Abstract
This article argues that constitutional engineering along consociational lines in Burundi – explicitly accommodating ethnicity rather than attempting to suppress it – was instrumental in reducing the political role of ethnicity, but that other endogenous and exogenous factors also played a role. After surveying developments since 1988, this article focuses on the 2005 polls. The outcome of the parliamentary elections suggests that the “disappearance of the ethnic factor,” extolled by many at the time, was achieved by constitutional constraints rather than by social or political dynamics. Nevertheless, with regard to the country's most important and lethal historical problem, the ethnic divide, constitutional engineering has proved hugely effective. Burundi's main cleavage is now between (and within) parties rather than ethnic groups, and when violence occurs it is political rather than ethnic. Burundi's current crisis is therefore not a failure of consociationalism but of democracy.
The 2015 electoral period has created instability and violence in Burundi, but – contrary to what many feared – it has not led to the reappearance of ethnic strife, at least for the time being. This article argues that it is democracy, not consociationalism, that has failed. This is shown by examining institutional engineering over an extended period of time, beginning in 1988. The article then highlights the effects of the 2000 Arusha Accords on both ethnic relations and democracy, along with the agreement's translation into the 2005 Constitution and its implementation with the 2005 elections. Later developments indicate a shrinking democratic space and increasingly poor governance, but the constitutional framework at the base of ethnic accommodation has proved remarkably resilient. At the moment, and contrary to the pre-Arusha past, conflicts are political rather than ethnic. In light of Burundi's history, this is a major achievement of institutional design.
Antecedents
Ethnic amnesia (Lemarchand 1992) characterised Burundi's constitutional and political dispensation up to the late 1980s. The ethnic issue was not addressed but ignored; indeed, it was forbidden to mention ethnicity. Just like in Rwanda since 1994, this served to hide the fact that power was exercised by an elite among the Tutsi ethnic minority. Nevertheless, ethnically inspired violence cost scores of lives, particularly in 1972 when the Hutu elites were decimated by a state-sponsored genocide. Hundreds of thousands more fled into exile.
This began to change after another round of violence caused thousands of victims in August 1988, less than a year after Pierre Buyoya came to power through a military coup. Under considerable international pressure, he embarked on a policy of reform aimed at breaking the cycle of violence. Without explicitly saying so, he introduced a quota system that enabled the Hutu majority to enter the political scene. The introduction of representational parity was to be the prudent beginning of the use of constitutional design to deal with the ethnic divide.
Initial measures were informal. In the autumn of 1988, the Commission Nationale Chargée d'Έtudier la Question de l'Unité Nationale (National Commission to Study the Question of National Unity) and the cabinet each comprised an equal number of Hutu and Tutsi. For the first time since 1965, a Hutu became prime minister. Hutu were increasingly represented equally with Tutsi in most bodies of the state. However, the armed forces and the police and intelligence services resisted change, remaining predominantly Tutsi, a fact that became brutally apparent in October 1993.
These mechanisms were progressively formalised, particularly in the 1992 Constitution, which called for broadly inclusive power sharing. For example, article 84 stipulated that “the government be composed in a spirit of national unity, taking into account the diverse component parts of the Burundian population,” a reference to the ethnic groups but also to regional balance. Likewise, article 101 required parties' lists of candidates for the parliamentary elections to be “composed in a spirit of national unity.” Such inclusivity requirements were incorporated in 12 of the 189 articles of the Constitution (Reyntjens 1992). Although the terms “Hutu” and “Tutsi” were not used, these provisions were clearly consociational mechanisms aimed at reassuring the minority Tutsi by offering them over-representation.
