Electoral violence and the legacy of authoritarian rule in Kenya and Zambia

Abstract

Why do the first multiparty elections after authoritarian rule turn violent in some countries but not in others? This article places legacies from the authoritarian past at the core of an explanation of when democratic openings become associated with electoral violence in multi-ethnic states, and complement existing research focused on the immediate conditions surrounding the elections. We argue that authoritarian rule characterized by more exclusionary multi-ethnic coalitions creates legacies that amplify the risk of violent elections during the shift to multiparty politics. Through competitive and fragmented interethnic relations, exclusionary systems foreclose the forging of cross-ethnic elite coalitions and make hostile narratives a powerful tool for political mobilization. By contrast, regimes with a broad-based ethnic support base cultivate inclusive inter-elite bargaining, enable cross-ethnic coalitions, and reduce incentives for hostile ethnic mobilization, which lower the risk of violent elections. We explore this argument by comparing founding elections in Zambia (1991), which were largely peaceful, and Kenya (1992), with large-scale state-instigated electoral violence along ethnic lines. The analysis suggests that the type of authoritarian rule created political legacies that underpinned political competition and mobilization during the first multiparty elections, and made violence a more viable electoral strategy in Kenya than in Zambia.

Introduction

The shift to multiparty politics in sub-Saharan Africa in the early 1990s raised hopes for democratization across a continent where most countries had limited experience of competitive elections. The transition gave rise to ‘founding’ elections, viewed as critical junctures on the path to democracy. In some of these elections, as in Malawi and Benin, there were only isolated incidents of violence, while electoral contests were far from peaceful in, for instance, Côte d’Ivoire and Nigeria. Why do the first multiparty elections after authoritarian rule turn violent in some countries but not in others?

We argue that an important step for understanding this underexplored variation lies in the varying strategies chosen by political leaders in how to consolidate political support during authoritarian rule. The institutional setup of the authoritarian regimes across much of Africa looked similar, for example, in the dominance of a single party. Yet, how leaders sought to retain the support of a ruling coalition within the authoritarian framework differed significantly. One salient aspect is to what extent ethno-regional interests were represented in the ruling coalition. A majority of states resorted to some form of ethnic inclusion to accommodate ethnic grievances and undermine regime challengers (Rothchild & Foley, 1988). However, regimes pursued inclusion to different extents, and we argue that the degree of inclusion matters. Specifically, in the context of multi-ethnic societies, leaders that rule by means of a narrower ethnic support base, what we refer to as exclusionary approaches, create interethnic relations characterized by competition and fragmentation. With the introduction of multiparty elections, such dynamics heighten the risk of electoral violence by foreclosing the forging of cross-ethnic elite coalitions and by contributing to more competitive intercommunal relations. In this context, leaders can gain politically by resorting to electoral tactics that rest on hostile group narratives. By contrast, strategies that rely on a more broad-based ethnic support base, here called inclusionary approaches, are more likely to produce interethnic relations that alleviate the risks of violence in founding elections by fostering inclusive inter-elite bargaining and by contributing to more amicable intercommunal relations, thereby lowering incentives for hostile out-group mobilization.

We explore this argument by comparing founding elections in Kenya (1992) and Zambia (1991), ending authoritarian single-party rule in both countries. This comparison poses a puzzle. In both countries, the leaders warned about the dangers of reintroducing multipartyism: Zambia’s president Kenneth Kaunda claimed that electoral competition would propel ‘Zambia into the “Stone Age politics” of ethnic violence’ (Bratton, 1992: 85); while Kenya’s president Daniel arap Moi declared that multiparty elections would ‘usher in tribal conflict and destroy national unity’ (Barkan, 1993: 90). However, while Kenya experienced electoral violence resulting in at least 1,500 deaths, Zambia’s elections were largely free from violence. Our analysis suggests that the different strategies used for building and maintaining a multi-ethnic ruling coalition during the authoritarian era represent one vital component for explaining this variation.

The contribution of this article is threefold. First, the focus on legacies from an authoritarian past advances research on electoral violence by explicating how electoral violence can have deep historical underpinnings. To date, most literature approaches the causes of electoral violence from a short-term analytical vantage point, focusing on the immediate dynamics or institutional context of electoral contests. ¹ Second, we tally with an emerging literature highlighting how institutional choices during authoritarian rule generate legacies that structure political choice when democratic openings occur. ² While previous research recognizes ethnic conflict as a critical challenge to democracy in multi-ethnic settings and unpacks different measures to counter such conflicts, it is less recognized that strategies to address conflict in heterogeneous societies were also prevalent during authoritarian rule. Third, by uncovering dynamics where the electoral contest remained peaceful, we contribute to the existing case study literature, which focuses mainly on violent elections. ³

Motivation

An extensive literature acknowledges the challenges of introducing elections in multi-ethnic societies with limited democratic experience (c.f. Mansfield & Snyder, 1995; Cederman, Gleditsch & Hug, 2013). Elections hold the potential for fueling violence in ethnically divided societies when leaders suddenly become reliant on securing popular support (Mann, 2005; Snyder, 2000), and the ballot box offers no institutional protection for the minority against majority rule. Electoral violence, levied to influence electoral results or to protest announced results, often erupts along ethnic lines as electoral dynamics tilt towards ‘ethnic head counts’ (Chandra, 2004; Kuhn, 2015). Many studies probe how institutional provisions may reduce the risk of violence in multi-ethnic states, such as formal power-sharing, minority guarantees, quotas, and electoral rules that secure broad political representation (Horowitz, 2000; Lijphart, 1977; Reilly, 2001; Sisk, 1996; Fjelde & Höglund, 2016). While this research recognizes the often deep historical roots of politicized ethnicity, it primarily explores institutional engineering within the framework of formal democratic institutions. ⁴ We know less about how the political dimensions of interethnic relations evolve from the period of authoritarianism to electoral democracy, and what consequences they have for violent or peaceful electoral conduct.

