INTRODUCTION
The power-sharing model for resolving conflict has recently come to prominence, having been employed to ‘end’ post-election crises in both Kenya and Zimbabwe. Minimally, power-sharing involves the construction of a more or less inclusive government that represents a broad range of concerned parties, but may also include provisions regarding the distribution of bureaucratic posts and new rules for the make-up of the security forces and their subsequent management. Supporters of power-sharing identify three main advantages over alternative strategies of conflict resolution. First, by providing all parties with access to power, it offers the shortest route to ending conflict. Second, by allowing for broad participation, power-sharing formulae promise to confer legitimacy on the government and its actions, facilitating a process of reconciliation. Finally, by providing political leaders with incentives to maintain their proximity to power, and hence moderate their stances, it is hoped that power-sharing will pave the way for the effective institutional reform necessary to diffuse the underlying roots of instability (for a more comprehensive overview see Mehler Reference Mehler2009). The Foreign Secretary of the United Kingdom, David Miliband, summarised the broad international support for power-sharing in Kenya when he urged the country's leaders ‘to agree a power-sharing Cabinet that will serve Kenyans effectively in order that the country can move forward and start to deal with the underlying issues that fuelled so much of the violence in the post-election period’ (DfID 2008). The desire of British and American governments to remove Robert Mugabe from power led them to oppose power-sharing in the Zimbabwean case. However, the much reported ‘success’ of the model in resolving the Kenya crisis empowered Thabo Mbeki and his allies within the Southern African Development Community (SADC) to adopt a similar approach, effectively legitimating a strategy that had been a cornerstone of Mbeki's ‘quiet diplomacy’ since the disputed 2002 elections.
The remarkable spread of the model is demonstrated by the range of countries where power-sharing arrangements have now been employed: in Africa alone, some form of inclusive government has been attempted in Angola, Comoros Islands, Burundi, Central African Republic, Congo, Chad, Côte d'Ivoire, Democratic Republic of Congo, Djibouti, Liberia, Mali, Senegal, Somalia, South Africa and Sudan. Moreover, in recent years similar strategies to those utilised in Kenya and Zimbabwe have been advocated to deal with political crises in cases as diverse as Afghanistan, Honduras and Madagascar. The dramatic rise in the popularity of unity governments is problematic because, to date, analysis of the conditions necessary for power-sharing to be effective, and the likely consequences if these conditions do not hold, has been largely superficial. Significantly, there has been little discussion of whether a model used to end civil wars is suitable for dealing with the very different challenges posed by cases of democratic deadlock. Yet the roots of political instability, and the barriers to reconciliation and state-building, are likely to be very different in these two sets of cases. Furthermore, little attention has been paid to variations in how power-sharing deals play out on the ground, and to the crucial question of whether or not power-sharing facilitates reconciliation and reform in the medium to long term. This paper aims to address this shortcoming by tracing the paths to, and impact of, power-sharing in Kenya and Zimbabwe.
VETO PLAYERS AND THE PROSPECTS FOR REFORM
Having first described the broader political context, we focus on explaining the emergence, disposition and strategic importance of the main veto players who emerged during the power-sharing negotiations. Veto players are actors who, through their formal constitutional powers or informal influence, are effectively able to reject policy proposals in a particular field. Following Tsebelis (Reference Tsebelis2002), we distinguish between institutional veto players, such as legislatures and judiciaries, and partisan veto players that operate within them, such as political parties. Because the influence of different actors typically varies across a range of issues, veto-players analysis must begin by specifying the relevant policy area. As this paper is driven by an overriding interest in the ability of power-sharing to deliver the sorts of far-reaching institutional change necessary to advance the process of democratic consolidation, we focus here on actors capable of blocking the reform of the constitution, electoral law, and the security services, leaving broader policy debates to one side. Explaining which actors enjoy such a veto requires us to look at Kenyan and Zimbabwean politics in the longue durée.
Most veto-player analysis proceeds on the assumption that in political systems where more veto players exist, more compromises are likely to be required in order to secure the necessary approval for any given piece of legislation or reform, thus reducing the prospects for radical change. While we agree that the different balance of partisan and institutional veto players in Zimbabwe and Kenya helps to explain the marked variation in the form taken by power-sharing arrangements in each case, our approach diverges significantly from this rather static framework, which typically focuses on identifying the relevant number of veto players at a given point in time, and rarely provides a historical explanation of why certain veto players emerge rather than others (see, for example Keefer & Stasavage Reference Keefer and Stasavage2003). In contrast, we are interested in tracing the historical roots of the key veto players who shaped how power-sharing agreements were implemented in Kenya and Zimbabwe. Having identified the main partisan veto players as the largest political parties, and the main institutional veto player as the military, our aim is to demonstrate the way in which historical patterns of civil–military relations and political competition shaped the strength and disposition of these veto players, and the consequences for the way in which power-sharing played out on the ground.
In Zimbabwe, the stronger barriers to power-sharing resulted from the combination of strong institutional and partisan veto players, which in turn reflected the capacity and willingness of military leaders to block the transfer of political power, and the refusal of senior Zimbabwe African National Union – Patriotic Front (ZANU-PF) leaders to countenance working side by side with their Movement for Democratic Change (MDC) opponents. In turn, the strength and intransigent nature of these actors can only be explained by considering two key developments during the late 1990s. First, Mugabe's increasing political vulnerability encouraged a progressive militarisation of government, which over time earned the security forces an effective veto. Second, ZANU-PF's deeply divisive use of an exclusive ‘patriotic history’ combined with the strategic use of political violence gave rise to intensely hostile elite relations. The ‘opposition's’ monopoly over victimhood, and the military's complicit responsibility in so much of the violence, served to harden political identities and to undermine the potential for common ground between ‘government’ and ‘opposition’.Footnote 1 Consequently, after the signing of the Global Political Agreement (GPA), ZANU-PF refused to make space for new political players, giving rise to the politics of continuity; the more things change the more they stay the same.
In contrast, in the Kenyan case the institutional veto did not exist because the exclusion of the military from politics ensured that any deal brokered by key civilian leaders could be expected to hold. Although the recent election controversy and civil conflict brought elite relations to an all-time low, Kenya's history of relatively inclusive single-party rule together with the complex nature of the conflict meant that there was significantly more intra-elite understanding and trust than in Zimbabwe. Significantly, because all ethnic groups could claim to some extent to be ‘victims’, and because members of all of the main parties stand accused of being ‘perpetrators’, it was in the interest of key veto players on both sides to prevent prosecutions for past crimes. Consequently, the attitude of political leaders was more accommodating than in Zimbabwe, and Kenya's political leaders realised that they could use the unity government established by the National Accord and Reconciliation Act to forge fresh alliances in order to protect their own positions, resulting in a form of power-sharing that we term the politics of collusion.