The effectiveness of this constitutional design was put to the test in the elections of June 1993. In the presidential race, the challenger Melchior Ndadaye won 65 per cent of the vote, versus 32 per cent for the incumbent Pierre Buyoya. Ndadaye's party, the Front pour la Démocratie au Burundi (Front for Democracy in Burundi, FRODEBU), captured 71 per cent in the parliamentary elections, while the formerly single party the Union pour le Progrès National (Union for National Progress, UPRONA) managed a mere 21 per cent. In ethnic terms, the membership of the Assembly was about 85 per cent Hutu and about 15 per cent Tutsi, which corresponded nearly perfectly to the demographic composition of the population. This confirmed Tutsi fears that democracy would transform the Hutu demographic majority into ethnic political domination. It also meant that even across party lines, Tutsi MPs could not secure a blocking minority of 20 per cent should they feel that a constitutional amendment threatened their vital interests. So the consociational objective of reassuring the ethnic minority was not achieved (on the elections and their aftermath, see Reyntjens 1993).
Ndadaye understood the discontent of many in the ousted politico-military elite very well, and he attempted to alleviate their fears. For example, although FRODEBU commanded an 80 per cent majority in Parliament, Ndadaye offered over a quarter of the positions in his cabinet to UPRONA, including the post of prime minister, to which the Tutsi technocrat Sylvie Kinigi was appointed. More than one-third of cabinet members were Tutsi. These concessions were to no avail, as during the night of 20–21 October 1993 army units carried out a coup without encountering resistance in Bujumbura. Ndadaye and some of his close associates were assassinated, others went into hiding. Violence erupted in the provinces, with tens of thousands of people killed in ethnic (Hutu vs. Tutsi) and political (FRODEBU vs. UPRONA) violence. The events triggered political stalemate at the centre, but also, and above all, the beginning of a decade-long civil war.
Arusha and Constitution Making
In June 1994, former FRODEBU leaders who rejected attempts at accommodation created the Conseil National pour la Défense de la Démocratie (National Council for the Defence of Democracy, CNDD) and its armed wing, the Forces pour la Défense de la Démocratie (Forces for the Defence of Democracy, FDD). Older rebel movements also again engaged in military operations. A “government convention” negotiated between FRODEBU and UPRONA in September broke down, and the army restored Buyoya to power through a coup in July 1996. As countries of the region imposed an embargo on Burundi, Buyoya was eventually forced to move towards negotiations with the opposition that started in June 1998, in Arusha, Tanzania, under the mediation of former Tanzanian president Julius Nyerere, succeeded after his death by Nelson Mandela. Pressure from the mediators and regional leaders eventually produced results in August 2000, when the Arusha Peace Agreement was signed. Its main provisions called for a consociational form of power sharing and for the deployment of international security forces. Most predominantly Tutsi parties among the 19 signatories agreed only reluctantly and expressed reservations about key provisions. They signed the accord, but did not embrace it sincerely. The rebel movements, all predominantly Hutu, were not involved in the talks and rejected the accord outright.
The implementation of this weakly supported deal took several more years. The post-transition Constitution was approved by over 90 per cent of voters at a referendum held on 28 February 2005. Translating the principles agreed upon in Arusha, it was markedly consociational (Sullivan 2005; Vandeginste 2009, 2014), attempting as it did to combine majority rule with minority protection. This was achieved by trademark instruments of consociationalism, such as quotas, minority over-representation, and minority veto. The president, likely to be a Hutu, was to be seconded by two vice presidents, a Hutu from a mainly “Hutu” party and a Tutsi from a mainly “Tutsi” party. The cabinet was to be composed of 60 per cent Hutu and 40 per cent Tutsi members, and 30 per cent of cabinet members were to be women. In light of Burundian history, the posts controlling defence and public order were particularly delicate. Therefore, the Constitution provided that the ministers for the army and the national police were to be of different ethnic groups.
The membership of the National Assembly was to also be 60 per cent Hutu and 40 per cent Tutsi. If this balance was not achieved through the ballot, the electoral commission was to ensure it by creating additional parliamentary seats and filling them with members of the insufficiently represented ethnic group – a process known as “co-optation.” A minimum of 30 per cent of MPs were to be women, and three slots were reserved for Twa (a group representing less than 1 per cent of the population), both goals to be ensured by co-optation if necessary. The Senate was to be composed of an equal number of Hutu and Tutsi, as were the army and the national police. 1
More details on the constitutional dispensation can be found in Reyntjens 2015.
