Power Sharing and Coups d’état in Postconflict Settings: Evidence From Burundi and Guinea-Bissau

Power Sharing and Incumbent Motives

First, power sharing can trigger fears about a decline in status among incumbents that increases the expected payoff to a coup relative to the status quo. While power-sharing provisions take several forms, those that divide executive or military power provide all parties the ability to veto state policy (Nomikos, 2020). Because such resources are typically finite, the allocation of some to former rebels reduces those available to incumbents. Fears about a decline in status are likely to be particularly acute in the context of ethnically divided states, where dominant groups often react violently to dismantling existing systems of privilege. As Cederman et al. (2013, p. 62) put it, “the shock of demotion is likely to trigger strong emotional reactions that spill over into violence to ‘reverse a reversal’” in status. While incumbents thus may agree, in principle, to share power, they regularly work to prevent power-sharing provisions from being implemented (Hoddie & Hartzell, 2003). ¹ In this context, a successful coup enables incumbents to forestall an anticipated reduction in their own power and resources.

Because the military serves as a particularly important source of employment and patronage (Harkness, 2018; Horowitz, 1985), incumbent soldiers throughout the ranks of the military may seek to block the entry of former rebels. Unless the size of the military is to be expanded, integrating rebels requires the demobilization of some government soldiers (White, 2020). Demobilization can result in a loss of income for individual soldiers as well as have broader economic consequences for the communities that soldiers help support (Harkness, 2018). Integrating rebels at higher ranks can also curtail promotions and generate resentment among existing soldiers forced to watch former enemies “leapfrog” them through the ranks (White, 2020). Yet power sharing within the executive branch can also result in a loss of status among incumbent elites. As Ottmann (2020, p. 618) describes, both executive and military power sharing provide elites “direct access to resources in order to maintain their armed forces and reward their constituents.” The division of ministerial portfolios and other positions in the executive branch reduces access to these resources. Of course, power sharing will not always trigger coup attempts from incumbents. Even where grievances are widespread, coup attempts remain risky—roughly half of coup attempts fail (Powell & Thyne, 2011). However, power sharing can create new grievances among incumbent elites that can make this risk seem worth it.

Power Sharing and Rebel Opportunity

Power sharing also creates new opportunities for coups by former rebels. To see why this is the case, it is important to understand how coup attempts unfold. In some cases, high-ranking military or civilian elites may be able to stage a coup without the use of force or visible movement of troops. Most of the time, however, coup plotters attempt to capture the executive and other high-ranking members of the regime; symbolic centers of political power, such as the presidential place or legislature; and broadcasting facilities needed to publicize their actions (De Bruin, 2019, 2020; Luttwak, 1979; Singh, 2014). In doing so, the goal is to convince others that victory has, in fact, already occurred, and thereby head off potential resistance (Geddes, 1999; Singh, 2014). To succeed, coup plotters must have information about the location of key coup targets, physical access to them, and coercive power to secure them. Coup plotters must also be able to anticipate and head off potential resistance to the coup.

Power sharing provides former rebels with the information, access, and—in some cases—additional coercive resources that facilitate the seizure of coup targets. Former rebels integrated into the executive branch typically live and work in the capital, providing them information about and access to symbolic centers of political power. Positions within the state also provide crucial information about sources of potential resistance to the coup and how to head them off. In the case of military power sharing, former rebels with the information and access to the regime acquire coercive power as well; as Harkness (2011, p. 7) describes, military integration “creates (at least temporarily) a dangerous spoiler problem by doing exactly that which it is designed to do: grant significant military capabilities to former combatants and their leaders.” Such integration is not a prerequisite for staging a coup; after all, potential coup plotters without positions within the state apparatus can recruit co-conspirators within it. However, integration into the state increases the resources available.

