Politics in Kenya remains vulnerable to ethnic tensions despite its openness and vibrancy, but it can also be argued that Kenyan politics is becoming increasingly mature. This article explains the political economy dynamics behind the first two orderly presidential successions in post-colonial Kenya. It proposes a conceptual framework that shows how instrumental ethnicity plays out in a quasi-differentiated society in which ethnic organizations are the key conduits for the flow of rents between political and economic elites. More specifically, it shows how the internal fragmentation of ethnic groups intensifies the structural uncertainties that are commonly associated with intra-elite pacts in weakly institutionalized polities. It is argued that the 1978 and 2002 presidential successions in Kenya were orderly, paradoxically, because some of the crucial political and ethnic organizations were fragmented to the extent that they created conditions of great uncertainty for the elite. In this context, the rule of law was upheld as a last-ditch strategy to mitigate uncertainties in the face of rampant fragmentation. This shows that ultimately elite fragmentation can generate political stability provided that there is enough at stake for the elites.

Horrifying images of the post-election violence in Kenya in 2007–8 have been used by some in the media to portray Kenya as a country teetering on the verge of tribal warfare.1 The fact that such a dramatic deterioration of order occurred in a country that was considered to be one of the most stable in the region has led many to examine the underlying fractures in the Kenyan state.2 Indeed, Kenya's plight attracted so much attention that subsequently the alleged perpetrators of the violence wereprosecuted at the International Criminal Court (ICC). However, one should not be quick to trivialize the role of ethnicity in Kenya by construing it simply as an inherently destabilizing force. Although ethnicity has led to political instability on many occasions, it has remained such an important part of the social contract in Kenya since before independence that it must have played its part in stabilizing the country on other occasions.3 Kenya, unlike many other African countries, has avoided a successful coup d'état or an all-out civil war throughout its post-colonial history. It is puzzling why Kenya remains vulnerable to ethnic tensions well into the twenty-first century despite its growing middle-class and relatively robust civil society. But it is equally important to examine the flipside of the puzzle: why has Kenya remained relatively stable – except for occasional election-related violence – through much of its independent history, despite the conspicuous ethnic fault lines?

One of the prominent features of Kenya's post-colonial elites is the extent to which they used the constitution to contend for power and privileges. These elites were composed of leaders of political parties, high-ranking bureaucrats, prominent business people, and traditional leaders. Successive regimes in Kenya have been adept at using the law as an instrument to impose the will of the winning coalition on the entire polity. The relatively high number of constitutional amendments undertaken around times of regime uncertainty demonstrates the emphasis Kenya's elites put on taking cover behind the law.4 Anyang Nyong'o traces back the practice of seeking legitimation through the law to the colonial history of the country.5 In fact, this is not a peculiarly Kenyan phenomenon. The need to maintain a semblance of legal order is part of the internal logic of neo-patrimonial regimes. Nicolas van de Walle points out that it is the ability of the elites to regulate their own tendency to subvert the rational-legal order needed to maintain their position that determines the degree of stability in Africa today.6 The broad literature on neo-patrimonial politics in Africa offers useful insights into how regimes sustain themselves using disorder and patronage.7 But there is a dearth of conceptual arguments explaining the coexistence of constitutionalism and despotism in some countries.8 The occurrence of rule-based transitions in otherwise weakly institutionalized polities is even less often explained.

This article aims to provide a theoretical framework for explaining the coexistence of political maturity and lingering fragility in societies such as Kenya. As Angelique Haugerud has argued, ‘at issue are fundamental questions about how to analyse social asymmetries that breed conflict and change, without neglecting or misrepresenting forces of cohesion and stability’.9 I then apply the resulting theoretical claims to analyse the first two orderly presidential successions in Kenya's history, which stand in stark contrast with the post-election crisis in 2007–8. Part of the challenge in explaining transitions emanates from the fact that they are steeped in the structural foundations of the polity concerned at the same time as they are predicated on the contingent decisions of key actors. Therefore, it might not be possible to capture the entire essence of a political transition using either interpretative or rational choice theory alone.10 This article attempts to straddle the two intellectual traditions in explaining political transitions in Kenya by focusing on the strategic choices that political actors make within the structural constraints set by group identity.

In line with the logic of neo-patrimonial polities, Kenya's political elites exploit disorder and exclusive access to critical information to maintain control over their clients.11 But what is more important is how they contain violence among themselves in the first place to be able to rule over their clients. The ‘violence and social orders’ framework is a useful analytical tool for understanding the interaction between violence and economic rents in the evolution of social orders including neo-patrimonial polities.12 The conceptual framework I introduce in the next section builds on the core argument of the violence and social orders framework to characterize a scenario in which intra-elite uncertainty could result in institutional stability in societies such as Kenya.