Scholarship focusing more generally on regime characteristics and political violence primarily highlights the capacity of authoritarian regimes to repress dissent, including ethnic challengers (e.g. Muller & Weede, 1990). This literature posits that the potential for ethnic violence surged as democratic institutions placed significant constraints on the ability of regimes to coerce multi-ethnic populations into compliance (Mousseau, 2001; Saideman et al, 2002). Yet, the large variation in ethnic violence following the introduction of multiparty rule suggests that ‘removing the repressive lid’ provides only part of the story of how intergroup relations evolve from authoritarian rule to multiparty electoral competition.

Across different authoritarian regimes, significant variation exists in how leaders manage threats and how they stay in power through maintaining a dominant coalition. In multi-ethnic societies, these choices will be structured along ethnic lines. We argue that the choice of ruling strategy leaves interethnic relations in markedly different places at the transition to multiparty politics, producing different baseline risks for founding elections to turn violent. Below we focus on one salient aspect of how autocratic leaders manage interethnic relations: the degree to which various ethnic groups are represented in the ruling coalition. Next, we draw the line from a more exclusionary system to a higher risk of violent multiparty elections.

Violent elections as legacies from authoritarian rule

Research on authoritarianism highlights the challenge of controlling the majority of the population excluded from power, but dictators cannot rule on their own. Instead, political leaders across the spectrum of regime types rely on the maintenance of ruling coalitions, whose support is critical to ensure regime survival (e.g. North, Wallis & Weingast, 2009; Svolik, 2012). In democratic contexts, elections and electoral rules predominantly shape the nature and composition of these ruling coalitions. In authoritarian states, ruling coalitions are largely determined at the discretion of the elite (Svolik, 2012).

Where ethnicity is politically salient, the task of coalition-building tends to focus on inclusion or exclusion of ethnic groups (Wimmer, Cederman & Min, 2009; Bormann, 2019). Across Africa’s post-colonial authoritarian regimes, building dominant coalitions required balancing the need for ethnic inclusion to ensure cooperation, with the risk of facing violent challenges from excluded groups (Roessler, 2016). In these weakly institutionalized states, elites sought to solidify their rule through political incorporation of ethnic groups that could otherwise destabilize the regime. By co-opting ethnic leaders, it was ‘possible to reduce the scale and intensity of their demands’ (Rothchild & Foley, 1988: 233). The ethnic configuration of the state, in turn, reflected elite bargains among a subset of all ethnic groups, agreeing to cooperate to share executive power and ‘rents that come from controlling the state’ (Roessler, 2016: 12).

Strategies to sustain ruling coalitions included the establishment of formal institutions, such as a strong regime party that could maintain a multi-ethnic support base through elections (e.g. Horowitz, 2000: 429). Most importantly, however, regimes rested on informal structures of patronage politics. At the top, regimes accommodated ethnic elites deemed central to their stability through the distribution of public offices and state resources (Arriola, 2013). These ethnic leaders, in turn, acted as brokers of patronage that could mobilize their own ethnic communities (Bayart, 1993; Bratton & van de Walle, 1994). The inter-elite bargains, thus, translated into societal stability through the ethnic intermediaries’ embeddedness in patron–client networks that extend through society as a whole (North, Wallis & Weingast, 2009; Roessler, 2016).

In heterogeneous societies without a majority group able to rule on its own, incumbents depended on the support of the leadership of other ethnic groups to ensure regime survival. Almost all ruling coalitions in Africa involved the inclusion of some ethno-regional intermediaries, but the degree of ethnic inclusion (both the size and ethnic plurality) varied significantly. ⁵ Authoritarian regimes can thus be placed along a continuum depending on the degree of intra-elite accommodation at the center, and incorporation of various ethno-political interests into the ruling coalition (Horowitz, 2000; Rothchild & Foley, 1988). These can be referred to as more or less exclusionary or inclusionary, depending on how broad the ethnic powerbase of the regime was (Horowitz, 2000; Huntington, 1970; Rothchild, 1997).

At one end of the continuum are regimes that Rothchild & Foley (1988) label ‘elite consensual systems’, characterized by broadly incorporative grand coalitions, where most major ethnic groups were allowed some participation in decisionmaking processes, typically through various cabinet or executive appointments to ethnic representatives (Arriola, 2009). Senegal under president Senghor’s one-party regime is a case in point, where the government relied on inclusion of a majority of Senegal’s ethno-regional and religious groups, and no single ethnic group faced discrimination.

At the other end of the spectrum are regimes that effectively monopolized state executive power at the hand of a single ethnic group (Wimmer, Cederman & Min, 2009). Examples at the extreme end include Burundi (1966–88), where the Tutsi minority formed the ruling coalition while the Hutu majority were excluded under the military regime, and the settler oligarchy of South Africa, where the minority Afrikaners governed while excluding the black majority. Beyond this extreme, we find regimes that survived through exclusionary approaches without monopolizing the state. These relied on support from constituencies within the ruling coalition, while simultaneously restricting political activity and resources for excluded groups (Huntington, 1970). Kenya, as the empirical analysis demonstrates, fits this category.

We argue that how leaders formed ruling coalitions during the authoritarian era influenced inter-elite dynamics, the relationship between the broader ethnic groups, and the relationship between ethnic leaders and their ethnic communities. The approaches adopted during authoritarian rule had long-lasting legacies by giving rise to different political incentives, for both the incumbent and opposition, when formal institutions shifted with transitions to multiparty politics.

