The Burundi Peace Negotiations: An African Experience of Peace–making

In August 2005, the former rebel leader, Pierre Nkurunziza, was installed as the president of a newly-elected Hutu majority government in the central African state of Burundi. This marked the culmination of almost nine years of a formal peace process that followed a trajectory of peace negotiations, peace and ceasefire agreements, a transitional government and democratic elections (Irinnews.org, 2002a & b; Reyntjens, 2005). Yet, it is an uneasy peace as ceasefire agreements were not reached with all warring factions and low-intensity violence and a culture of impunity pervades Burundi society.

Since independence from Belgian colonial rule in 1962, Burundi has been highly unstable with six governments between 1962 and 1966, the abolition of the monarchy (1966), four successful coup d’états (1965, 1976, 1987 & 1994), and the assassination of its first democratically-elected president, Melchoir Ndadaye, in October 1993. Like their neighbours in Rwanda, the people of Burundi have been subjected to episodes of genocidal violence; an estimated 200,000 people were killed in 1972 and a further 20,000 in August 1988, and since Ndadaye's assassination in 1993, warfare waged by the military and government-backed militias against rebel groups and their supporters has killed some 200,000 people and forced over 350,000 into exile (ICG, 1998; Arusha Peace and Reconciliation Agreement for Burundi, 2000).

The Burundi peace negotiations originated essentially from a regional thrust for peace and has been represented as a successful attempt by Africans at regional peace-making (Bentley & Southall, 2005; Mpangala et al. 2004). Peace negotiations started in 1996 and culminated in the signing of the Arusha Peace and Reconciliation Agreement in August 2000. Rebuffed by the rebels and signed reluctantly by the political parties, it became a peace agreement without a cessation of hostilities. Fighting continued even after the 2005 elections, as it took a further year before a ceasefire agreement was signed with the last remaining rebel faction.

Scholars have long challenged the appropriateness of the western-derived peace imposed on Africa, and more recently, have criticized the promotion of a ‘liberal peace’, structured by hegemonic neo-liberal political economic practices (Mafeje, 1995; Hansen, 1987; Richards, 2005; Willett, 2005). Using the Burundi peace negotiations as an example, this article examines the prevailing concept of peace that informs contemporary conflict resolution in Africa, especially the ways in which it has been structured by neo-liberalism and the relationship between peacemaking and protracted wars. Also of importance is how different interest groups (local, regional and international) shape the outcome and, in essence, define the peace appropriate to Africa, such that accords can be signed whilst fighting continues; in effect normalizing extreme forms of violence.

This article draws on research conducted since 1998 with Burundi politicians in Belgium, Tanzania and Burundi, and with members of the facilitation team from the Mwalimu Nyerere Foundation. Archival data from the Arusha negotiations were gathered at the Foundation and from newspapers, news websites and reports of international NGOs.

The article argues for peace negotiations to be seen, not primarily as attempts to end warfare and promote social justice, but as arenas of political struggles, beyond that envisaged between the belligerents, due to the prevalence of a multitude of supporting actors seeking to promote vested interests. The resulting peace agreement is not necessarily consensual or reflective of a compromise for the sake of peace; it marks, essentially, a temporary stalemate between the manoeuvrings of international, regional and local actors. The Burundi peace process represents struggles between different competing visions of peace; the neo-liberal one supported by western donors, and that of regional actors, seeking to establish a new mode of politics on the continent. Unable to incorporate the perspectives of civil society and without international political clout and financial resources, regional actors conceded to the imposition of a ‘liberal peace’, which, while promoting ethnic equity through power-sharing among the elites and democratic elections, leaves the extant social system in tact and is not conditional on the cessation of direct violence.

The Peace Problematic in Africa

War and peace are complex terms. Their meanings depend largely on how we conceptualise violence. War refers to the use of violence to settle political differences, in effect, the struggle for state power. Galtung (1969) distinguishes between ‘direct violence’ (fighting) and ‘structural violence’ (indirect forms of violence stemming from exposure to conditions of poverty and powerlessness). He argues that ‘negative’ and ‘positive’ peace result from the ending of direct and structural violence respectively, even though the latter is difficult to achieve, being reliant on efforts to end social injustice. Useful as it may be in articulating the brutality of poverty, this dichotomy has become less so in our quest to make sense of the persistence of direct violence after peace agreements have been signed and democratic elections have been held to resolve political differences. ‘Negative peace’ seems far from being achieved in many post-conflict countries in Africa. Recently, scholars of African warfare have questioned this dichotomy of war and peace. Richards (2005:4-5), for example, asks us to see war and peace as existing along a continuum, defining war as a ‘long-term struggle organised for political ends’. Thus, he contends we can understand why ‘peace can often be more violent and ‘dangerous’ than war’. However, history has shown that states can achieve ‘negative peace’ and can go some way to address social injustice. Why not contemporary Africa? Steans (1998:106-7) refers us to classic realist thought in which ‘hegemonic states dominate international institutions and by doing so ‘manage’ international security’, thus, ensuring that war and peace have different meanings in hegemonic and weak states.

During the Cold War, proxy wars fought on the continent of Africa helped to maintain the peace in Europe and North America. In post-colonial Africa, ‘peaceful’ states could include dictatorial, militaristic or one-party regimes. Indeed, the formula for a peaceful society in Africa seems to involve the tolerance of higher levels of direct and structural violence. The end of the Cold War and the rise of economic liberalism produced new explanations for warfare and new frameworks for peace. Conflict resolution models tend to employ the popular discourse that represents African wars as essentially primordial, often arising from age-old innate and irrational enmities and, even when the interpretations attempt to avoid ethnic reductionism, they draw on two prevailing academic discourses. The first links ethnicity with notions of greed or grievance harboured by conflicting parties (Berdal & Malone, 2000). And the second represents war in the post-Cold war period as ‘new’, non-ideological and fuelled by the avariciousness of tribal ‘warlords’, aiming to capture state resources for private gain (Kaldor, 2001). ‘New wars’ are presented as being of greater complexity and in need of multifaceted solutions. Such approaches interpret war and peace as confined within the boundaries of the nation-state, thus failing to consider the ways in which global geo-politics and market-based economic systems influence regional and local politics and exacerbate the conditions for war. Post-Cold War transformations in conflict resolution under neo-liberalism are well-documented (Duffield, 2001; Clapham, 1998; Richards, 2005; Tull & Mehler, 2005). These include an increased role for non-state actors in the negotiation process, a proliferation of mediators, and power-sharing, instead of winner-takes-all elections.