Specifically, I argue that elites in weakly institutionalized – but sufficiently consolidated – polities abide by de jure institutions such as a constitution in order to mitigate extreme uncertainty. A given polity can be judged to be sufficiently consolidated if the formal institutions of the state are well developed enough to outperform anarchy in the eyes of powerful elites. One of the sources of political uncertainty in such countries is the ex ante fragmentation of elite blocs that would normally guarantee the enforcement of an intra-elite pact. The existence of political uncertainty implies the lack of predictability of the general distribution of power and one's future position in the distribution. ‘Intra-elite pact’, in the current context, refers to the ever-changing set of tacit agreements among the elites regarding the distribution of positions of power and privilege. Such pacts are normally enforced through the influence of elite blocs that are able to credibly threaten ‘mutually assured destruction’.

So long as those blocs of elites with the capacity to mobilize a critical threshold of collective action hold together, individual elites can rest assured that a predetermined level of minimum bargain will be delivered, even though the bargain may sometimes involve violence among non-elites, or between the elite and other groups. Uncertainty abounds when critical elite blocs fragment to the extent that ex post enforcement of a minimum bargain seems no longer guaranteed. Provided that a given polity is sufficiently consolidated and the elites have too much to lose in case of a full-blown crisis, de jure institutions will be reserved as a failsafe option to mitigate the inherent uncertainty of self-enforcing elite pacts. I employ the available historical evidence to demonstrate that elites in Kenya settled for rule-based transitions in 1978 and 2002 because attempting to renegotiate the intra-elite pact at those times was too risky for the elites, given the fragmentation of the dominant ethnic groups as well as major political coalitions in the country.

I proceed to operationalize the key theoretical claims of the article in a few simple steps. First, I provide a bird's-eye view of the political history of Kenya in the first two decades after independence in an attempt to assess the degree of political and economic consolidation as well as intra-elite differentiation. Second, I identify the major elite blocs that were instrumental in maintaining the intra-elite pact in the lead-up to the Kenyatta and Moi successions. Third, I assess the degree of fragmentation of the identified elite blocs and the uncertainty surrounding the two orderly transitions in contrast with the 2007 elections.

Kenya scholars have been cognizant of elite fragmentation in Kenya. Anyang Nyong'o acknowledges that the paralysis among various contending factions in the nationalist coalition contributed to the orderliness of the Kenyatta succession.13 Covering more recent developments, Daniel Branch and Nic Cheeseman attribute the post-election violence in 2007–8 to a long-winded process of elite fragmentation.14 However, Branch and Cheeseman fail to account adequately for the orderliness of the 2002 elections in the context of continued elite fragmentation. Nor do they recognize the role of elite fragmentation in the Kenyatta succession, presumably because they claim that the elite alliance began to fragment following the accession of Daniel arap Moi to the presidency. Moreover, both sets of authors refrain from pursuing the ethnic dimension of elite fragmentation explicitly. Rather, Branch and Cheeseman argue, ‘the increased salience of ethnicity is better understood as the outcome of changes in institutional context and the decision-making matrix facing political leaders, rather than their cause’. This seems like an accurate evaluation of the emergence of ethnicity as a politically salient factor in the formative years of post-colonial Kenya. However, once ethnicity has already assumed a key role in the political calculus of the society to the extent that it was exploited to generate instability around a number of elections, one should be able to take account of the instrumental role ethnicity plays in intra-elite bargaining.

Violence, economic rents, and intra-elite pacts: introducing a theoretical framework15

The bulk of the literature on neo-patrimonial politics in Africa seems to have been preoccupied with the failure of most African polities to attain a Weberian state.16 However, a Weberian state cannot be attained without escaping a Hobbesian trap, which is still holding back several African countries. According to the violence and social orders framework, societies develop political order when powerful actors with the capacity for violence find it more beneficial to make peace and share the rents that come with the resulting mutual third-party enforcement than to go it alone and gamble for a more uncertain gain.17 Such states, in which order is maintained by the consent of the elites instead of impersonalized institutions, are called natural states. The overarching organization of the elites is called a dominant coalition. The dominant coalition in natural states is an umbrella organization built on incentive-compatible allocation of rents among elites and is sustained without outside third-party enforcement. Note that the dominant coalition may consist of rival elite groups as long as such groups abide by an implicit contract not to resort to violence.