The Proof is in the Pudding: The 2005 Elections
As subsequent polls have been atypical, as explained later, the functioning of this institutional design will be checked here against the outcome of the 2005 elections. This was an electoral marathon, with municipal, parliamentary, senatorial, and presidential elections, but this assessment limits itself to the 4 July polls for the National Assembly, a process by and large considered free and fair by international and domestic observers (Union Européenne 2005; COSOME 2005). At the end of the process, including the co-optation process, the composition was as shown in Table 1.
Full Composition of the 2005 National Assembly
The membership of 118 (up from the elected 100 members) met the constitutional requirements: 69 Hutu (60 per cent), 46 Tutsi (40 per cent), 36 women (31 per cent), and 3 Twa. Overall, the CNDD-FDD secured 54 per cent of the seats, a clear majority but by no means a blank check. Indeed, under the Constitution, a two-thirds majority was needed to pass legislation in a number of important areas (so-called “organic laws”), such as defence and security. Amending the Constitution required a four-fifths supermajority. The CNDD-FDD would therefore need to develop post-election alliances across political and ethnic lines – an explicit goal of consociational design.
In addition, the CNDD-FDD, generally portrayed as a “Hutu” party, emerged as the most interethnic, because nearly one-third of its MPs were Tutsi. By contrast, UPRONA and the Mouvement pour la Réhabilitation du Citoyen-Rurenzangemero (Movement for the Rehabilitation of Citizens – Rurenzangemero, MRC) confirmed their “Tutsi” nature, as no Hutu were elected on their lists. However, this apparent variation in the multi-ethnicity of the parties was actually the consequence of constitutional design, not of political, sociological, or ethnic choices. Article 168 of the Constitution required that the lists of candidates proposed by the parties in each province “be multi-ethnic in nature and take into account gender balance”:
Of every three candidates proposed one after the other on a list, only two shall belong to the same ethnic group, and at least one out of every four shall be a woman.
In addition, the closed-list proportional representation system precluded ethnic voting. Thus, while all parties proposed lists with mixed ethnicities, it was the absolute number of seats they won per province that determined the ultimate ethnic distribution of their seats in Parliament. Because UPRONA failed to secure more than one seat in any province, it could elect only the top person on any provincial list, always a Tutsi, at the expense of Hutu placed lower on the lists. The two large “Hutu” parties – CNDD-FDD and FRODEBU – likewise typically listed Hutu at the top of their provincial lists (although the CNDD-FDD placed a Tutsi on top of its list in one province). Had these Hutu parties garnered only one seat per province, their parliamentary representation would have been almost exclusively Hutu. But because they won more seats per province, they had to delve deeper into their lists, which as required by the Constitution included Tutsi. Thus, the CNDD-FDD's parliamentary delegation emerged as the most multi-ethnic simply because the party gained the most electoral support. Its victory was by no means evidence that Burundians voted for a party that was inherently multi-ethnic. The claim that Burundi's election hailed the “disappearance of the ethnic factor” was therefore premature at best.
The Constitution stipulated that the first post-transition president was to be selected by the National Assembly and the Senate sitting in joint congress, so this election became a mere formality. The CNDD-FDD held 96 of 166 seats (58 per cent), just five votes short of the two-thirds majority needed to elect the president, so it was the only party to propose a candidate: Pierre Nkurunziza, its chairman. On 19 August, he was elected with 151 votes. Although the vote was secret, this overwhelming support, representing 90 per cent of the Congress, revealed that some members of UPRONA and most or all members of FRODEBU and the CNDD had voted for him.
