Abstract
The purpose of international conflict-resolution efforts is, in the short term, to bring an end to violent armed conflict, and, in the medium to longer term, to prevent the revival of conflict. However, at least one of the mechanisms often utilised in conflict resolution and peace agreements, power-sharing, may not only prove problematic in early negotiation and implementation, but may potentially be at odds with the longer-term goal of preventing resurgence of conflict. Why might this be the case? Longer-term peacebuilding seeks to prevent conflict in part by building strong and sustainable states. Such states should be able to avoid reverting to armed conflict because they would be more responsive to grievances and more effective in dealing with violent dissent. However, power-sharing arrangements may undermine such efforts by placing in power individuals and groups not fully committed to, or unable to take part in, governance for the benefit of the entire populace; in part because it necessarily places in power those who have engaged in significant violence to achieve their ends. This is likely to create less democratic states, although we do not insist that democracy is or should be the only goal of peacebuilders. Rather, we suggest that power-sharing arrangements may tend not only towards undemocratic states, but towards states which are not responsive to the needs of the citizenry for security in ways which may undermine human security and state legitimacy.
The purpose of international conflict-resolution efforts is, in the short term, to bring an end to violent armed conflict, and in the medium to longer term to prevent the revival of conflict. However, at least one of the mechanisms often utilised in conflict resolution and peace agreements, power-sharing, may not only prove problematic in early negotiation and implementation, but may potentially be at odds with the longer-term goal of preventing resurgence of conflict. When and why might this be the case? Longer-term peacebuilding seeks to prevent conflict in part by building strong and sustainable states. Such states should be able to avoid reverting to armed conflict because they would be more responsive to grievances and more effective in dealing with violent dissent. However, power-sharing arrangements may undermine such efforts by placing in power individuals and groups not fully committed to, or unable to take part in, governance for the benefit of the entire populace; in part because it necessarily places in power those who have engaged in significant violence to achieve their ends (Mehler 2009). This is likely to create less democratic states, although we do not insist that democracy is or should be the only goal of peacebuilders; we simply assert that, although many see democracy as the ultimate guarantee of sustainable peace, power-sharing arrangements often tend to result in undemocratic states. We suggest that power-sharing arrangements may tend not only towards undemocratic states, but towards states which are not responsive to the needs of the citizenry for security in ways which may undermine human security and state legitimacy.
Power-sharing arrangements are commonly built into peace agreements to provide critical assurances to negotiating parties and induce them to sign and implement them. Such arrangements seek to provide incentives for warring parties to remain committed to the peace process and to assuage their security concerns. Despite the frequency of their usage, the efficacy and durability of power-sharing arrangements vary greatly. Indeed, while power-sharing has provided stability for Switzerland, similar arrangements have failed to prevent Lebanon from descending into civil war. 1 Furthermore, even when power-sharing arrangements may help to assure parties in the short run, they are not designed necessarily to create robust, sustainable, non-violent states in the longer run.
Our focus here is power-sharing arrangements in peace agreements specifically, not those which have solely been written into legislation or the constitution, or which emerge from political transitions, as in, variously, Belgium, Switzerland, and South Africa. Hence, for example, the 1943 power-sharing arrangement that accompanied Lebanon's accession to independence falls outside the purview of this article; the 1989 modification of this arrangement which was negotiated at Ta'if to bring the Lebanese civil war to an end falls in the universe of our cases.
In spite of the diversity of settings in which power-sharing arrangements have been put in place, three key factors seem to affect their ability to ensure a sustainable transition from war to peace and, subsequently, to provide a framework for the conduct of democratic politics: the nature of the state – more specifically the extent of its centralisation and/or repressive power; the capacities and interests of armed group(s) involved in, and excluded from, the agreement; and the nature and/or robustness of the involvement of external actors in either supporting or undermining the agreement.
In many instances, such factors converge to create situations not only unfavourable to the success of power-sharing arrangements, but in which such arrangements create new risks. These include the creation of new grievances and perverse incentives, the risk of importing conflictual behaviour into weak state institutions which are unable to manage them, and alternatively including new participants in governance who lack the background and capacity to function properly, or who may be sidelined by embedded political and bureaucratic practices. These risks, and their consequences, often result in a situation where the short-term reliance on power-sharing runs at odds with the long-term objective of creating robust, sustainable, non-violent states.
In this article, we argue that the risks associated with power-sharing arrangements are particularly acute in the context of weak or collapsed states such as exist in the African context. In conflict-affected countries, the new grievances and perverse incentives often associated with power-sharing further decrease the state's ability to effectively deter and assure (Saideman and Zahar 2009). States whose institutions cannot easily accommodate the grievances of new players (assurance) will often fall back on (more or less violent) methods of exclusion which, in turn, increase the potential for the emergence of violent challenges. And states that cannot effectively wield the threat of legitimate violence (deterrence) are ill-equipped to address these challenges. In other terms, in such conditions, power-sharing arrangements might in fact speed the decline of the state. In this article, we elaborate on these challenges with illustrative lessons from a number of African states that have been the sites of power-sharing arrangements, including Burundi, the Democratic Republic of the Congo, Côte d'Ivoire, Liberia, Rwanda, Sudan, Sierra Leone, and Zimbabwe. While African states have been the subject of an unprecedented range and number of peacemaking and peace-building experiments that involve power-sharing, this is not only a risk in power-sharing arrangements in Africa; our argument is founded on broader comparative analysis of such arrangements in Bosnia, Nepal, Colombia, Lebanon, and elsewhere (Sriram 2008). Based in part on the results of earlier in-depth research of a number of power-sharing agreements (Sriram 2008, 2010; Zahar 2005 a and b, 2009 a and b), and using our findings inductively to look at other cases and develop theoretical propositions, this study is a first step for future research by ourselves and others. As such, it utilises a wide range of comparative examples rather than large-N analyses or a number of in-depth case studies. This enables us to develop a number of theoretical critiques and analytic categories which may be tested in subsequent research.
Why Power-Sharing?