A dominant coalition of elites reigns over society by maintaining a double balance in the political and economic spheres. The attainment of a double balance refers to the existence of symmetry between distribution of rents and distribution of political power. ‘In a world of disorder there is a premium both on the vertical and personalized infra-institutional relations through which the “business” of politics can be conducted and on access to the means of maximizing the returns which the “domestication” of such disorder requires.’18 Primarily, the bargaining power of the political elites emanates from the brute force they are able to wield through mobilizing their clients. The elites are supposed to secure a share of the collective pie for their clients in return for their allegiance. Hence, at the most basic level of political development, each individual elite will use the threat of violence to gain direct access to economic rents that they will then redistribute to their clients.

With economic development comes differentiation. In the same way as an expanding market stimulates division of labour, a more consolidated polity encourages specialization, creating a distinction between political and economic elites. Consolidation provides sufficient stability for the elites to invest their time, influence, or money in either political or economic organizations with a longer time horizon in mind. The more initial endowments an elite has in political activities, the further they specialize in political organizations as the polity becomes consolidated. As a result, political elites maintain privileged access to political organizations whereas economic elites will have privileged access to economic organizations. The two types of organizations have a symbiotic relationship in natural states. Economic organizations extract existing rents from non-elites. They also create more rents as their productivity rises. In the process, economic organizations share parts of the rent collected with political organizations in exchange for the latter's support. Political organizations use their capacity for violence to guard the rents that are available to economic organizations. Specialization in economic organizations entails relinquishing direct access to the means of violence. Likewise, specialization in political organizations means giving up direct access to economic rents. Therefore, as the degree of specialization increases, there should be better coordination and enforcement mechanisms for the exchange of rents and protection among the elites.

Ethnic allegiances can provide the foundation on which formal and informal organizations facilitate the exchange of rents and protection between political and economic organizations. Robert Bates notes that the power of local administrations over the distribution of resources in colonial times, coupled with the correspondence between administrative and ethnic boundaries, created an incentive for the post-colonial elites to maintain politically cohesive ethnic groupings.19 On top of utilizing ethnic allegiances to buy support from their co-ethnic clients, the elites in such relatively modern – and therefore differentiated – polities as Kenya could employ ethnic identity to build unified blocs of political and economic elites. These blocs will then be used as leverage in the dominant coalition. ‘Tribalism is thus perceived less as a political force in itself than as a channel through which competition for the acquisition of wealth, power and status is expressed.’20

What keeps a weakly institutionalized polity stable is the presence of a dominant coalition. What keeps a dominant coalition stable, in turn, is the existence of an intra-elite pact enforced through the threat of ‘mutually assured destruction’. Now, let me briefly demonstrate the intra-elite dynamics in a multi-ethnic dominant coalition characterized by a given degree of specialization (see Figure 1). Suppose there are two major ethnic groups in a given society. Let NPi be the number of elites belonging to ethnic group i who have a comparative advantage in political organizations. Likewise, let NEi be the number of elites hailing from ethnic group i who have comparative advantage in economic organizations. A group of elites will have a comparative advantage in either of the two realms if their initial endowment of talent and influence in that particular realm is relatively higher than that of the rest of the elites. Figure 1 displays a two-by-two matrix representing the intersections between political-economic organizations and ethnic organizations. The four clusters of elites represented by the circles may be viewed as interdependent centres of power, the influence of which depends on the size of NPi or NEi. The circles are bound vertically by ethnic allegiances, whereas the horizontal bonds are formed by political and economic organizations.

Figure 1.

The dominant coalition as a matrix of political, economic, and ethnic organizations.

Figure 1.

The dominant coalition as a matrix of political, economic, and ethnic organizations.

Consider each pair of circles bound by a vertical or a horizontal band to be a conveyor belt of rents. The vertical conveyor belts redistribute rents among co-ethnic elites, whereas the horizontal conveyor belts are used to share rents within political or economic organizations. A political party is the most common example of a political organization. A corporation, on the other hand, is an example of an economic organization. Given the necessity of rent sharing to maintain the stability of dominant coalitions, the present framework conceives of ethnic organizations as conduits for the flow of rents between political and economic organizations.

Note that the internal structure of dominant coalitions in natural states is always in flux. There is constant renegotiation of the pecking order of elites in the dominant coalition, depending on their access to means of violence and economic rents. The fact that violence is avoided does not imply that there are no losers and winners in terms of the continuous reallocation of positions of power and privilege. The de facto distribution of power and privileges might not necessarily match the pattern of allocation dictated by de jure institutions. Suppose a certain exogenous change induces a fragmentation of elite blocs, and therefore renders the existing intra-elite pact unstable. When do elites settle for the default option of de jure institutions in response to fragmentation? In the following paragraph, I use the analogy of the conveyor belt to look into the effects of various scenarios of intra-elite fragmentation.