At the elite level, more inclusionary regimes facilitated the growth of a relatively cohesive ruling class, united by a shared interest in accessing state resources on which their positions depended (Bayart, 1993; Arriola, 2009). By extension of patron–client relations, a larger number of groups perceived themselves as ‘within the ruling coalition’, which promoted intra-elite accommodation and integration. In such systems, clientelism ‘provide[s] the cement by which ethnic identities are amalgamated within the boundaries of a more inclusive political system’ (Lemarchand, 1972: 70). What actually trickled down to the ethnic constituencies was not necessarily more in regimes with inclusive ruling coalitions. However, rejecting the predominance of a single ethnic cleavage, and inter-elite accommodation, still nurtured accommodative policies to manage intercommunal relationships.

By contrast, regimes that monopolized state power at the hands of a single group, or relied on a strategy of narrow elite co-optation, produced elite dynamics characterized by fragmentation and competition. Exclusionary regimes engineered ethnicity through ‘divide and rule’ strategies that induce collective action problems among those excluded from power (Acemoglu, Verdier & Robinson, 2004). Restricting access to state resources and repression reinforced the divide and armed leaders from excluded groups with a discourse of historical injustices, political discrimination, and distrust (LeBas, 2006).

The degree of inclusion in coalition-building fostered not only different legacies for inter-elite dynamics, but also how ethnic communities are mobilized and, in turn, their relationship to each other. The elites’ bargaining power within the regime coalition largely depended on the strength of the constituencies mobilized. Elites therefore had strong incentives to cultivate their clientelist base. Relying on narratives that invoked out-group resentment and a sense of group entitlement was one way of doing this (Straus, 2015: 58). As individuals perceived access to state resources to depend on having a coethnic in power, they had incentives to rally behind their in-group coalition and request access to patronage as members of particular ethnic groups (Bates, 1983; Posner, 2005; Wimmer, 1997). Thus, both elite-driven top-down and community-driven bottom-up processes reinforced the political salience of ethnicity.

These processes were mirrored by similar counter-mobilization efforts from ethnic groups outside the ruling coalition (Wimmer, Cederman & Min, 2009). The competitiveness of the mobilization was higher with more exclusionary coalitions. The stakes increased with the number of groups excluded from power, fostering a perception of politics as a zero-sum game. More inclusionary coalition-building did not feed into competitive mobilization of ethnic groups in the same way. The shares of groups in the regime coalitions were larger, reducing ethnic elites’ political premium from marshalling their own constituents to signal strength.

What were the implications of these characteristics for the transition to multiparty rule? We argue that the political legacy of more exclusionary systems, with fragmented and competitive interethnic relations, made it difficult to counter the risk of violence in the transition, since strategies of ‘divide and rule’ closed the door for broad-based elite bargaining. The exclusionary context also provided fertile ground for politicians to capitalize on ethnic cleavages to solidify their electoral agenda. The strategically induced, competitive mobilization of ethnic groups under more exclusionary regimes fueled a perception of politics as a zero-sum struggle that persisted through the transition. These parameters shaped the perception of both threats and opportunities in the transition to multiparty rule, and thereby the opportunity structures for both incumbent and the opposition in terms of how political influence could be pursued (Kirschke, 2000; McAdam, Tarrow, & Tilly, 2001: 46–47). The legacy from exclusionary regimes allowed hostile ethnic narratives to become powerful tools for electoral mobilization. Elites had large payoffs from resorting to exclusionary appeals and extremist rhetoric focusing on out-group threats, which fed into the violent mobilization of ethnic constituencies (Horowitz, 2000; Rabushka & Shepsle, 1972).

In a polarized electoral environment, violence can take several forms (e.g. Staniland, 2014). First, ethnic rhetoric can legitimize harsh security crackdowns and repressive policing against members of other groups, measures that can be used to harass voters, target opposition candidates, and violently coerce voters along ethnic lines (e.g. Taylor, Pevehouse & Straus, 2017). Second, it can entail the formation of ethnic militia groups resorting to violence to intimidate and demobilize opposition supporters to change the ethnic power-balance on the ground (e.g. Raleigh, 2014). Third, it can manifest as intergroup violence if opportunistic politicians provoke ethnic riots to solidify their own electoral base (e.g. Wilkinson, 2004).

More inclusionary regimes faced transition to multiparty politics with less competitive, antagonistic, and divisive patterns of interethnic relations both at the elite and communal levels. They were thus less likely to experience electoral violence. The opportunity structure engendered by authoritarian rule created more options for elites to forge cross-ethnic coalitions, as well as different incentives for intragroup mobilization. Specifically, the absence of blatant discrimination of certain groups reduced zero-sum perceptions of politics and opened up for interethnic elite negotiation. It also made it easier for elites to close political deals during transitions (Bratton & van de Walle, 1997: 87). Such dynamics reduced leaders’ incentives to resort to ethnic outbidding and exclusionary appeals and made it easier for followers to resist such rhetoric. In turn, the conditions for violent ethnic mobilization were less ripe.

Case selection and sources ⁶

We explore this argument by comparing founding elections in Kenya (1992) and Zambia (1991). These elections display considerable variation in the main outcome of interest: the presence of electoral violence during founding multiparty elections.

Zambia and Kenya share several characteristics that make the comparison relevant. In both countries, the regime introduced single-party rule against the backdrop of escalating ethnic tensions and the ‘inability of multiethnic parties’ to maintain multi-ethnic support (Horowitz, 2000: 429). The two countries had shifted between periods of single-party rule and multiparty elections, with similar electoral systems (single-member plurality), and both are ethnically diverse with Kenya commonly described as having 42 different ethnic communities and Zambia 73 (Dresang, 1974; Posner, 2007). Ethnic groups live relatively segregated geographically and dominate specific regions. How ethnicity nests with electoral politics is the subject of this study; here, it is sufficient to note that ethnic identification is important in both countries. The cases also differ on important accounts (colonial experience, population density, and the land question). We return to these factors in the analysis.

