While ethnic power-sharing has been an enduring institutional feature of Burundian politics over the last two decades, key political and military leaders who decisively marked Burundi’s negotiated transition to peace have died. Pierre Nkurunziza (Hutu/CNDD-FDD) and Pierre Buyoya (Tutsi/Uprona), both former presidents, passed away in 2020. Earlier, former military strongmen Adolphe Nshimirimana (Hutu) and Jean Bikomagu (Tutsi) were assassinated in the aftermath of the 2015 political crisis. Nonetheless, ethnic power-sharing between Hutu and Tutsi has (seemingly) stood the test of time.
So how can we explain the resilience of power-sharing beyond the disappearance of some of its key architects?
What is power sharing?
Power-sharing is a common war termination tool in conflict situations subject to a negotiated settlement (rather than a military victory). Parties obtain a guaranteed share in the exercise of political power, at the level of the executive, the legislature, the security sector and/or other institutions. Some power-sharing agreements are deliberately short-lived, conceived as transitional arrangements paving the way for elections. Other power-sharing agreements, however, also contain a longer-term blueprint of the state’s post-conflict institutions. Thus, in order to remedy old and prevent new grievances, a guaranteed political representation of societal segments is combined with various types of autonomy and veto power arrangements for minorities.
Burundi is a case in point.
For years, the country was the scene of a civil war fought along ethnic lines. In 2000, a peace agreement was signed in Arusha between a group of Hutu dominated parties and a group of Tutsi dominated parties. Quotas to ensure the representation of Hutu (an estimated 85% of the population) and Tutsi (an estimated 14%) were a cornerstone of the peace agreement. And, a unique case on the African continent, the ethnic quotas – at the level of the executive, the national assembly and the senate, but also the police, the army, the local administration and state-owned companies – were included in the constitution.
In a recent paper, we look at the real life implementation of the ethnic quotas in national government over the past two decades (2001-2020). Were ethnic quotas (60% government ministries for Hutu; 40% for Tutsi) respected? And was there an evolution in the allocation of ministerial portfolios?
Burundi’s power-sharing design does not specify which ethnic group is entitled to which ministerial portfolio. This is surprising because not all ministerial portfolios are equally salient and attractive.
We may expect that the ethnic group that, over time, becomes more powerful, grabs the most salient ministerial portfolios (while complying – or not – with the initial power-sharing agreement in terms of the number of ministerial positions). In our paper, we also analyzed whether we see an evolution in the allocation of provincial governor positions, where ethnic quotas do not apply.
Ministerial portfolio salience, a complicated story
Ministerial portfolios are generally divided into three categories—high, moderate and low salience. In the literature, ministerial portfolios that have under their purview hard political matters – diplomacy, defense, security and finance – are considered high salience. We conducted a survey with former ministers and prominent political analysts to define portfolio salience in the specific context of Burundi.
It turned out that ministerial portfolio in Burundi is highly salient in three instances. First, when it concentrates the state monopoly on violence, an important resource for power conquest and conservation. Second, when the portfolio comes with significant international funding, allowing the minister to channel procurement opportunities to political allies. Third, when the ministry has a strong local presence throughout the country, enabling the minister to sustain a network of political clients.
Quotas across critical junctures
In Africa in general, and in Burundi in particular, presidents are the most prominent political actors. We may expect the president’s co-ethnics to enjoy important privileges, including access to the most salient ministerial portfolios. After the Arusha peace agreement, Burundi initially had a Tutsi president (October 2001 – March 2003) and, since April 2003, a Hutu president (first Domitien Ndayizeye, followed by three presidential terms of Pierre Nkurunziza, CNDD-FDD). We used the start of a new presidency as a critical juncture, thus considering five presidential terms as units of analysis. While taking into consideration their number of days in office, all ministers holding a high salience portfolio were coded in terms of their ethnic identity.
The findings are shown in the following graph (note that the figure shows the ratio of Burundi and Hutu individuals, not the percentage of any one group in such positions):
Source: Authors’ own calculations
First, throughout the entire period, ethnic quotas were – to the extent mathematically possible – strictly respected. Second, under the first two Hutu presidencies, the incumbent president attributed less than 60% of the high salience portfolios to co-ethnics, which may seem counterintuitive. We found that high salience ministerial portfolios were used to secure support from the rival ethnic group. Third, from 2010 onwards, we see a clear and growing deviation from the quotas that regulate ministerial portfolios in general (but not the category of high salience portfolios in particular). The control over high salience ministerial portfolios by the president’s co-ethnic took time. It happened only when the risks involved were reasonably marginal. Given Burundi’s 1993 experience with a Hutu presidential election victory being annulled by a coup d’Etat, this was when full control of the army commandment was guaranteed. In 2009, for the first time ever, a Hutu (from the ranks of President Nkurunziza’s CNDD-FDD) was appointed as army chief of staff. In addition, we also found that the allocation of high salience portfolios has served to weaken intra-group cohesion within the Tutsi ethnic group. The Tutsi who were appointed to high salience ministerial portfolios were generally from the northern or central provinces (rather than from southern Burundi, which for decades had dominated political and military institutions) and were mostly members of the CNDD-FDD party, often well connected to influential army generals.
For provincial governorships, our findings are as follows (note that the figure shows the percentage of individuals from each ethnic group holding such positions):
Source: Authors’ own calculations
The increase in the president’s co-ethnics share in provincial governor positions – not regulated by ethnic quotas – is more immediate and more outspoken. The absence of quotas is one explanatory factor. In our paper, we argue that this also due to the expectations ruling elites have in this type of positions. Provincial governor positions are a tool for in-group cohesion. They are carrots for influential subnational politicians whose defection would be detrimental to the ruling party interests.
The grey zone in power-sharing institutional design as a blessing
Our case-study shows that quotas matter for sustaining power-sharing deals. The absence or vagueness of quotas, however, at the same time also contributes to the functionality of power-sharing. Indeed, omissions in Burundi’s power-sharing design enhanced its adaptability, thus prolonging its lifecycle. They enabled Burundi’s dominant political actor to translate the new balance of power in the distribution of the most juicy positions without annulling the power-sharing institutions, which might potentially have re-ignited inter-ethnic tensions.
Reginas Ndayiragije is a PhD candidate & Associate Researcher at the Institute of Develpment Policy (IOB), University of Antwerpen.
Stef Vandeginste (@ArushaExit) is an Associate Professor at the Institute of Development Policy (IOB), University of Antwerp.