In case of the fragmentation of both political and economic organizations, the vertical set of conveyer belts stand as long as the political and economic centres of power within each ethnic group are cohesive enough to generate rents to keep the belts running. Such cases might be characterized as autarky or ethnic confederation. Accordingly, even if both political and economic organizations fragment to the point of pushing the coalition into fragility, the elites could retreat into their ethnic organizations to renegotiate the intra-elite pact. On the contrary, the horizontal conveyor belts cannot operate independently of each other since rents should be funnelled from economic to political organizations through some form of intermediary group.

Another scenario features the weakening of the ethnic ties in either of the groups while political and economic organizations remain sufficiently cohesive. Under this scenario, the dominant coalition remains stable since rents will be delivered effectively using the three standing conveyor belts. A critical situation arises if one of the ethnic conveyor belts fails at the same time as either of the political or economic conveyor belts breaks down. In this case, one centre of power would be left completely out of the loop, losing access to any means of mutual enforcement that normally forecloses intra-elite violence. Faced with the uncertainty of unenforceable renegotiation, and therefore violence, the elites settle for de jure institutions to deal with any impending change.

The following claims summarize the main propositions of the preceding analysis. First, elites are likely to vie to renegotiate intra-elite pacts as long as ethnic ties hold together, providing the necessary economic rents and violence capacity to ensure a minimum guaranteed bargain in the form of autonomous ethnic blocs. Second, the rule of law will be employed as a failsafe strategy when at least one major ethnic bloc is fragmented at the same time as other non-ethnic political alliances are fractured. The level of fragmentation within an ethnic group is significant enough to trigger the above reaction if it involves the emergence of multiple self-sufficient elite clusters. In order to have such an effect, an ethnic group must serve as a lynchpin that unites other ethnic groups, either via shared heritages with it or shared mistrust against it. As a general condition, the polity under consideration must be sufficiently consolidated for the second claim to hold.

Political and economic consolidation in early post-colonial Kenya

At independence, the new political elites of Kenya inherited a state that was strong and viable enough to emerge as the centrepiece of the dominant coalition. The colonial state bestowed the new elites with the tools of economic coercion such as labour repression that had been applied to facilitate settler accumulation.21 On the political front, the colonial state had already created and deployed the instrument of large-scale political and military coercion to quash the Mau-Mau rebellion. Therefore, the post-colonial Kenyan government inherited the foundations of a relatively consolidated natural state from the colonial system. Nic Cheeseman summarizes the pre-independence influence on the construction of Kenya's post-colonial natural state as follows: ‘In part, the impetus to regulate political space so tightly came from Kenya's white settlers, who made the first breakthrough in creating new arenas of political space, and subsequently set about defending their right to access these arenas exclusively.’22

On top of being sufficiently consolidated, the dominant coalition was also differentiated in terms of political and economic organizations. ‘In contrast to the continental picture, a class of national entrepreneurs seems to have emerged in Kenya, distinct from both the power holders of the State and what is called for convenience the “petty bourgeoisie”.’23 The state apparatus was used to create and protect rents for powerful elites. Land in the former “white highlands” constituted the core of the rent pool in the first decade after independence.24 Access to capital markets was also systematically restricted, first, by the informal networks of institutional investors that had emerged from the colonial system, and then by governmental institutions such as the Capital Issue Committee.25 Trade was another area where the state machinery was used to ensure exclusive access for the elites. The control of trade was cloaked in the Africanization drive, which has seen the promulgation of the Trade Licensing Act (1967) and the Import, Export, and Essential Supplies Act (1967).26 The elites would then share the rents through informal mechanisms such as kickbacks and quasi-formal arrangements – such as harambee – that they jointly utilized to deliver patronage to their clients.27

The claim of intra-elite differentiation should not give the impression that Kenya became a functionally differentiated society soon after independence. On the contrary, straddling continued to be a means of securing some level of direct access to economic rents for the political elites. Civil servants being the key actors in the de facto one-party state of the Kenyatta era, the Ndegwa Commission provided for them to engage in private sector activities by ruling out concerns of conflict of interest as long as minimal safeguards were in place.28 The Kenyatta regime managed to consolidate the control of violence relatively well, precluding the need for the economic elites to have direct access to the means of violence. With the diffusion of violence that ensued towards the beginning of the Moi era, the need for every elite to maintain direct access to both the means of violence and economic rents increased.29 In what follows, I demonstrate the role ethnicity played in holding the dominant coalition together by facilitating the exchange of rents and protection.