The analysis builds on secondary sources and interviews with academics, politicians, government officials, NGO representatives, traditional leaders, and local residents. We conducted 40 interviews in Kenya (2011) and 45 in Zambia (2014 and 2017). The case analysis evolves over three parts: (1) elite strategies during authoritarian rule and their effect on interethnic relations; (2) the election and outcome; and (3) the implication of authoritarian rule on violent versus peaceful elections.

Kenya’s authoritarian legacy and violent 1992 elections ⁷

In December 1992, Kenya held its first competitive multiparty elections after three decades of single-party rule. While the election created a democratic opening, large-scale violence (state-sponsored, but with clear ethnic connotations) killed approximately 1,500 people and displaced 300,000 (Africa Watch, 1993: 1). The ruling party, the Kenyan African National Union (KANU), emerged as the sole winner, with Daniel arap Moi retaining the presidency.

Politics during single-party rule

Kenya’s first president Jomo Kenyatta established de facto one-party rule in 1969, which continued when Moi came to power in 1978. During this era, political leaders used ethnic coalition-building as a strategy to maintain their powerbase, but access to state power and resources was reserved for only a few ethnic groups. Ethnicity was a key asset in political mobilization and shaped voting patterns (Hydén & Leys, 1972). Overall, Kenyans perceive national politics as a rivalry between Kikuyu, Kalenjin, Luo, Luhya, and Coastal groups (Posner, 2007). ⁸ However, ethnic coalitions are not fixed and shifting alliances characterized the single-party era, where Kenyatta and subsequently Moi sought to promote their own ethnic interests using ‘similar political methods of factional manipulation to achieve these ends’ (Throup, 1987: 34; see also Omolo, 2002).

Exclusionary ethnic coalition-building was not a prevalent trait in the immediate post-independence period. In 1963, KANU won the first free elections, and was initially able to unite all ethnic groups in the ruling coalition. The main opposition party, the Kenyan National Democratic Union (KADU), was co-opted and eventually merged with KANU in 1964. While KANU and KADU had similar political agendas, their support bases were different: KANU drew support mainly from the Kikuyu community, based in Central Province and part of Rift Valley, and some communities in the Nyanza and Eastern Provinces. KADU’s prime ethnic base was the Luhya (Western Province), Kalenjin and related groups (Rift Valley), and some communities from the Northeastern and Coast Provinces (Barkan, 1993). Thus, in the initial post-independence era, a broad multi-ethnic coalition formed the ruling coalition at the center (Throup, 1993).

Nevertheless, the Kikuyu community came to dominate the civil service, and state policies favored the Central Province economically (Barkan, 1993: 87). When Kenya became independent and Kenyatta (of Kikuyu origin) came to power in 1963, expectations were that the Kikuyu would gain a privileged position. Although the independence movement (the Mau Mau) was multi-ethnic, the Kikuyu dominated it and faced repression, regardless of whether they were part of the uprising or not (Throup, 1993). This underpinned a sense of entitlement among Kikuyu, seen in land claims in the Rift Valley made by Kikuyu migrant laborers (Kanyinga, 2009).

Land redistribution, particularly in the fertile Rift Valley region, soon became a major political issue. KADU, supported by white settlers, opposed land-buying programs which would accommodate Kikuyu land claims. In this context, the KANU–KADU merger was an interethnic elite ‘pact’, where ‘in exchange for political power KADU dropped its opposition to settlement schemes in the Rift Valley’ (Klopp, 2001: 477). Kenyatta’s land-buying program led to the migration of many Kikuyu farmers to the region (Harbeson, 1973) and laid additional foundation for land conflict.

After the KANU–KADU merger, the ethnic ruling coalition grew increasingly exclusionary. In the process of consolidating power, the regime ostracized another community, the Luo, and singled them out as second-rank citizens, invoking smear campaigns and marginalization of Luo leaders (van Stapele, 2010: 115). The ethnic divide widened as Kenyatta demoted Vice-President Oginga Odinga (a Lou). ⁹ While the conflict originated in ideological disagreement, it intertwined with ethnicity (Ogot, 2003: 33). In 1969, the government banned Odinga’s party, the Kenya People’s Union (KPU), and one-party rule became de facto. Originally, both Luo and Kikuyu MPs supported KPU, but it became branded as a Luo party and members associated with it faced punishment, including loss of jobs, social rejection, and difficulties in obtaining government loans or licenses (Mueller, 2014: 5).

After Kenyatta’s death in 1978, Moi came to power and ruled in a paranoid and kleptocratic manner (Barkan, 1993: 87–88; Throup, 1993: 385). Moi inherited a state in the hands of patrons that had been loyal to Kenyatta, and, as he was Kalenjin, he lacked support from a dominant ethnic group. To ensure control, Moi raised the level of state patronage transferred through individuals and made access to patronage conditional on loyalty to the executive. These strategies served to increase ethnic salience (Cheeseman, 2006: 79). More specifically, the regime’s powerbase became increasingly exclusionary geographically and ethnically. Moi began to favor formerly ‘disadvantaged groups’, including Kalenjin and allied tribes, while many Kikuyu had to leave the civil service (Kahl, 1998: 113). Public investments were reoriented from the Central Province to Moi’s main support base, primarily the Rift Valley (Throup, 1987: 34). Thus, ‘over time, through reorganizing national alliances and patronage networks to ensure patrimonial control, KANU alienated many within Kikuyu and Luo constituencies’ (Klopp, 2001: 277). However, the Kikuyu remained economically influential, placing Moi in a vulnerable position (Holmquist & Ford, 1995: 177).

