Abstract
For the past twenty years, Burundi has experimented with power-sharing as an instrument of political liberalisation, democratisation and conflict resolution. This contribution analyses the different meanings the concept of power-sharing has had throughout Burundi's recent and extremely violent political transition, in particular during the lengthy peace process. It shows how national and international actors have found inspiration in the toolbox of consociationalism to negotiate and design the Arusha Peace and Reconciliation Agreement for Burundi signed in August 2000 and its post-transition Constitution. Power-sharing has been instrumental in achieving the – short-term – objective of war termination. It has also de-ethnicised political competition and reduced the (potentially) destabilising effect of elections. Measured against more ambitious state-building objectives (democracy, rule of law, accountable and effective governance), power-sharing has (so far) not been able to make a difference. Several factors and developments threaten the “survival” of the power-sharing model in Burundi.
With an estimated population of around 8 million inhabitants – composed of roughly 85% Hutu, 14% Tutsi, 1% Twa and some Ganwa – Burundi is a densely populated, land-locked and poor country that gained independence in July 1962, after several decades of German and Belgian occupation. In April 2009, Burundi's last armed rebel movement was registered as a political party and laid down its arms. This was nearly thirty years after the creation, in 1980, of the Palipehutu (Parti pour la libération du peuple hutu) movement among Hutu refugees who had left Burundi after the large-scale massacres (sometimes classified as genocide) of 1972 and lived in exile in Tanzania. Also, this was more than fifteen years after the collapse of a political transition which, for not more than five months in 1993, was portrayed as a model to the rest of Africa. In June 1993, multi-party general elections were won by the predominantly Hutu Frodebu (Front pour la démocratie au Burundi) party, with Melchior Ndadaye defeating and replacing the incumbent President Pierre Buyoya (of the predominantly Tutsi Uprona, Parti de l'Unité pour le Progrès national). In October 1993, President Ndadaye and several other dignitaries were assassinated during a (failed) coup attempt, leaving Burundi in a political and institutional imbroglio and setting the scene for the massacres of tens of thousands of Tutsi and Hutu civilians and a civil war from July 1994 onwards.
With local, parliamentary and presidential elections scheduled for 2010, Burundi faces the next crucial step in its complex transition process. It remains highly uncertain whether the 2010 elections will mark a next step towards democratisation or towards an increasingly (electoral) authoritarian political regime. Neither scenario can be ruled out (Vandeginste 2009). Today, Burundi is more peaceful than during the past fifteen years. This is not the outcome of a military victory by one of the parties to the armed conflict. It is the result of a negotiated settlement with, at its core, a multi-layered and multi-faceted power-sharing arrangement that emerged after a lengthy peace process. The resulting political and constitutional regime is today rightly considered to be “as close as any African state has come to implementing Lijphart's consociational formula” (Lemarchand 2006). International diplomats have constantly encouraged Burundi's political actors to continue following “the path of dialogue, power-sharing and consensus” (United Nations Security Council 2005).
Burundi's experience begs a number of questions. When, why and how did power-sharing emerge as a key instrument in the toolbox of political, military and diplomatic actors? What was the stated purpose of power-sharing during the different stages of Burundi's transition process? Who is sharing power with whom and which spheres of the public sector are affected by the power-sharing arrangements? To what extent does power-sharing respond to the root causes and driving forces of the violent conflict? To what extent is Burundi, after twenty years of trial and error, a success story?
Power-Sharing before and after the Collapse of the Democratisation Process in 1993
Burundi's current power-sharing system is often said to be rooted in South Africa's role as lead mediator in the peace negotiations process from the end of 1998 onwards (with the replacement of Julius Nyerere by Nelson Mandela) (Bentley and Southall 2005). It is beyond doubt that South Africa had a decisive impact on the arrangement agreed upon in August 2000 (in the Arusha Peace and Reconciliation Agreement for Burundi), which continues to shape Burundi's current constitutional and political landscape. Although he was never directly and personally involved in the Burundian peace process, one might argue that indirectly, Arend Lijphart, who decisively inspired the drafters of South Africa's post-apartheid Constitution, also held the pen when the Arusha Agreement and the Burundian Constitution of 18 March 2005 were drafted. An irony of history is that Lijphart composed his model – to which he also gave normative value – on the basis of an empirical study of, among other countries, Belgium's power-sharing (or so-called pacification) model (Lijphart 1977, 1999; Huyse 1970). These indirect Belgian roots may help to explain why Burundi's political system today, despite its cumbersome complexities, is so “familiar” to Belgian diplomats who continue to support the Burundian power-sharing model more than, for instance, their French colleagues do. The Belgian/South African coalition of “believers” in Burundi's power-sharing model, combined with the diplomatic weight of both countries in Burundi, is most probably one of the factors that explain the “survival” of the system, despite (as detailed below) the dominant party's reservations and outright objections to it.
However, it would be erroneous to consider power-sharing in Burundi as a mere “invention” during the peace process. Three developments in Burundi's recent political history further clarify the introduction of power-sharing as a mode of governance: (i) the reform process initiated under President Pierre Buyoya, (ii) the Constitution of 13 March 1992, which introduced a multi-party system, and (iii) the deals struck between political parties in the aftermath of the failed coup d'état of 21 October 1993. Each of these historical roots in its own way has clearly affected Burundi's current power-sharing regime and, therefore, offers important analytical insights. Taken together, they illustrate Burundi's struggle with the tensions between the three cornerstones of its political liberalisation process: power-sharing, democracy and elections.
