiler motivation and behavior to be negotiable—a judgment that confirms the continuing relevance of their role as peace maker. They tend to seek out any evidence that confirms the basic willingness of the parties to still reach agreement and ignore compelling evidence that suggests one of the parties may reject peace completely. They tend to grab at any straw that seems to hold out the promise of a settlement. Even when confronted with compelling evidence of bad faith and the preference of one or more of the parties for war, they are likely to insist that there are no alternatives to negotiation. In some cases the mediators or special representatives seem unwilling to place the responsibility for continued hostilities on the parties themselves and instead blame their own organizations for not providing the one request that would have made the difference between war and peace. Finally, there is the perverse tendency of custodians to so value an agreement that they blame the victim rather than the spoiler.
The Need for International Agreement and Coordination
A common denominator among the successful cases of spoiler management is unity and coordination among external parties in defining the problem, establishing legitimacy for the strategy, and applying the strat-
egy. Although this is not a surprising finding, it is nonetheless a robust one, and external parties to a peace process ignore it at their peril.
Spoilers often exist because external patrons provide them with guns, ammunition, capital, and sanctuary. External patrons may also help internal spoilers survive by supporting their claims to legitimacy—support that can play havoc with a strategy such as the departing train, where the key to putting pressure on the spoiler is to declare its grievances illegitimate and to insist that the peace process itself embodies the best chance for resolving the conflict. In the cases of successful spoiler management, external support for the spoiler had either dried up or been severely curtailed. Inducement succeeded with RENAMO because its external patron, South Africa, wanted a peaceful settlement to the war. Unlike UNITA or the KR, RENAMO did not have easy access to illegal markets to finance a continuation of the war.
International unity and cooperation require cultivation, time, resources, and pressure. It is therefore better if both are institutionalized in the peace process, as with the Core Group in Cambodia.67
The Legitimating Function of Spoiler Management
Another key aspect to spoiler management is the development of an international consensus about what are legitimate and illegitimate solutions to a country’s civil war. The successful strategies of spoiler management all have in common a unified stance by external actors about which spoiler demands should be met and which should be rejected. For example, the departing-train strategy in Cambodia depended on the willingness of international actors to define limits of accommodation for the spoiler. In Cambodia even the former patrons of the KR agreed that their client’s demands were illegitimate and that the peace process could move forward without them. In Mozambique external actors agreed to legitimate RENAMO as a nationalist party, socialize it into an agreed set of rules of behavior, and establish limits on how far it would be appeased.
By contrast, in the failed cases of spoiler management, no international consensus formed about legitimate and illegitimate solutions to the civil wars. In Rwanda external actors failed to create a common stance toward the Hutu extremists and wavered about the content of the Arusha Peace Accords. In Angola little attempt was made to rally international support against UNITA’s return to war. Indeed, the strategy that emerged from the United States insisted that legitimacy was irrelevant to ending the war and proceeded to pressure the party that had won the election.
The finding that legitimization is an integral part of spoiler management is important in two regards. First, it counters the adage that solutions to internal conflicts must come from the participants themselves. In
this study successful management of internal conflict has resulted from the willingness of external actors to take sides as to which demands and grievances are legitimate and which are not. Second, it shows that when external consensus is used in conjunction with a coherent larger strategy, the setting of a normative standard can be an effective tool for conflict management. This last point is crucial. For all the lip service they pay to the power of norm setting, when it comes to protecting peace and managing spoilers the member states and many UN personnel seldom act like they mean it.
How confident can policy makers and theorists be in the conclusions presented above? Although the set of cases is small and necessarily incomplete, I believe the conclusions derived here are robust. The successful cases of spoiler management were tough cases. Few analysts in 1991 predicted that the KR would be so weakened that it would cease to exist seven years later; similarly, most experts on Southern Africa were skeptical that RENAMO could be transformed from a killing machine into a loyal democratic opposition party.
While more work needs to be done on refining the strategies of spoiler management presented above, and while more cases can add to our knowledge about necessary conditions for such strategies to succeed, I believe the analysis of peace making here is sound. The period after a peace agreement is reached is a time of uncertainty and vulnerability for peace makers and citizens alike. International actors who seek to bring deadly civil wars to a close must anticipate violent challenges to peace processes. Instead of thinking generally about the possible threats to peace, they must ask, “Who are the threats to peace?” The custodians of peace must constantly probe the intentions of warring parties: they must look for evidence that parties that sign peace agreements are sincere in their commitment to peace, and they must search out and make good use of intelligence about the warring parties’ goals, strategies, and tactics. The emphasis of this paper on individual typing of leaders and attributing permanent traits to some of them raises important questions about the weight of personality and situation as motivations of spoiler behavior. But I believe that the analysis begins to correct an analytic trend that asserts that all behavior by leaders in civil war is situationally determined.
The research presented here strongly suggests that international consensus about norms and coordination behind a strategy of aggressive management of spoilers can provide the difference between successful and failed implementation of peace agreements. Custodians must judge
what is right or wrong, just or unjust, and fair or unfair in peace processes. They can do so either explicitly by creating an international consensus about what is appropriate for the warring parties or tacitly by not taking action in the face of violent attacks and spoiler behavior.