Although there was some pre-election violence, the polls were generally considered as free and fair, and there was no violent backlash. This outcome, as opposed to the one seen in 1993, can be explained by several factors. Constitutional design played a major role, as the Tutsi elites were now more reconciled to the victory of a “Hutu” party, in what was seen as a competition between the CNDD-FDD and FRODEBU. Indeed, as some Tutsi leaders told me at the time, the 2005 electoral contest “was not their concern, but that of the Hutu.” The fact that pre-election violence occurred mainly between these two “Hutu” parties demonstrates how the design contributed to a marked decrease in the salience of ethnicity. But the peaceful outcome of the elections also has other endogenous and exogenous explanations. In contrast to 1993, most leaders of the political parties and civil society in 2005 knew first-hand the cost of civil war and so were willing to accept compromise to avoid its resumption. These leaders also learned during the transition period to work together and acknowledge the “other” ethnic group's concerns. The hate media, so prevalent in 1993 and during the following years, had given way to (in part internationally funded) peace media. Finally, regional leaders and the international community played crucial roles. In particular, the preventive deployment of peacekeepers helped integrate former Hutu rebels into Burundi's army, while greatly reducing the risk of a violent backlash from the army's traditionally Tutsi officers and soldiers, thereby paving the way for the implementation of the peace agreement. The seemingly successful transition was therefore due to a unique combination of factors, but it was at the same time fragile, particularly with regard to its democratic component.
Successful Consociationalism, Failed Democracy
The CNDD-FDD consolidated its hold on power rapidly, and it did so by violating principles of good governance in ways that pervade much of Africa: committing financial abuses and engaging in corruption, as well as suppressing the opposition, the press, and civil society (Human Rights Watch 2009, 2010). Lacking experience in running a state, the party reflexively embraced such anti-democratic practices, despite and sometimes in response to criticism from Burundi's urban elites and the international community. Yet, the government retained popularity with its rural constituents by providing social services, such as free primary education and healthcare for the most vulnerable. The former rebels' long experience in the bush had attuned them to the needs of ordinary Burundians.
Domestic politics became increasingly fragmented and partisan, often leading to institutional stalemate. The main schisms were not along ethnic lines, however, but between various “Hutu parties” (and even within them, but also within “Tutsi” parties), often arising more from personal differences than political ones. Howard Wolpe, the late US special envoy to the region, blamed all sides:
The CNDD-FDD never fully transitioned from its status as a rebel group to that of a democratic political party. It has therefore carried its historical paranoia and authoritarian tendencies into its style of governance – intimidating opponents, harassing opposition political parties, torturing its political enemies. At the same time, the opposition parties were simply unprepared to accept their electoral defeat, regardless of the broad international consensus that the elections were legitimate and an accurate reflection of popular sentiment. (Wolpe 2011: 62)
Curtis (2013) finds one of the causes of this illiberal evolution in the contradiction between international rhetoric about peacebuilding, which emphasises liberal governance and inclusive participation, and narrower conceptions of peacebuilding that view its key goal to be the maintenance of stability and control. Thus, encounters between international, regional, and local actors produced governance arrangements where violence, coercion, and militarism remained central.
Despite this failure to develop a culture of democracy, Burundi's last rebel movement, the Forces Nationales de Libération (National Liberation Forces, FNL), did register as a political party on 22 April 2009, belatedly formalising the end of a civil war after more than 15 years. Political activities during the remainder of 2009 and the first half of 2010 focused on the forthcoming second post-Arusha elections – municipal, parliamentary, and presidential – slated for mid-2010. The results of those elections – namely, the relatively peaceful but banal perpetuation of one-party dominance – highlighted again both the capabilities and limits of constitutional engineering. Consociational guarantees had successfully diminished ethnicity as an electoral issue – no small achievement in light of Burundi's history – but failed to produce better governance than in most African countries.