Despite the very real differences between Kenya and Zimbabwe, there are good reasons to think that the prospects for reform are equally bleak in each case. The politics of continuity in Zimbabwe has been characterised by a lack of progress on constitutional and institutional reform, revealing the sham nature of power-sharing in that context. By contrast, the politics of collusion in Kenya has resulted in a government which on the surface appears to be more functional than its Zimbabwean counterpart, and has consequently retained international confidence. However, in reality this reflects not a common desire to reform, but the ability of key Kenyan political leaders to use unity government as a screen, behind which they have cultivated an ‘anti-reform’ alliance that sounds the death knell for attempts to end the culture of impunity. Both cases therefore reveal the danger that power-sharing arrangements simply turn back the clock to the days of the one-party state and the politics of forced inclusion, so common throughout Africa in the post-independence era. Rather than create space for reform coalitions, power-sharing can be manipulated by incumbents desperate to retain their positions in the face of electoral defeat, undermining the prospects for reconciliation or institutional regeneration. The lesson to be taken from the experience of Kenya and Zimbabwe thus far is bleak; power-sharing serves to postpone conflict, rather than to resolve it.
THE ROOTS OF CRISIS IN KENYA AND ZIMBABWE
While both Kenya and Zimbabwe developed relatively capable states, the process which led to political crisis was profoundly different in the two cases. Most obviously, while politics in Kenya has tended to revolve around inter-communal competition among a far more ethnically diverse and rural population, in Zimbabwe class and geography have been more important than ethnicity. Furthermore, the variation in the way that things ‘fell apart’ in the two countries reveals that political stability and disorder in each country rested on very different foundations.
During the 1970s and 1980s, the stability of the Kenyan one-party state was underpinned by a high level of elite consensus, in which a system of inclusion ensured that it was in the interest of a range of elite actors to secure the allegiance of their communities to the state and to demobilise popular protest (Branch & Cheeseman Reference Branch and Cheeseman2006). However, in the 1980s, the elite alliance that had served as the foundation of the Kenya African National Union (KANU) government was undermined by President Daniel arap Moi's use of divide-and-rule strategies which alienated rivals and heightened the salience of ethnic identities. The return to multiparty politics, combined with Moi's resort to increasingly repressive and exclusionary strategies, resulted in growing divisions among the elite, with political leaders becoming progressively more willing to marshal their supporters, and occasionally militia groups, against their rivals. At the same time, a process of state informalisation epitomised by rampant theft including the infamous Goldenberg scandal led to a weakening of institutions that had previously exerted some constraints on the executive, generating widespread corruption that eroded the basic capacity of the state to provide services (Branch & Cheeseman Reference Branch and Cheeseman2009).
The political crisis that engulfed Kenya in 2008 was triggered by the apparent refusal of President Mwai Kibaki to relinquish power after Raila Odinga's supporters had already begun celebrating victory, but its roots lay in the divisive politics of the 1980s and 1990s, which both increased opposition to the state and undermined its capacity to maintain order (Cheeseman Reference Cheeseman2008). This is not to deny the significance of short-term events; of course, the flawed process of counting ballots, the farcical rush to swear Kibaki in, evidence of malpractice raised by European Union Observers, and statements from the Chairman of the Electoral Commission that he did not know who had won the elections, all served to intensify opposition suspicions and increase the prospects for unrest (Cheeseman Reference Cheeseman, Mehler, Melber and Walraven2009). However, a full understanding of the violence that followed the disputed polls and its meaning for contemporary Kenyan politics must recognise that the Kenya crisis represented the explosive interaction of complex historical processes and short-term political machinations (Mueller Reference Mueller2008).
In 2002, the erosion of support for Moi, his failure to hold together KANU's remaining elite alliance, and a rare moment of opposition unity, finally led to the party's removal from power by the victory of the National Rainbow Coalition (NaRC), led by Mwai Kibaki and Raila Odinga. NaRC's multiethnic coalition collapsed after Kibaki refused to honour a pre-election memorandum of understanding that the post of Prime Minister would be created for Odinga. A process of political realignment ensued, in which the cabinet split over the question of constitutional reform, with Odinga and Kibaki actively campaigning against each other. Significantly, following this period of political musical chairs, leaders who had previously stood on the same platform now found themselves heading competing parties, blurring the line between rival camps.
When the Orange Democratic Movement (ODM) was formed to oppose the government's favoured constitutional draft, Kibaki was left increasingly isolated as Odinga recruited a number of powerful ethnic patrons to join his ‘pentagon’, knitting together an effective coalition of the dispossessed. Kibaki's new electoral vehicle, the Party of National Unity (PNU), seemed to have a rather narrow support base (Cheeseman Reference Cheeseman2008). Odinga's supporters were united by a belief that they had been denied ‘their turn to eat’ as a result of Kikuyu domination of land, political power, and economic opportunity. The fragmentation of the ODM with the creation of Kalonzo Musyoka's ODM-Kenya notwithstanding, the way in which senior political figures coalesced around these two groupings raised the salience of a range of historical grievances, notably land claims, which stretched back through the Moi era to the colonial period. ODM supporters were attracted by a promise of majimbo (regional) government, a policy interpreted by Kibaki's supporters as a direct threat to their livelihoods and personal security because of the association of the majimbo policy with elite orchestrated ethnic violence that KANU had utilised in the build up to the 1992 and 1997 elections (Anderson Reference Anderson2003; Lonsdale Reference Lonsdale2008). When civil conflict erupted in the wake of the election results, these fears were realised.
However, the violence in 2008 was far from straightforward, largely as a result of the rotation of the elite between parties, and the politicisation of militias during the 1980s and 1990s. On the one hand, a number of non-state gangs allied to the ODM perpetrated attacks on groups assumed to have supported Kibaki, notably in the Rift Valley. Conversely, a high number of casualties occurred as a result of direct state repression, notably in Nyanza where the police were responsible for a large proportion of the fatalities (Cheeseman Reference Cheeseman, Mehler, Melber and Walraven2009). By the end of the violence it was increasingly clear that few Kenyan politicians could emerge from the crisis with any credit, and that many had blood on their hands.