The dominant coalition in post-colonial Kenya has always mirrored the distribution of power among ethnic leaders in the country. The first two national parties, the Kenya African National Union (KANU) and the Kenya African Democratic Union (KADU), were organized as strategic teams of bigger and smaller tribes, respectively. In spite of an ostentatious declaration of commitment to Kenyan nationalism, Kenyatta himself could not help but acknowledge the influence of ethnic identity on nation building: ‘To him (the African), the traditional Tribal Council – which occasionally met other Tribal Councils through nominated representatives, or in times of war through intermediaries – was at once a Government and an expression of the very personality of each and every citizen.’30 To be sure, Kenyatta was cognizant of the importance of trans-ethnic alliances for stabilizing the young dominant coalition. He used co-option to win over leaders of other ethnic groups while employing appeasement to prevent his own ethnic base from fragmenting.

Although the dominant coalition of the Kenyatta era was stabilized on a multi-ethnic foundation, it essentially remained a collection of tribal patrons. Ethnic associations were used to leverage access to important sources of rent such as land. In many of the cases, the land-buying companies, such as the powerful Ngwataniro Corporation, were headed by ethnic leaders who gained access to land and finances by virtue of their political position within the dominant coalition.31 Ethnic associations such as the Luo Thrift Union had been in existence since before independence. In the first decade and half after independence, patronage resources were relatively abundant. Therefore, positions of ethnic leadership were sometimes used to share rents with leaders of other ethnic groups in exchange for political support. As part of their harambee obligations, local politicians sometimes had to seek sponsorship from politicians from other ethnic groups and regions to complement their financial contributions.32

The role of ethnicity in rallying otherwise differentiated elites in response to perceived threats from other ethnic groups was evident as early as 1969 when the Kikuyu intensified their oath-taking rituals in the face of public unrest following the assassination of Tom Mboya.33 There is probably no starker manifestation of the ethnic nature of the core of Kenya's dominant collation than the elites' reaction to the prospect of electoral competition in the 1990s. Daniel arap Moi was frank enough to argue publicly that multi-party democracy would inevitably degenerate into tribal bloodshed.34 He was alluding to the fact that the elites would retreat to their ethnic enclaves as soon as the implicit pact to stabilize the dominant coalition was compromised by the prospect of open competition. When the elections finally happened, outright violence was used to gerrymander and suppress opposition votes.35 The question, then, is: why did the ethnic belligerence of the late 1960s give way to the orderly transition in 1978? By the same token, why did the peaceful transition of 2002 happen in spite of the ethnic fortification of the 1990s? The next section employs the key theoretical arguments laid out in the previous sections to examine the intra-elite dynamics surrounding the first two presidential successions. The core of the analytical exercise lies in investigating whether the fragmentation of major elite blocs and the concomitant uncertainty have contributed to the orderliness of the transitions.

The Kenyatta succession

The political scene in 1970s Kenya was characterized by a weak ruling party and strong individual politicians. With no ideological support provided by a viable party, the political organizations of the young nation were held together by the highly centralized executive and a well-oiled patronage machinery.36 As the official trans-ethnic ruling party, KANU, was becoming dormant, a more formidable quasi-political organization representing the dominant Central province groups was rising to prominence. That was the Gikuyu, Embu, and Meru Association (GEMA). Although the organization was established with the mandate of a social welfare association, most of GEMA's leadership was composed of political heavyweights including several cabinet ministers. Apart from political elites, GEMA's membership list ‘reads like Who's Who in the world of finance and business in Kenya’.37 GEMA was an organizational manifestation of the role of ethnic ties in enforcing the intra-elite pact in the de facto one-party state.

In spite of the combined effect of a weak KANU and a rising GEMA, the executive bureaucracy continued to hold the core of the dominant coalition through the 1970s. Ethnically, the executive was clearly dominated by the Kikuyu. As of 1972, 41.2 percent of top-level government officials hailed from the Kikuyu ethnic group, which constituted about 20 percentof the population.38 However, Kenyatta had sent a public signal that he intended to keep the executive out of the sphere of influence of tribal organizations such as GEMA.39 This may have emboldened the Attorney General of the day, Charles Njonjo, to draw on the privileged access he had to the powerful presidency to maintain a belligerent stance against GEMA.40 In spite of his Kiambu origin, which was similar to most of the leadership of GEMA, Njonjo always showed contempt toward overt ethnic manoeuvring.41 Therefore, he took advantage of his position as Attorney General in using the law to maintain his own axis of influence.