Power-sharing means dividing the institutions of governance between political parties and rebel movements, in the context of a new constitution and democratic elections. Theoretically, it is appealing as the lack of equity among groups is seen as a significant factor in African conflicts and it also reflects the sidelining of state sovereignty inherent under neo-liberalism (Bangura, 1994; Clapham, 1998; Tull & Mehler, 2005). Power-sharing can satisfy greed and grievance while still retaining ‘democratic principles and procedures’, making it ‘compatible with democracy while diminishing its most destabilising side effects’ (Spears (2000:105). Indeed, many political parties, especially the weaker ones, prefer to gain constitutionallyprotected political advantage through negotiations before committing themselves to the electorate (Bangura, 1994). Tull & Mehler (2005) note that power-sharing agreements, by enabling rebel leaders to gain state power, can lead to the reproduction of ‘insurgent violence’, as rebellion becomes the means to gain political leverage and international acceptance, thus making access to political power difficult for those who champion non-violence. Power-sharing can legitimise and normalise violence as part of the political discourse. Theoretically, multipartyism and democratic elections should lead to greater representation and accountability, but experience suggests that, in the context of neo-liberalism, they are easily manipulated by the elite, and can lead to xenophobia and sectarian violence.

The proliferation of non-state organisations and consultants as mediators has turned peace into a veritable industry. In addition to the UN, the African Union and western governments, numerous international, regional and local NGOs have been active in formulating and promoting peace initiatives. The nature of the peace problematic is such that the motivations of individual actors beg investigation, especially since the proliferation of mediators seems to protract warfare rather than lead to swift resolutions. As Clapham (1998:209) notes, with respect to the Rwandan war of the 1990s,

The goal of the ‘liberal peace’ is to enforce rapid modernisation of African societies. This is to be accomplished by ‘transform[ing] the dysfunctional and war-affected societies that it encounters on its borders into cooperative, representative and, especially, stable entities’, and involves ‘reconstructing social networks, strengthening civil and representative institutions, promoting the rule of law, and security sector reform in the context of a functioning market economy’ (Duffield 2000:11). Ultimately, the ‘liberal peace’ necessitates the entrenchment of neo-liberal political economic practices, which, hitherto, have undermined state sovereignty, focusing as they do on non-state actors; the market for economic development and personal liberation; civil society and/or external (‘non-territorial’) actors for humanitarian and welfare intervention, and multipartyism for diffusing political tensions and mediating competition.

Richards (2005) contends that ‘liberal peace’ takes wars out of their social contexts, and argues that instead of conflict resolution, one should focus on conflict transformation which would re-socialise war and re-direct ‘the social energies deployed in war to problem-solving ventures on a co-operative basis’ (Richard, 2005:18). As radical as it may seem, Richard's project is based on advocating a more participatory vision of peace which could fit well with neo-liberal conceptualisation of civil society. The importance of collective responsibility for long-term peace is indisputable. Peace must rest with the return of agency to the people affected. But the shortcomings of Richards’ analysis lie with his focus on peace as resting primarily within the capacities of local communities.

The ‘liberal peace’ envisages a process of ‘creative destruction’ commonly known as post-war reconstruction, which has become a major policy arena for development agencies, multinational corporations, non-governmental organisations (NGOs) and private military contractors; two key policy areas are demobilisation and reconciliation. However, without the space for alternative and progressive conceptualisations of peace, demilitarisation and reconstruction often result in a shift in the nature of direct violence, from military combat to low-intensity violence (human rights violations, including sexual violence) on a generalised scale (Amnesty International, 2004). Consequently, post-war reconstruction has had limited success in Africa because it tends to reinforce the social system within which violence and inequalities are embedded (Mafeje, 1995; Spears, 2002; Tull & Mehler, 2005). Rather than, as Duffield (2001:11) claims, ‘liberal peace reflect[ing] a radical development agenda of social transformation’, it constitutes, instead, a form of ‘negative peace’ that normalises extreme violence.

In sum, the liberalisation of peace-making has resulted in the proliferation of mediators, the inclusion of non-state actors, the sharing of power between belligerents, the privileging of violence as a mechanism for political inclusion, and what amounts to the de-politicisation of warfare (Clapham, 1998; Tull & Mehler, 2005; Richards, 2005). Thus, the dominant conflict resolution model, in spite of its liberalisation, upholds a non-transformative concept of peace.

The situation in Burundi epitomises the inconclusiveness of neo-liberal peacemaking. A detailed examination of the Burundi peace negotiations and subsequent events (1996-2005) enables us to identify, more specifically, the inherent limitations in neo-liberal conflict resolution. This is achieved by constructing a narrative of the peace process that examines the impetus for peace, the key issues that stalled the talks and the contributions of a multiplicity of actors: regional states, mediators, especially the leadership (Julius Nyerere and Nelson Mandela), the protagonists (Burundi government, political parties, rebel movements, and their negotiating positions) and western donors, multilateral organisations, peace consultants, international NGOs and civil society organisations, especially women in light of UN Security Council Resolution 1325. The actualities of peace-making are investigated through a focus on the key issues debated and the tenets of the peace accord. Implementation of the accord, post-Arusha, reveals the temporality as well as the contradictory outcomes of the ‘liberal peace’, that whilst appearing to open the political space for marginalised ethnic elites and women, for example, it eschews genuine public participation and institutionalizes violence as mode of political contestation.