The pathway to upheaval ran differently in Zimbabwe, where the crisis has strong historical roots centred on intricate interactions between land grievances, the constructions of nationhood and citizenship, state formation, and the centralised use of political violence (Hammar et al. Reference Hammar, Raftopoulos and Jensen2003; Raftopoulos & Savage Reference Raftopoulos and Savage2005). Zimbabwe's land problem, so central to politics, must be seen against the background of a long and shifting tradition of land contestation going back to the colonial period. State policy changed from one that acted against illegal farm settlers (then referred to as ‘squatters’) in the 1980s, to one that actively encouraged and was complicit in the violent seizure of white-owned commercial farms in 2000 (Alexander Reference Alexander2006). The key turning point in this transition occurred in 1997, when the state agreed to meet long-neglected demands for financial compensation and land allocation from war veterans of Zimbabwe's liberation struggle (Alexander Reference Alexander, Mustapha and Whitfield2009). This development resulted in the creation of an alliance between the state and war veterans. This new political axis proved crucial in the state's response to its defeat in the 2000 constitutional referendum, when ZANU-PF's proposed new constitution was actively opposed by civic groups such as trade unions, students and churches, all falling under the banner of the National Constitutional Assembly (NCA) (Raftopoulos Reference Raftopoulos, Moyo, Makumbe and Raftopoulos2000). Significantly, in 1999 the constitutional reform debate also gave rise to a new opposition political party that opposed ZANU-PF's plans, the MDC, which drew some of its membership from the NCA.
ZANU-PF interpreted the 2000 referendum result as a worrying victory for the MDC, coming as it did just months before scheduled parliamentary elections. The defeat was particularly significant given the onset of economic crisis resulting from the government's adoption of a disastrous Economic Structural Adjustment Programme (ESAP) in the 1990s, the costly and unbudgeted payout of war veterans' compensation grants, corruption by ZANU-PF elites, and the country's 1998 involvement in the Democratic Republic of the Congo (DRC), where ‘in six months the government spent more money on the DRC military venture than it had spent on land purchases since 1980’ (Kriger Reference Kriger2007: 70). The prospect of electoral defeat drew a violent response from the government. The alliance with the war veterans was mobilised to forcefully seize white-owned commercial farms, and violently crush MDC support and party structures. The military and Central Intelligence Organisation (CIO) provided logistical support and weapons to the war veterans, while the police turned a blind eye to the violence (ZI 2004). Ideologically, ZANU-PF made use of Zimbabwe's multifaceted liberation history to construct a narrative called ‘patriotic history’ to shore up its waning legitimacy, branding opposition supporters as sell-outs (Tendi Reference Tendi2010).
Mugabe narrowly won the elections, but failed to stem the atrophy of support for the party. Although there was no let-up in the violent and divisive politics employed by the government, in March 2008 Tsvangirai defeated Mugabe in the first round of the presidential elections. In order to retain control of the powerful executive presidency in the June presidential election run-off, ZANU-PF and the Joint Operations Command (JOC), comprising the heads of Zimbabwe's army, air force, police, CIO and the prisons service, responded with an intensity of violence rarely surpassed in the history of independent Zimbabwe. The violence forced Tsvangirai to withdraw from the run-off, leaving Mugabe the victor in an election devoid of international and domestic legitimacy, triggering renewed attempts at international mediation.
CIVIL–MILITARY AND INTRA-ELITE RELATIONS
Operationalising a veto-players framework demands that we pay attention to the source of a particular actor's veto (informal or formal), its effectiveness (partial or complete), and the factors underpinning the willingness of an actor to exercise their veto (short and long-term costs and benefits). Due to lack of space, we focus on the main institutional and partisan veto players that demanded the most concessions in the negotiations, namely the military and the main political parties. Although other institutions such as the judiciary have played a significant role during the crises in Kenya and Zimbabwe, they did not raise any significant challenge to a power-sharing deal. Similarly, while the presence of a number of smaller parties complicated both the elections and their aftermath, it quickly became apparent that they were happy to join any power-sharing deal that offered them representation. It was therefore the way in which the different historical contexts led to variations in the nature of civil–military relations and political competition that shaped the practice of power-sharing in the two cases (Brown Reference Brown2009; Cheeseman Reference Cheeseman, Mehler, Melber and Walraven2009; Chitiyo Reference Chitiyo2009; Matyszak Reference Matyszak2009).
In Zimbabwe, the central role accorded to the military, and the deep divide which developed between MDC and ZANU-PF leaders, made securing agreement on a power-sharing formula a formidable task, resulting in the politics of continuity, in which power-sharing did little to change the dominant political dynamics. In contrast, because Kenya has a history of more inclusive power-sharing deals, and because civilian leaders are isolated from pressure from military and police leaders, partisan and institutional veto players were less of an immediate barrier. However, in part because political parties are less disciplined in the Kenyan context, the main veto players are significantly more fragmented, with authority within each political party divided between a number of semi-autonomous rival political leaders, none of whom is clearly dominant. It is therefore necessary to recognise the existence of ‘secondary’ partisan veto players, to accommodate factional leaders such as William Ruto and Uhuru Kenyatta who, while not the heads of their respective coalitions, enjoyed sufficient independence and influence to be considered veto players in their own right. For example, while the leader of the PNU, Mwai Kibaki, was clearly the primary veto player within parliament, Uhuru Kenyatta deserves to be seen as a secondary veto player because in his capacity as KANU Party Chairman he controlled a small but significant party that, while it sat within the PNU alliance, had not surrendered its independent status and did not have to follow the PNU line. It was the ability of these factional leaders to create cross-coalition ‘anti-reform’ alliances which gave rise to the politics of collusion, and made the Kenyan experience so different from the Zimbabwean case, where deep trenches existed between ZANU-PF and MDC leaders.
Partisan veto players
Explaining the attitudes of partisan veto players towards power-sharing deals requires us to consider the history of relations between political leaders, the lessons learned from previous attempts at political accommodation, and the thorny question of how different members of the political elite (here understood to refer to MPs and senior party officials) conceptualise their self-interest.