It is important to understand power-sharing in context. In particular, one must understand the impetus behind it and other incentives used to address the vulnerability that groups may feel in disarming as part of the peace process, as well as the rationale for including power-sharing arrangements in peace agreements, and the purported process by which such institutional arrangements are to prevent conflict renewal. We recognise that many of the short- and medium-term impediments to the negotiation and implementation of power-sharing arrangements may also be impediments to other peace agreements. However, the particular ways in which power-sharing arrangements may therefore fail is specific to their provisions. Furthermore, power-sharing has become one of the most frequently adopted approaches to ending civil war. Therefore, a separate analysis of challenges to such agreements, distinct from other types of peace agreements, is important.
Security, Peace Agreements, and Peace Implementation
Power-sharing incentives are but one set of tools through which the international community and third-party mediators may facilitate negotiations. Although the presence of strong third-party guarantors may help to prevent cheating, often such guarantors are not available (Walter 1997, 1999). Thus, parties to peace negotiations face a security dilemma. They require reassurance that, should they negotiate and compromise, their security and survival will not also be compromised (Snyder and Jervis 1999; Posen 1993; Hartzell, Hoddie and Rothchild 2001; Collins 1998). While the parties to an internal armed conflict might in principle become more secure by negotiating and disarming, each fears the defection of the others, and the high costs of being unprepared for surprise attacks. Similar concerns were often cited by analysts of Northern Ireland as one of the main stumbling blocks to the full decommissioning of the Irish Republican Army without which the Unionists refused to implement power-sharing provisions. Yet negotiating and implementing a peace agreement generally requires that one or more parties give up their means of self-defense, while still occupying the same territorial space as their adversaries (Hoddie and Hartzell 2003: 315; Walter 2001). If each party fears being disadvantaged or even destroyed, it may become very difficult to achieve agreement on sensitive security issues such as disarmament and, more generally, to achieve agreement on future power structures (Hartzell, Hoddie and Rothchild 2001). It is for this reason, according to many scholars, that power-sharing and third-party guarantors are so important for peace processes, as they lessen the fear and costs of defection, in the short term, and institutionalise a balance among mistrustful parties in the longer term (Walter 2002; Sisk 1996).
Peace negotiations and agreements seek not only to address the original sources of conflict, and grievances that may have developed over the course of the conflict, but also the acute security concerns of the key parties, particularly, although not only, nonstate armed groups (Walter 2001: 129; Walter 1997: 340). While failure to alleviate security concerns is not the sole reason that peace negotiations and implementation may break down, it is a critical one, because, according to peace agreements, parties that were previously at odds are expected not only to risk disarming, in whole or in part, and therefore risk their survival, but also to work together, or at least to peacefully coexist, in a postconflict (usually) unitary state. Such a state would generally not only have a monopoly over the use of force, but would also have primary decisionmaking power over the distribution of political and economic resources. Thus armed groups will want to ensure that their interests are protected, and that the resources of the state are not used against them (Hartzell 1999: 5; Weber 1946: 78). Commitments to cooperate and disarm must be credible, and institutional protection through power-sharing is only one tool to increase credibility. As we shall see, external guarantors and institutional design are thought to be credible because they can impose costs on possible defectors.
The presence or absence of third-party guarantees, while not the primary focus here, is an important intervening factor in the success or failure of the arrangements examined in this article. The nature and extent of third-party involvement can of course vary significantly. The presence of a third-party guarantor may only ensure compliance during a transitional period, but it may also mask the real intent of parties who comply just to satisfy external actors. If this is the case, behaviour is likely to change once external actors leave, and long-term implementation and peace consolidation may be less successful (Hoddie and Hartzell 2003: 316).
Equally important, third-party guarantees are not always available, and parties to a conflict, as well as the mediators, may have little capacity to induce such guarantees. The experience of international involvement in and support for the Bicesse Accords in Angola is a case in point (Angstee 1993). In many instances, agreements will be negotiated and may even be implemented in the absence of third-party guarantees. Given the mistrust between the parties and the security dilemma described above, the agreement may be designed to lock in certain protections for each of the parties, particularly through power-sharing, which is believed to help to support stable peace-building in its medium to long term. While much of the literature focuses only on the political dimensions of power-sharing – as sharing of access to political power through electoral arrangements, appointments, or secured civil service posts – we treat power-sharing as having four dimensions. These are sharing of control over or access to benefits of: security, territory, politics, and economic resources. Before addressing these dimensions, it is important to discuss the rationale for including power-sharing arrangements in peace agreements, and the purported link between such arrangements and conflict renewal.
Power-Sharing as a Solution to Internal Conflict
Much has been written about the use of power-sharing as a tool to mitigate or prevent conflict in ethnic, plural, or divided societies. Indeed, Ted Gurr includes power- and resource-sharing as part of what he terms an emerging “doctrine” in the international practice of managing conflicts. Such political arrangements are expected to ensure that grievances that could promote conflict are addressed through nonviolent means (Gurr 2002: 28; Fayemi 2006: 171). While power-sharing solutions and theories emerge from work on ethnically divided societies (McGarry and O'Leary 1998), in this article we examine the use of power-sharing in both ethnic and non-ethnic conflicts. However, it is worthwhile to first examine the reasons why power-sharing is generally thought to be a useful tool in ethnic or identity conflicts. The case was first cogently made in Arend Lijphart's Democracy in Plural Societies (1997). Though Lijphart's work focussed on mitigating tensions in democratic societies, his analysis remains relevant even for undemocratic societies emerging from conflict. Specifically, the consociational arrangements he advocates may help to mitigate risks of conflict, because decisions are taken by consensus, which may help to prevent majorities from controlling minorities (Mukherjee 2006; Schneckener 2002: 204–5).
Dealing with the demands of plural rival groups is clearly a challenge in democratic societies, but it is obviously a far greater challenge where groups have engaged in armed conflict. This is true whether the conflict is based on ethnic rivalries or not. Peace processes and transitions are often more unstable and insecure than even the preceding periods of conflict, in situations where fighting may have reached a stalemate. Indeed, transitions may provoke the fears and insecurities that helped to spark and maintain conflict initially – the fear that if a party makes any concessions, the adversary will take advantage of them. As a result, both governments and armed groups are often wary of making concessions, and in particular of laying down arms, unless they believe that their own interests will be protected in any new institutional arrangements. Parties thus need, and negotiations often result in, confidence-building measures such as power-sharing. In general, power-sharing is thought to offer parties institutionalised insurance that they will not face future policies that are discriminatory, retributive, or otherwise harmful to their interests. 2 At the same time, it has increasingly been acknowledged that while these tools have their merits, they also have their limits (O'Flynn and Russell 2005; Roeder and Rothchild 2005), which we will elaborate upon throughout this article.