The Burundi Peace Process: A Regional Thrust for Peace

The Burundi peace negotiations, spearheaded by Tanzania, arose from a regional initiative by neighbouring states concerned about the protracted nature of the conflict and its destabilising impact on the region. This initiative was given impetus through the actions of the US-based Carter Center which sponsored the first two preliminary meetings on Burundi and the region. It took four years from the beginning of the talks (April, 1996) before the peace agreement was reached (August, 2000) – aspects of which are still being implemented by the democratically-elected government that took office in August 2005.

Warfare in Burundi is commonly represented by the media and by the international community as an age-old ethnic conflict between the majority Hutu and minority Tutsi ethnic groups – the latter having dominated the state, almost uninterrupted, since independence from Belgian colonialism in 1962. Alternative explanations, while recognizing that violence manifests itself largely along ethnic lines, focus either on the prevalence of ethnic inequalities in a rentier and neo-patrimonial state or on the failure of the postcolonial state to transcend discriminatory colonial practices based on racial and social hierarchies (Ngaruko & Nkurunziza, 2000; Daley, 2006). The first two perspectives provide the underlying discourse of the peace negotiations and provided the foundations for the neo-liberal governance and economic reforms outlined in the peace agreement. However, the process of negotiations and the implementation of reforms reflect the tensions between neoliberalism and ethnic-based ideologies, and attempts to chart a new mode of politics in the region. These tensions highlight the competing perspectives on peace arising from the contradictions inherent in neo-liberalism, as its simultaneously opens and narrows the political space.

The anti-state critique at the foundation of neo-liberalism manifests itself in the promotion of a multiplicity of actors in peace-making. In the case of the Burundi peace negotiations, states retained key roles in the procedures. Financial support for the talks came from international donors (the UN, EU, US, Canada etc.), regional states (Uganda, Tanzania, Kenya, Zaire [DRC] and later South Africa) acted as overseers, supporting the mediatory roles of Julius Nyerere (April 1996-October 1999) and Nelson Mandela (1999-2000).¹ Regional states sought to define the nature of the peace process and to exert influence on the agreement. After regional summits in June and July 1996, they issued communiqués advocating a negotiated settlement to the crisis in Burundi and national reconciliation. They called for an arms embargo and the denial of visas to those opposed to peace. The Tutsi officers’ coup of 25 July 1996 that returned Pierre Buyoya, a former Tutsi President and coup-leader, to power was condemned and regional states called for the immediate restoration of the National Assembly and the lifting of the ban on political parties.

There were two mechanisms that regional states sought to use to exert influence over the peace process. The first was the proposal for the establishment of a regional peace-keeping force, which, although requested by the interim Burundian (Hutu) President, Sylvestre Ntibantunganya, and his Tutsi Prime Minister, Antoine Nduwayo, met with considerable opposition from the military and the Tutsidominated parties (Daily News, 19 January 1996, Bunting et al. 1999). Protests by members of the Tutsi political class and their supporters threw the country into the chaos that culminated in the July 1996 coup d’etat.

Opposition to the regional peace-keeping force, especially one which involved troops from neighbouring states, arose partly from intermittent proposals by a variety of actors for the incorporation of Burundi into a neighbouring state (Tanzania) in order to provide a geographic solution to the problem of ethnicity. It was mainly for this reason that many Burundians, Hutus and Tutsis, as well as external observers, questioned the motivations behind the involvement of the Tanzanian state in the peace negotiations and accused them of bias. In fact, Nyerere, who was close to the first elected leader of Burundi, Prince Louis Rwagasore (the Prince was part of the decolonisation movement), was opposed to coups, and was sympathetic to the plight of the Burundian Hutu refugees who sought asylum in Tanzania in large numbers after the 1972 genocide, and after 1993. Given the narrow political base of the military leadership of Burundi under Buyoya, any attempt to foster peace and negotiations was interpreted as support for the rebels. The convergence of neo-liberal peace and neo-liberal democracy undermined elements in Tanzanian society who used xenophobic attacks on the refugees to bolster their political ambitions.

The second mechanism by which regional states sought to exert pressure on the Burundi regime was regional sanctions. The economic embargo that was in place from October 1996 to January 1999 was not effective, in the sense that it was difficult to seal road, rail and air links to land-locked Burundi. Consequently, there was widespread violations and profiteering by regional elites (The East African, 11 March 1998). Furthermore, the lack of support from western governments meant that goods and weapons continued to be transported via daily flights to and from Belgium, France and other EU countries. Despite its ineffectiveness, opposition to the sanctions were such that regional states were forced to ease them twice to enable the importation of essential goods, including fuel, medicine, foodstuffs, educational and agricultural goods.

The Burundi military regime launched an international campaign against the sanctions, arguing that it was the poorest of the poor who suffered most from sanctions. The government's ‘diplomacy’, which was according to Bunting et al. (1999: not numbered), ‘dynamic, aggressive and effective’, had the support of the international community. The World Bank, the UN, Western countries and international NGOs, such as the International Crisis Group and Action Aid, were very critical of the regional initiative (Irinnew.org, 16 December 1998; ICG, 1998; Mthembu Slater, 1998). The French ambassador to Tanzania, Jacques Migozzi, was reported to have said ‘the blockade does not resemble those placed against Iraq and Libya. This was decided by heads of state in the region. They lack global status’ (Baraka & Mfinange, 1998). France's objection to the embargo seems to rest on the fact that it was a purely African initiative, but also sanctions undermined neo-liberal free market ideology and affected French and Belgian businesses.

International donors sought to punish the peace-makers by withdrawing their financial support. The European Union, for example, called for an audit of monies spent on the talks and reduced their financial support Tanzania. Nyerere criticised the neo-liberal approach to peace, when he argues:

However, it was the influence of South Africa, through its mediator, Nelson Mandela, that forced the peace agreement. After Nyerere's death, the Burundi government, along with international peace advisers and the US, supported a South African mediation headed by Mandela (Irinnews.org, 19 October 1999). An African mediator, with limited knowledge of Burundi politics, was also preferable to the military wing of the Burundi's political elite, which argued that ‘a South African mediation would correct ‘a number of weaknesses observed in the methodology and management’ (Irinnews.org, 19 October 1999).