Kenya's political elite has a history of relatively cohesive co-existence. The one-party state presided over by Jomo Kenyatta was one of Africa's more inclusive regimes, notwithstanding the assassination of Pio Pinto and J. M. Kariuki (Widner Reference Widner1992). Since the late colonial period Kenyan politics has taken the form of ‘boss’ politics, with a series of ethnic patrons presiding over highly personalised political machines (Gertzel Reference Gertzel1970). Patron–client networks, running from the executive, through regional Big Men and MPs, to local communities, served to connect the periphery to the political centre. The personal provision of patronage, controlled by Kenyatta and later Moi, served as the glue of national integration. Significantly, early in KANU rule, leaders from a broad range of communities were invited to share in the fruits of independence so long as they did not seek to challenge Kenyatta's personal hegemony. Following the ‘voluntary dissolution’ of the opposition Kenya African Democratic Union (KADU) in 1964, the ruling KANU effectively assimilated KADU leaders. As a result, what had been a predominantly Kikuyu and Luo party gained a significant proportion of Kalenjin, Maasai and Luhya leaders. A number of these went on to be extremely successful within the party, most obviously Daniel arap Moi, who succeeded as president on Kenyatta's death in 1978 (Throup Reference Throup and Schatzberg1987). Although the Kenyatta succession revealed the tensions within KANU, ‘once the succession was decided, the elite, and the bourgeoisie as a whole, had an overriding interest in stabilizing the regime upon which they thrived’ (Tamarkin Reference Tamarkin1978: 33).
From 1978 onward, President Moi's increasingly authoritarian rule undermined this elite cohesion, with a growing number of prominent patrons excluded from access to state resources. Nonetheless, even in the 1992 ‘founding’ multiparty election, the party returned MPs from an impressive kaleidoscope of ethnic groups, and during the 1990s the need to form multiethnic alliances in order to effectively compete in national elections compelled political leaders to enter into a range of coalitions (Throup & Hornsby Reference Throup and Hornsby1998). This history of relatively inclusive politics and alliance formation shaped elite understandings of how a power-sharing solution might be expected to work, and the mutually advantageous possibilities that it would generate.
However, in order to fully understand the role of partisan veto players in the Kenyan case, it is important to recognise the importance of factional politics within each broad political alliance. Under multipartyism, Kenyan politics has been characterised by periods of concentration, as leaders sought to broker viable electoral pacts, and of fragmentation, as these alliances quickly fell apart after the polls (Cheeseman Reference Cheeseman2008). The consequence of this process of elite musical chairs is that few political leaders have not, at one point or another, worked as colleagues. In turn, this continual reshuffling of the elite pack has meant that corrupt figures from previous regimes are not contained within any one political organisation, but rather tend to be distributed throughout the party system (Murunga & Nasong'o Reference Murunga and Nasong'o2006). As a result, senior figures within both the PNU and the ODM have something to fear from launching investigations into past corrupt activities. The presence of Musalia Mudavadi in the ODM ‘pentagon’ of leaders is a perfect example of this phenomenon: as a member of the NaRC government, Mudavadi personally signed some of the most important documents in the Anglo Leasing scandal (ibid.).
The ‘Kenya crisis’ in part resulted from mounting distrust between Kibaki's allies and the opposition leaders grouped around Raila Odinga (Branch & Cheeseman Reference Branch and Cheeseman2009), but did not change this underlying reality. Although the violence committed in the aftermath of the elections shocked the nation and served to harden battle-lines and entrench communal identities, the extent to which this actually created cleavages within the elite was undermined by divisions within the ODM leadership, the history of alliance formation, and the involvement of a wide range of actors in the violence. Within the ODM, the various leaders and the communities they claimed to represent had little in common bar their common demand that ‘it is our turn to eat’. Although they sided with Odinga in 2007, many senior ODM figures had actually spent longer working alongside PNU leaders. William Ruto, for example, was a protégé of Daniel arap Moi, who formed an electoral pact with Kibaki in 2007. Musalia Mudavadi was another former Moi loyalist who found himself within the ODM during the Kenya crisis. Similarly, Charity Ngilu, recruited to the ODM to attract Kamba voters, had previously been Chairperson of NaRC. Given the deep connections that cross-cut the government/opposition divide in 2007/8, it is misleading to see the Kenya crisis as occurring between two coherent and unified political blocks. Consequently, the violence did not serve to straightforwardly harden the position of ‘government’ and ‘opposition’ partisan veto players in the way that it did in Zimbabwe.
The relationship between factionalism, elite relations and repression was compounded by the complex nature of the violence in the Kenyan case, where both sides were implicated in atrocities. On the ODM side, William Ruto is closely connected with many of the individuals thought to have been prominent in the organisation of the worst of the civil strife in the Rift Valley. During Moi's time, Ruto personally oversaw the use of militias to carry out ethnic cleansing and create a ‘KANU zone’ in the 1992 and 1997 elections, and there is considerable evidence that similar networks were activated in 2007 (Lynch Reference Lynch2008). On the PNU side, Kibaki and the Minister for Internal Security, George Saitoti, were ultimately responsible for the activities of the police force, which has come in for the most strident criticism from the Waki Commission into post-election violence (Cheeseman Reference Cheeseman, Mehler, Melber and Walraven2009). More significantly, prominent Kibaki allies, such as Uhuru Kenyatta, are believed to have connections to the violent mungiki gang, which engaged in revenge attacks on communities assumed to have supported the opposition following the first wave of violence. As a result of the diffusion of violence, neither side was in a position to claim a monopoly over victimhood, and both sides had reason to fear post-election prosecutions. This common interest, combined with the history of elite inclusion, grounded the willingness of key veto players to allow a more functional power-sharing agreement, and underpinned the emergence of the politics of collusion within the unity government.
Zimbabwe has no history of de jure single party rule and elite consensus has been secondary to forced inclusion. The early 1980s saw the opposition Zimbabwe African People's Union (ZAPU) depicted as a party bent on disrupting the country's national security by fomenting dissident elements in the Matabeleland province. From 1981 onward, the operation of ex-ZIPRA deserters from the Zimbabwe National ArmyFootnote 2 in Matabeleland provided ZANU-PF with a pretext to launch the Gukurahundi campaign to eliminate ZAPU, laying the foundation for single-party rule (CCJP 2002). Under the guise of crushing dissident activity, which in reality comprised a ‘poorly armed group of less than 400 at their peak who survived mainly by avoiding confrontation’, the military was deployed to Matabeleland, where it carried out a campaign of intimidation that left up to 20,000 dead (Alexander & McGregor Reference Alexander, McGregor, Allen and Seaton1999: 251). The Gukurahundi campaign forced ZAPU leader Joshua Nkomo to dissolve his party and join ZANU-PF under the Unity Accord of 1987, leaving Zimbabwe a de facto one-party state. Shortly afterwards, a discourse of permanent unity began to be propagated, and all opposition was cast as treason. Thus, previous experience of supposedly inclusive government in Zimbabwe has not served to develop trust or mutual understanding between partisan veto players on either side of the political spectrum.