These are not the only reasons put forth for institutional reform. A recent World Bank study concluded that improvement in institutions reduced the risk of civil war (Djankov and Reynal-Querol 2007).
Power-sharing arrangements in peace agreements include at least one of four dimensions – security, territory, politics, and economics – with many processes involving more than one (Hartzell and Hoddie 2007; Sriram 2008). Each of these is thought to be particularly helpful in addressing specific past grievances and subsequent concerns over the future. Political power-sharing, or the allocation of specific posts in government or specific proportions in the executive, legislature, or judiciary, is but one type of power-sharing. Power-sharing in the security realm is specifically geared to reassuring groups that have experienced violence or repression at the hands of the state and its agents. Groups that experience repression are particularly likely to eventually decide to take up arms; in the absence of reassurance they may be wary of laying those arms down. Territorial autonomy has historically been used to address the self-determination demands of minorities that are both ethnically, linguistically or religiously distinct from the rest of the country and geographically concentrated (Lapidoth 1996). Economic power-sharing is often used where discrimination has resulted in differential distribution of state resources, and economic development among the various regions of a country, and particularly where grievances over maldistribution of resources have generated conflict. We do not explore in detail here specifics of the four types of power-sharing, which have been discussed sufficiently in the literature. However, as becomes evident in our discussions of short-, medium-, and long-term challenges for power-sharing arrangements further below, each specific type may face particular unique challenges, and indeed in some circumstances different types may offer different incentives for key players and even work against one another. Their capacity to function effectively may also be affected by the three contextual factors we discuss next. Thus, for example, governance or political power-sharing incentives may prove difficult to implement where an armed group is insufficiently capable of operating in competitive politics or complex bureaucracies, or where the state apparatus is successfully resistant to the incorporation of armed groups. At the same time, political or security power-sharing provisions may be less appealing where there is a real prospect for territorial autonomy or independence.
The Context of Power-Sharing: Three Factors
There are, as discussed at the outset, three key factors which may affect the creation and implementation of power-sharing arrangements and their durability: the nature of the state, the nature of the armed group(s), and the degree of third-party engagement. The first has not been sufficiently discussed in the power-sharing literature, ironically despite the centrality of the state to conflict resolution efforts and in particular to the institutional bargains offered through power-sharing. The second has been mostly discussed from the perspective of spoiling, or the danger posed by armed groups to the negotiation and early implementation stages of the process (Stedman 1997); while the third has been covered most extensively.
First, the nature of the state is critical. In particular, two dimensions have empirically emerged as particularly relevant for our purposes. The first is the extent of state centralisation or the state's relative ability to project power beyond the capital city. This is particularly relevant in the African context where, as Jeffrey Herbst (2000) has thoughtfully argued, the state is often absent the further one moves away from the center. The second is the state's repressive capacity. As was cogently argued by Nazih Ayyubi in a study of Arab regimes (1996), and more broadly illustrated by the history of the former Soviet Union, states may be able to function for a long time with very little internal legitimacy as long as they maintain effective control over their security forces thus making rebellion prohibitive. Where the state's reach and repressive capacity are strong, conflicts are unlikely to erupt. The very nature of the state's repressive capacity might need further unpacking. For example, if the military is not subject to civilian rule, or has been an autonomous political player, it might resist or alter changes outlined in power-sharing deals (Chinchilla 2008).
Where the state's reach and/or repressive capacity are weak, there may be an increased likelihood of internal conflict and subsequent power-sharing arrangements. Where a state and state institutions have been quite weak, the burden of power-sharing arrangements can be more than they can bear in the immediate post-conflict period. As a set of rules and institutions intended to frame the behaviour of formerly warring factions in such a way as to prevent the recurrence of violence, power-sharing is only as efficient as the capacity of state institutions to enforce the rules and punish would-be transgressors. The weaker the institutions intended to function as the guardians of the system, the less likely it is that they will have the capacity to perform their monitoring and enforcing functions properly, and the less likely that they can accommodate the destabilising effects of competition for primacy between former enemies within them. Alternatively, a state might be relatively centralised, strong and repressive. A repressive state might have the power to prevent cheating by partners, but it might equally be able to abandon its own commitments to its power-sharing partners, or formally adhere to them while resisting any genuine participation. In such instances, the efficiency and durability of power-sharing might depend almost exclusively on the ruling group's willingness to abide by its commitments.
Second, the nature and interests of the armed group(s) may be important. This is not to recapitulate the spoilers debate, but rather to make a different point. We seek to draw attention to another dimension of the politics of non-state armed actors: the way in which the intersection between their objectives and their level of institutionalisation may either facilitate or hinder the transition from war to peace. Insurgents who pursue political agendas and seek to either participate in or replace incumbent governments are more likely to develop proto-state structures (Zahar 2000). Some even develop into fully-fledged quasi-states, such as Abkhazia and Nagorno-Karabagh. But, in many instances, armed groups may have no experience or expertise in governance, and find it difficult to engage, particularly if the state structures and officials resist or undermine their participation. The lower the level of group institutionalisation and previous experience with governance, the more difficult they will find it to function within the often complex maze of rules and institutions that make up contemporary power-sharing arrangements. Former rebels turned politicians sometimes need to be trained in the most basic functions such as their role as parliamentarians or the fundamentals of the legislative process. In places such as Afghanistan where warlords can also be illiterate, the extent of the challenge becomes extremely clear. Alternatively, their interests and goals may not be well-identified or properly addressed in agreements, undermining their commitment to implementation (Dudouet 2009). Thus an agreement which emphasises national power-sharing where a group has sought greater local autonomy might offer the right general incentive – political governance – but in the wrong form. Alternatively, groups which were not included may become more vocal.