In the municipal elections of 24 May 2010, the CNDD-FDD received 64 per cent support, with voter turnout at 91 per cent. Although international and domestic observer missions noted a number of irregularities, they found the polls, by and large, to be free and fair. The CNDD-FDD won just 2 per cent more of the vote than it did in 2005, rather than winning by a huge margin, which would have signalled manipulation. The geographic distribution of reported votes was also plausible, as other parties won strong support among disgruntled residents of Bujumbura, while the CNDD-FDD enjoyed a landslide in the rural areas that were its base. Nevertheless, the losing parties claimed that the municipal elections were riddled with massive fraud, and a dozen of them established an opposition front called the Alliance Démocratique pour le Changement – Ikibiri (Democratic Alliance for Change – Ikibiri, ADC-Ikibiri), which boycotted the subsequent legislative and presidential elections.
Nkurunziza was thus the only candidate in the presidential race to be determined on 28 June: a direct election by universal suffrage, unlike his indirect election by Parliament in 2005. Nkurunziza won 92 per cent of the vote, but the boycott suppressed turnout to only 77 per cent, a sharp decline from the municipal elections. Voter apathy grew even stronger during the parliamentary elections, on 23 July, when turnout fell to 67 per cent, and the CNDD-FDD garnered 81 per cent of the vote. Two of the other parties that had not boycotted also received some support: 11 per cent for UPRONA, and 6 per cent for FRODEBU-Nyakuri, a splinter of FRODEBU close to the CNDD-FDD. To satisfy constitutional quotas for Tutsi and for women, additional appointments were made by co-optation, resulting in 81 seats for the CNDD-FDD, 17 for UPRONA, 5 for FRODEBU-Nyakuri, and 3 for Twa, totalling 106 seats. Due to the boycott, the CNDD-FDD obtained 76 per cent of the seats, exceeding the two-thirds supermajority needed for parliamentary votes on organic laws. Even more significant, by teaming with the allied FRODEBU-Nyakuri, the dominant party was able to surpass the 80 per cent threshold required to amend the Constitution. Thus, the government now had the power to abolish the consociational guarantees for minority ethnic and political groups that were the basis of the Arusha Accords. The electoral boycott had thus weakened the balances enshrined in the Constitution, backfiring on the opposition. However, the CNDD-FDD was unlikely to tamper with the Constitution's essential guarantees, even though the party did not participate in the Arusha talks. Indeed, the ruling party's own Tutsi MPs and senators would oppose any such change that could reignite ethnic tensions, threaten vital Tutsi interests, and end their political careers.
The outcome of the elections and the opposition's boycott cast a long shadow on political developments over the following years. The CNDD-FDD gradually reinforced its stranglehold on the political landscape, as Burundi developed into a de facto one-party state. The maintaining of its maquis mentality and sense of being besieged by everyone and everything on the outside was proof of its inability to transition from rebel force to governing party. As early as 2005, and strongly reinforced after 2010, the CNDD-FDD put in place a system of patronage and violence that eventually covered the entire country. Several opposition leaders went into exile (though most returned). Civil society and the independent media were increasingly threatened, particularly by the intelligence service (Service National de Renseignements, SNR). Arbitrary arrests, “disappearances,” and extrajudicial executions were on the rise. Incipient rebel movements started operating, mainly from the Democratic Republic of the Congo, but their actions were no genuine threat to the regime. The Imbonerakure, the youth wing of the ruling party, increasingly engaged in violence against supposed opponents, but the latter, too, maintained a capacity for violence. Both the ruling party and the opposition thus contributed to increased militarisation.
Political tensions rose during 2014 in anticipation of the third post-transition electoral round, scheduled for June–July 2015. While the CNDD-FDD's announcement on 25 April 2015 that Nkurunziza would be its presidential candidate for an unconstitutional third term was the proverbial last straw, in reality the political crisis had begun well before then (Thibon 2014; International Crisis Group 2015; Nindorera 2015; Grauvogel and Simons 2015; Van Acker 2015). 2
The conflict around presidential terms limits is discussed by Stef Vandeginste (in this issue; Vandeginste 2016).