The idea that power sharing facilitates coups by former rebels is perhaps an obvious one to many rulers. After all, in many cases, rulers excluded rebels or the populations they represent from the state precisely because their inclusion was seen as constituting a coup threat. As Roessler (2011, 2016) has shown, a common coup-prevention strategy in Africa has been to use ethnicity as a shortcut to identify who might be disloyal—and to then to exclude non-coethnics from the military and other state institutions to limit the threat of a coup. Where power-sharing provisions are implemented, the risk of a coup from those previously excluded thus returns.

Addressing Counterarguments

It is possible that power sharing affects former rebels and incumbent elites in additional ways. For example, where fewer rebels are integrated into the state than initially anticipated, or where power sharing locks some rebels out of power, it may generate new grievances among rebels, in addition to those among incumbents, which motivate coups. However, as the large literature on power sharing emphasizes (e.g., Gent, 2011; Jarstad & Nilsson, 2008; Nomikos, 2020), power sharing is designed as a substantial concession to rebels previously excluded from power—one that should reduce grievances overall. Moreover, to the extent rebel grievances stem from exclusion from the state, those rebels that are dissatisfied would also face much greater obstacles to coordinating a coup than those that are not. It is also possible that where power sharing proceeds, it also influences the ability of incumbents to stage a coup. For example, where rebels have been integrated into the state, they may be better positioned to intercept incumbent coups. While plausible, this argument would suggest fewer coups occur once power sharing has been implemented—an expectation at odds with the empirical record. ² For these reasons, I expect power sharing primarily to affect rebel opportunity and incumbent motive.

While I have argued that power sharing creates new incentives for coups, another potential counterargument posits that power sharing instead reduces the salience of political and ethnic divisions—and thus the impetus for coups as well. However, integration into the state does not guarantee that former combatants are reconciled. Examining 11 cases of military integration, for instance, Krebs and Licklider (2016) find power sharing often further entrenched divisions. Alternatively, it could be that where rebel victory appears imminent, incumbent elites may welcome power sharing as a means to preserve some of their status. Yet even in these circumstances, the expected payoff to a coup is likely to remain larger than that of the power-sharing status quo. As Bell and Sudduth (2015) find in their analysis of coup attempts during the civil war, the prospect of a rebel victory tends to prompt incumbent elites to stage riskier coups rather than to refrain from them. A final possibility worth considering is that both power-sharing provisions and rebel coups are driven by another factor, such as rebel strength; some research, for instance, has found that power sharing is more likely as the relative strength of the rebel group increases (Gent, 2011). While this possibility cannot be ruled out, it is one that can be examined in the case studies that follow.

Summary of Expectations

In sum, I expect power-sharing provisions to provoke at least two different types of coup attempts. The first are perpetrated by incumbent elites, for whom power-sharing agreements create fears about a decline in status that provides new motives for coups. The second are perpetrated by former rebels, for whom power sharing provides resources that increase their ability to stage coups successfully. The different causal mechanisms linking power sharing to coups need not be conceived of as competing ones; indeed, it is quite plausible that power sharing represents a compromise that neither side is fully satisfied with. However, the discussion above does suggest that coups by incumbents will be most likely to occur prior to or during the implementation of agreement provisions, to prevent full implementation and a resulting loss of status, while coups by former rebels are more likely to occur after power-sharing provisions have been implemented, when former rebels have access to new resources. It also suggests that coups by former rebels will be rarer in part because incumbent elites frequently succeed in blocking the implementation of power-sharing provisions that would bring rebels into the state. ³

The 2001 Coup Attempts in Burundi

The ethnic composition of the government has been a source of conflict in Burundi since its independence in 1962. Although in the minority, Tutsis had effectively ruled Burundi for all but 4 months since then. With political power also came a disproportionate share of economic resources. At independence, Burundi’s army was largely Tutsi, although there were also some Hutu officers. After a failed coup by Hutu officers in 1965, Hutu military and political leaders were executed and purged from the military. This, in turn, provoked an uprising in central and northern Burundi. Growing divisions between the king and army led to another coup attempt in 1966, by Tutsi-Hima officers, which brought Michel Micombero to power (Harkness, 2011; Lemarchand, 1996; Loft & Loft, 1988).