Coming from a regional superpower, Mandela was considered to have brought ‘a lot of clout’ to the negotiations, as South Africa ‘had the military capacity to support any agreement’ (interview with member of facilitation team, 2005). One manifestation of the clout that Mandela brought was that numerous presidents, including the then US President Clinton, went to the signing. However, the limited regional knowledge of the South African Ministry of Foreign Affairs enabled the Burundi military elite to turn the negotiations to their advantage. The South Africans made concessions to the Buyoya regime without consideration of their full consequences.

The Actors in the Drama of Liberal Peace

The peace industry in Africa has set up an array of donors and peace consultants. Those with vested interests in the region, such as the former colonial power, Belgium, and francophone countries– France, Switzerland & Canada; major donors such as the US, the European Union and the UK sent special envoys or supplied facilitators to play leading roles on committees. Non-governmental organisations, such as the London-based International Alert, Washington-based Search for a Common Ground, Rome-based Comunità di Sant’Egidio, and South Africa's ACCORD, became active participants in the peace process.

While the western media criticised the regional states, this same media directed little criticism to the activities of western observers – as representatives of donor countries and multilateral blocks – who attempted to steer the negotiations to suit their interests. According to the reports by the Tanzanian Facilitation team, the special envoys from outside the region ‘displayed a tendency to want to dominate and control the process’, holding alternative ‘secret’ peace talks concurrently in Europe (Burundi Peace Negotiations, 1998). These were meant to undermine the Tanzanian initiative, but their impacts were greater in Burundi, where they led to an intensification of the violence, as those members of the political elite who were not invited to the talks felt marginalised and were worried about exclusion from any peace pact. These alternative talks also conveyed international legitimacy on the Burundi regime and the president's refusal to accept regional mediation. As well as institutions, individual international ‘consultants’ also tried to influence the peace process; some provided advice to the belligerents and colluded in attempts to shift the talks away from Tanzania's control.

Political Parties, Ethnicity & Governance Reform

In line with neo-liberal governance reforms, all political parties were invited to the negotiation table – those that fought the 1993 elections and those formed after the elections. Giving them equality of status led to a proliferation of parties and may well have contributed to factionalism within the rebel movements. Many parties lacked a recognised constituency and had not tested their legitimacy with the disenfranchised Burundi population. A similar situation existed in the Rwandan peace talks of the early 1990s where, according to Clapham (1998:205), any protagonist ‘who could muster evident support … had to be admitted … on terms of broad equality with existing regimes’ and were given a status ‘that only very inadequately reflected their popular support or military strength’.

In all, there were 84 delegates from the Burundi political parties, the government and National Assembly. The key political parties to the negotiations were Front Pour la Démocratie au Burundi (FRODEBU), which won the democratic elections of June 1993, whose President was assassinated, and whose dominance of a temporary coalition was overthrown by the July 1996 coup; Union Pour le Progrès National (UPRONA), the ruling party, aligned to the coup leaders, and Conseil National Pour le Défense de la Démocratie (CNDD), a breakaway party from FRODEBU. Representatives of the military participated as part of UPRONA or the government delegation and as consultants to one of the committees. Those rebels, PALIPEHUTU-FNL (Parti Pour la Libération Du Peuple Hutu /Forces National de Libération) and CNDD-FDD (Conseil National Pour le Défense de la Démocratie / Forces pour le Défense de la Démocratie), who continued to fight, were not recognised by Nyerere, who refused to acknowledge leaders who ‘gained power by force’ (African Confidential, 21-22 October 1999). They were, however, invited by Mandela to join the negotiations.

Ethnic ideology was a determining factor in the alignment of the political parties at the negotiations. They organised themselves into two broad interest groups: Group of 7, comprising Hutu-dominated parties, and Group of 10, made up of Tutsidominated parties with the addition of the Government of Burundi and the National Assembly. By enabling participants to articulate a common position on key issues, these groupings may have increased the effectiveness of the debates; nevertheless, they reinforced ethnicity as the basis of political discourse and obscured the fact that not all parties were ethnically exclusive in their membership.

The degree of ethnic allegiance varied from the integrated FRODEBU and UPRONA to the Tutsi extremist and ironically named, Parti Socialiste et Pan-Africaniste (INKINZO) and the purely Hutu, Parti Pour la Libération Du Peuple Hutu (PALIPEHUTU). Individual parties articulated moderate to extremist ethnic ideologies. And within the broad groupings, there were considerable internal divisions. Disagreements over approach created leadership crises and breakaway movements. The partnership of the Government of Burundi, National Assembly and UPRONA (G3) were often at variance with the rest of the G10. These differences among the Tutsis often reflect divisions, among the political, military and entrepreneurial elites from the province of Bururi and between the Bururi elite and those from Muramyva and other regions. Amongst the Hutus divisions were manifested mainly over the issue of whether to negotiate with the Buyoya government and the power sharing arrangements, leading to the creation of internal and external wings of the various Hutu-dominated political parties. Radical Hutus were often critical of those with allegiance to UPRONA, and some rebel movements sought negotiations not with political parties but with the military – considered to be the real power in the regime.

In the discussions at the peace negotiations, there was widespread agreement that the cause of warfare was not ethnicity. Protoco1, Article 4 states; ‘the parties recognise that the conflict is fundamentally political, with extremely important ethnic dimensions; It stems from a struggle by the political class to accede to and/or remain in power’ (Arusha Peace and Reconciliation Agreement for Burundi, 2000:16).

Yet, the realm of politics was confined to political parties that were perceived as being defined largely by ethnicity. This may be because that, apart from those expressing extreme ethnic ideologies, there were very little differences between the parties. Few had a conceptualisation of democracy that extended beyond the demand for equality between Hutus and Tutsis. Democracy simply amounted to equal guaranteed access to state institutions and ethnic quotas in the judiciary and army. Participating in the negotiations and signing the agreement secured parties a stake in government, which they would not necessarily have achieved through democratic elections.