A faltering economy gradually undermined Moi’s patronage-based regime and contributed to increased marginalization of certain groups (Mueller, 2014; Throup, 1993). A coup attempt in 1982 (allegedly led by Luo officers, but also involving Kikuyu), further engrained the regime’s repressive approach to dissent (Barkan, 1993; Mueller, 2014). In the wake of the coup, Moi disproportionally promoted Kalenjin officers within the security forces and reinforced the paramilitary police (Roessler, 2005: 213). In this context, a perception of ‘Kalenjinization’ grew and the Kalenjin became associated with the regime’s repressive measures (Lynch, 2011: 140).

The election

The transition to multiparty politics took place in a context of rising inflation and food shortages. In August 1991, a wide coalition of clergy, lawyers, and opposition groups formed the Forum for the Restoration of Democracy (FORD), challenging the political and economic state of the country. The government responded with force and arrested several FORD leaders. The internal pressure, combined with donors’ suspension of foreign aid (Kirschke, 2000), led to legalization of multiparty politics in December 1991.

Moi won the presidential election with 36% of the vote, and KANU secured 57% of the seats in parliament because the opposition was severely divided (Mueller, 2014). ¹⁰ FORD was initially a cross-ethnic coalition, including Luo and Kikuyu. Yet within months, several rival alliances were formed along ethnic lines, where each faction fielded its own presidential candidate. FORD broke into two factions: FORD Asili, led by a Kikuyu (Kenneth Matiba), and Ford Kenya, led by a Luo (Raila Odinga) (Oyugi, 1997: 47–48). Partly, the fragmentation of the opposition was a consequence of Moi’s early strategy to channel patronage via powerful individuals serving as ethnic patrons to secure support. The approach had raised ethnic salience, fostered a perception of politics as zero-sum, and created strong center–locality relations where individual patrons had the capacity to mobilize mass support and deliver their ethnic constituencies (Cheeseman, 2006: 302–303). By being both wealthy and having an established ethnic support base, the main political leaders knew that ‘even if they suffered electoral defeat, the strength of their personal support would ensure that they continued to be major players in Kenya’s political landscape’ (Cheeseman, 2006: 303).

Large-scale violence and coerced displacement accompanied the election. As a direct response to the democratic transition, violence began over a year prior to the elections. It intensified in the months preceding the election and continued after. While people were killed and displaced in several parts of Kenya, the violence centred on the Rift Valley Province (and along its border with Nyanza and Western Province), which saw the largest share of parliamentary seats and where mass support for Moi was crucial (Kahl, 1998: 111–112). Violence was levied strategically to establish KANU dominance, by playing on prevailing land conflicts between the ‘indigenous’ groups (Kalenjin, Maasai, Samburu, and Turkana) and more recent settlers in the region. The violence targeted mainly Kikuyu, Kisii, Luhya, and Luo – groups associated with the opposition. It served to appropriate the promised (land) resources for KANU-supporting communities, to secure the support of the Kalenjin community, and punish opposition voters (Boone, 2011: 1328; Throup & Hornsby, 1998: 195–197).

Moi’s regime portrayed the violence as ‘ethnic clashes’ (Barkan, 1993: 90; Throup, 1993: 390). While the ethnic dimension was clear, the government was involved in fomenting the violence. High-ranking government officials seeking re-election supported the training and arming of militias, mainly Kalenjin (Africa Watch, 1993; NCCK, 1992). Moreover, the regime ordered the security forces not to intervene and perpetrators enjoyed impunity (Throup & Hornsby, 1998: 64). By contrast, revenge attacks targeting the Kalenjin were less organized (Africa Watch, 1993: 2).

Explaining electoral violence

How did authoritarian rule influence Kenya’s 1992 election? The mode of coalition-building and diversion of opposition during single-party rule contributed to electoral violence, via legacies relating to the character of interethnic relations at the elite and communal level. These shaped political elites’ options for political mobilization. Two processes were particularly important.

First, the legacies from authoritarian rule created fragmented and competitive elite relations. The opposition came primarily from outside of Moi’s main circle of power and was initially united by having been excluded from power and state patronage (due to their ethnicity). But it quickly disintegrated along ethnic lines. Because ‘ethnic brokerage’ was central for how leaders mobilized political support, it was neither feasible nor desirable for the opposition to strive for cross-ethnic support (Cheeseman, 2006: 302; LeBas, 2011). In addition, no inclusive elite pact, which could have reduced the risks leaders faced, was on the table (Kirschke, 2000: 396). Instead, Moi continued a previous tradition of divide and rule to secure victory, and ‘introduced new means of coercion to assert political control’, where ethnic militias played a key role (Kirschke, 2000: 397). In this process, the security forces, where Kalenjin officers allegedly had been promoted disproportionately, remained loyal to Moi and did not halt the violence (Roessler, 2005: 214).

Second, legacies from the authoritarian era made electoral rhetoric exploiting ethnic cleavages a viable electoral tactic. ‘Exclusionary ethnicities’ had been entrenched by Moi’s concentration of patronage to his support base and increased stigmatization of other groups (Lynch, 2011: 9; Mueller, 2014: 8). Following this logic, antagonistic, intergroup competition underpinned the electoral rhetoric and tied in with deep-seated land grievances, exploited by politicians to garner electoral support. ¹¹ KANU politicians, seeking re-election, were the prime instigators. The purpose was to punish pro-opposition supporters or prevent them from voting through intimidation or displacement. After land was cleared, it was used as patronage to reward militants and secure electoral support (Boone, 2011: 1328). In this process, the politicians resorted to ethnic outbidding: for instance, at political meetings politicians called for the eviction of ‘aliens’ and ‘foreigners’ (mostly Kikuyu, but also Luo) from the land, so that native ownership (to primarily Kalenjin and Maasai) could be restored (Africa Watch, 1993: 18; Klopp, 2001). These communities, in turn, feared expulsion if KANU lost the election (Kahl, 1998: 109). ¹²

Zambia’s authoritarian legacy and peaceful 1991 elections

Zambia became a single-party state in 1972 when the government banned all parties except Kaunda’s United National Independence Party (UNIP). In October 1991, the first multiparty elections in 30 years were held. Kaunda and UNIP were massively defeated, Kaunda stepped aside, and Zambia experienced a largely peaceful transition to multiparty politics (Bratton, 1992).