Military governments in the 1970s and 1980s maintained Tutsi political and economic dominance. Micombero established a single-party state, abolished the monarchy, and purged remaining Hutu officers from the military, replacing them with Tutsis. A successful coup ousted Micombero from power in 1987 (Loft & Loft, 1988). In the decade that followed, Burundi’s economy grew, but Tutsis continued to hold positions of political, social, and economic privilege. The instability of global prices for coffee and tea, combined with a balance of payment deficit, threatened the country’s modest economic progress. In the 1980s, the International Monetary Fund imposed a structural adjustment program that led the government to cut spending (Watt, 2008). In 1987, Major Pierre Buyoya seized power in a coup. Under increasing international pressure, he began a series of democratic reforms. Another rebellion broke out in 1988 (Loft & Loft, 1988).

That same year, Burundi began to experiment with power sharing. In the face of increasing domestic unrest and international pressure, “Buyoya believed that a gradual extension of patrimonial privileges and access to the state was a way to maintain control” (Curtis, 2013, p. 82). He reshuffled his cabinet, appointing an equal number of Hutu and Tutsi ministers. Burundi’s first multiparty elections, in June 1993, brought Melchior Ndadaye, the Hutu founder of the newly formed Front for Democracy in Burundi (FRODEBU), to power (Reyntjens, 1993). Yet many Hutus were concerned that the military and other government posts remained dominated by Tutsis. Soldiers from the army’s second commando battalion staged an abortive coup attempt, but it failed. Under competing pressures, Ndadaye appears to have believed that “limited power sharing was a pragmatic response to his vulnerable position vis-à-vis the dominant Tutsi elite and military, on the one hand, and the new FRODEBU politicians and supporters with their high expectations on the other” (Curtis, 2013, p. 82). He announced plans to reform the Tutsi-dominated military and called for the rapid promotion of Hutus through the military ranks.

Power-sharing proposals were not well received among Tutsis inside or outside the government. As Samii (2014, p. 215) describes, many portrayed “Tutsi control over the armed forces as a necessary protection against ‘genocidal’ or ‘revolutionary’ tendencies among the Hutu masses.” Power sharing also threatened to undermine their privileged economic positions. In 1993, Tutsi officers staged another coup that killed Ndadaye—a “watershed event,” which undid Burundi’s brief period of interethnic consensus (Lemarchand, 1996, p. xiii). In response, Hutu militias perpetrated massacres of Tutsis, which were followed closely by indiscriminate killings of Hutus by Tutsi army units (Des Forges, 1994, pp. 206–207).

In the civil war that followed, which ran from 1993 to 2000, some members of the FRODEBU government split off to form a National Council for the Defense of Democracy (CNDD), along with its military wing, the Forces for the Defense of Democracy (FDD). As Samii (2014) describes, the dismantling of the Tutsi-dominated army was a central aim of the rebellion. International efforts to mediate began almost as soon as the conflict itself did. In April 1994, the UN Special Representative to the Secretary-General for Burundi spearheaded an effort to broker a power-sharing agreement between FRODEBU and the opposition. A series of agreements were reached in 1994 that called for executive power sharing. However, ethnic militia, including Tutsi militia in the capital, repeatedly spoiled the peace process (Curtis, 2013). Violence between opposing ethnic groups occurred frequently between 1994 and 1996 (Salehyan et al., 2012). In July 1996, Buyoya seized power again. During his second tenure, he brought Tutsi militia under control, in part by incorporating them into the national army. However, the economic situation deteriorated. In 1998, economic pressures propelled Buyoya to return to the negotiating table, and another peace process began in Arusha, Tanzania (Watt, 2008).