Civil Society Representation

Current neo-liberal governance reforms argue for the promotion of civil society actors. In peace-making, the proliferation of non-state actors implies a greater role for civil society participation. In practice, however, peace-making continues to deny representation and thus political agency to local civil society groups campaigning for peace. At Arusha, two reasons were put forward against their participation. First, a majority among the Burundian negotiators were reluctant to have civil society groups as independent members. They cited the rules of procedure agreed in 1998, that any new group would need unanimous support in order to be admitted. This, apparently, was to prevent the Burundi military class bringing in carefully selected civil society representatives. The second reason was that Burundi's civil society organisations, many dominated by Tutsis, were not sufficiently independent from the state and representative of ‘all groups’. The peace accord notes, ‘the notion of civil society [in Burundi] is in fact a new one and is not well understood by the population, just as civil society itself does not understand its own mission’ (Arusha Peace & Reconciliation Agreement for Burundi, Report of Committee IV, Reconstruction and Development: 125, Para 2.5.6.1). This seems a coded reference to the lack of ethnic inclusiveness in many civil society organisations, and the inevitable dominance of these organisations by the educated Tutsi elite. In the end, civil society representatives were requested to participate as part of the delegation of the political parties they were affiliated to. However, their exclusion as independent participants reinforced the idea that peace making is solely the prerogative of political parties and rebel movements, not the collective responsibility of the people.

The role of Burundi women in intervening in the peace process has been lauded by numerous academics and NGOs (Burke et al. 2001). However, a detailed examination of their involvement proves instructive. Women for Peace, a Burundi organisation, was established in 1993, well before the peace negotiations began (Burke et al. 2001). However, the women's campaign only gained momentum when it was coordinated under an umbrella organisation, CAFOB (Collectif des Associations et ONGs Féminines du Burundi), and attracted international support, although its effectiveness depended on its co-operation with women's organisations in the region. In 1998, a delegation of women from Uganda, Rwanda and Tanzania visited Buyoya, Museveni and Nyerere, seeking explanations for the exclusion of women's organisations from the talks; this intervention led to concessions by the parties. Three women were appointed to the negotiations by the government and three by FRODEBU. Representatives of women's organisations were allowed to attend as observers in October and December 1998 (Arusha III), but their status and credentials were questioned; namely, were they representing women or existing political parties. Reportedly, the women's viewpoints did not differ from that of other party delegates and some were wives of colonels. The inclusion of women, even as tokens was beneficial, in that, as a group, women are probably the most victimised in war time and patriarchal practices might be difficult to defend in their presence.

The Burundi's women's movement also benefited from a new policy framework, supported by Security Council Resolution 1325 (2000), which stresses the importance of women's ‘equal participation and full involvement in all efforts for the maintenance and promotion of peace and security’ (para.5). With the support of UNIFEM and the Mwalimu Nyerere Foundation, an All-Party Burundi's Women's Peace Conference was held in Arusha from 17 to 20 July 2000, just over a month before the signing of the peace agreement. The fifty Burundi women who participated came from all social groups, including the diaspora and refugee camps. The conference drew up proposals to engender the draft peace agreement outlining how specific gender-related clauses could be incorporated (Burke et al. 2001). Virtually all of these gender-specific changes were later accepted by the parties to the agreement because, as, according to one member of the facilitation team, ‘at that stage they did not matter’.

The Negotiations: Schedules & Outcomes

The details of the negotiations have been drawn out to show that peace is an industry in the West and how the consultants, international NGOs and peace industry experts sought to manipulate the process for their own purposes. Experts on Burundi are familiar with these processes, but these details are repeated here in order to drive home the point of how western peace concepts did not validate the lives of Africans. In 1996, the US created the Africa Crisis Response Initiative in order to stop genocide in Burundi, but it was the same government of the US that supported the military coup of Buyoya when he came to power in July 1996. Buyoya, who had just returned from an internship in the US, had been given money to establish a ‘peace foundation’.

The first All Party Peace Talks (Arusha I) took place on 21 June 1998; all parties except those fighting attended, including the special envoys from the EU, US, Canada, OAU, Switzerland, Communita di Sant’Egidio (a Rome-based Catholic NGO) & the UN. The multinational team reflected the interests of the international community and, as such, the talks were structured ‘within the broad contextual framework of the international community’ (Butiku, 2004). Five broad areas of concern were identified: the nature of the conflict, democracy and good governance, peace and security, reconstruction and development and guarantees of implementation of the agreement and five committees were established to negotiate the issues and reach an agreement on each of them.

At the second All Party Peace Talks on 20 July 1998 (Arusha II), each committee drew up its agenda and worked independently of the others with regular consultations with the Facilitation Team (Bunting et al. 1999). Competing interests dogged the selection of committee chairs and vice-chairs; western donors, international NGOs and the Burundi government sought to influence the appointments. The Facilitation team was concerned that some committee chairs and vice-chairs, who originated from outside of the continent, did not fully support the peace process and were obstructive, often failing to perform their duties (Burundi Peace Negotiations, 1998).

Four more peace talks and numerous intercessional consultations took place before a draft agreement was submitted to all parties in March 2000. In April, Mandela made a statement saying that the draft agreement would not be changed. However, external pressure forced him to make changes, after which, according to one observer, ‘the draft agreement started to unravel’ (interview, member of facilitation team, 2001).

The first change was to Protocol I following a request from Louis Michel, the Belgium Prime Minister and Minister of Foreign Affairs, who, on a visit to Arusha in July 2000, asked Mandela for two paragraphs implicating the Belgian Colonial state in the murder of the democratically-elected Prime Minister, Prince Louis Rwagasore in 1961, to be removed.² The circumstances surrounding the death of Rwagasore were one of the few points on which all the Burundi parties agreed. Mandela went against the wishes of the Burundians and proceeded to remove the paragraphs, unilaterally, with the justification that Burundi will need aid from Belgium. This act may have undermined the legitimacy of the agreement and led to parties not keeping their commitments. After this, various groups were able to put pressure on the facilitator for changes to the draft document.