Power-Sharing as Political Liberalisation under the Single Party State
When Pierre Buyoya seized power, deposing President Jean-Baptiste Bagaza in September 1987, there was every reason to believe that continuity would prevail, i.e. a concentration of political, military and economic power in the hands of an elite of Tutsi of the Hima clan from the southern province Bururi. Like Buyoya, his predecessors Micombero (1966–1976) and Bagaza (1976–1987) were military from the commune of Rutovu in Bururi. As before, violent struggles to maintain or obtain political power on the basis of ethnic, clan and regional alliances and cleavages were the likely result. However, some five years later, Burundi's political regime had dramatically changed.
The government army's ruthless suppression of a Hutu uprising in the north of the country (Ntega and Marangara communes) in the summer of 1988 was, contrary to earlier massacres, strongly condemned by Burundi's international partners. This pushed President Buyoya to embark on a process of political liberalisation. At the heart of the major dilemma he faced lay the question of power-sharing. On the one hand, as a member of an ethnic minority group and from a particular region and clan, he had every interest in highlighting the unity of the country and in denying the political relevance of Burundi's segmentation (like President Kagame's regime in Rwanda does today). On the other hand, political liberalisation inevitably meant addressing Hutu grievances – including their lack of political representation – and regional imbalances.
Initially, Buyoya's process of political liberalisation amounted to a policy of political reconciliation and inclusiveness, but without democratisation. In October 1988, he created a National Commission in charge of studying the question of national unity, composed of twelve Hutu and twelve Tutsi members. He also established a government of national unity, composed of an equal number of Hutu and Tutsi and led by a Hutu prime minister from the central province Muramvya. In Buyoya's own words, this historical initiative “inaugurated a policy of power-sharing” (Buyoya 1997: 82). In his opinion, this policy of power-sharing was necessary both as a remedy for past injustices – i.e. the legacy of exclusion on ethnic and regional grounds from the structures of power – and in order to build lasting peace in Burundi.
In summary, Burundi's first exposure to and experiment with power-sharing was a matter of political inclusiveness vis-à-vis those groups that, because of ethnic, regional and/or clan related reasons, had for several decades been excluded from political power. It was, at that time, not related to consociational constitutional multi-party democracy, nor to a negotiated settlement between armed opponents. This may well remain an important factor in the political psychology of Burundi's elites today. Indeed, seen from this historical angle, power-sharing and a single party (or dominant party) system can go hand in hand.
Power-Sharing and Democratisation
The post-Cold War wind of change that affected the wider African continent pushed President Buyoya to go further. A new Constitution was adopted in March 1992, laying the foundations of a new democratic order, including the principle of separation of powers, progressive human rights protection provisions and multi-partyism.
The elections of June 1993 were the culmination of the democratisation process. They replaced the legitimacy of the drum (under the monarchy, until 1966) and the gun (under the subsequent military regimes, between 1966 and 1993) by the legitimacy of the ballot box. But electoral competition also meant ethnic competition (Gahama 1995). The overwhelming electoral victory of the predominantly Hutu party Frodebu and its presidential candidate Melchior Ndadaye were too threatening for the vested interests of the incumbent elite. On 21 October 1993, a military coup d'ètat – which seemingly failed, but, in the end, has probably proved to be one of the most successful failed coups in history (Reyntjens 2000) – put an end to the short-lived legitimacy of the ballot box. The assassination of Ndadaye, the speaker and deputy-speaker of the National Assembly, led to a power vacuum and an institutional imbroglio. The country was in turmoil, with some 50,000 civilians killed and hundreds of thousands displaced or forced into exile in the immediate aftermath of the coup, making the organisation of new elections impossible.
Was it, among other things, a lack or rather an excess of power-sharing that derailed the democratisation process? On the one hand, Burundi's Constitution of 13 March 1992 was to some extent inspired by consociational-ism (see also Reyntjens 1992). The Constitution repeatedly expressed a belief in national unity but at the same time requested that the diversity of the various components of the Burundian population be taken into account (which was mirrored in the standard clause “en tenant compte des diverses composantes de la population burundaise”). This applied to the registration and functioning of political parties (who were not allowed to identify themselves with an ethnic group, a religious belief, a region or a gender group) and their lists of electoral candidates (art. 56, 57 and 103). It also applied to the nomination of a presidential candidate, which required the support of a group of 200 persons “created in the spirit of national unity but reflecting the diversity of Burundi's population” (art. 67) as well as to the composition of the government (art. 86). And, in actual practice, Ndadaye's government was indeed composed of “no more” than thirteen ministers (of a total of twenty-three) from his own party (despite Frodebu's dominance at the National Assembly in which it held around 80% of the seats). Nine ministers, including the prime minister (Sylvie Kinigi of Uprona), were Tutsi and five were from Bururi province. The Ndadaye government could therefore – in con-sociational terms – be qualified as a grand coalition. On the other hand, the degree of consociationalism in 1992 was far less extensive than under the Constitution of 18 March 2005 thirteen years and a peace process later (as we will indicate in more detail below). No power-sharing provisions regarding the judiciary, the military, the provincial administration, the diplomatic representation, the state owned companies, et cetera were included. Moreover, the sweeping electoral victory of Frodebu – in combination with a highly ethnic interpretation of the elections – was largely perceived as the replacement of Uprona rule by a demographically majoritarian and democratically legitimised Frodebu rule (almost as if elections had been held under a purely majoritarian system). And, in actual practice, a policy of what became known as frodébisation indeed affected many spheres of the public sector. Provincial governors, commune administrators, senior civil servants, school directors, diplomats – many of whom were Tutsi of Uprona obedience – were replaced by mostly Hutu newcomers of Frodebu obedience. As a result of the elections, the incumbent regime had clearly lost control over its own reform process, a loss which brought forth violent resistance. Indeed, the October 1993 coup attempt was aptly summarised by Sullivan as the active use by (part of) the defeated elite of its de facto veto right (Sullivan 2005). As a pre-emptive strike against a frobébisation of the military, it decided to use its quasi-monopoly on the army, clearly hoping to thus turn back the democratisation clock.