The Arusha Accords, signed in 2000, provided for a rotating presidency and multiparty elections, along with integration of the military to bring it to a 50–50 Hutu–Tutsi split. The agreement explicitly described power sharing as necessary “to prevent acts of genocide and coups d’état” (Arusha Accords (2000), Protocol II, chapter 1, article 11). War criminals and prior coup plotters would be excluded, but remaining members of the rebel factions participating in the Arusha process could join the military. At the same time, some portion of both the army and the rebel forces were also to be demobilized (Samii, 2014; Watt, 2008).

Tutsi political and social leaders remained staunchly opposed to efforts to share power. While the FDD had important successes on the battlefield, several prominent Tutsi politicians believed that a military that had not been more thoroughly defeated should not have to reform. In February, Tutsi extremist parties organized a rally against peace talks (Salehyan et al., 2012). At the same time, the two largest Hutu-led rebel groups thought the reforms did not go far enough and refused to sign. There was widespread concern that the allocation of government positions was a result of “elite office trading” rather than a reflection of popular grievances and concerns (Curtis, 2013, p. 85). Rumors of a coup circulated during negotiations and in the run-up to implementation of the power-sharing provisions. While some former rebels were successfully integrated into lower ranks of the military, there was strong opposition to integrating them at higher ranks (Daley, 2006, p. 675).

Eventually, Tutsi officers staged two coup attempts with the aim of blocking integration into the officer coups. Concerns that implementing power-sharing provisions would result in a decline in status were a central motivation for both coup attempts (Africa Confidential, 2001; Chin et al., 2021; Harkness, 2011). During the agreement negotiations, Buyoya had worked to “actively reassure the Tutsi community, indicating that, under no circumstances, will the Arusha accord be implemented, or the reform of the army or judicial system be initiated” (International Crisis Group [ICG], 2001a, p. 17). Despite these reassurances, however, the accord was being implemented. Within the army, one army officer voiced what were widespread concerns that the government was working to demobilize Tutsis within the military, “throwing them out into the street” to be replaced by “genocidal” rebels (ICG, 2001a, p. 16). Junior officers and enlisted men in particular were concerned about a loss of status (Watt, 2008).

The first failed coup attempt occurred on April 18. Tutsi officers in Burundi, led by Lieutenant Pasteur Ntakarutimana, staged a coup attempt to halt the integration of Hutu rebels into the armed forces. As Chin et al. (2021, p. 368) summarize in their narrative of the coup attempt, “the Tutsi mutineers sought to strengthen and Tutsi-dominated regime and prevent the incorporation of rebel Hutu elites into the regime.” Ntakarutimana’s troops seized control of the national radio and television station in downtown Bujumbura. Coup plotters announced the removal of President Pierre Buyoya, the dissolution of the National Assembly, and the closure of the airport. Some 200 officers from the Higher Institute of Military Cadre moved to join the coup (ICG, 2001a, pp. 13–14). Forces loyal to Buyoya—including the gendarmerie, presidential guard, and 12^th parachute regiment of the army—eventually suppressed the coup (Africa Confidential, 2001; Watt, 2008).

In the wake of the failed coup, negotiations continued. Buyoya announced a new agreement for the formation of a protection force for exiled political leaders who were meant to return and join the transitional government—an agreement that angered many Tutsi elites and soldiers who could not understand why concessions were made without a seize-fire commitment from the rebels. Concern about potential job losses was also rampant. As the ICG (2001b, p. 14) summarized:

the privates cannot understand that they are asked to mobilize against the rebels while at the same time plans are being made to demobilize them and to give jobs to those whom they are fighting. They see themselves as the victims of the reform of the army, while their officers would remain in their posts.

In response to the continued implementation of power-sharing provisions, Tutsi officers in battalions responsible for security in the capital thus staged a second coup on July 23 (Harkness, 2011; ICG, 2001b; Lacey, 2001). This time, coup plotters kidnapped the army chief of staff, while supporters attacked army headquarters and the homes of senior officers. The coup failed when broader support for the failed to materialize. As Chin et al. (2021, p. 368) describe, the July coup was “an attempt to prevent a new transitional government from forming and thus blocking entry of Hutus into the regime.” This is also the interpretation reached by Harkness (2011, p. 23) about both coup; she writes,

initially, the existing government forces allow [sic] integration to move forward—recruiting Hutus into the rank-and-file. Officer integration, however, proved far stickier: initially, army officers blocked the recruitment of Hutus into their ranks and then staged coup failed coup attempts in 2001.