In its criticisms, the Burundi government said the draft agreement was not practical – it was ‘replete with confusion, ambiguity and double standard’ (Irinnews.org, 4 August 2000). For the government and the Tutsi elite, reform of Tutsi-dominated state institutions, such as the military and the gendarmerie, were the most contentious. Retaining the status quo was seen by many Tutsis as the guarantor of their security. Hutus, on the other hand, saw their repression as being linked to the mono-ethnic security services and demanded the ‘restructuring and retraining of the army’ (FRODEBU, 6 ^th plenary, 19 October 1998). The armed Hutu groups, excluded initially from the talks, made themselves significant players in less than one year. From June 1999 to January 2000 they became a force to be reckoned with if peace was to be achieved. In late 1999, CNDD-FDD conducted a scorched earth policy in the southern and eastern Burundi provinces of Bururi, Makamba and Ruyigi. Human Rights Watch (2000) reports that FDD insurgents not only killed or kidnapped civilians, but looted and burnt houses, schools, health centres, communal offices, in Kinyinya, Gisuru and Nyabitare. In response, the government intensified its repressive tactics using the military and the gendarmerie, while Tutsi militias attacked civilians, as a means of expressing their dissatisfaction with the negotiations.

The Burundi regime sought to by-pass the Arusha committees by negotiating directly with Mandela in South Africa. Buyoya proceeded to manipulate the process to obtain a result that disproportionately favoured the Tutsis and secured the continuation of his presidency during the first interim period of the transitional government, when key issues relating to the ceasefire negotiations and governance would be decided (Irinnews.org, 4 August 2000; ICG, 2000).

In June 2000, Mandela held closed door talks with the rebels and representatives of the Burundi government in South Africa. While the rebels sought to renegotiate the whole agreement, the Burundi military elite made a deal with the South Africans regarding the ethnic balance in the transitional government and the army. What became known as ‘the Pretoria compromise’ was supposed to ensure the security of the Tutsi. Buyoya emerged with assurances that Tutsis would have 50 per cent of the Senate, 60 per cent of the government and political parties with five per cent or more of the vote would gain a place in government. Tutsis would also make up 50 per cent of the army. The government may have hoped to exclude Hutu rebels, because as soon as the agreement was reached, the military started to recruit Hutus into the lower ranks.

The Arusha Peace and Reconciliation Agreement for Burundi was signed on 28 August 2000 amidst disagreements. The pressure to reach an agreement was so intense that articles relating to the establishment of the Senate and the composition of the armed forces were added to the draft agreement on the eve of signing, after last minute concessions were made by FRODEBU, without the approval of the other G7 parties. Agreement was not reached on several issues: leadership of the Transitional Government, arrangements for a ceasefire, the composition of the armed forces and the treatment of political prisoners.

A scaled down version of the agreement was signed by Buyoya and by the G7 (Hutu parties). The G10 (Tutsi parties) claimed they signed ‘under extreme international pressure and with reservations’. The rebel movements, CNDD-FDD and PALIPEHUTU-FNL, refused to sign. Consequently, it was a peace agreement without a ceasefire.

The main tenets of the peace agreement focused on the establishment of a three-year transitional government; with the transitional period divided into two 18 months intervals headed alternatively by a Tutsi and a Hutu; legislative power was to be exercised by a National Assembly of at least 100 members and a Senate comprising of two delegates from each province (one Tutsi one Hutu). During the transitional period a new constitution was to be produced and approved by the Senate and National Assembly, and an independent electoral commission established to organize elections. Refugees were to be repatriated and reintegrated into their former communities and the monitoring and guaranteeing of the process would be the job of an Implementation Monitoring Committee (IMC), supported by the presence of an international peace-keeping force.

Arguably, the peace agreement followed a standard formula aimed at instituting a system for power-sharing between political parties (rebel movements would register as parties) and ensuring the presence of ethnic and gender plurality within the political structures. Power-sharing is aimed at rectifying ethnic inequalities without the displacement of those already in power, since, the proposed size and membership of the National Assembly and the Senate allowed for the incorporation of past presidents and representative of most factions. Furthermore, the stipulation of the electoral commission that parties must be multi-ethnic forced the main political parties to recruit members from other ethnic groups. Ethnic competition is not anathema to neo-liberalism, which promotes group and individual rights in the context of the free market.

Post-Arusha Developments

The incompleteness of the Peace Agreement in 2000 (which was witnessed by so many luminaries) resulted in its slow implementation. It took a year before the Transitional Government was constituted and five years from the signing of the Agreement before a democratically-elected government was in situ. In between there were two attempted coups – in April and July 2001. Some elements of the Burundi ruling class with an ethnic base used their political resources to delay implementation. Once the political leaders understood the machinations of western donors, they sought to rearrange their politics to suit.

Faced with numerous petitions for and against candidates for leadership regional states decided that Buyoya should be Transitional President for the first eighteen month period with Dometien Ndayizeye (G7 & FRODEBU) as Vice-President (Chhatbar, 2001). They threatened further sanctions if Buyoya failed to abide by the conditions; even then, he later sought to change the clause in the Constitution that prevented him from running as President. The transitional government sworn in on 1 November 2001 had 26 Cabinet posts, of which 14 went to Hutus; Tutsis retained the key ministries of defence, foreign affairs and finance. Four women party members were made ministers.

In the introduction, a distinction was made between the instant peace accord of neoliberal peace and the long drawn out concept of African peace. The contradiction between these views became manifest in the establishment of the regional peacekeeping force. By this time the AU had taken a firmer position against genocide in Africa. Despite considerable Tutsi opposition to the force, AMIB (African Union Mission to Burundi) started on 18 October 2001 and was made up initially of 701 South African troops. Following UN Security Council Resolution 1554, AMIB was converted to ONUB (United Nations Mission in Burundi) in June 2004 and increased to 5,650, incorporating troops from Ghana, Nigeria and Senegal; all countries external to the region. ONUB's presence helped to maintain a semblance of stability during the implementation of the accord, but did not stop the fighting and human rights violations by all sides.