ZANU-PF's reliance on repression rather than consensus to gain compliance intensified as the challenge to the party's hegemony grew. Consequently, Zimbabwean politics has been polarised since the MDC's formation in 1999. At the time of the 2008 presidential election run-off, violence was perpetrated by the military, the police, war veterans, ZANU-PF youth militia and supporters, and targeted at the ‘sell-out’ opposition (HRW 2008). In contrast to Kenya, ethnicity was not a factor. The aim of the violence, codenamed Operation Makavhoterapapi (meaning ‘where did you place your vote?’), was to decimate the MDC's party structures through arrests, beatings, torture, targeted assassinations and disappearances, and to break its support through pungwes (all-night indoctrination vigils), intimidation, public beatings, and displacement (SPT 2008). While war-veterans invaded the few remaining white-owned commercial farms, the state controlled media blamed the opposition for the violence (Zimbabwe Online 2008b). Alongside these repressive strategies, ‘patriotic history’ was employed to isolate and deter the emergence of resistance by dividing Zimbabweans into ZANU-PF ‘patriots’ and opposition ‘traitors’. In elections, Zimbabweans were urged to defend the country's sovereignty, depicted as being under threat from Western powers seeking ‘regime change’ in collusion with ‘colonial’ white farmers and local ‘sell-outs’ (Mugabe Reference Mugabe2001: 81). ZANU-PF's fusion of violence and rhetoric encouraged a process of political polarisation that established physical and mental barriers to negotiating with the ‘enemy’, thereby entrenching the intransigence of key veto players. As a result (Alexander & Tendi Reference Alexander and Tendi2008: 12),
this was an ideologically driven battle, as much of the post-2000 violence had been: ideas mattered, something that seems to be almost entirely absent in explanations of Kenyan violence. The foregoing accounts for differences in both the practices and language of violence and the far harder boundaries between parties in Zimbabwe. Where the opposition is characterised as traitorous to the nation, in league with foreign powers, and in effect beyond the protection of the law, the easy side switching seen among Kenyan politicians is difficult.
While the JOC announced that all ‘those who have been breathing fire about Kenyan-style violence should be warned that violence is a poor substitute for intelligence’ (Zimbabwe Online 28.3.2008), the unidirectional nature of the violence ensured that the Kenyan context could not be repeated in Zimbabwe. In contrast to the confused and complex picture in Kenya, human rights groups estimate that only 1% of violent incidents in Zimbabwe were committed by the MDC (Alexander & Tendi Reference Alexander and Tendi2008). Significantly, and in contrast to the ODM, the MDC has a monopoly over victimhood, and has continued to demand justice despite attempts by ZANU-PF and the JOC to bury the issue of state-sponsored violence during the elections (Zimbabwe Online 2009).Footnote 3 The question of prosecutions remained topical because, in contrast to Kenya, political parties have no mutual interest in hiding the issue. The willingness of the MDC to speak out against perpetrators of violence has bred fear among the ZANU-PF elite, solidifying distinct and fiercely opposed partisan veto players united by their common intransigence. Antagonistic elite relations prevented the sort of musical chairs witnessed in Kenya, and laid the foundations for the politics of continuity.
Institutional veto players
Although arduous and protracted, the power-sharing negotiations in Kenya were simplified by the predominantly civilian nature of politics. While a failed coup attempt in 1982 played a major role in shaping the paranoid approach of the Moi regime, the security forces in Kenya have on the whole been keen to stay out of politics. Kenya has no history of including military figures in the cabinet, nor is there any tradition of allowing military leaders to have a deciding say in either domestic policing or foreign policy. Successive presidents have appointed key allies to head the military in order to ensure compliance among the top brass, but this has been the extent of the politicisation (Tamarkin Reference Tamarkin1978). During the Kenya crisis, both military leaders and President Kibaki appeared keen to ensure that the military did not become embroiled in the conflict, resisting calls to utilise the military's organisational capacity to restore order. Heightened political interference in military affairs in the wake of the crisis, including the abandonment of rules which had regulated length of tenure and promotions, and were credited with contributing to the professionalisation of the military, suggest that the decision to insulate the military from the crisis was not simply taken in the national interest. Rather, it seems likely to have been inspired by a common fear among PNU and army leaders that the multiethnic nature of the army rank and file meant that, should the army became involved in direct action on Kenyan soil, senior officers might not be able to maintain control over their juniors (Branch & Cheeseman Reference Cheeseman2008).Footnote 4
In contrast, in Zimbabwe an ostensibly civilian regime has been progressively militarised since the late 1990s (Chitiyo Reference Chitiyo2009). In the Zimbabwean context, ‘military’ refers to the JOC and the entire military leadership. Militarisation denotes the incremental appointment of retired and serving military leaders to key state institutions and to positions within ZANU-PF, resulting in an increasingly blurred distinction between ZANU-PF, the military, and the state (ibid.). Although the origins of this process are obscure and the precise timings of the transformation hard to pin down, the military first became openly political on 9 January 2002 when, flanked by other members of the JOC, Commander Vitalis Zvinavashe declared that the military ‘will only stand in support of those political leaders that will pursue Zimbabwean values, traditions and beliefs for which thousands of lives were lost in the pursuit of Zimbabwe's hard won independence, sovereignty, territorial integrity and national interests’ (BBC News 9.1.2002). Zvinavashe's statement came two months before the 2002 presidential election, which, in the shape of the MDC and Morgan Tsvangirai, presented Mugabe with his sternest challenge to date. Tsvangirai, with his lack of liberation war credentials, was the target of Zvinavashe's intervention. Later, the military mobilised to ensure victory for Mugabe, and has subsequently reiterated its commitment to ZANU-PF on the eve of every national election.