Is the post-Cold War international environment conducive to forging such consensus and coordination? The answer is mixed. A large gap currently exists between international proclamation of norms—human rights, rule of law, democracy, and good governance—and behavior in warfare that makes a mockery of such norms—wanton killing of noncombatants, ethnic cleansing, war crimes, and atrocities. The establishment of international consensus about norms writ large seems disarmingly easy; in the cases presented above the establishment of consensus about norms in a specific case was much harder. States seem much more likely to proclaim norms when they view compromise on their freedom of maneuver in the abstract; in actual cases, national foreign policies still define situations differently, ascribe motivations differently, judge self-interest differently, and waver in applying universal normative standards.
Likewise the present age presents difficult problems for international coordination in applying a strategy of spoiler management. Economic globalization and the weakening of state control over trade and borders weaken state control over private economic actors who trade with and profit from potential spoilers in civil wars. Cutting off spoilers from resources and capital may be more difficult in the post-Cold War era than previously. Similarly, a proliferation of nongovernmental organizations involved in the implementation of peace accords in civil wars increases the difficulty of establishing and coordinating strategies toward spoilers.
I thank the following people for their comments, criticisms, and suggestions: Howard Adelman, Nichole Argo, Michael Brown, Cynthia Chataway, Juergen Dedring, Michael Doyle, Daniel Druckman, William Durch, Page Fortna, Alexander L.George, Charles L.Glaser, Robert Jervis, Elizabeth Kier, Stephen Low, Dianne McCree, James Morrow, Michael O’Hanlon, Jerrold Post, Tonya Putnam, Donald Rothchild, Timothy D. Sisk, Janice Gross Stein, Paul Stern, and Saadia Touval. I also thank the current and former policy makers and diplomats who spoke with me off the record about their peace-making experiences. A slightly different version of this paper has been published in the fall 1997 issue of International Security (vol. 22, no. 2, pp. 5–53).
figure for the respective populations is 85 percent Hutu and 14 percent Tutsi. Based on new calculations, Howard Adelman estimates that the percentage of Tutsi was greatly underreported, hence the 70 and 30 percent figures here (private communication, Howard Adelman, October 10, 1996).
Howard Adelman and Astri Suhrke, with Bruce Jones, The International Response to Conflict and Genocide: Lessons from the Rwanda Experience, Study 2, Early Warning and Conflict Management (Copenhagen: Joint Evaluation of Emergency Assistance to Rwanda, March 1996), p. 25.
Howard Adelman, “Preventing Post-Cold War Conflicts: What Have We Learned? The Case of Rwanda,” paper presented to the International Studies Association Meeting, San Diego, April 17, 1996, p. 7.
Prunier, The Rwanda Crisis: History of a Genocide (N.Y: Columbia University Press, 1995), pp. 198–203.
Adelman et al., The International Response, p. 68.
Ibid., p. 203.
Anthony Marley, U.S. Department of State, presentation at the Fourteenth Annual Africa Conference, Paul H.Nitze School of Advanced International Studies, Johns Hopkins University, Washington, D.C., April 7, 1995. Lieutenant Colonel Marley (retired) was the U.S. military attaché to the Arusha peace process.
Adelman et al., The International Response, p. 32.
Michael Barnett, “The Security Council, Peacekeeping, and Indifference in Genocide in Rwanda,” Cultural Anthropology, vol. 12, no. 4 (December 1997), pp. 551–578. At the time, Barnett was a Council on Foreign Relations fellow with the U.S. Mission to the United Nations.
Stephen Solarz, “Cambodia and the International Community,” Foreign Affairs, vol. 69, no. 2 (Spring 1990), pp. 99–115.
See especially Steven Heder, “The Resumption of Armed Struggle by the Party of Democratic Kampuchea: Evidence from National Army of Democratic Kampuchea ‘Self-Demobilizers,’” in Steven Heder and Judy Ledgerwood, eds., Propaganda, Politics, and Violence in Cambodia: Democratic Transition Under United Nations Peacekeeping (Armonk, N.Y.: M.E.Sharpe, 1996), pp. 73–113.
Although the Paris Peace Accords were signed in October 1991, the operational plan for UNTAC was not presented to the UN Security Council for approval until February 19, 1992. On February 28 the Security Council approved the mission, and on March 15 the secretary-general’s special representative to Cambodia, Yasushi Akashi, arrived in Phnom Penh. A small UN holding operation was deployed as a bridge between the signing of the Paris Peace Accords and the arrival of UNTAC.
James A.Schear, “Beyond Traditional Peacekeeping: The Case of Cambodia,” in Donald Daniel and Bradd Hayes, eds., Beyond Traditional Peacekeeping (New York: St. Martin’s, 1995), p. 253. Schear was an assistant to Akashi in Cambodia.
Trevor Findlay, Cambodia: The Legacy and Lessons of UNTAC (Oxford, U.K.: Oxford University Press, 1995), p. 51.
Under the UN Charter, a Chapter 7 operation permits enforcement against identified threats to peace and does not require the warring parties’ consent.
Steven R.Ratner, The New UN Peacekeeping: Building Peace in Lands of Conflict After the Cold War (New York: St. Martin’s Press and the Council on Foreign Relations, 1995), pp. 170–171; Michael Doyle, UN Peacekeeping in Cambodia: UNTAC’s Civil Mandate, International Peace Academy Occasional Paper Series (Boulder, Colo.: Lynne Rienner, 1995), p. 67; and Findlay, Cambodia, pp. 37–38.