With the benefit of hindsight, a case could now be made to argue that a more elaborate consociational power-sharing arrangement (like the one negotiated in Arusha in 2000, see below) would probably have made the elections less threatening for the incumbent elite and its (typically neo-patrimonial) interests, and therefore less destabilising for the country and more beneficial to the maintenance of peace and security. The price to pay – i.e. the democratic deficit on how the country was governed produced by reducing the impact of free and fair elections – would arguably have been much more affordable than the years of civil war that followed.
In summary, Burundi's second experience with power-sharing – (too) mildly consociational and based on a “belated and reluctant recognition of the importance of ethnic identity” (Lemarchand 1994: 161) – was not able to contain the resistance generated by the “revolutionary” effects of the democratisation process, in particular the consequences of the multi-party elections of June 1993 for the vested interests of the incumbent elite.
Power-Sharing after the Collapse of the Democratisation Process
In its initial reaction to the coup attempt of 21 October 1993, the United Nations requested the immediate restoration of democracy and the constitutional regime. When, however, on 25 November 1993, the Special Representative of the UN Secretary-General, Ahmedou Ould Abdallah, took up his duties, his immediate objective was to restore order and security, to bring an end to the politico-ethnic violence and to prevent further political and institutional destabilisation. His strategy consisted of power-sharing negotiations between what became known as the majority (the FCD, Forces du Changement Democratique, a coalition of predominantly Hutu parties led by Frodebu) and the opposition (the CPPO, Coalition des Partis Politiques de l'Opposition, a coalition of predominantly Tutsi parties led by Uprona). Through the successive Kigobe Accord of 19 January 1994, the Kajaga Accord of 4 February 1994, the Rohero Agreement of 12 July 1994 and, finally, the Convention of Government of 10 September 1994, a power-sharing deal was struck. Rather than substantive policy documents, these successive deals were little more than a catalogue of positions to be filled and how these were allocated to either the FCD or the CPPO. It was agreed to establish a coalition government composed of 55% of FCD ministers and 45% of CPPO ministers. For each ministry, the posts allocated were specified. A similar arrangement was negotiated for the provincial and local administration and for the diplomatic representations. Although the Convention of Government itself stipulated that it merely supplemented the Constitution of 13 March 1992, some provisions clearly had more far-reaching effects. For instance, the National Assembly could no longer exercise its constitutional powers with regard to the dismissal of the government (art. 33). The legitimacy of the ballot box was henceforth formally replaced by the legitimacy of the power-sharing consensus (art. 50).
The Convention of Government defined itself as a blueprint for a new, consensus-based governance and Ould Abdallah praised the “power-sharing arrangements between the representatives of the two main communities and the commitment to develop policies based on consensus” (Ould Abdallah 2000: 73). In actual practice, there was, at that time, very little elite consensus among the signatories and very outspoken resistance to the power-sharing arrangement by increasingly extremist politicians on both sides. On the Tutsi side, the agreement was not signed by a number of smaller parties, including the PARENA (Parti pour le Redressement National) party of former President Bagaza. They were a driving force behind the so-called “dead city” campaigns of youth gangs and vigilante groups that through violence stirred up insecurity in Bujumbura. They rejected the idea of sharing power with those Hutu political leaders whom they considered responsible for the large-scale massacres (sometimes classified as genocide) of Tutsi civilians. On the Hutu side, an important dissident faction within Frodebu, including former minister Leonard Nyangoma (a Hutu from the southern province Bururi), rejected the power-sharing arrangements as well as those political adversaries they considered responsible for the coup d'état and the assassination of President Ndadaye. In June 1994, Nyangoma established the National Council for the Defence of Democracy (CNDD, Conseil National pour la Défense de la Démocratie). In one of the rare substantive political documents ever published by a Burundian political movement, the CNDD denounced the “anti-democratic thesis called consensus and power-sharing between winners and losers of the elections” as a violation of international law (CNDD: 1995). 1 With radical opponents on both sides, consensus-minded moderates became increasingly marginalised. Burundi became the scene of a civil war, with security forces, largely controlled by the old Tutsi elite, fighting the Hutu rebel movements Palipehutu-FNL (Forces Nationales de Libération) and CNDD-FDD (Forces pour la Défense de la Démocratie). A military coup brought former President Buyoya back to power on 25 July 1996. Buyoya himself justified the coup (which he described as “an act of rescue for a people in danger”) by denouncing the failure of the power-sharing regime established under the Convention of Government to restore security and by referring to the imminent threat of a foreign military intervention (Buyoya 1997: 118–123). Other observers interpreted the events of 25 July 1996 as a mere formalisation of a creeping coup that had started (but seemingly failed) nearly three years earlier (Reyntjens 2000: 16).
The CNDD-FDD split in different factions over the years. After several internal leadership battles, Pierre Nkurunziza (a Hutu from northern Ngozi province) emerged as the leader of the politically and militarily most important faction. In August 2005, he was elected president of the Republic after his party had won the parliamentary elections. Leonard Nyangoma continues to lead his “own” CNDD party, which is represented in Parliament but not very popular outside Bururi province.