Politics during single-party rule

After a relatively peaceful path to sovereignty, Zambia gained independence in 1964 and Kaunda became the first president. ¹³ The independence movement, led by Kaunda and UNIP, started with passive resistance, but later reverted to active resistance including the burning of bridges and government buildings (Larmer, 2011: 40; Rasmussen, 1974). Although regional divisions existed, no particular ethnic group dominated and UNIP was ‘committed to obtaining total support from the whole country’ (Roberts, 1976: 221).

Ethnicity has been important in political mobilization in Zambia and has shaped voting patterns since independence (Osei-Hwedi, 1998; Posner, 2005). At the center, political competition is primarily between four broad ethno-linguistic groups – Bemba, Nyanja, Tonga, and Lozi – but these are all composed of different ethnic groups. Political parties do not fully align with a region or group, but have generally been considered to represent a particular region and the ethnic group dominating it. While Zambians generally perceived UNIP as a Bemba-leaning party at independence, it later became associated with the Nyanja (Posner, 2005: 107–108).

Post-independence, UNIP remained the political powerhouse. However, internal tensions of an ethnic nature ravaged UNIP and in 1966, Lozi speakers left and formed the Unity Party (UP). Factional fighting continued in 1967: the Bemba–Tonga alliance won over the Lozi–Nyanja alliance and Bemba took up many senior government positions. Despite allegations of Bemba hegemony, the government structures remained ethnically diverse and involved other groups (Dresang, 1974: 1609–1611).

Two years after its formation, UP was banned due to agitation for increased autonomy of Barotseland in the west. The ban signalled the government’s increasing sensitivity to competition. To reduce the risk of ethnic contestation, the regime championed the motto of ‘One Zambia, One Nation’, intended to foster national, not ethnic, identity (Burnell, 2005). Politics came to evolve around ethnic balancing: a range of policies, from land distribution to ministerial appointments, depended on balancing ethnic representation rather than on merit (Dresang, 1974: 1615).

Single-party rule was a reaction to challenges arising from within UNIP (Gertzel, Baylies & Szeftel, 1984: 7). In 1971, the United Progressive Party (UPP) emerged as a UNIP-splinter group. UPP was a Bemba-dominated party, led by former vice-president Simon Kapwepwe. UPP threatened Kaunda, as it represented long-term UNIP representatives and drew considerable support from UNIP strongholds like the Bemba-dominated Copperbelt and the north (Burnell, 2001; Posner, 2005). This was in sharp contrast to the African National Congress (ANC), the main opposition party from independence until the introduction of single-party rule, which had its prime support base in the Tonga-dominated south. An increasing number of interparty violence incidents served as a pretext for UNIP to act on single-party rule. In February 1972, the government banned UPP and detained Kapwepwe together with over 100 leading party members. The same year, the regime co-opted the ANC leadership and a constitutional amendment passed in December 1972 made Zambia a de jure single-party state (Gertzel, Baylies, & Szeftel, 1984: 16–17).

During the first years of single-party rule, Zambia was riding high on revenues from copper. Kaunda began to promote national unity based on socialism and Christian values (Burdette, 1988: 77–78). For Kaunda, it was a political necessity to prevent ethnicity from becoming the only political identification, since he was partly an outsider to the ethnic game: he was born and raised in a Bemba-speaking area in northern Zambia, but his parents were from Malawi. ¹⁴

Under Kaunda’s rule, the political elite was structured according to ethnic balancing. ¹⁵ Kaunda deliberately balanced ethnic and regional interests, including political compromises and power-sharing in political, economic, and military realms (Bratton, 1992: 83; Lindemann, 2011). ¹⁶ While his policy of ‘appointing members of all the country’s major ethno-regional groupings to the national structures […] reinforced growing tensions within the party’ (Larmer, 2011: 56), he intentionally deployed it to reduce the risk of any particular group feeling excluded and the political salience of ethnicity (Dresang, 1974: 1609). Furthermore, the regime introduced a rotational system, where public servants would never work in or represent their original constituency, but moved around to different parts of the country. Both these factors contributed to interactions within the political elite based on a culture of cooperation and coexistence, rather than rivalry founded on ethno-regional differences. ¹⁷

In contrast to Kenya, regional favoring was not evident during Zambian single-party rule. ¹⁸ Kaunda combined formal policies (like public spending on education) and informal neo-patrimonial interventions to uphold loyalties across ethno-political elites. These elites became part of a system where cooperation was sufficiently inclusive to foster integration (Burnell, 2005: 119). While economic recession in the mid-1970s increased the population’s disillusionment with the regime, it did not skew regional distribution of state resources, which could have created ethnic grievances among disadvantaged groups. Instead, the different regions all took a reasonable part of the losses (Burdette, 1988: 124).

During single-party rule, UNIP remained a nationalist party and no major destabilizing ethno-regional political movements existed (Burdette, 1988: 160–161). Still, UNIP faced political dissent and one issue of discontent was alleged underrepresentation by some regions in the Central Committee (Gertzel, Baylies, & Szeftel, 1984: 101–102). Kaunda responded by reshuffling the Central Committee leadership. Although such efforts aimed to reduce imbalance, accusations of bias were common. Kaunda’s early cabinets were perceived as Bemba-dominated, but later considered to favor Nyanja-speakers (Posner, 2005: 98).