Despite international condemnation, the splitting of the ruling party, the violence, the aborted putsch, and the fleeing into exile of hundreds of thousands as a consequence of Nkurunziza's disputed bid for a third term, elections went ahead more or less as scheduled, and the outcome respected the institutional design. Since co-optation, the National Assembly has been made up of 60 per cent Hutu, 40 per cent Tutsi, and three Twa, and over one-third of MPs are women. The CNDD-FDD maintained its dominant position by capturing over 70 per cent of the seats. Nkurunziza was re-elected with almost 70 per cent of the vote, and the two vice presidents along with those elected to other government positions abided by the constitutional requirements. This was thus a remake of the 2010 electoral exercise, only worse, as the potential for violent destabilisation was now far more serious, and as regional dynamics had entered into the equation. Most opposition parties boycotted the polls, though contrary to 2010 they had good reason to do so. The CNDD-FDD was the comfortable winner, claiming institutional “normality” and engaging in “business as usual” even in the face of international sanctions. “Business as usual” included major human rights abuses by the security and intelligence services (see, for instance, Amnesty International 2015).
However, despite international fears of genocide and attempts by some players to inject ethnicity into the conflict, the consociational logic prevailed. Some Hutu and Tutsi sided with Nkurunziza, and some Hutu and Tutsi opposed him. The bi-ethnic army remained by and large united (although this coherence has come under increasing strain), and both Hutu and Tutsi were involved in the attempted coup, which was defeated by other Hutu and Tutsi. As the main divide and the violence remain political rather than ethnic, at least for the time being, the constitutional engineering has proved remarkably resilient. As noted by Vandeginste (2015: 636), the pacification model has “prevented electoral violence from transforming into ethnic violence” (emphasis in original). But this achievement cannot be taken for granted. Just like ethnicity can be engineered away, it can be engineered back in, not per se through institutional means, but by political statements, insidious suggestions on the internet, or the redeployment of Tutsi officers of the former Forces Armées Burundaises (FAB), to quote just some examples of recent worrying developments. 3
Monitoring of ethnically based hate speech by La Benevolencija Burundi found that about 38 per cent of such speech came from public authorities at all levels of the state, 25 per cent from opposition leaders, and 37 per cent from youth (20 per cent of youth were linked to the regime, 17 per cent to the opposition) (La Benevolencija Burundi 2016). A declaration by the CNDD-FDD dated 3 October 2015, reacting to sanctions taken against individuals by the EU, contained nine references to ethnic groups, all suggesting that the EU measures displayed a bias against Hutu.
These dynamics show that while consociationalism has managed the ethnic divide up to now, it has not brought liberal democracy, good governance, or respect for human rights, and it has actually increased intra-ethnic tensions. Indeed, other lasting achievements of the Arusha Peace Agreement and the 2005 Constitution relating to legitimacy and democracy have not been observed. From a country constantly on the brink of massive ethnic confrontation and indeed genocide, Burundi has thus become a banal dictatorship characterised by standard conflicts over power, one that can be found in many African countries. In 2015 alone, Congo-Brazzaville and Rwanda introduced constitutional amendments allowing sitting presidents to run for additional terms, and South Sudan extended the tenure of its president, Salva Kiir, by three years. Ugandan president Yoweri Museveni was re-elected in early 2016 (term limits disappeared in 2005), giving him a fifth term in office, in addition to the 10 years he was in power unelected. A little later, Djibouti's Ismaïl Omar Guelleh ran for a fourth term, after having sworn in 2011 that his third was to be the last. At around the same time, Niger's president, Mahamadou Issoufou, was re-elected after a campaign that his main opponent had to run from his jail cell. But these outcomes are not inevitable, as shown by the recent examples of Nigeria, Benin, and Cape Verde, where opponents carried the day and former incumbents graciously (though sometimes grudgingly) acknowledged defeat.