In summary, between the failed coup d'état of 21 October 1993 and the successful coup of 25 July 1996, Burundi experimented with an internationally mediated transitional power-sharing arrangement that was, however, never firmly rooted in the domestic political scene, because of both the distrust among its signatories as well as the fierce opposition by other political forces. Political violence rather than political dialogue prevailed. This experience may well remain an important factor in the political psychology of some of Burundi's political elite today. Indeed, seen from this historical angle, a case could be made to argue that power-sharing, first of all, constitutes a “betrayal” of the people's will as reflected through democratic elections. And, secondly, practice has shown that power-sharing is an inherently weak source of political authority, as it can easily be done away with through armed violence.
Burundi's Constitution of 18 March 2005 and the many consociational features of its current political regime are strongly rooted in the peace process that started in June 1998. In a way, also the political resistance against implementing and maintaining the consociational power-sharing equilibrium can be explained by taking into account the positions of the different political and military actors during the peace process.
Used as the main conflict resolution tool during Burundi's peace process, power-sharing clearly had two different dimensions, which only partly overlapped and which also determine the structure of this section. On the one hand, power-sharing refers to the consociational political organisation of Burundi's plural society on the basis of a formalised co-habitation of segments (primarily of an ethnic and regional nature) that are considered to be politically relevant. By attempting to redress horizontal inequalities through the use of power-sharing, most notably to accommodate the desire for more political representation expressed by Hutus, Burundi's peace process definitely addressed one of the root causes of armed conflict. This was done primarily through the Arusha Peace and Reconciliation Agreement of 28 August 2000 (although it took almost five years to transform this political agreement into a post-transition constitution). Quite paradoxically, the Arusha Agreement was not negotiated nor signed by the Hutu rebel movements. On the other hand, both (to a minor extent) in the Arusha Agreement and (to a very large extent) in the later peace agreements (signed in 2003 and in 2006), the notion of power-sharing mainly referred to the dividing of the cake between competing political elites and their networks, in particular the distribution of posts (at different political, military, senior administrative, diplomatic and economic levels) between an incumbent government (initially the interim government, later the democratically elected government of President Pierre Nkurunziza) and the two rebel movements.
This has – at least – two important consequences for the acceptance and perception of power-sharing in Burundi. Firstly, the two types of power-sharing were sometimes difficult to combine. Posts had to be awarded to members of the predominantly Hutu rebel movements without undermining the earlier consociational power-sharing arrangement negotiated in Arusha. Secondly, the former rebel movements (today transformed into political parties) frequently challenge and reject the Arusha Agreement because they did not negotiate it (although, when it serves their interests, they do not hesitate to call upon “Arusha”).
Burundi's Peace Process
The table on the following page gives an overview of the (main) peace agreements concluded to put an end to the civil war in Burundi.
In response to Pierre Buyoya's coup d'état of July 1996, a group of neighbouring countries imposed sanctions on Burundi, that aimed at pushing the Buyoya regime towards restoring the 1993 constitutional order. They designated former Tanzanian President Julius Nyerere as facilitator. The Arusha peace negotiations process started in June 1998. In approximately two years, some ten rounds of negotiations were held. Five different working groups were established, in charge of drafting the above-mentioned protocols. In October 1999, Nyerere died and was replaced by former South African President Nelson Mandela as facilitator. On 28 August 2000, the Arusha Peace and Reconciliation Agreement for Burundi was signed by the government, the National Assembly and 17 political parties, subdivided in two groups, one predominantly Hutu (also known as the G7) and one predominantly Tutsi (also known as the G10). The Arusha Agreement was greeted with significant scepticism because of: (i) the important reservations expressed by a number of signatories, (ii) the important matters that were left unresolved, and (iii) the absence of the (predominantly Hutu) rebel groups CNDD-FDD and the Palipehutu-FNL among the signatories. As a result of the latter deficiency, the peace agreement was even unable to secure an immediate cease-fire, let alone sustainable peace. However, while no more than a first step, the Arusha Agreement clearly shaped the framework for the continued mediation of the conflict and in the end decisively shaped Burundi's current Constitution of 18 March 2005.
Burundi: (Main) Peace Agreements
On 16 November 2003, a Global Ceasefire Agreement (GCA) was signed between Burundi's transitional government – established as a result of the Arusha Agreement and headed, initially, by Pierre Buyoya and, later, by Domitien Ndayizeye (Hutu, Frodebu, from Kayanza province) – and the CNDD-FDD rebel movement, which, as a result, joined the National Assembly and the government. The GCA stipulated that, while revoking possible conflicting provisions, it was an integral part of the Arusha Agreement. But the CNDD-FDD's actual political commitment to implement the Arusha Agreement was variable, in particular insofar as it felt taken hostage by a complex set of power-sharing provisions that de facto reduced the impact of its overwhelming electoral victory in 2005.
Talks between the Palipehutu-FNL and interim President Ndayizeye formally started in January 2004, but were suspended following the massacre of Banyamulenge (Eastern Congolese ethnic Tutsi) refugees in Gatumba in August 2004, for which the FNL claimed responsibility. After a first Agreement of Principles towards Lasting Peace, Security and Stability in June 2006, a Comprehensive Ceasefire Agreement (CCA) was signed in September 2006 between the new government of President Pierre Nkurunziza and the Palipehutu-FNL. Implementation of the CCA remained problematic for almost two and a half years. It was not until April 2009 that Burundi's last armed rebel movement (after changing its name to FNL, leaving out the ethnically exclusive, original part “Palipehutu”) was registered as a political party and laid down its arms.