Ceasefire negotiations between the Transitional Government and the Hutu rebel movements proceeded, with the Facilitation team headed initially by the then deputy president of South Africa. The commitment of African leaders meant that what was a long and tortuous process ended with binding agreements. The lengthy process was due to a) the intervention of too many mediators (holding meetings in Rome, Libreville, South Africa & Dar es Salaam); b) factionalism in the rebel movements; CNDD-FDD had split into two factions headed by Jean-Bosco Ndayikengurukiye and Pierre Nkurunziza, so did PALIPEHUTU-FNL headed by Alain Mugabarabona & Agathon Rwasa) and c) the intensification of the war – both the government and rebels used the post-accord period to rearm. Even FRODEBU, as late as November 2002, was reported to have established a rebel group, Ramico-Pax (Tindwa, 2002).

The smallest factions of the two rebel movements (Ndayikengurukiye's CNDD-FDD & Mugabarabona's PALIPEHUTU-FNL) were the first to reach a ceasefire agreement with the government on 7 October 2002. A year later, on 16 November 2003, Nkurunziza's CNDD-FDD and the government agreed to the Pretoria Protocol on Political, Defence and Security Power Sharing in Burundi, which gave CNDD-FDD 40 per cent of the integrated general staff and the officer corps in the army. In the new army, Burundi National Defence Force, military command posts would be shared 50/50 on an ethnic basis. CNDD-FDD also signed a ceasefire agreement on 3 December 2002 in which it would become a party and be allocated two ministerial posts. Nkurunziza was made Minister for Good Governance. The largest branch of PALIPEHUTU-FNL, headed by Agathon Rwasa, joined the ceasefire talks in April 2005 after regional heads of state declared the FNL to be a ‘terrorist organisation’. FNL increasingly relied on terror tactics against civilians to strengthen its negotiating position. For example, in August 2004, FNL massacred 152 Congolese refugees at Gatumba camp for no discernible reasons (HRW, 2004).

After a turbulent drafting phase, on 28 February 2005, a referendum on the new Constitution showed overwhelming support from 91.2 per cent of voters. UPRONA's central committee and three other Tutsi parties had urged a ‘no’ vote and refused to sign a code of good conduct during the elections.

Thirty-four political parties fought the communal and parliamentary elections that were finally held from June to August 2005. CNDD-FDD was confirmed winner of municipal communal (senatorial) polls held on 3 June; of the 3,225 seats, CNDDFDD won 1,781, followed by FRODEBU with 822, UPRONA came third with 260 seats and Movement for the Rehabilitation of Citizens (MRC) gained 88 seats (Irinnews.org, 16-24 June 2005). This pattern was repeated in the elections for the National Assembly, held on 4 July 2005: CNDD-FDD gained 58.55 per cent of the votes, FRODEBU 21.69, UPRONA 10 per cent, CNDD four per cent and MRC two per cent. At the end of August 2005, Pierre Nkurunziza, as the head of the largest elected party, was chosen as President by the National Assembly and Senate. Tutsis were co-opted onto the National Assembly and the Senate in order to reflect the 60/40 ethnic quota of the Peace Agreement and the Constitution. Reyntjens (2005:130) points to the ‘unconstitutional’ composition of the Cabinet, having representation from parties that did not gain five per cent or more of the vote: MSP-Inkinzo and PARENA have one seat each, plus two ministers are without party affiliations. Both UPRONA and FRODEBU are under-represented and CNDD-FDD have proportionately more seats than they hold in the National Assembly. Women obtained 35 per cent of the Cabinet seats, achieving greater representation than envisaged in the peace agreement, and than acquired in the local elections, where they gained 5.4 per cent of the communal presidency and 33.3 per cent of the vice-presidency (ONUB, 2006b).

CNDD-FDD's victory was no surprise for those in the region, and for external observers. The party's growing popularity after February 2005 meant that Nkurunziza was invited to visit several western countries. As soon as the Arusha agreement was signed, CNDD-FDD presented itself as the legitimate opposition to UPRONA, wooing, challenging and even kidnapping FRODEBU members of the government. It is difficult to know how CNDD-FDD garnered popular support; reports that voters were intimidated, especially in the areas they controlled, are balanced by those that point to admiration for their long struggle (Ligue ITEKA, 2005). Many voters may have opted for the party that was more likely to bring an end to the warfare. However, violence, ranging from massacres, summary executions, scorched-earth policy, bombing and rape, was used by all factions to articulate political differences and to influence the accord.

As noted, a neo-liberal interpretation of the Burundi conflict explains the multiplicity of political parties as a reflection of competing entrepreneurial cliques, both among the military and political elite (Nkurunziza & Ngaruko, 2002). Burundi's heavily aid dependent and indebted economy has been liberalised since 1988. The impact of the war meant that structural adjustment was suspended between 1993 and 2004. A ‘greed and grievance’ argument supports intense competition among ethnic elites who, primarily, rely on the state for accumulation in the context of a limited resource base and diminishing rents. Ngaruko & Nkurunziza (2000:384-385) claim that ‘state predation by power holders who share its rents has led to rebellions by those excluded, triggering, in turn, repression by the army, whose primary role has appeared to be the defence of the system of predation’. While this is a plausible explanation, their argument that neo-liberal economic reforms would reduce elite hold on state resources is not proven in Burundi, where the assassination of the democratically-elected president was considered by some to have been linked to his attempts at reforming the activities of companies operating in the free trade zone.