Parliamentary backbenchers were junior partners in the dominant coalition of the 1970s. Nevertheless, they continued to pose a challenge to the government occasionally.42 They often organized on a trans-ethnic basis. The most prominent example in this regard is the rise of J. M. Kariuki, a fiery Kikuyu backbencher and advocate of a more equal wealth distribution. However, politicians like Kariuki did not quite fit the prevailing intra-elite pact, which was founded on a highly unequal distribution of rents and ethnic mobilization. Kariuki was assassinated in an attempt to restore the enforcement of the intra-elite pact.43 However, parliamentary backbenchers remained important ‘swing’ actors, with a potential to sway the power balance in the dominant coalition in alliance with figures like Njonjo or Kariuki.

In terms of organized economic interest, the Kenya Association of Manufacturers (KAM) represented manufacturing interest, which was dominated by foreign capital. On the commerce side, the Kenya National Chamber of Commerce (KNCC) represented local trading interests. Most economic organizations did not need direct access to the means of violence due to the presence of a sufficiently consolidated executive, which they lobbied constantly. GEMA was probably the only major exception that seemed to have offered the means of violence to co-ethnic economic elites at an organized level outside the state apparatus.

The GEMA hegemony seemed invincible until class tensions within the Kikuyu community began to surface in earnest towards the mid-1970s. The assassination of Kariuki, who belonged to a northern Kikuyu clan, was the last straw that broke the back of Kikuyu unity, severing the elites of the privileged Southern Kikuyu clan from the rest of the tribe.44 In the face of simmering revolt following Kariuki's assassination, the Kikuyu elites could not resort to oath taking, as they did following the death of Mboya. In a dramatic display of reliance on the coercive power of the state – as opposed to ethnic allegiances – Kenyatta presided over a massive military parade days after Kariuki's body was found.45 The implied shift in the locus of power towards the central state machinery afforded the Njonjo axis further advantage over the Kikuyu nationalists.

The final blow to the dominance of concentrated Kikuyu centres of power came with the failed attempt by GEMA's leadership to get the constitution amended to bar Vice- President Moi, an ethnic Kalenjin, from acceding to the presidency. Ironically, the real power implications of the “Change the Constitution” confrontation had more to do with conflict between Moi's Kikuyu supporters and his Kikuyu challengers than with the vice-president himself. The campaign pitted Charles Njonjo, who teamed up with Finance Minister Mwai Kibaki, in support of the vice-president, against most of GEMA's leadership. Joseph Karimi and Philip Ocheng claim that the civil service ‘remained completely loyal to the vice-president, the Attorney General and the Minister of Finance and Economic Planning’.46

After the core of the Kikuyu political hegemony had fallen out, the ripple effect triggered the break-up of other political alliances as well. The silent fragmentation of the political elites continued until the eve of the 1977 KANU election, which was later cancelled at the eleventh hour because the president was alleged to have fallen ill. In the lead-up to the cancelled polls, uncertainty over the ability of the existing intra-elite pact to guarantee future returns or to impose future sanctions became intense. This uncertainty persuaded those with votes to ‘trade promises for future concessions to whomever offered the highest price … without regard for previous commitments made’.47 More importantly, even the outwardly unified GEMA elites were fragmenting behind the scenes as they weighed their options of whom to field as a candidate for the chairmanship of KANU.48 The fragmentation was not severe enough to threaten the overall stability of the dominant coalition at that point because the Kenyatta persona was still there as the ultimate safety valve. Even though Moi did not seem to be his chosen heir, Kenyatta opted not to stray too far from de jure institutions when he insisted that the Change the Constitution group bring their agenda before Parliament.49

Admittedly, the constitutional power transfer could not have been taken for granted based only on the political factors described above. There was a critical exogenous factor that had tilted the balance in favour of constitution-based transition in the face of an otherwise indeterminate situation. That was the fact that Kenyatta died in Mombasa following a normal day at work. After the transition had been secured, the Moi government exposed plans that had been hatched to prevent Moi from taking over the presidency by military means.50 Although systemic factors assured the flimsiness of the plot in the first place, good luck could also take some credit in delivering Moi's succession.

Reinforcing the effects of the systemic factors, the occurrence of the coffee boom in the year leading up to the transition was another exogenous component. Kenyatta's ability to incorporate new elites into the fold towards the end of his reign, predicated on a strategy of ensuring eventual stability, was fuelled by patronage resources earned through the coffee windfall.51 There is also another twist to how the coffee boom might have affected the Kenyatta succession. Normally, the direct effect of the coffee boom would have empowered the Kikuyu elites since earnings would be concentrated in the coffee-growing country of Central Province. However, in 1977–8 there was widespread smuggling of coffee from the then dysfunctional Uganda for export via the port of Mombasa. Such activities helped to spread the benefits of the boom to elites in the rest of the country, especially in the Western and Rift Valley districts bordering Uganda.52