In return for its support, the military has been granted an increasingly important role within the state and ZANU-PF. On retirement, senior military figures such as Commanders Solomon Mujuru, Josiah Tungamirai (both 1992) and Zvinavashe (2003) have been appointed to the ZANU-PF politburo, the party's most powerful decision-making body. The process extends to the permanent secretaries of key ministries such as Energy, Industry and International Trade (Chitiyo Reference Chitiyo2009), while George Chiweshe, the chairman of the Zimbabwe Electoral Commission (ZEC) that ran the controversial 2008 elections, is a retired Brigadier General.
Significantly, given their increasingly central political role, military actors are among the most intransigent when it comes to the question of power-sharing. Drawing on their service as guerrilla fighters in Zimbabwe's liberation war, the JOC see themselves as the ‘guardians of Zimbabwean sovereignty’ and refuse to countenance the defeat of ZANU-PF, the ‘deliverer of Zimbabwean independence’ (Chiwenga 2006 int.). At key moments following the 2008 elections, military figures closed off opportunities for compromise and reconciliation. Tsvangirai (BBC News 17.4.2008) claims that Mugabe's first response to his loss in the March 2008 polls was to initiate meetings between the MDC and ZANU-PF to discuss conceding defeat and the possibility of power-sharing. However, this conciliatory moment passed quickly as ‘hard-liners’ in the ZANU-PF politburo and, crucially, the JOC set about a violent crackdown against the opposition (ibid.; Washington Post 5.7.2008). From this point on, the ZEC delayed the release of election results, while the JOC co-ordinated Operation Makavhoterapapi (HRW 2008).
The JOC found a strong ally in a hard-line ZANU-PF faction led by Emmerson Mnangagwa, which had aligned itself with Mugabe's 2008 candidature at ZANU-PF's December 2007 congress in order to thwart a rival faction headed by Mujuru (Alexander & Tendi Reference Alexander and Tendi2008). In the wake of the March polls, the Mnangagwa faction took the opportunity to expand their influence within the party, working tremendously hard to secure Mugabe's re-election. Tellingly, Mnangagwa was transferred from the obscure Rural Housing and Social Amenities ministry to the State Security ministry, where he was responsible for overseeing JOC activities and reported directly to the president (HRW 2008). Similarities between the pattern of violence rolled out by the JOC and Mnangagwa and previous waves of intimidation were ‘no coincidence: several members of the JOC – most notoriously Perence Shiri and Mnangagwa – had been directly involved in the 1980s violence’ (Alexander & Tendi Reference Alexander and Tendi2008: 11). Through the alliance between Mnangagwa and the JOC, partisan and institutional veto players joined forces, fusing together a number of strands of resistance to any form of political accommodation with the MDC. The involvement of notorious Gukurahundi figures in the 2008 violence further strengthened the logic of the politics of continuity, as members of the JOC sought to guarantee their immunity over human rights violations past and present.
THE DYNAMICS OF POWER-SHARING IN KENYA AND ZIMBABWE
The different evolution of veto players in the two countries was directly reflected in the key players of the power-sharing negotiations. While the Kenyan talks were predominantly civilian, and largely revolved around the main partisan veto players who had played a prominent role in the election campaign, in Zimbabwe members of the JOC took part in the unity government negotiations, an example being the talks of 11 August 2008 when Mugabe arrived at Zimbabwe's Rainbow Hotel in the company of General Chiwenga (UK Guardian 14.8.2008). However, it is difficult to assess the impact of institutional and partisan veto players on the negotiation process itself for two reasons. First, the talks were held behind closed doors, and no reliable transcripts of the debates and the various positions adopted by different actors are available (although some second-hand accounts exist, see ibid.). Second, the deals were negotiated under great international pressure and were signed largely for international consumption; consequently, the actual text of the documents should not be taken at face value. In both cases, incumbent governments were willing to sign unity deals precisely because they recognised that so long as they retained the all-powerful presidency they would continue to be able to effectively veto reform by simply refusing to implement the clauses of the agreements they found most problematic.
It was thus not in the negotiation process, but in the implementation period, that the impact of the different veto players at work in the two cases came to the fore. Indeed, although roughly similar power-sharing deals were signed in Kenya and Zimbabwe, they gave rise to markedly different political dynamics within months of their inception. The similarities are most apparent in the infrastructure of power-sharing: in each case incumbents were forced to distribute cabinet posts roughly fifty-fifty between the ‘government’ and ‘opposition’, but refused to give up the presidency and sought to maintain control over the main levers of coercion, including Foreign Affairs, Home Affairs, Defence, and Internal Security. The opposition therefore had to be content with the creation of a new post of Prime Minister to accommodate Tsvangirai/Odinga, and a number of ministries less integral to the maintenance of political control, such as Health and Education. The only significant difference was that Zimbabwe's desperate economic plight, combined with the clear preference of Western governments for the MDC, forced ZANU-PF to allow opposition leader Tendai Biti to take up the powerful position of Finance Minister, in a bid to turn the taps of international financial assistance back on.
However, despite the many similarities, the important differences between the two cases became clear once the attention of international actors had moved elsewhere, leaving domestic veto payers to struggle over how the new rules of the game would work in practice. In Zimbabwe, the combination of institutional and partisan veto players ensured that the power-sharing government was stillborn; from the very start, ZANU-PF and military hard-liners had no intention of respecting the unity deal. In contrast, in Kenya the history of elite cohesion and the absence of institutional veto players allowed for a more cohesive government, precisely because MPs from rival parties shared experiences and interests. However, it is important to recognise that, this variation notwithstanding, a common thread continues to underpin the fates of the two counties: power-sharing in both cases occurred in the absence of a viable pro-reform alliance within the parliament, and without an elite consensus on the need for institutional change. As in the bad old days of the one-party state, unity government appears to have enabled presidents to retain control of the political agenda and to marginalise reformers. Consequently, in both cases the prospects for genuine constitutional change and democratic consolidation remain bleak.
The politics of continuity in Zimbabwe
The willingness of partisan and institutional veto players to obstruct and subvert reform in Zimbabwe has resulted in painstakingly slow progress, with the deep divide separating ZANU-PF and the MDC overtly expressed in the content of the GPA. Terence Ranger (Reference Ranger2008) notes that prior to signing the agreement, Mugabe ‘spoke to an Assembly of Zimbabwean Chiefs …. He told them that ZANU-PF and the MDC were completely different from each other: as different as fire and water.’ Ranger continues: ‘When one reads the agreement it certainly does not sound as though the parties are speaking a common language. One can see seams of fire and rivulets of water running side by side, and hardly ever mingling, throughout the whole document. The fire is insurrectionary, anti-colonial, “patriotic history”, focussed on sovereignty and isolationism. It uses the familiar language of Mugabe-ism. Side by side with it runs the universal language of democracy and human rights and development’ (ibid.). Given this continuing ideological division, it is unsurprising that ZANU-PF and its intellectuals continue to make use of a divisive ‘patriotic history’ to polarise the political system.