Finally, where the commitment of external actors to monitoring, or enforcing, peace agreements is weak, or perceived by one or more parties to be weak, this will affect the viability of these arrangements. Recent work has highlighted the importance of third-party credibility in determining the success or failure of troop deployment in peacekeeping situations. This is all the more important, it is argued, because peacekeepers succeed if they can sustain a deterrent posture: forcing the protagonists not to resort to weapons without having to fire a single shot in the process (Martin-Brûlé 2008). In some instances external actors may actively seek to undermine arrangements, or, through their own instability, passively undermine arrangements. While most research on external actors takes the United Nations as its point of reference, one should not forget that outsiders are involved either directly or indirectly in many of the world's internal conflicts. The involvement of third-parties in what has become known as internationalised civil wars does not stop, nor should it be expected to, with the signing of a peace agreement. Of course, in many contexts it will be the interaction of these factors that create serious challenges.
Our development of these three factors as an analytic device is based on inductive reasoning, as we are interested in theory development, and future research may help to refine or indeed challenge the salience of each. However, initial observations across a wide range of cases suggest that the constellation of these three factors creates a range of challenges for negotiation and implementation of peace agreements relying upon power-sharing. These contextual factors intersect with the limitations of each of the four types of power-sharing in complex ways which are illustrated in our discussions of negotiation and implementation below. We turn first to the short-term challenges in negotiation and then implementation, and then to the difficulty of creating sustainable states with something more than a negative peace.
Power-Sharing: Short-Term Satisficing Solution or Long-Term Plan for Sustainable State-Building?
For practitioners involved in negotiating power-sharing agreements, these arrangements are designed to achieve a set of immediate security demands in the short-term, and are seldom designed to create sustainable, strong states. They are, in a sense, satisficing approaches. The goal is that of negative peace – halting current fighting and limiting the risks of its return. That, however, is not the view espoused by many analysts of power-sharing, especially its consociational variant, who see such arrangements as the starting point for not only lasting peace but democratic sustainability (Taylor 2009). Notwithstanding the potential utility of power-sharing arrangements in the short term, it is essential to understand their impact on post-conflict states in the longer term. The effects of power-sharing are not as uniformly positive as enthusiasts, who market the consociational experience of Northern Ireland as “best practice” and see it as a template for ending conflicts and bringing about democracy, contend (McGarry and O'Leary 1998, 2004). They may not only fail to create stronger, democratic, sustainable states, but embed arrangements which make achieving such states far more difficult. While it is clearly right that those negotiating power-sharing arrangements are concerned primarily with conflict mitigation and termination, not state-building, it is critical that they understand the potential unintended consequences of the short-term bargains they support for longer-term state structures. However, we contend that even the short-term goals of power-sharing may prove difficult to achieve, first because agreement may prove difficult to reach and even more difficult to implement. In the following sections, we turn first to the limits of power-sharing in the short to medium term before addressing its longer term consequences.
Limits in the Short Term: Challenges of Negotiation and Immediate Implementation
Challenges of Negotiation
Negotiating peace agreements of any sort is clearly difficult, given that parties will have very different interests and agendas, and understandings about why armed conflict was waged and what might constitute a just solution, as well as their mistrust of one another and security fears. Therefore, in a general sense, the challenges of negotiating peace agreements with power-sharing incentives are not necessarily greater than negotiating ones without them. However, we do identify here the particular ways in which power-sharing incentives may be difficult to negotiate. While power-sharing incentives may logically offer a way out of the security dilemma faced by armed groups, mistrust among them may result in fear of disarming. This is particularly likely to be the case where one group is left in a privileged position, as the SPLM (Sudan Peoples' Liberation Movement) was by the CPA (Comprehensive Peace Agreement), in relation to all other armed groups in the South of Sudan, which were to be disarmed while the SPLM became the government and sole security provider in the South (Sriram 2008: 107–142). Governments may also be unprepared to offer such incentives, believing either that armed groups will use the negotiation or cease-fire period to rearm, or that, once empowered with such access to governance, the groups will harm those currently in power or the state itself. Here the nature of the state and of the armed group may combine to limit options.
Furthermore, such incentives may foster trust only if a third-party guarantee is also present. The negotiation of Sudan's CPA was arguably bolstered by the presence of strong regional mediators and the prospect of a United Nations peacekeeping force, while the limited participation in the Darfur negotiations may reflect the poor prospect for a robust international force there. 3 In Mozambique and Angola, active international involvement in negotiations was clearly a factor in explaining the timing of the Rome Agreement and the Bicesse Accords respectively. Likewise, analysts have made much of the difference in the robustness of UNAVEM I (the first United Nations Angola Verification Mission) and UNOMOZ (United Nations Operation in Mozambique) to explain the failure of the Angolan peace process and the success of its Mozambican counterpart. And in Nepal, the presence of the relatively weak UNMIN (United Nations Mission in Nepal) was less of a robust deterrent. While we as analysts or advocates might wish that a third-party guarantor always be available, we know that often it is not.
There has yet to be a comprehensive peace agreement encompassing all key parties to the conflict in Darfur.
It may also simply be the case that the wrong incentives are being offered. For example, some groups value territorial autonomy and governance of resources, but have no particular interest in governing at the national level. These groups might not be interested in signing an agreement that offers them, say, participation in national security forces or in parliament. In Sudan, the SPLM/A (Sudan People's Liberation Movement/Army) had the most clearly articulated demands for power- and resource-sharing, and territorial autonomy. It had less apparent interest in inclusion in security forces, which is at odds with future autonomy, so perhaps it should be no surprise that the creation of joint integrated units has been halting, at best (Sriram 2008: 107–142). Where the nature, capacities, and interests of the group are not fully taken into account, negotiations will clearly be hampered.
It may be the case that these incentives alone are not enough to convince armed groups to sign any peace agreement: they will insist that some of their original demands, those that drove them to take up arms, be met as well. Where incentives such as autonomy are part of their original demands, this may facilitate agreement. The SPLM/A's platform had for some time alternately included fundamental change in the national government and regional independence; it achieved the first at least formally, and the prospect of the second through a referendum in the CPA (Sriram 2008: 107–142). The failure to acknowledge key demands or interests of a group may make agreement less likely.