In the absence of an institutional foundation transcending the personality cult of Jomo Kenyatta, the political elites were not sure if their privileges would be preserved under the next president. At the same time, no single faction was strong enough to circumvent the constitutional path. The Kikuyu elites had been able to manipulate the potential path of succession on at least two earlier occasions – in 1964 and 1968 – using constitutional amendments.53 Their inability to do the same in 1976 exposed their fragmentation, forcing the crucial civil service – particularly the intelligence services – to throw in their lots with Njonjo and Moi.54 Given the apparent subordination of other political formations to the executive and the predominance of the Kikuyu in that body, the point of crisis that forced the elites to resort to the rule of law would not have arrived if the Kikuyu hegemony had not fragmented. The rents enjoyed by the elites were too large for them to try a risky bargain. Moreover, the state was consolidated enough to enforce a minimum guaranteed bargain. Under those conditions, the elites chose to resort to the default option of adhering to the constitutional provisions to manage the transition.

The Moi succession

The onset of multi-party competition in the early 1990s unleashed a proliferation of non-programmatic “briefcase parties” in Kenya.55 KANU maintained its stranglehold by responding to the decentralization of de jure competition with the decentralization of de facto power – in other words, the public deployment of physical force. The violence that was perpetrated around the 1992 elections was as much an instrument solidifying the bond between the Kalenjin sub-groups as it was a tool of coercion directed towards other ethnic groups residing in the Rift Valley.56 In this way, the Rift Valley elites within KANU entrenched their position as a major political bloc at the centre of the dominant coalition throughout the 1990s, using an eclectic mix of state and localized violence.57 Until the late 1990s, when the government was forced by the international financial institutions gradually to relinquish its gatekeeper role in the finance sector of the economy, most economic elites had no choice but to remain allied with KANU as the main dispenser of economic rents.58

The opposition parties began the decade as personality cults embedded in ethnic allegiances. The same familiar names from the independence era, such as Jaramogi Oginga Odinga and Mwai Kibaki, re-emerged as key opposition figures to dominate the political scene at least in the first half of the 1990s. The intra-elite pact continued to be renegotiated using violence as leverage. When Moi blamed the post-election violence in 1997 on Kibaki's threat to contest the election results in court, he was unwittingly admitting to the use of violence as a means of renegotiating the intra-elite pact.59 Towards the second half of the decade, opposition leaders scrambled to rectify their organizational deficiencies at the same time as they pushed for constitutional review to create a more level playing field.60

In the aftermath of the 1997 elections, the ruling party was implicitly divided along the lines of KANU-A and KANU-B. However, the opposition remained so fragmented that the prospect of any shift in the balance of power depended on what would happen inside KANU.61 The challenge of forging a viable opposition alliance persisted because whether the leader of a given ethnic group would be willing to run as number two on a party's ticket depended on the distribution of executive power ex post. That is one of the reasons why the opposition decided to take advantage of Moi's lame duck status to shift the battleground to the constitutional review process before the 2002 elections.62 However, even the attempt to kick-start the review process did not escape the mire of ethnic problems. The original movement of civil society organizations to facilitate the process was portrayed by KANU loyalists and Odinga's colleagues as a front for Kikiyu power mongering.63 In effect, the longstanding latent rift between the Central Province communities, on the one hand, and the Rift Valley and Luo ethnic groups on the other resurfaced in an anticipation of impending contest for de jure institutions. The supra-ethnic alliance of the 2002 elections can be seen as simply an interlude before the same rift re-emerged with the constitutional referendum in 2005.

The constitutional review process could no longer serve as an instrument to forge alliances once it was externalized as a largely independent commission headed by a renowned constitutional scholar, Yash Pal Ghai.64 For most of the 1990s, the political elites were vying for power and privileges within the dominant coalition by using their ethnic blocs as leverage to negotiate within a given constitution. This time around, the constitution itself became an object of negotiation. However, the negotiation was removed – to a significant extent – from the traditional centres of intra-elite contest into the hands of a ‘third-party arbiter’ in the form of Ghai. In the absence of a clear timeline for delivering a new constitution, however, this new development created uncertainty regarding the de jure distribution of power and privileges that would be applicable after the elections.65 The uncertainty had an upside, though. It mellowed the winner-takes-all mentality of the political elites, which otherwise would have consigned them to their ethnic fortresses.