Over time, power became concentrated within UNIP’s leadership and increasingly with the president (Bratton, 1992: 83; Gertzel, Baylies & Szeftel, 1984: 102–104). Political participation was encouraged within UNIP, while the party repressed all opposition. Kaunda viewed ethnic polarization as a threat to his power and delegitimized all dissent by characterizing any criticism as tribalism (Larmer, 2011: 65). While Kaunda’s repression was not on par with Moi’s, disappearances or suspicious deaths silenced many political opponents, and expressed opposition often resulted in a state reaction, such as demotion or loss of employment (Burnell, 2005; Phiri, 2006).

In the second half of the 1980s, UNIP further restricted political liberties, for example, by outlawing strikes (LeBas, 2011: 98). After food riots in the 1980s, Bemba were at a higher risk of clampdown by the security forces, but this was mainly due to protest starting in Bemba-dominated towns, not ethnic bias (Lindemann, 2011: 1859; Scarritt, 1993: 269).

The election

The transition originated in the 1980s, when discontent with the regime grew widespread. The trade unions became the main champions of the reintroduction of multiparty politics. Faced with widespread rallies in favor of change, Kaunda agreed to introduce multiparty elections in 1990. The Movement for Multiparty Democracy (MMD) registered as a political party, led by Frederick Chiluba, a Bemba-speaking Lunda from Luapula Province, who gained a prominent role due to his position as trade union leader (Cheeseman, 2006: 306). MMD drew support from many different groups in society, including labor and business, and several top-level politicians left UNIP for MMD (Bratton, 1992: 92; Phiri, 2006: 167). The opposition was able to unite political leaders from all major ethnic groups, and the leaders refrained from appealing to ethnic and regional loyalties (Osei-Hwedi, 1998: 232).

While internal divisions existed within MMD, it remained united. One reason was that ‘senior figures within the MMD had little to gain and everything to lose from splitting with the party’ (Cheeseman, 2006: 306). UNIP’s power had largely rested on indirect support from urban workers, which gave the unions a prominent position in Zambian politics. When the labor unions turned against UNIP in the late 1980s, individual political leaders held neither the networks nor economic power for mass mobilization (Cheeseman, 2006: 78).

Although the threat of violence was looming during the transition, the elections remained relatively peaceful. There were allegations from both parties about threats and intimidation, but only a few instances of violence occurred, such as youth groups harassing traders who did not have UNIP member cards. The regime deployed paramilitary forces to townships, but there was no real attempt to use the security forces for UNIP’s purposes (LeBas, 2011: 217–218).

Much of the tension ahead of the election concerned disagreement over electoral rules and monitoring. Furthermore, the opposition accused the government of misusing state resources in the campaign (Bjornlund, Bratton & Gibson, 1992; Bratton, 1992: 90). While thousands of voters were excluded from the voters’ registry, these flaws were caused by mismanagement, rather than deliberate manipulation (Rakner & Svåsand, 2005: 91).

The 1991 election resulted in a landslide victory for MMD. Chiluba gained almost 76% of the presidential vote and MMD 125 of 150 parliamentary seats (Burnell, 2001: 240). Fears that Kaunda, who controlled the security forces, would use the army to remain in power existed, but he handed over power peacefully to MMD. ¹⁹

Explaining the absence of electoral violence

The political legacy from the period of single-party rule shaped political mobilization ahead of the 1991 elections and had implications for how the transition unfolded. Kaunda’s deliberate policy to prevent ethnic dominance – politically, economically, and socially – was essential in countering ethnically based electoral violence at the re-introduction of multiparty politics.

First, ethnic balancing during authoritarian rule fostered a culture of cooperation within the Zambian elite, rather than rivalry based on ethno-regional differences. Not only was elite bargaining common, it was also inclusive by involving elites from a broad set of ethnic groups (Burnell, 2005; Lindemann, 2011). The legacy of elite cooperation had been particularly strong during political crises. In 1991, Chiluba and Kaunda met in a series of church-led meetings (Bartlett, 2000) and followed a longstanding tradition of elite negotiations during turbulent times. These meetings settled many contentious issues. Most importantly, the incumbent and the opposition leader agreed to concede to electoral defeat (Bjornlund, Bratton & Gibson, 1992: 413).

Second, inclusionary governance during single-party rule made interethnic relations less fragmented. This contributed to the opposition’s ability to form a broad-based coalition across ethnic divides, with leaders united in their interest in removing Kaunda from power, not an interest in placing their ethnic group in power (Osei-Hwedi, 1998: 232). In fact, 20 MMD candidates were ‘former or sitting UNIP MPs and 12 had been cabinet ministers or central committee members’ (Baylies & Szeftel, 1992: 83). Post-independence state policies that had created well-organized and centralized labor movements composed of different ethnic identities also contributed to the broad-based cross-ethnic opposition (LeBas, 2011: 246).