Finally, these incentives can seldom eliminate the “spoiler problem”, and may actually help to create opposition to the peace process, or at least consolidate the opposition of groups that were previously positive or neutral towards it. Peace agreements can seldom be comprehensive, and opposition groups, armed or otherwise, as well as government hard-liners, may oppose peace agreements that exclude them or that they feel give away too much. Disenfranchised groups may even take up arms if they see agreements progressing that appear to exclude them and divide the pie of power and resources in a permanent fashion. Thus, the completion of the CPA in Sudan heightened demands for similar incentives by rebel groups in the East and Darfur (Sriram 2008: 107–142). A similar pattern appears to have emerged in the Terai region of Nepal. 4 And, of course, the Arusha Accords were, analysts argue, directly responsible for fostering Hutu extremism in Rwanda (Paris 2004: 70–71). Agreements might then fail not simply because they haven't acknowledged the nature or demands of existing armed groups, but also haven't recognised potential demands of excluded populations with the capability to become organised groups and challenge any agreement.
The Agreement between the Government of Nepal and the United Democratic Madhesi Front (28 February 2008), online: http://www.unmin.org.np/downloads/keydocs/2008-02-28-Agreement.SPA.Govt.UDMF.ENG.pdf, was negotiated to address the demands of one key group.
Challenges of Implementation
As with the challenges of negotiating power-sharing arrangements, implementing power-sharing arrangements is not necessarily more difficult than implementing other types of peace deals. Indeed, while experts disagree on the precise statistics, a significant percentage of negotiated settlements fail within five years. However, there are specific types of challenges which are specific to the implementation of power-sharing arrangements, which we illustrate here. One or more parties may sign an agreement but easily violate it, either because the incentives were not important to it/them, or because they can achieve key goals through violence. An example is the 1999 Lomé Accord in Sierra Leone, which provided the RUF (Revolutionary United Front) leadership with access to control over resources, directly and indirectly. Not only was Foday Sankoh made a vice president and a minister with control over diamond mines, but fellow RUF leader Johnny Paul Koroma was also made the head of the government commission for peace consolidation, and provisions were made for the incorporation of some RUF members into the state security forces. Yet, as we know, the RUF rapidly returned to fighting – control over diamonds was an insufficient incentive, as the RUF already had attained de facto control over many mines. And evidently, broader political rule was not of significant interest to the RUF either. In Zimbabwe, in spite of the power-sharing deal between the ruling ZANU-PF (Zimbabwe African National Union-Patriotic Front) and the opposition MDC (Movement for Democratic Change), senior ZANU-PF officials have baulked against the deal and worked to undermine it using violent and non-violent means, including the arrest of MDC ministerial nominees and the detention, and torture, of human rights activists and MDC sympathisers. According to some analysts, resistance to the deal within ZANU-PF is partially due to the fact that power-sharing threatens economic benefits that senior officials within the party have secured from their hold onto power. 5 Likewise, it became apparent in 1992 that UNITA (União Nacional para a Independência Total de Angola) leader Jonas Savimbi would not settle for anything less than the presidency of Angola when his failure to achieve this objective in the first post-conflict electoral contest resulted in his decision to revoke his commitment to the terms of the Bicesse Accords. This illustrates how the nature, interests, and capacities of the state or of the armed group may affect the success of implementation.
Zimbabwe's Power Sharing Government: The Struggle Goes On, The Economist, 4 June 2009, online: http://www.economist.com/world/mideast-africa/displayStory.cfm?story_id=13788284; see also, Alex Perry and Simba Rushwaya, Mugabe Pushes Back Against Zimbabwe Power Sharing Deal, Time, 7 October 2008, online: http://www.time.com/time/world/article/0,8599,1847979,00.html.
Parties may sign an agreement involving incentives that are of lesser interest to them, or that they are unable to partake of adequately. This will particularly be the case where armed groups are not structurally prepared to participate in governance activities. In Sudan, for example, many suggest that the SPLM really was interested in the possibility of complete independence, interim autonomy, resources, and autonomous security forces. The CPA gave it all of these things, as well as proportional participation in the central government, and in joint security forces. But many in the SPLM/A resist the idea of participating in joint security forces and being redeployed outside what they expect to be their own country in the future. Many of the SPLM members who have taken up posts as heads of government ministries appear to have been unprepared for their roles, or are unable to actually direct embedded bureaucracies (Sriram 2008: 107–142). In Colombia, when the rebel group M-19 (Movimiento 19 de Abril) disarmed and formed a political party, it was considered unusually prepared, as a group led by many urban educated people, to function in politics. Yet while a few individual politicians were able to thrive, the party failed, an object lesson to future groups considering negotiation and disarmament (Sriram 2008: 143–180). The nature, interests, and capacities of an armed group may affect the viability of many provisions designed to include them in governance. In such instances, the asymmetry in capacities between the state and its power-sharing partners, as well as its interest, or lack thereof, in ensuring the smooth functioning of power-sharing will play a crucial role in the success and/or failure of the power-sharing experiment.
Furthermore, old patterns of mistrust and cheating are sometimes simply transported into institutions of governance, new and old. In Côte d'Ivoire, the Marcoussis power-sharing agreement did little to overcome mistrust between southern and northern political leaders. The former perceived it as
“a French-mediated sellout to terrorism, while northern leaders claim that [Prime Minister Laurent] Gbagbo reneged on private and public promises he made regarding the powers to be exercised by the Prime Minister and appointments to the cabinet” (Rothchild 2008: 143).
In Sudan, many of the institutions developed by the peace agreement to ensure implementation of specific governance incentives – the petroleum commission, the DDR (Disarmament, Demobilisation and Reintegration) commissions, and the like – have not been created, or have been highly dysfunctional. Furthermore, when SPLM members were placed at the head of key ministries, many of the functions of those ministries were first transferred to presidency, which was dominated by the NCP (National Congress Party) and the old government. Such cheating and manipulation of institutional structures can increase mistrust and, where it is severe, provoke a revival of tensions or even conflict. Nascent state institutions may be unable to manage conflict, or conflict may be managed in repressive ways. The result could be the collapse of a peace agreement, the failure of state institutions, or a more violent resolution of disputes. The nature of the state, either due to strength and resistance to accommodating new participants, or due to the weakness of institutions, may thus compound the effects of the previous challenge, where state institutions resist accommodation and groups are not well equipped to challenge them or fully function within them.