In sum, military power-sharing provisions triggered two coup attempts in Burundi in 2001, as junior officers and enlisted men concerned with protecting their status attempted to block rebel integration into the higher ranks of the military.

The 1999 Coup in Guinea-Bissau

While the 2001 coup attempts in Burundi were staged by incumbents seeking to prevent a decline in their status, the 1999 coup attempt in Guinea-Bissau was staged by former rebels exploiting the resources power sharing provided them. It represented the latest iteration of an intraelite conflict to control the state that began when the country gained independence from Portugal in 1974. Under colonial rule, Cape Verdeans, who were a minority in Guinea-Bissau, were given educational advantages and employed at higher rates. The struggle for independence, led by the African Party for the Independence of Guinea and Cape Verde (PAIGC), was dominated by Guineans; some 80% of those who fought were Balanta in particular. ⁴ In the first years following independence, however, Cape Verdeans continued to dominate top government and party positions. After the death of Amílcar Cabral, the charismatic founder of the PAIGC, his half-brother, Luis Cabral, became Guinea-Bissau’s first president, leading a one-party regime (Ferreira, 2004; Forrest, 1987).

Despite the prominent role that Balanta members of PAIGC played in the country’s independence, few were promoted to higher office. Cabral instead focused on elevating those he perceived as personally loyal (Forrest, 1987). His administration was politically fragile and disconnected from social formations in rural areas (Forrest, 2002). Over time, Cabral became locked into a struggle for political power with the chief of army staff, João Bernardo Vieira, a member of the small Papel ethnic group. In 1980, Cabral proposed a new constitution in which Vieira’s position would be abolished and the president would become the head of the armed forces. In response, Vieira deposed Cabral in a successful coup (Forrest, 1987). Vieira blamed Cabral for the country’s economic weakness and dependence on foreign aid, but these problems did not disappear when he took power (Munslow, 1981).

Once in office, Vieira purged Cape Verdeans from the government and formed a new Revolutionary Council dominated by Balanta and Papel officers (Embalo, 2012). Vieira also promoted Balanta men rapidly through the ranks in the military. At the same time, however, he sought to prevent subsequent coups by restricting military resources. By June 1983, shortages of supplies in the army were rampant and junior officers were threatening a coup attempt. Vieira responded by ordering loyal soldiers to create checkpoints around the capital and granting the armed forces “first call” on food and supplies. Vieira also tried to reduce the role of the military in politics by strengthening the PAIGC as a civilian counterweight (ICG, 2008). Vieira’s efforts provoked a series of coup attempts in 1982, 1983, 1985, and 1987—all of which failed. After each, Vieira purged political opponents, increasingly targeting Balanta officers and government officials. By 1985, most Balanta officers had been arrested, expelled, or executed. In their places, Vieira primarily promoted Papel officers, as well as some Fula and Mandinga (Embalo, 2012).

By the mid-1980s, an increasingly dire economic situation in the county prompted economic reforms, including efforts to abolish state monopolies, cut public spending, liberalize trade, and establish a market-oriented price system. The first structural adjustment program began in 1987. However, it failed to improve welfare for many, and resulted in disillusionment with the regime (Rudebeck, 1990). The second, launched in 1989, was put on hold due to poor performance. There was widespread public opposition to efforts to reduce public service employment and cut defense spending (Zejan & Kokko, 1998).

In 1991, Vieira bent to increasing international and domestic pressure to democratize. A new constitution marked the end of one-party rule. However, the PAIGC remained the only party with sufficient infrastructure and organization to win the country’s first elections in 1994. The military, which had become an important source of resources and status, fought to maintain a central political role (ICG, 2008, p. 10). The size of the army was reduced by nearly half between 1991 and 1997, generating a great deal of resentment within the military (Africa Confidential, 1998; Embalo, 2012).