In several respects, Burundi's constitutional and institutional setting stands out as a model of consociationalism, even though, in actual political practice, important gaps remain between the model and its implementation (Vandeginste 2008). We will analyse how two of the classical consociational characteristics as identified by Lijphart apply to Burundi (its peace agreements, its Constitution and its actual practice), while briefly also referring to the question of veto rights. The element of segmental autonomy is not present (and hardly relevant) in the case of Burundi where segments are not territorially separated and do not have the classical distinctive features (a different language, culture, religion, etc.). 2
Given the absence of those features, the existence of ethnic groups in Burundi has given rise to major controversy, with a clear clash between a primordialist and an instrumentalist approach to ethnicity (Kadende-Kaiser and Kaiser 1997; UN CERD 1997). The decisive, subjective element in affirming the existence of ethnic groups in Burundi is self-identification on the basis of ethnic affiliation.
The Arusha Agreement stipulated that the President of the Republic shall be assisted by two Vice-Presidents, belonging to different ethnic groups and different political parties. This is confirmed and further refined in the Constitution. When appointing the Vice-Presidents, attention must be paid to the dominant ethnic composition of their political party (art. 124, para. 2). 3 This provision requires some further clarification. It was inserted into the Constitution in order to avoid that a Tutsi and a Hutu Vice-President be chosen from two different, but both predominantly Hutu parties. This marked a compromise that was reached after lengthy discussions on a crucial aspect of Burundi's consociational power-sharing. In fact, this is the only provision where the Constitution makes a difference between “predominantly Tutsi” and “predominantly Hutu” parties. Throughout the peace negotiations, and also at the time of drafting the post-transition constitution, the G10 coalition of predominantly Tutsi parties had strongly insisted that only Tutsi politicians belonging to a Tutsi party could truly represent the Tutsi segment in society. There were two main reasons for favouring that approach, one very pragmatic, the other more conceptual. Firstly, most of the Tutsi parties feared that, as a result of the elections, they would no longer be represented in the institutions unless their presence was guaranteed by the Constitution. Given their electoral defeat in 2005, that fear turned out to be justified for most of them. Secondly, from a theoretical consociational perspective, the case could indeed be made that continued dialogue, bargaining and compromising between segmental elites requires a clear separation between these elites, in accordance with the lines of separation between the societal segments. However, in the end, Burundi opted – more in line with the position favoured by the G7 coalition of predominantly Hutu parties – for “consociational political parties”, thus including a crucial centripetal element in its power-sharing design. Political parties must not be established on the basis of ethnic or regional exclusivity (art. 78 of the Constitution) and, quite importantly, each list of electoral candidates must reflect Burundi's ethnic and gender diversity. Out of three successive candidates proposed by a party on a blocked list, at least one needs to be of a different ethnic group. The reasoning behind this approach was that, while Burundi's segmental cleavages must be explicitly and formally recognised at the political level, it was at the same time necessary to reduce the conflict potential of ethnicity by encouraging political parties to be ethnically inclusive. It is generally acknowledged that this approach has – at least temporarily – been quite successful. Today, political competition in Burundi no longer coincides with ethnic cleavages. Furthermore, the dominant party CNDD-FDD, while rooted in a Hutu rebel movement, is no longer perceived as an exclusive Hutu party. In fact, most Tutsi members of parliament are member of the CNDD-FDD and many presidential advisors are Tutsi (but, quite interestingly, not from Bururi).
In practice, since the 2005 elections, the First Vice-President has indeed been a Tutsi member of Uprona whereas the Second Vice-President has been a Hutu member of CNDD-FDD.
Regarding the composition of the government, the Constitution requires that a maximum of 60% of the ministers are Hutu and a maximum of 40% of the ministers Tutsi. At least 30% must be women (art. 129). The Minister of National Defence and the Minister in charge of the National Police must belong to different ethnic groups (art. 130). Furthermore, political parties that obtain 5% of the votes cast at the parliamentary elections are entitled to participate in a coalition government, with the number of ministerial posts that is proportionate to their seats at the National Assembly. In practice, these provisions have been more or less respected, though not without difficulties. Apart from the CNDD-FDD (which held 64 of the 118 seats in the National Assembly after winning 57.8% of the votes cast during the 2005 elections), only Frodebu (21.6% of the votes cast; 30 seats in the NA) and Uprona (7.1% of the votes cast; 15 seats in the NA) were able to claim a number of ministerial posts. Throughout the legislature, the “cohabitation” between CNDD-FDD and its coalition partners knew many ups and downs, with tensions and repeated government reshuffles seriously affecting the effectiveness of the government (Vandeginste 2008). Essentially, the 2005 electoral results were not propitious to a smooth implementation of the power-sharing arrangement. The CNDD-FDD, despite winning an absolute majority at the elections, felt taken hostage by power-sharing arrangements laid down in a peace agreement it had not even negotiated itself (but of which it – reluctantly – endorsed the outcome) and by the losers of the elections. For Frodebu and Uprona, the CNDD-FDD mostly behaved like the new single party, reproducing many of the bad governance practices of the old days. 4
In Burundi, the CNDD-FDD is indeed mostly referred to as “the ruling party” (le parti au pouvoir) whereas Frodebu and Uprona, despite being part of the grand coalition, are seen (and portray themselves) as the main opposition parties.
