ACPED also represents an improvement on existing data through including multiple forms of identity on top of ethnicity, namely regional background and political affiliation. Previous studies have shown that the importance of ethnicity varies across contexts (Bratton et al., 2012; Cheeseman and Ford, 2007; Wahman, 2017; Basedau et al., 2011). This variation means that other cleavages, such as regional background or political party affiliation, can take precedence in the regime’s calculations of which elites to include the government.
Burundi illustrates how different types of cleavage become important over time. The Burundian Civil War (1993-2005) pitted a Tutsi-dominated government and military against a collection of competing Hutu insurgents (Colombo et al., 2019). During this conflict, the international community and
reformist Tutsi leaders attempted to end the conflict through incorporating more Hutus into the government (Minorities at Risk, 2003). Nevertheless, the Tutsi community continued to occupy nearly 50 percent of the cabinet in spite of making up only 13.6 percent of the population (see figure 3.6). The war ended in late 2005 when Hutu former-rebel, Pierre Nkurunziza, was elected president by the two chambers of the Burundian parliament. Figure 6 shows a definite increase in Hutu cabinet ministers from 2006 onwards.
Figure 3.6: Burundi Cabinet 1997-2017 Ethnicity
With the movement to multiparty democracy, the Nkurunziza regime faced electoral competition from predominantly Hutu opposition leaders (Colombo et al., 2019). The country was plunged into crisis after Nkuruziza argued that he should stand for a third term in the 2015 elections as his first term had been decided by parliamentary, as opposed to a popular, vote. Many of the figures opposing
Nkurunziza’s decision, including a general who launched an attempted coup against the regime, were Hutu (Africa Confidential, 2015).
Figure 3.7: Burundi Cabinet 2006-2017 Hutu Ministers Regional Background
Faced with an increasing level of political agitation from Hutu co-ethnics, Nkurunziza increasingly appointed ministers from his home region of Ngozi, an area of reliable political support (Africa Confidential, 2015a). Figure 3.7 shows the percent of Hutu ministers who come from Nkurunziza’s home region. A similar situation occurred in Rwanda where president Habyarimana concentrated power in the Akazu, an inner circle of Rwandan elites predominantly from his home region (Human Rights Watch, 1999). This regional aspect of changes in cabinet make-up would not be revealed by the FRT dataset.
Political party affiliation is another form of relevant identity which is overlooked in current datasets. The role of political parties in elite power sharing varies significantly across different countries in Africa. In the post-independence era, many African states were run by autocratic single parties which acted as a means to integrate an ethnically and ideologically varied elite (Van de Walle, 2007). Some of these parties – such as Tanzania’s Chama Cha Mapinduzi (CCM) or the Botswana Democratic Party (BDP) – have ruled their countries for around half a century. Since the widespread adoption of multiparty elections, some parties – both ruling and opposition - have drawn on specific ethnic and regional clusters, despite many countries imposing bans on parties which promote an
ethnic/religious/regionalist agenda (Cheeseman and Ford, 2007; Wahman, 2017; Bogaards et al., 2010). The alignment between ethnic identity and political support has been prominent in Kenya,
Ghana, Ivory Coast and Sierra Leone (Langer, 2005; Posner, 2007; Jockers et al., 2009; Kandeh, 1992). Figure 3.8 illustrates these two different interactions between ethnicity and political party.
Figure 3.8 - Sierra Leone and Tanzania – Intersection of Political Affiliation and Ethnicity
In Sierra Leone, the Sierra Leone People’s Party (SLPP) is perceived as a party representing Mende interests while the All People’s Congress (APC) draws support from northern groups such as the Temne and Limba (Kandeh, 1992). Both parties have ruled Sierra Leone at different points covered by the ACPED data.19 Though neither APC nor SLPP appointed ministers from a single ethnic
constituency, 40 percent of the minsters when the SLPP was in charge were Mende although the group accounts for approximately a third of the national population. Similarly, nearly half of the ministers belonging to the APC led regimes are either Temne or Limba compared to 42 percent of the population. Conversely, in Tanzania the ruling CCM functions as a broad-based coalition, drawing from a wide range of ethnic groups without any particular patterns of ethnic dominance.20 This
difference in party composition influences the dynamics of etho-political competition within the two
19 An SLPP party member occupied Sierra Leone’s presidency from 1998 to 2007, while an APC candidate
occupied the presidency from 2007 to 2017.
20 The exception may be the high representation of Zanzibari ministers in the ruling party, perhaps reflecting the
states. While in Sierra Leone ethno-political competition happens during the main elections between parties, in Tanzania ethnic competition happens within the party during electoral primaries (Egboh and Aniche, 2015; Kandeh, 1992).
The heterogeneity of political parties in the cabinet can also be used to estimate the ruling party’s dominance over the political landscape. While Tanzania and Botswana are politically dominated by a single ruling party, states such as Kenya are ruled by multi-party coalitions of convenience liable to internal rupture over differences in policy, questions of leadership and disagreements over the division of power (Lynch, 2006; Arriola, 2013). In highly contested states - such as Central African Republic, Zimbabwe or Sudan - power is divided among regime and opposition parties or armed movements either as externally mandated unity governments or strategic alliances formulated by the leader (Bariagaber, 2013; Cheeseman, 2011).
Current datasets rely purely on ethnicity to approximate regime strategies of elite management, limiting their capacity to examine how regime survival strategies affect multiple dimensions of representation. Consequently, ACPED represents a major improvement on currently available data.