Power-sharing may also fail to create a grand coalition and instead create incentives for extremism and ethnic or political hard-liner outbidding. Centrist parties may be pulled to one extreme and find themselves able to reach or maintain power only by forming coalitions with extremist parties, or by taking more extreme stands themselves. The Arusha Accords which distributed power in cabinet equally between Rwandan President Juvénal Habyarimana's MRND (Mouvement Révolutionnaire National pour le Développement) and the RPF (Rwandan Patriotic Front) were perceived so negatively by Hutu extremists that Habyarimana was forced to pursue what Alan Kuperman described as a two-track strategy: seeking to coopt Hutu moderates while working to develop a forceful option with extremists (Rothchild 2008: 253). Likewise, in the Sudan, parties organising for the upcoming parliamentary elections to be held in 2010 have clustered in loose coalitions. A closer look at the coalitions headed by the NCP and the SPLM – coalitions that are necessary if these parties are to win a majority of votes nationally for the NCP and in the South for the SPLM – indicates that the smaller parties tend to hold more extremist views. For example, the NCP is often outflanked on religious issues by its partner the PCP (Popular Congress Party). Smaller parties in the South, particularly those that are not currently represented in government or in parliament, tend to make more extremist demands in pursuit of their objectives. 6 The nature of both state elite groups and armed groups may thus have a devastating effect on politics.
Source removed for purposes of anonymity.
Territorial power-sharing arrangements may encourage secessionist tendencies and the breakup of the state. Alternatively, such strategies may encourage the homogenisation of populations. Where preconflict populations were mixed and were displaced into new largely single-identity communities by conflict, they may not remix after the conflict, or may do so very slowly; if internal borders are made more significant, people may not return to areas they previously inhabited, which may have been more plural. Or population shifts may be more coerced, with groups compelled out of fear to migrate to regions identified as belonging to “their” group. Furthermore, autonomous territories may not be sustainable without serious resource commitments from the central government or the international community. Obviously, central governments, which are likely to be dominated by a group that is ethnically, linguistically, or religiously distinct from the majority of the autonomous region, may not be committed to supporting that autonomy financially. This may promote, not the breakup of the state, but simply the failure of autonomy arrangements. Some analysts suggest that this explains the breakdown of Sudan's 1972 Addis Ababa Agreement; conversely, the presence of oil and the development of oil extraction in Southern Sudan today could, in theory, assist its government in developing its own capacity over time (Atlas and Licklider 1999: 50).
Power-sharing arrangements are unlikely to be sustainable where parties have committed themselves to power-sharing arrangements out of short-term pragmatism rather than as part of long-term policy. If they see governance arrangements as mere tools to achieve power through nonviolent means, they may be unlikely to comply with the agreement if they feel the strategy is failing – for example, because an adversary is likely to do better outside elections. Thus, in Liberia, the leaders of the LURD (Liberians United for Reconciliation and Democracy) and of MODEL (Movement for Democracy and Elections in Liberia) resisted the disarmament of their militias although they were members of the transitional cabinet. This reluctance was partially attributed to their desire to “keep their militias intact should their services be required again by their respective sponsors in Guinea and Côte d'Ivoire” (Africa Confidential 2003 cited in Rothchild 2008: 153). This may particularly be the case if power-sharing is viewed as imported or externally imposed, as was arguably the case in Bosnia-Herzegovina. Where politics is historically clientelistic, corrupt, personalistic, or absolutist, parties committing themselves to democratisation and power-sharing may do so for cynical or short-term ends, and will eventually seek to obtain absolute power. Again, the nature and interests of state and/or armed group(s) will shape prospects for implementation of agreements.
Power-sharing arrangements run the risk of the creation of new conflicts, or the stoking of existing conflicts, when they exclude significant parties or interests. The pragmatic choice to include only warring parties, or only powerful parties, at the negotiating table spawns potential new spoilers. The CPA in Sudan appears to have incited groups in the east and Darfur to use violence to get their demands heard regarding land, resources, and discrimination, out of concern that the north, under the peace agreement, will divide the power and resources of the country while keeping them marginalised. In 1966, the Nigerian military government sought to placate minorities in the secessionist region of Biafra and elsewhere by dividing the country into twelve states. This ushered in a process of conflict-management by territorial fragmentation. In 1979, the amended Nigerian Constitution introduced the notion of “indigeneity” to refer to those citizens of a Nigerian state whose parents or grandparents historically originated from a community within that state. Indigeneity has since become the source of new and increasingly violent local conflicts between “indigenes” and “settlers”, particularly in the Middle Belt region. Here it is less the nature and interests of groups included in the agreement than of those excluded that may shape the fate of the agreement.
After an agreement, violence may erupt not among former enemies but among former allies. This is not a problem unique to power-sharing arrangements, but may well emerge in them. Peace agreements generally seek to address the grievances that initiated the original conflict, and power-sharing seeks to address the fears and demands of the original combatants in relation to each other. However, in either a former government or a former rebel group, one or more factions may be privileged over others in the division of the “spoils” of peace. If this is the case, fighting and factional divisions among former allies, or rifts within groups, may result (Atlas and Licklider 1999: 35–54). Groups, whether state elites or armed groups which were held together in conflict while in fact rather diverse in a range of ways, may splinter if an agreement is tailored which fails to take account of this.
Finally, implementation of power-sharing may be difficult if not impossible in “bad neighbourhoods”. Neighbouring states may foster or destabilise power-sharing in multiple ways. Neutrality has often been invoked to explain the success of the Swiss experiment; likewise, the intrusions of Syria and Israel into the internal politics of Lebanon have been partially blamed for the instability of its power-sharing experiment (Zahar 2005). When geostrategic interests or ethnic kinship considerations draw neighbouring states into the politics of a power-sharing state, they might destabilise the internal pacts. Thus scholars have pointed to the destabilising role of neighbouring states and refugee flows in Rwanda, on power-sharing arrangements. Here it is not just the commitment (or lack of it) by external actors to the agreement that matters; some external actors may actively undermine it.
Limits in the Longer Term: Building Sustainable and/or Democratic States
Power-sharing is thus a set of arrangements that seek to create “good enough” arrangements in the short-term to enable war to peace transitions, but which may, as we have seen, face significant challenges even in this. A greater problem arises as we expand our time horizon, or broaden our expectations about what kind of peace, and what kind of state, are being built. To some, this might look like unfair criticism, holding power-sharing to such exacting standards that it cannot but fail. However, there are serious concerns in the academic and policy communities for the sustainability of peace agreements. And most of the recent peace agreements that have been negotiated under international auspices have included one or more dimensions of power-sharing. In the following section, we argue that power-sharing agreements, even whilst deemed successful in ushering in a transition from war to peace, 7 run the risk of creating states unable to effectively respond to governance challenges that they may face in the future.