The Arusha Agreement did not include any explicit provision requesting ethnic, regional or gender balance in the National Assembly. The Constitution, however, stipulates that the Assembly should be composed of 60% Hutu and 40% Tutsi members, with a minimum of 30% women (art. 164). As noted above, each list of electoral candidates for the National Assembly also needs to reflect ethnic and gender diversity. If, as was the case in 2005, the outcome of the elections is not in accordance with these constitutional proportions, additional members are to be co-opted by the Electoral Commission.
The Arusha Agreement stipulated that the Senate was composed of two delegates from each province, of different ethnic communities and (indirectly) elected by the commune councils of that province. The Senate may co-opt up to three Batwa so as to ensure representation of this community. These principles of regional representativeness and Hutu/Tutsi parity in the Senate are confirmed by the Constitution (art. 180) and scrupulously respected in practice.
Quite importantly, these quota are combined with qualified majority requirements. In order for the National Assembly to enact legislation, two thirds of its members must be present and laws can be enacted only with a two thirds majority (art. 175, para. 1). As a result, the Hutu majority of MPs cannot pass legislation without seeking allies across ethnic lines. This grants a de facto veto right to Tutsi MPs, despite the fact that Burundi has no formal, explicitly recognised veto rights to protect the vital interests of minorities. Also, despite its absolute majority (both in terms of the votes cast and of its number of seats), the CNDD-FDD party has to seek support from other parties in the National Assembly. This combination of quota and qualified majority requirements is firmly in line with the consociational spirit that underlies Burundi's negotiated settlement. In actual practice, also as a result of a major internal split within the CNDD-FDD in early 2007 and dissensions within Frodebu and Uprona, the complexity of the consociational rules and the reluctance of several players to abide by them gave rise to considerable instability. Throughout 2007 until June 2008, this resulted in an almost complete standstill of the legislative activities of the National Assembly.
The principle of proportionality (but with minority over- representation) also applies to other levels. Firstly, in order to ensure that commune councils reflect the ethnic diversity of their constituencies, the Electoral Commission can order the co-optation of up to one fifth of the members of the commune council from an under-represented ethnic group (art. 266, para. 1). Taken together, no more than 67% of all commune administrators should belong to the same ethnic group (art. 266, para. 2). Secondly, regarding the provincial administration, the Constitution merely requires that it reflects Burundi's diversity (art. 143), without further specifications, contrary to the initial G10 request that each provincial governor be assisted by a vice-governor of a different politico-ethnic family. Furthermore, regarding appointments in public administration and in diplomatic posts, the government must keep in mind the need to preserve an ethnic, regional, political and gender balance (art. 135 of the Constitution). As far as the leading positions in state-owned companies are concerned, appointments should be carried out on a 60% Hutu/40% Tutsi basis (art. 143). Finally, the Arusha Agreement stipulated, concerning defence and security forces, that
“for a period to be determined by the Senate, not more than 50% of the national defence force shall be drawn from any one ethnic group, in view of the need to achieve an ethnic balance and to prevent acts of genocide and coups d'état” (see also article 257 of the Constitution).
However, in light of the earlier experience under President Ndadaye in 1993, the drafters of the Arusha Agreement were also aware of the fact that changes in the composition of the defence and security forces were to take place gradually and not abruptly. Therefore, in addition to the principle that long-standing political, ethnic, regional and gender imbalances must be redressed, it was also stressed that the “correction of the imbalances shall be approached progressively, in the spirit of reconciliation and trust in order to reassure all Burundians” (see also art. 258 of the Constitution). The ethnic parity in the composition of the Burundi's National Defence Force and the National Police Force was also reconfirmed in the Pretoria Protocol on Political, Defence and Power-Sharing of the 2003 GCA. It is widely acknowledged that the successful power-sharing at the level of the defence and security forces – both in ethnic terms and, as noted below, between the former government army and the former rebel movement CNDD-FDD – constitutes the major cornerstone of Burundi's transition towards peace. Any attempt to revise this delicate equilibrium would inevitably carry the major risk of jeopardising the whole process. In 2009, the Senate established commissions to study the implementation of the principle of ethnic parity in the army and the police, but showed no intention of revising the principle as such.
Sharing Power with the Rebel Movements
Burundi was exposed to yet another type of power-sharing when negotiations with the two main rebel movements were entered into.
CNDD-FDD
Under considerable pressure from the regional peace initiative and with South African facilitation, a first cease-fire agreement between the transitional government and the CNDD-FDD rebel movement of Pierre Nkurunziza was signed in December 2002. Three fundamental principles were agreed upon by its signatories: (i) Burundi must be governed by the rule of law and on the basis of national unity, democracy, pluralism and respect for human rights, (ii) new defence and security forces must be put into place, composed of members of government forces and of combatants of armed political movements, and (iii) a power-sharing system must be established in the framework of an inclusive transitional government. The remainder of the peace process, completed through the signature of the GCA in November 2003, essentially concerned the specification of the principle of power-sharing in the executive branch and in the defence and security forces. Power-sharing, in this case, was a matter of finding the right balance between posts held by the transitional government (itself the outcome of the power-sharing deal struck in Arusha in 2000) and those allocated to the rebel movement. Indeed, except for the explicit confirmation of ethnic parity in the command posts of the army, in the police force and in the intelligence service, the GCA and its different protocols were silent on consociationalism. (Indirectly, the agreements reached in Arusha were reaffirmed insofar as it was stipulated, in its article 2, that the GCA constituted an integral part of the Arusha Agreement.)