It is worth recalling that Tsvangirai was in favour of swift negotiations, not because he thought power-sharing was a panacea, but because ‘the people have suffered enough’ (UK Guardian 14.8.2008). Yet the 2008 violence and human rights abuses are rooted in the country's complex unresolved legacies of impunity, intolerance and pseudo-reconciliation. The diplomatic rush to reach a power-sharing deal papered over the need to resolve these issues. It also resulted in the failure to craft an arrangement that would guarantee real sharing of executive authority between the ‘government’ president and ‘opposition’ prime minister. Indeed, Mugabe has retained much of the executive power he wielded before (Matyszak Reference Matyszak2009), and the repeal of repressive legislation requires Mugabe's consent and the support of ZANU-PF in parliament (ibid.). Furthermore, the security apparatus remains under Mugabe's control and there is no timetable for security sector reform. More generally, the Institute of Democracy in Africa (IDASA 2009) and Sokwanele (2009) have reported continuous breaches of the GPA by ZANU-PF. These include Mugabe's unilateral and hence illegal appointment of ZANU-PF members and the military to positions such as permanent secretaries to ministries, ambassadors, Governor of the Reserve Bank, and Attorney General (see also Sky News 7.10.2009).
Of course, the significance of the military to the dynamics of power-sharing in Zimbabwe is far from unique. Sullivan (Reference Sullivan2005: 88, 93) has described a ‘lethal military veto’ in the Burundi case, arguing that the decision not to over-represent minorities in the armed forces exacerbated Tutsi fears that military reintegration would undermine their ability to translate military domination into an informal veto over political decisions. This, in turn, undermined the ‘sense of security’ that military control had previously provided, led to the emergence of a ‘use it or lose it’ attitude towards military power among the Tutsi military elite, and so contributed directly to the attempted coup and the failure of the power-sharing in 1993. What is remarkable about the Zimbabwe case is that the military have come to enjoy such a direct impact on the fate of power-sharing in an ostensibly civilian regime that has not witnessed civil conflict on anything like the Burundian scale. The Zimbabwean experience, and its significance for contemporary events, thus demonstrate the value of moving away from a static operationalisation of the veto-players framework in order to understand the identity and disposition of veto players in the longue durée. It is only when the role of the military is seen through the lens of the liberation war, ‘patriotic history’, and ZANU-PFs struggle to retain coercive control, that the importance of the military veto to the failure of power-sharing in a ‘civilian’ regime can be fully explained (Tendi Reference Tendi2010).
The ‘opposition's’ response to the politics of continuity has been one of ‘protest and capitulation’ over a range of issues including the renewed seizure of white-owned farms and human rights violations (Matyszak Reference Matyszak2009: 8). For example, ‘the GPA presented to parliament in 2009 contained an 18-month schedule for the drafting of a new constitution’, but on signing it into law ‘Mugabe quietly (and without objection from the MDC) dropped this schedule’. Consequently, ‘there is no constitutionally binding timetable for the introduction of a new constitution’ (ibid.: 11). Remaining in the government is a better option than leaving for MDC leaders, because it offers the hope that the ‘opposition’ can improve service delivery, win over ZANU-PF voters, and see to it that a constitution guaranteeing free and fair elections is somehow drafted and adopted. As in Kenya, remaining in the unity government is also attractive because it allows the ‘opposition’ access to state resources which have been used for personal enrichment and to strengthen its patronage network (see The Herald 15.9.2009; Great Zimbabwe News 24.9.2009; Zimbabwe Independent 16.4.2009; Zimbabwe Times 30.9.2009).
Yet there is little evidence that the MDC's presence in government is having any concrete effect. Power-sharing has done nothing to halt the militarisation of the state: in September 2009, Mugabe appointed eight retired military officials as new board members to parastatals under the Information and Publicity ministry, where they are likely to be stumbling blocks to the democratic media reforms that the MDC is attempting to implement (SW Radio Africa 2.10.2009). The commitment of the military to continued ZANU-PF rule is solidified by the lavish lifestyle that leaders have managed to carve out for themselves within the power-sharing administration. In October 2008, one month after the GPA was signed, the military seized control of the Marange diamond fields in eastern Zimbabwe, engaged in forced labour of children and adults, and tortured and beat local villagers (HRW 2009). ‘Army brigades have been rotated into Marange to ensure that key front-line units have an opportunity to benefit from the diamond trade’, helping to sustain the military and ensure unity amid desperate economic conditions (ibid.). Tsvangirai's demands for an investigation of the military's human rights abuses in Marange and for the military to leave the diamond fields have been ignored by the generals.
It is thus unsurprising that ZANU-PF has not abandoned its strategy of repression. Human rights defenders, journalists and lawyers continue to be intimidated, harassed, threatened, detained and charged, often in malicious prosecutions (AI June 2009). Human rights NGOs persistently highlight state-led political violence against MDC activists and supporters that are reminiscent of the atrocities that occurred at the time of the 2008 elections (SPT 30.6.2009; ZHRF 2009). Senior MDC politicians continue to be intimidated: in 2009, the Tsvangirai MDC finance minister Tendai Biti received a letter containing a live 9 mm bullet and a warning to ‘prepare your will’ (UK Guardian 27.7.2009). Legislators from the Tsvangirai-led MDC have been arrested, and some convicted, on a range of charges including rape, electoral fraud, kidnapping, and inciting public violence (Ibid. 2.8.2009). One MDC MP was arrested for playing music that ‘denigrates’ Mugabe (SW Radio Africa 31.7.2009). At the time of writing, fourteen MDC MPs and senators are facing charges, which is designed to reverse the ‘opposition's’ parliamentary majority and cause humiliation (BBC News 14.10.2009). This is a direct product of the politics of continuity, and provides strong evidence of the slim prospects for genuine reform.