Most analysts define success as the ability to avoid a relapse into violent conflict in the five years following the signing of a peace agreement.
At least two features are commonly expected in post-conflict states by those who seek to assist them. The first might be considered a more sustainable peace – the ability of the institutions of state to withstand challenges and crises and manage the potential for violence in a peaceful, constitutional fashion. This will entail strong institutions capable of convincing those with grievances that resorting to violence would be counterproductive and punished by the state. The second expectation is that of a more positive peace, and often one that entails, if not a democratic state, one that is relatively representative of, and responsive to the needs of, the whole population of the state (Jarstad 2008). This entails a state capable of convincing those with grievances that their concerns will be addressed in a non-violent manner.
The creation of a robust state capable of containing resorts to violence, and a responsive, non-violent state, are broader goals of peacebuilders, but are not goals explicitly built into power-sharing arrangements. Nor are they foremost in the minds of researchers as they evaluate the success or failure of power-sharing schemes. Yet, if we are to take seriously the findings of research that suggest that state policies are the prime predictor of violent conflict 8 and that fragile and/or non-democratic states are more likely to resort to either repressive or discriminatory policies in dealing with the grievances of communities in their midst (Murshed 2008), then concerns for the long-term consequences of power-sharing should be front and centre in our analyses.
This is the main finding of the Minorities at Risk project at the University of Maryland. See generally http://www.cidcm.umd.edu/mar/.
Beyond simple stability, for some the goal of peacebuilding and state-building should go further, creating democratic sustainability. What does democratic sustainability entail and how can one think about this in relation to power-sharing? Democratic sustainability can be thought of as the right balance between deterrence and assurance. States need to be able to deter potential rebels in their midst. Unless a state is able to wield an effective monopoly over the use of violence, it will not be able to deter potential rebels. And, as Herbst has argued, African leaders face a peculiar set of problems related to patterns of demographic settlement, geography and ecology when considering the cost of expanding their domestic power infrastructure (Herbst 2000: 11–13). The central state has often been much more visible in and around the capital than in the hinterlands, as has been the case in, for example, the Democratic Republic of the Congo. This has facilitated the rise of armed rebellion, often supported politically and financially by neighbouring states.
But deterring rebellion is but one side of the democratic sustainability coin. The state also needs to assure groups in its midst that it will give a fair hearing to their grievances and that it will not use its military might abusively against them. Indeed, assurance is an essential part of a successful deterrence strategy. Decision-makers must thus tread carefully as they rebuild security institutions, for fear that these will be perceived as a threat and thus spur counter-reactions amongst those groups in society who hold grievances against the state. They must also work to provide other outlets for the expression and management of grievances and conflicts.
A further precondition for successful assurance and deterrence is the basic ability of the state to function. This harkens back to Weber's discussion of the bureaucratic state and of the legitimacy derived from the basic ability to administer. A dysfunctional state, unable to provide basic services to its population, supporters and opponents alike, is likely to lose all credibility in the eyes of its population. As discussed earlier in this article, credibility is a cornerstone of deterrence. It is also intimately tied to perceptions of state legitimacy as recent research by Susan Woodward has demonstrated (Woodward forthcoming 2010). Thus, the basic ability of states to function is equally important for the state's ability to both deter and assure.
Beyond functioning and service provision, the sustainability of the state may depend on a distinct element of legitimacy – the perception within a significant portion of the population that the state operates according to a clear set of rules and principles of fairness and transparency. This depends upon the presence of rule of law, whereby the rules of the road are set forth in a process which is previously defined, by individuals chosen according to an agreed process (Hart 1961). This need not be a democratic rule of law, although democracies most frequently operate according to such principles. Socio-legal scholars have found that citizens are far more prepared to accept government edicts and laws, even ones with which they disagree, if they conceive them to have been reached through a fair process (Tyler 2006). In societies emerging from conflict the real and perceived presence of the rule of law is now viewed as critical by the United Nations (United Nations 2004; Sriram, Martin-Ortega and Herman 2010). However, power-sharing arrangements, by inserting persons into various levels of governance by dint of their former coercive roles, may create institutional structures incapable of winning public confidence and securing longer-term legitimacy.
The delicate balance between deterrence and assurance is particularly difficult to achieve in those African situations where the state itself remains rather strong, but prone to discrimination and repression, as in Sudan and Angola. In Angola, and in spite of the Bicesse Accords which called for the merger of the MPLA (Movimento Popular de Libertação de Angola) and UNITA's forces into the FAA (Forças Armadas de Angola, the new Angolan armed forces), neither party was really interested in integration, hoping to ultimately prevail militarily. And when the two protagonists signed the Luena memorandum, following the death of Jonas Savimbi and the military defeat of UNITA at the hands of the MPLA, all that the rebel group could get were “promises that existing integration provisions would be implemented” (Brown and Zahar 2008: 80). The problem is not limited to those states that appear strong because of their military capabilities and/or their willingness to use force to quash dissent. It is also evident in other cases where strength is less a function of military might and more a function of one party's effective control of, or near monopoly over, sensitive state institutions. In Mozambique, which is commonly referred to as a success story, the state's deterrent capability remains a sensitive issue. Mozambican politicians and analysts agree that RENAMO's (Resistência Nacional Moçambicana) influence in the security sector, which was guaranteed in the Rome Agreement under provisions for security power-sharing, is being eroded by the ruling FRELIMO (Frente da Libertação de Moçambique) party (Brown and Zahar 2008: 81). While it is too early to speculate on more recent cases, there are signs that similar difficulties can be expected in locales such as the Sudan and Zimbabwe. Based on the experience of these two countries to date, there are equally concerns that, should the state renege on its commitments and attempt to use violence against its opponents, the international community's willingness to intervene may be extremely limited.