Under the Protocol of 8 October 2003, four ministries in the transitional government – including the crucial and newly created post of Minister of State, whom the President was required to consult “on all key matters” – were allocated to the CNDD-FDD, as well as fifteen seats in the National Assembly, three provincial governors, thirty communal administrators, two ambassadors and the direction of twenty percent of state owned companies. The Protocol, further specified through the Forces Technical Agreement of 2 November 2003, also provided for power-sharing in the command structures of the new defence and security forces: 60% government/40% CNDD-FDD in the new National Defence Force and 65% government/35% CNDD-FDD in the national police and intelligence service. The remarkably smooth and successful integration (or, at the very least, coexistence) of the government forces and the rebel movement at all hierarchical levels was most probably the main stabilising factor for Burundi. Together with the deployment of a peace-keeping force of the African Union – later replaced by a United Nations operation (ONUB, Opération des Nations Unies au Burundi) – this paved the way for the local, parliamentary and presidential elections that were held between June and August 2005, a few months after a new, post-transition Constitution had been adopted. After its overwhelming victory in the elections, the CNDD-FDD did not question the delicate equilibrium in the National Defence Force but, in practice, increasingly gained control over the police and intelligence service.
Palipehutu-FNL
Attempts at negotiating with the Palipehutu-FNL were unsuccessful until the end of 2003. Founded by Hutu refugees in Tanzania in 1980 with the ideology of defending the cause of Hutu peasants, the movement requested that negotiations be held between Hutu and Tutsi, not between ethnically mixed parties and institutions (like in the Arusha peace process). Also, while calling for a new “social contract” between Hutu and Tutsi, it insisted that the Hutu demographic majority be “truly democratically” reflected in a political majority, which was hard to reconcile with the consociational spirit of Arusha.
Quite noteworthy, that, in addition to other important factors like internal divisions and increased military weakness, it was the successful negotiation (and subsequent implementation) of the earlier power-sharing agreements (Arusha and the GCA) that forced Palipehutu-FNL to moderate its position and to enter into negotiations from early 2004 onwards. With the agreed ethnic parity in the defence and security forces and the representation of Hutu at all levels of the state, in particular after the 2005 elections, the movement had lost much of its ideological raison d'être. What remained was essentially a strategic game by its leadership, aimed at securing the largest possible piece of the cake to be divided among Burundi's elite, typical of a neo-patrimonial exercise of political authority. After many ups and downs, peace agreements were signed in June and September 2006 but it was not until April 2009 that the peace process with the FNL was successfully completed.
Compared to the CNDD-FDD, the FNL has, so far, been “poorly rewarded” for years of armed struggle. According to the International Crisis Group (ICG), the 2006 agreements looked like a “surrender” of the FNL (ICG 2007: 2). Regarding the political organisation of Burundi's segmented society, it was agreed in the Common Country Assessments (CCA) that a commission of experts should be created to rewrite the history of Burundi, in particular on the ethnic question, and that popular consultations should be organised from the grassroots level to the top, leading towards what is vaguely referred to as a social contract. Three years later, none of this has happened. In addition to an immediate cease-fire, the agreements also included rather vaguely worded principles on the return of refugees, the release of FNL supporters under a regime of provisional immunity and the demobilisation of FNL combatants or their integration in the new defence and security forces. Most observers agree that – with the CNDD-FDD precedent in mind – the prospect of posts in the administration, the police and the army and of Disarmament, Demobilisation and Reintegration (DDR) packages seduced many unemployed youth (including demobilised CNDD-FDD members) to join the Palipehutu-FNL in 2007 and 2008. However, many were left behind disappointed, at all levels of the movement. In April 2009, the earlier agreements were specified. While the Palipehutu-FNL claimed it had 21,000 combatants, it was in the end agreed to integrate 3,500 in the defence and security forces and to offer DDR assistance to 5,000 combatants. Some thirty-three, relatively minor posts were promised to the FNL leadership. In June, FNL leader Agathon Rwasa became the head of the Institut National de Sécurité Sociale (INSS), spokesperson Pasteur Habimana was appointed as councillor at the Burundian embassy in India, two others were appointed as provincial governors, another became a commercial director at the National Tea Office and several others were appointed as advisors in different ministries and public services. The main questions for the movement, now registered as the party FNL but still facing – like the CNDD-FDD – the crucial challenge of transforming itself into a political party, is whether it will be able to curb internal dissidents and attract voters at the 2010 elections like the CNDD-FDD did in 2005. One year before the elections, the leadership of both former rebel movements is remarkably confident of winning the elections.
Some Concluding Observations, Including on the Way Forward
At various moments during its violent process of political transition over the past twenty years, Burundi has experimented with power-sharing. Power-sharing has been used as a tool
to initiate political liberalisation after decades of one-party rule,
to temporarily preserve a minimum of institutional stability after democratic elections – the effects of which provoked a military coup,
to address long-standing grievances of politically under-represented segments in society,
to negotiate peace with rebel movements.
For each of these, the objectives, dynamics, actors, driving forces and sources of resistance that shaped the reality of power-sharing varied considerably. An earlier experiment with power-sharing inevitably shaped the expectations, perceptions and use of power-sharing in a later one. The Burundian case clearly shows how complex realities have an overwhelming impact on the effects of power-sharing and how, vice versa, drawing conclusions on the merits and limits of power-sharing is extremely hazardous when making an abstraction of the specific setting in which the notion is being implemented.