The civil war in Guinea-Bissau, which lasted from 1998 to 1999, was precipitated by a personal falling out between Vieira and Ansumame Mané, who was army chief-of-staff at the time. In February 1998, Vieira suspended Mané, a member of the Mandinga ethnic group, from his post for allegedly selling arms to separatist rebels in Senegal. While Mané had fought with Vieira against Portuguese rule and helped engineer the coup that originally brought him to power, he had subsequently come out against Vieira’s economic reforms as undermining the socialist tradition of Amílcar Cabral (Forrest, 1987; ICG, 2008). When Vieira moved to appoint a new chief-of-staff in June 1998, troops loyal to Mané seized control of army barracks in Bissau. Two days later, Mané announced the formation of a rebel “military junta” (Herbert, 2003; Levitt, 2012).

In the ensuing civil war, which involved 11 months of fighting, primarily in and around the capital of Bissau, Vieira retained the loyalty of his presidential guard and some army battalions comprised of Papels (Dias, 2013). Most other soldiers, however—an estimated 2,000 out of 3,500 total in the armed forces—defected to Mané (Forrest, 2002). Vieira was forced to call in reinforcements from abroad; within 2 days, 1,300 troops arrived from Senegal, then another 500 from Guinea. Despite these reinforcements, Mané’s forces controlled an estimated 80% of the country by October (Herbert, 2003; Kovsted & Tarp, 1999).

A peace accord, the Abuja Agreement, signed on November 1, 1998, called for political power sharing in the form of a national unity government that would take office the following February. Reflecting the military balance of power, Vieira’s government relinquished half its ministerial posts—including the powerful ministries of defense, economy and finance, and internal administration—to Mané’s rebels. The agreement also called for the freezing of military positions, withdrawal of all foreign troops, and deployment of peacekeepers from the Economic Community of West African States (ECOWAS). It did not include the demobilization of Mané’s forces or a broader military restructuring (Abuja Agreement, 1998). ECOWAS peacekeeping troops were deployed, and a timetable for elections was set. In February 1999, a new interim national unity government was sworn in. The rebels held ministerial posts in the ministries of defense and internal administration; Francisco Fadul, who had been an advisor to Mané, led the government as Prime Minister.

In May 1999, however, the peace agreement broke down. Early that month, several donor countries convened in Geneva, where the government secured US$200 million to support rebuilding. But tensions remained over Vieira’s resistance to dismantling his 600-person presidential guard. As Embalo (2012, p. 266) describes, “instead of disarming his guard as required in the Abuja Peace Agreement, the president had reinforced it with Papel men and his allies,” who also were forming new ethnic militias. When a Guinean warship arrived off the coast of Bissau during the Geneva meeting, rebels were concerned it was delivering additional weapons for the presidential guard and militias (Africa Research Bulletin [ARB], 1999).

On May 6, former rebels that had been reintegrated into the unity government opted to attempt a coup to force Vieira from office rather than reopen the civil war. Under the direction of Mané, the coup plotters seized weapons that had been stockpiled at the airport (Herbert, 2003). Mané reportedly insisted to the ECOMOG commander guarding them that if the presidential guard was not going to be disbanded as agreed, his own troops should have access to an equal number of weapons (Amnesty International, 1999). With weapons in hand, the former rebels surrounded the Presidential Palace and attacked the homes of other senior government officials loyal to Vieira. The presidential guard resisted but was eventually overpowered. Several hundred soldiers and civilians were killed in the fighting. On May 7, Vieira surrendered (ICG, 2008). “Power sharing under the Abuja Agreement,” Levitt (2012, p. 190) concludes, “seems to have reinforced the belief of would-be coup plotters in the Bissauan armed force that violence was the most expedient way too effectuate change.” ⁵ It also provided former rebels with resources in the form of access to centers of political power and information about the whereabouts of weaponry and coup targets.