The politics of collusion in Kenya
In Kenya, power-sharing has facilitated the creation of new alliances, as MPs have sought to exploit the new arrangements for their personal gain, while safeguarding themselves and their allies from prosecution. Although the election proved extremely confrontational, and relations between Odinga and Kibaki were poor, the willingness of Kenyan leaders to put aside the national interest for personal gain quickly overcame their mutual animosity. Partisan veto players have been willing to allow a more harmonious and, on the face of it, effective unity government to emerge, safe in the knowledge that they remain capable of exercising their veto when it truly matters, most notably with regards to potential investigations into human rights violations and corruption.
The most visible sign of this politics of collusion is the recruitment of forty ministers and fifty-two assistant ministers to the unity cabinet, representing the largest – and most expensive – cabinet in the country's history, with the bill for salaries and allowances topping an estimated US$15 million per year (Cheeseman Reference Cheeseman, Mehler, Melber and Walraven2009). Indeed, despite the fierce competition between parties and factions for positions within the new government, Kenyan MPs quickly managed to find common ground by devoting their efforts to maximising their own conditions of service. That all MPs, as members of the ‘government’, are included in the flow of largesse and hence stand to gain by defending the system through which they secure access to resources is one of most damaging consequences of unity government. This is most obviously apparent in the willingness of MPs to provoke widespread condemnation by first refusing to pay tax on their generous allowances, and then awarding themselves a handsome pay increase, despite the perilous economic context facing the country (ibid.).
However, the shared greed of legislators is hardly a new development, and the distinctive features of the politics of collusion as it plays out in Kenya are best seen in the way in which the creation of a unity government has brought new opportunities for anti-reform elements from a number of parties to co-operate. It is through this subtle process that constituencies for reform in Kenya have been progressively undermined since the signing of the unity accord. The prospect of being prosecuted by domestic or international tribunals for involvement in election violence and corruption as part of the ‘agenda four’ of the unity agreement has provided anti-democratic factions with strong incentives to mobilise their cross-party influence in order to form a mutual protection society across party lines. At the same time, the organisational shell of the unity government has enabled these alliances to be built out of the public eye. The combination of incentive and opportunity explains the emergence of what on the face of it appears to be a remarkable alliance between the PNU's Uhuru Kenyatta and the ODM's William Ruto (Daily Nation 20.2.2009). The communities which these leaders ‘represent’ – Kikuyu (Kenyatta) and Kalenjin (Ruto) – fought against each other in some of the worst post-election violence of early 2008.
Despite deep cynicism towards such an alliance among the core support bases of both men, the fear that their names may feature in a secret ‘envelope’ of individuals due to be prosecuted for involvement in the post-election violence, and the large size – and hence electoral significance – of their home communities, have encouraged both leaders to engage in a process of alliance formation that can be traced back at least to February 2009. The deal appears to have been privately endorsed by President Mwai Kibaki, who is widely believed to have chosen Kenyatta as his successor within the PNU (The Standard 10.3.2009). Kibaki's backing would effectively mean that a Ruto/Kenyatta alliance would enjoy the benefits of incumbency, and thus have an incentive to prevent constitutional reform that would significantly decentralise power away from the presidency.
The precise form of the deal agreed by Ruto and Kenyatta is unclear, and both leaders have publicly stated their commitment to their current parties, but it is clear that each man is prepared to marshal his troops in order to protect the other. Consequently, when Ruto faced a motion of censure regarding corruption allegations, PNU MPs joined their ODM counterparts in refusing to vote against him (The Standard 21.2.2009). Similarly, when PNU's energy minister faced a similar motion of censure relating to yet another corruption scandal, Ruto and a sizeable faction of ODM MPs refused to support the vote (Daily Nation 23.1.2009; The Standard 21.2.2009). In this way, secondary-level partisan veto players in both parties effectively closed off opportunities to expose and tackle corruption. The murky nature of the politics of collusion makes it difficult to trace the likely impact of such back-room deals on the prospects for the punishment of violations of democratic norms and effective constitutional change. Nonetheless, Ruto and Kenyatta have a clear common interest in preventing reform in key areas. The detrimental impact that power-sharing has had on the willingness of ‘opposition’ leaders to oppose the ‘government’ over corruption scandals, and vice versa, is therefore likely to extend to broader issues of prosecution and constitutional review. As a result, prospects for genuine reform in Kenya in the short term are bleak.
For all the criticisms of power-sharing arrangements offered in this paper, it is clear that the question of how to resolve conflict in Africa remains extremely complex, and there may be good reasons for thinking that in some cases the benefits outweigh the costs. After all, power-sharing is usually justified principally in terms of the number of lives it is likely to save in the short term. However, in order to make accurate decisions as to when these benefits outweigh the costs, it is essential to fully recognise the barriers that unity government may create to genuine reform. To date, international actors and the academic community have been slow to consider how significant these barriers may be.
Even if power-sharing arrangements do deliver greater peace and stability in the short term, the deep flaws documented in this paper suggest that it should only be used as a last resort. Power-sharing is a generous title for a process which has condemned opposition parties to accept inferior positions within the government, despite their success at the ballot box. Not only has this revival of one-party rule demoralised opposition activists and made it extremely difficult for opposition leaders to justify their decisions to their own supporters; it has enabled authoritarian elements to maintain the benefits of incumbency which will no doubt be wielded with added ferocity in future elections. The implications for the prospects of democratic consolidation are clear. Thus, while periods of power-sharing may have been impossible to avoid in the DRC or Liberia, we should be extremely cautious before we advocate a similar strategy in countries engaged not in civil war but in democratic deadlock, such as Kenya and Zimbabwe. In these cases it is less clear that the benefits of power-sharing outweigh the costs, especially if unity governments turn out to represent little more than a reversion to the illusory inclusiveness of the one-party state.
Given this, the spread of the model in recent years is a cause for concern. By supporting the extension of power-sharing arrangements to Kenya, the international community inadvertently backed a form of government extremely unlikely to deliver reform, while simultaneously legitimating an incumbent regime that had demonstrated its contempt for the democratic process. The ability of SADC leaders to push for a similar outcome in Zimbabwe, enabling Mugabe to remain in power, suggests the rise of a dangerous precedent that may encourage vulnerable leaders in fragile democracies to engage in strategies to produce deadlock, in order to legitimate the creation of a unity government within which they are allowed to retain executive power. In this way, the spread of the power-sharing model has generated incentives for anti-democratic behaviour at a time when the progress of many of Africa's new multiparty systems towards democratic consolidation remains partial at best.