Third, the interethnic relations in place lowered incentives for violent ethnic mobilization. The political rhetoric did not revolve around exclusionary ethnic language, but widespread discontent with Kaunda and a need for change. Although the rhetoric was hardline, neither Kaunda nor MMD appealed to regional or ethnic loyalties, and both sides drew cross-ethnic support (Lindemann, 2011: 1860; Osei-Hwedie, 1998: 232). Kaunda’s history of ethnic balancing, combined with his own fluid ethnic identity, meant that it was difficult for him to draw on ethnicity to consolidate electoral support. In fact, ‘[b]oth UNIP before the 1990s and […] MMD […] in the 1990s were successful in polling support in every province and from across the entire ethnic spectrum’ (Burnell, 2005: 115). Thus, violent ethnic mobilization would have been a complete break with history, and it would have been difficult to win the elections with more narrow ethnic appeals. ²⁰

Divergent outcomes of electoral violence

Why did Kenya’s founding multiparty election turn violent, while Zambia’s did not? Whereas the transitions in Kenya and Zambia diverged on many dimensions, Zambia also had a potential for violence. The prelude to the reintroduction of electoral politics faced several crises, and Kaunda (like Moi) warned that political liberalization would have violent consequences. To understand the different outcomes, our analysis highlights how the regime in each country used varying strategies to forge ruling coalitions during authoritarian rule. While Kenya’s single-party rule contained a relatively exclusive approach, based on a narrow support-base and active suppression of those excluded from power, Zambia’s single-party rule was more inclusive, based on a broader ethnic support base and with deliberate efforts to counter ethnic divisions. While Kenya was not at the extreme end of the exclusionary spectrum, it was significantly more exclusionary than Zambia. Importantly, the mode of coalition-building was also underpinned by different logics: ‘[w]hereas in Kenya political life has been characterized by open deployment of ethnic self-interest, in Zambia it has been more covert and intermingled with other forms of allegiance – for example, on material grounds’ (Larmer, 2011: 274). Both approaches were strategies for regime survival and help to explain why electoral violence occurred in Kenya, but not in Zambia. In particular, the political legacies from authoritarian rule work through two main pathways.

First, legacies structure options for cross-ethnic coalition-building and cooperation, making electoral violence more or less likely. By emphasizing ethnicity over other political cleavages, exclusionary coalition-building engenders interethnic relations characterized by fragmentation and competition. Thus, while Zambia’s opposition drew support from all ethnic groups, Kenya’s opposition was fragmented and polarized along ethnic lines. Furthermore, in Zambia the legacy of cooperative interethnic elite relations reduced the perceived risks associated with the transition and enabled inter-elite bargaining that solved several contentious issues.

Second, political legacies place constraints on electoral mobilization. Specifically, more exclusionary ruling coalitions create ethno-political dynamics where narratives exploiting ethnic cleavages become a powerful mode of political mobilization for elites in the pursuit of electoral gains. LeBas (2006: 422) notes how ‘political incentives must align with existing identities, rather than encourage the formation of identities that transcend them’. Polarization and group boundaries are upheld and reproduced by political elites and create situations where space for bargaining shrinks (McAdam, Tarrow & Tilly, 2001). In Kenya, the electoral rhetoric played on historical injustices and ethnic divisions, and the violence served both to solidify the incumbent’s support base and to punish opposition voters. In Zambia, the use of an ethnically hostile rhetoric would have signalled a break from the rhetoric used during authoritarian rule.

A key mechanism for explaining electoral violence, thus, is the propensity of exclusionary regimes to cultivate perceptions of politics as a zero-sum game. When such perceptions outlast the authoritarian era and transfer into the electoral arena, an elevated sense of threat fuels electoral violence.

Developments prior to single-party rule clearly influenced politics in Kenya and Zambia. Without victimization of one ethnic group in Zambia’s independence struggle the policy of ethnic balancing was buttressed, whereas the Kikuyu victimization during Kenya’s liberation laid the foundation for exclusive ethnic politics. Relatedly, appeals to land grievances rooted in Kenya’s colonial and post-colonial policies, less prominent in Zambia, played a significant role in the 1992 violent elections and have contributed to a territorialization of ethnicity (Jenkins, 2012; Klaus, 2020). Such historical developments explain why the Rift Valley was the epicenter of the violence (Boone, 2011; Kahl, 1998).

Conclusion

In this article, we show how political legacies from distinct authoritarian regimes contribute to explaining violent versus peaceful multiparty elections in multi-ethnic states. We argue that policies enacted during authoritarian rule to uphold the regime’s powerbase shape interethnic relations at the elite and communal levels, and their interaction. Different ethno-political dynamics following from more exclusionary versus inclusionary approaches influence the mobilization strategies leaders adopt when competitive, multiparty elections are (re)introduced. Our conclusions endorse research underscoring the significance of historical grievances in violent electoral mobilization (e.g. Klaus, 2020; Klaus & Mitchell, 2015), and complement studies that emphasize the importance of perceptions of politics as zero-sum (e.g. Fjelde & Höglund, 2016), leaders’ threat perceptions (e.g. Hafner-Burton, Hyde & Jablonski, 2014), and political elites’ instrumental use of ethnic cleavages (e.g. Berenschot, 2020; Wilkinson, 2004). While these studies focus on immediate conditions surrounding the elections, we highlight how historical legacies shape political mobilization in the era of multiparty politics by creating different ethno-political incentive structures during the transition.

For future research, key topics are the mechanisms explaining self-reinforcing tendencies of conflict, and how previous experiences with violence shape future electoral politics. In this regard, it is pertinent to analyze how legacies from founding elections influence subsequent elections. It is noteworthy that electoral violence has been a pervasive feature in most Kenyan elections after 1992, while Zambian elections were largely free from violence after 1991 and until 2015. Zambian politics has become more volatile, with instances of electoral violence in 2016 (Goldring & Wahman, 2016). Yet, violence remains at a low level with few fatalities.

This analysis is rooted in the historical and contemporary realities of sub-Saharan Africa, marked by the colonial and post-colonial experience, and characterized by a heterogeneous ethnic landscape, pervasive patronage politics, and weak formal institutions. The findings carry particular relevance for countries in the region. However, in other parts of the world electoral regimes face comparable challenges due to high social fragmentation and past authoritarian rule. Thus, the political dynamics created by prior governing strategies are also likely to be of consequence for electoral violence in unconsolidated regimes in the Middle East and Asia.