By and large, when measured against the objective of war termination, the use of power-sharing can – so far – be considered to be a success story in the case of Burundi. At the same time, when measured against more ambitious state-building purposes (rule of law, human rights, democracy, effective and accountable governance), Burundi clearly has a very long way to go and consociational power-sharing has so far not been able to make a difference. This suggests, somewhat paradoxically, that the toolbox of consociationalism – composed on the basis of successful attempts to organise stable power-sharing democracies in divided societies – has been instrumental in ending violent conflict in Burundi, but not (or not yet) in preventing or reversing the trend towards increased electoral authoritarianism and the use of (although selective, relatively small scale and well-targeted) violence as a way of exercising political authority. Beyond the (exceptional) Burundian case, this also begs the question whether – and, if so, under what circumstances – complex, trial and error based and gradually fine-tuned power-sharing arrangements can indeed offer a more sustainable way of conflict resolution than the more frequently used “emergency power-sharing” deals, such as the ones struck in Kenya and Zimbabwe in 2008.
A conclusion the Burundi case-study clearly suggests is that its political actors have major difficulties in “reconciling” power-sharing with (supposedly democratic) elections. For some, power-sharing seriously distorts the outcome of elections and is therefore undemocratic. For others, power-sharing is just a temporary, transitory arrangement, paving the way towards elections but without durable impact after them. Yet others view power-sharing as the indispensable stabilising element that reduces the stakes involved in the elections for the neo-patrimonial elites competing for the limited pieces of the cake. There is probably some truth in all of these views. The latter view in particular reveals the essential challenge for consociational power-sharing in complex post-conflict settings like the one of Burundi. How much democracy can consociationalism possibly offer in situations of extreme violence and immense distrust, in which the short-term “ambition” of power-sharing must almost inevitably be limited? In turn, this reflects a more structural debate (see, i.a., Paris 2004) about how to promote and sequence objectives of peace, democracy and stable and effective governance.
How much power-sharing is desirable in Burundi's near future? On the one hand, advocates highlight the need to maintain the fragile equilibrium as a precondition for a more sustainable consolidation of peace in Burundi. Burundi's experience seems to contradict the classical criticism that consociational power-sharing “freezes” people's identities and therefore deepens the segmental cleavages and divisions (Roeder 2005). Instead, the acknowledgement and institutionalisation of the segments' political relevance may be seen as a first and necessary (though by no means final) step in the process of de-ethnicising political competition and of overcoming decades of politico-ethnic violence. (In which case neighbouring Rwanda still needs to embark on its own consociational journey, presumably after a next round of politico-ethnic violence). Also, there is little doubt that excluding certain elites from political power, even as a result of supposedly free and fair elections, may lead to considerable instability. Among the threats outgoing elites may have to face, one is particularly worth mentioning. Burundi has a longstanding tradition of impunity, both for gross and systematic human rights violations as well as for economic and financial crimes. Today, the interests of many leading Uprona, Frodebu, CNDD-FDD and FNL members are served by their alliance of impunity. Seen from that angle, it is important to lower the stakes involved in the 2010 elections in Burundi. Its current, admittedly complex power-sharing arrangement, which guarantees even losers of the elections a part of the cake, is therefore certainly a stabilising factor. (Conversely, another lesson learned from twenty years of trial and error with power-sharing in Burundi is that the prospect of judicial prosecution may incite those who lose the elections to become conflict entrepreneurs who, after a next round of peace talks, can always negotiate themselves back into power.) However, opponents, in particular leading figures in the CNDD-FDD, favour a reduction of the consociational complexity of Burundi's governance structures. While no doubt largely inspired by political opportunism (possibly short sighted, in light of the uncertain political landscape after the 2010 elections), they claim that power-sharing has served its purpose and that ethnicity will no longer lead to renewed violent conflict in Burundi. While today ethnicity is indeed no longer the main determiner of political divisions and competition in Burundi, it is very uncertain whether its conflict potential has been irreversibly wiped out. Reformists also refer to the complexity and ineffectiveness of political decision-making and its negative effects on the performance of (technocratic) governance, which is a classical criticism of consociationalism that indeed applies to the case of Burundi so far. This begs a challenging question for consociational engineers. What type of reforms might at the same time reduce the complexity of Burundi's power-sharing regime (and the efficiency costs this entails), while firmly safeguarding its consociational foundations (and the benefits that go with them)?
More generally, irrespective of their desirability, it is clear that several factors and developments threaten the future of consociational power-sharing in Burundi. Firstly, most of the favourable conditions Arend Lijphart identified are not fulfilled in the case of Burundi (Vandeginste 2006). Also, from 2010 onwards, with the direct election of the president, Burundi's political system will become somewhat more presidential and less parliamentary than today. Secondly, the international pressure (and third party enforcement) which has decisively contributed to the design and initial implementation of the power-sharing arrangement (and to successive rebel movements buying in to the principle of power-sharing) is no longer there since Burundi gradually “regained its sovereignty” following the end of the transition. Like in Rwanda, the international community may well be willing to accept a more authoritarian and less consensus- and dialogue-based governance style, as long as this does not threaten short-term stability in Burundi and the Central African region. Thirdly, although its consociational power-sharing regime is firmly entrenched in the Constitution, Burundi may here also – as in many other spheres – face the limits of the law. In particular the recent record of adjudication by the Constitutional Court, which, strictly legally speaking, could operate as an authoritative watchdog to ensure respect for the power-sharing arrangement, raises considerable doubts about its independence and its capacity to resist the imperatives of political expediency. Fourthly, the nature and functioning of political parties in Burundi remains problematic. Rather than setting a forum for exchange between segmental leaders who represent their followers and liaise with organised social movements, the microcosm of political parties in Burundi (which today total more than forty) primarily constitutes an unstable scene of ever-shifting alliances built around the neo-patrimonial interests of their leaders. The corporatist foundations and ramifications of classical consociationalism are clearly lacking.