Deterrence without assurance may thus result in deepened authoritarianism of the sort likely to foster grievances and ultimately result in renewed violence. Likewise, assurances without a concomitant deterrent capability raise a different set of concerns about the sustainability of peace. Indeed, where the state's deterrent capability has not been rebuilt, the ability to maintain gains achieved in the political sphere depends in great part on the stability of the regional environment and/or the commitment of external actors to shield the country from destabilising influences. Such, for example, is the case in the Democratic Republic of the Congo (DRC) where the 2002 peace agreement has survived against all odds. While progress in the political sphere includes the holding of the first post-war elections in 2006, the DDR programme, while it achieved some of its objectives, did not result in the creation of an integrated Congolese army capable of standing on its own. Indeed, the exclusion of armed groups in the Ituri region and the inability to secure the demobilisation of Rwandan armed groups operating on DRC territory raise concerns about the ability of the country to weather armed challenges if and when the MONUC (Mission de l'Organisation des Nations Unies en République démocratique du Congo) withdraws. The experience of Bosnia and Herzegovina, where recent developments have raised the renewed spectre of violence, should the EUFOR (European Union Force) withdraw, should serve as a warning. Indeed, the fourteen years since the Dayton Peace Agreement have seen many improvements in the country but these were not sufficient to ward off the risk of destabilisation caused, in part, by the weakness of the central state's deterrent capability (Zahar 2009).
Finally, analysts and practitioners should also worry about the debilitating effects of power-sharing where the challenges of negotiation and implementation outlined earlier weaken further already fragile post-conflict states, leaving them without the ability to deter or the capacity and/or willingness to assure. Often functioning without appropriate resources and with little or no external protection, these fragile post-conflict countries run the risk of being buffeted by instability stemming from inside and outside their borders. Still in the throes of conflict and while no power-sharing agreement is under consideration to bring it out of war, Somalia stands as the best warning of the fate that such countries may encounter. Further away, in the Middle East, Lebanon stands as another omen of what can befall countries that fail to reassure and deter simultaneously; a country that had rebuilt itself over fifteen years before seeing most of its accomplishments wiped out in six weeks of regional war followed by four years of intense political crisis which brought the country to the brink of renewed civil war (Zahar 2009). But although Lebanon's geographic position bordering Syria and Israel has prompted Western countries to intervene and commit 15,000 troops to give new life to the United Nations Interim Force in Lebanon as well as financial and political support to revive the fledgling state institutions, credible international commitment to African countries threatened with the collapse of their power-sharing structures might be more difficult to summon.
Conclusions
This cursory review of the short- and longer-term challenges of power-sharing has implications for both theory and policy. It was not designed as a comprehensive review of all power-sharing arrangements, but as an exercise in inductive theory-development to underpin future research. Theoretically, it obliges researchers to ask new questions about the assumptions underlying attempts to rebuild war-torn states and societies. In terms of policy implications, there are specific areas of concern to which practitioners ought to pay greater attention.
First, the inclusion of power-sharing arrangements in peace deals is premised on a belief that the problems (and the solution to these problems) lie in the institutional framework of the state, that it is sufficient to build states properly to ward off renewed violence and (hopefully) entrench democratic practices. This does not sufficiently acknowledge two important caveats: that groups within and outside the state may have interests that are not addressed by the power-sharing institutions and/or have reasons to continue fearing threats that cannot be simply addressed by institutional tinkering. This theoretical conclusion dovetails nicely with recent research that has raised questions about the adequacy of power-sharing, indeed more generally about the enterprise of international involvement in the reconstruction of post-conflict states. It also goes a long way to explain one of our empirical findings: that much depends on the nature of the groups, understood not so much as their capacity to spoil, but on a deeper level that captures their social basis, their interests and their ability to effectively represent and pursue both.
Second, power-sharing illustrates a paradox in the process of post-conflict reconstruction as in many, if not the majority of, instances, the state is both identified as the problem and the solution. States are the main source of threat to groups in their midst. Indeed, quantitative research on state violence suggests that state violence is not only more prevalent but more lethal than the violence meted out by rebel groups (Armstrong and Davenport 2008). Can states be reconstructed in ways that limit their repressive capacities? And, if so, is power-sharing the way to go? Much, as we have argued above, depends on the nature, capacities, and interests of the state and of the groups which agreements seek to include in governance. In other words, the extent of the transformation required to ensure a sustainable transition from war to peace and from violent to non-violent conflict management will greatly vary depending on starting conditions, something that current approaches, which have been criticised for adopting a cookie-cutter formula, do not sufficiently acknowledge. Much also depends on the manner in which power-sharing arrangements, and the risks associated with them, end up affecting the basic ability of states to provide services. Beyond its ability to deter the use of violence in its midst and/or its capacity to assure citizens of its own restraint, a state is a set of institutions of governance that is supposed to provide, at a minimum, a basic framework for the rule of law, as well as for the provision of those basic services without which the most fundamental requirements of human security are at risk. However, to date, most discussions of the success or failure of power-sharing arrangements do not take such considerations into account. The few notable exceptions do not address power-sharing per se; rather they discuss service provision in the more general context of the success or failure of peace agreements (Woodward forthcoming 2010).
Third, power-sharing arrangements may be appealing options for mediators, but often fail to do what they purport to do, which is provide greater security for the people. Even where they do provide stability at the national level, they are unlikely to address local-level violence sufficiently (Mehler 2008). Others have also argued that real human security can only be achieved through democratic practices and that power-sharing thwarts deep democratisation (Roeder and Rothchild 2005). Others still have argued that post-conflict reconstruction requires both top-down and bottom-up approaches if peace is to be entrenched (Lederach 1995, 1997). In other words, the discursive arguments often used to justify foreign intervention in post-conflict reconstruction – the so-called drive to humanitarian intervention – stand in stark opposition to the short- to medium-term empirical results of the institutional policies favoured by outside interveners. While some have argued that these policies are driven by the international community's obsession with exit strategies, the fact remains that some learning has occurred on this front. The most obvious change has occurred in the electoral realm, with the earlier obsession with “speedy” post-conflict elections replaced by a growing realisation that such elections cannot be successfully held, nor can they achieve their intended objectives, if some minimal conditions are not met. And while it is too early to pass judgement, there are at least discursive changes in the international community's commitment to stay the course in post-conflict countries, especially in countries such as Afghanistan.