Furthermore, although current economic liberalisation programmes may claim to have an anti-poverty focus, the economic recovery strategies proposed in the peace accord have led to greater hardship at a time when people are attempting to rebuild household and community institutions.³ The privatisation of the state marketing boards and the cutbacks in state expenditure has had negative implications for the employment and the provision of social welfare policies, without displacing the power of ethnic elites. In 2002, at a time of acute demand in health and other social services, the government introduced policies of cost-recovery in the health centres. Consequently, people have been detained in hospitals because of non-payment of hospital fees. HRW (2006) reports that 1,076 people were detained in 2005 of whom, 621 were held in the principal hospital. Since 2000, public sector workers, such as medical staff, teachers and the judiciary, have taken strike action for increased wage rates and the payment of overdue salaries (Irinnews.org, 7 March 2005).

The Disarmament, Demobilisation and Reintegration (DDR) programme, which started in December 2004, was administered by the World Bank as part of their Central African Multi-Country Demobilisation and Disarmament Program (MDRP) and implemented by NGOs with the assistance of the National Ceasefire Commission. At an estimated cost of US $84.4 million, the DDR programme aimed to demobilize the estimated 70,000 ex-combatants by giving them direct monetary payments. Such payments varied according to rank, ranging from US$500 for ordinary ranks to $3,000 for the generals. The prospects of de-mobilisation with remuneration provided by the international community encouraged many young Burundian men to join the rebel movements and the army during the implementation of the peace agreement (Human Rights Watch, 2003). However, DDR was not a smooth process. Former rebels were reluctant to move into cantonments, and demobilisation payments were not always forthcoming, causing protests from combatants and state-sponsored armed groups (Irinnews.org, 24 June 2005). By September 2005, only 16,634 were de-mobilised and the new army was to be limited to between 28,000 and 29,000. Many combatants were de-mobilised without being disarmed.

The issue of impunity was on the agenda when it affected the ability of politicians to return and participate in the elections. Buyoya was reluctant to set up the provisions allowing for the return of political exiles to participate as members of the transitional government, despite the request from the IMC (Irinnews.org, 7 June 2001). A temporary immunity law was eventually passed covering ‘crimes with a political aim committed after 1 July 1962 to the date of promulgation (27 August 2003)’, amidst considerable opposition from human rights groups, Tutsi opposition parties and extremist organisations (Irinnews.org, 3 September 2003).

With regard to the Truth and Reconciliation Commission (Protocol I:6), the political leaders were keen to protect themselves from any future criminal charges. On 1 September 2004, the National Assembly passed a law allowing the UN to set up a non-Judicial Truth Commission, with a substantial international component and the creation of a special chamber within the Burundi justice system to tackle crimes against humanity (Irinnews.org, 29 March 2005). It is unlikely that these bodies will receive meaningful support from the government, political parties or regional leaders. As one key participant claimed: ‘truth and reconciliation is no longer considered to be necessary … realizing that wrong things were done in past, but the future does not involve exclusion’ (interview with member of facilitation team, 2005). This viewpoint is in sharp contrast to that presented in post-election reports by human rights organisations citing the continuation of summary execution and torture by Burundian soldiers, intelligence agents and FNL rebels; rape, thefts and murders by criminal elements, especially ex-paramilitary armed gangs (Human Rights Watch, 2005; Ligue ITEKA, 2005).

Conclusion

A close examination of the procedures, actors and their positions during the Burundi peace negotiations reveals neo-liberal peace-making as a contested terrain, in which long-term cessation of hostilities may not be top of the agenda. Thus, the commitment of many of the actors to the ending of direct violence appears questionable. The victory of CNDD-FDD reinforces the notion that the political kingdom can be won through violence. On the part of the belligerents, fighting intensified during and especially after the peace agreement. The Buyoya government used violence against civilians and, with the help of international actors, attempted to control the negotiations until the outcome was beneficial to the Tutsis and to the President. CNDD-FDD, the stronger of the rebel movements, gained the upper hand by perpetrating warfare in order to attain an agreement that benefited it. FRODEBU was perceived to have made too many compromises and lost credibility.

Many parties signed the agreement out of political and economic opportunism nurtured by external actors. Power-sharing arrangements seemed to legitimize opportunistic behaviour. The immediate financial gains were from donor funds allocated to the various post-conflict reconstruction programmes. While peace negotiations may provide an ideal opportunity for tackling social injustice, in the case of Burundi, inequalities were interpreted as the problem of access by ethnic elites to state institutions, which could be resolved largely through power-sharing, despite the awareness of some participants of the endemic structural conditions that reproduce inequalities and violence.

Although it was right not to treat women as an essentialist category, the absence of a gendered understanding of the state and of the nature of violence has contributed to the proliferation of sexual violence even in areas considered peaceful and ‘safe’ (ONUB, 2006a; Ligue ITEKA, 2005). Moreover, the narrow reading of war as fighting between the state and rebel movements has meant that Hutu and Tutsi militias, criminal gangs and others who use the state of war to perpetuate violence have been able to continue to wreak havoc in many communities.

Finally, the regional initiative marked a significant and progressive step to the ending of genocidal violence and the abuse of human rights in Africa, yet it did not have the full backing of international forces for peace. From the perspective of the region, two lessons can be learnt from the Burundi peace negotiations. First, the potential exists for a major commitment by regional leaders to bring an end to protracted conflicts in Africa, especially those that are genocidal and can have regionally destabilising effects. Second, that with collective leadership, African leaders have the capacity to promote a new mode of politics, one that says no to coups and the overthrow of democratic regimes, and that this capacity can be mobilised and supported largely by African professionals. However, the asymmetrical power relations between western donors and NGOs and regional and local actors, and their multiple and often competing interests resulted in the Arusha Agreement reflecting not just the minimal conditions for peace, but one which institutionalises violence as part of the political discourse. Creating such spaces of peace requires a radical transformation of the state, much more than that envisioned in contemporary approaches to power-sharing and reconstruction programmes. It is not just sufficient to ensure that peace remains on the agenda of regional and international organisations, a perspective on peace must be articulated, one which is long-term in its outlook, opposed to all forms of violence, is inclusive and gives agency to a multiple-voiced political community.