When Moi finally picked Uhuru Kenyatta – son of Jomo Kenyatta – as his successor, the four-decade-old party started to unravel in earnest. Only two months before the 2002 elections, several splinter groups, including Odinga's bloc, deserted the KANU camp to join the opposition umbrella alliance. To put it bluntly, the National Alliance Rainbow Coalition (NARC) was nothing more than an emergency assembly point for a disparate group of politicians who were faced with a fast-changing political landscape shaken up by the fragmentation of KANU.66 The lack of political and organizational cohesiveness of NARC was exposed during the nomination process, which was marred by violence and the defection of disgruntled nominees.67 Although most of the Central Province was clearly behind Kibaki, the fragmentation of the Kikuyu elites between the two presidential candidates who both had Kikuyu credentials took much of the ethnic heat out of the contest symbolically.68

What was critical for the orderliness of the transition was the incentive and capacity of the incumbent party to use violence to try and renegotiate the results after the event. In the first place, KANU had disposed of a large part of the ethnic munitions it used to incite violence through the 1990s as soon as it nominated a Kikuyu presidential candidate. Even if the KANU elites had decided to incite violence as in previous elections, they would not have had the help of key Rift Valley politicians such as William ole Ntimama and Kipruto Kirwa, who had already joined the opposition.69 KANU could not have relied even on its newfound allies in the Mungiki sect, considering that they had already fallen out before the elections.70

Although there was subtle bias in the allocation of state resources favouring Kenyatta's campaign, Moi refrained from using his position to rig the elections.71 The intra-elite pact, which had been shaky ever since the reintroduction of multi-party elections in 1992, could not be sustained with the threat of ‘mutually assured destruction’ once KANU had fragmented, followed by a discord in the Kikuyu and KAMATUSA camps.72 Economically, the tense relationship between the government and international financial institutions, as well as the plummeting index of the Nairobi Stock Market, exposed the precarious rent base of the dominant coalition a few months ahead of the elections.73 In the absence of unified political and ethnic blocs to ensure the mobilization of rents and violence in case of confrontation, the elites chose to honour the electoral laws.

Conclusion: Kenya at the crossroads

By way of drawing together the central argument and implications of this article, I shall briefly contrast the orderly transitions of 1978 and 2002 with the post-election crisis of 2007–8 in terms of the different forms of uncertainty that might have affected the intra-elite pact. The first of these is the fragmentation of major elite blocs. Before both orderly transitions in Kenya, a major axis in the dominant coalition fragmented, triggering further fragmentation along other faultlines. Within the one-party state of the Kenyatta era, the fragmentation of the Kikuyu hegemony between 1975 and 1977 gave rise to the fracturing of KANU. The sequence of fragmentation was reversed at the time of the multi-party elections of 2002. Then the fragmentation of KANU led to the loosening of ethnic fortresses. The ethnic blocs had enough time to regroup after the 2002 elections, especially during the 2005 constitutional referendum, which effectively served as a dress rehearsal for the 2007 showdown.74

The second form of uncertainty relates to exogenous developments. In the case of both orderly transitions, elite fragmentation, which was the source of endogenous uncertainty, was intensified by uncertainty emanating from external factors. In 1978, the health of Kenyatta was the main source of external uncertainty. In 2002, the constitutional review process, which was delegated to a relatively independent commission, generated external uncertainty. By contrast, in 2007, it was obvious that Kibaki would run for a second term. Moreover, the uncertainties related to constitutional review had been laid to rest when the draft constitution was defeated in 2005.

The final form of uncertaity is economic: the Kenyatta succession occurred at a time when the global economy was sandwiched between two oil shocks and recessions. According to Table 1, real GDP in Kenya grew at an average rate of 4.2 percent in the three years preceding the 1978 transition. However, the windfall gain from the 1977–8 coffee boom masks many cracks, such as high inflation and the widening budget and current account deficits.75 Towards the end of the Moi era, Kenya's economy was in the doldrums with a three-year average growth of 1.6 per cent. The economy recovered significantly during Kibaki's first-term in office. But more importantly, growth volatility declined. As Table 1 shows, the coefficient of the variation of economic growth was significantly higher before the two orderly transitions than at the time of the 2007 elections. As a result, the elites would have been more uncertain about the availability of rents ahead of the 1978 and 2002 transitions.

Table 1.

Economic growth and volatility ahead of the three crossroads

Period Average GDP growth Coefficient of variation of GDP growth 
1975–7 4.2 1.1 
2000–2 1.6 1.1 
2005–7 6.4 0.1 
Period Average GDP growth Coefficient of variation of GDP growth 
1975–7 4.2 1.1 
2000–2 1.6 1.1 
2005–7 6.4 0.1 

In summary, elite fragmentation appeared to be both a source and a result of uncertainty around the 1978 and 2002 transitions. Violence needs some level of organization, which in turn needs some degree of foresight. Kenya's elites were not sure when to return to their ethnic trenches because of external uncertainties. But the more they stayed outside their comfort zones, the more uncertain they became. That is when they realized that the rule of law was actually the safest spot to congregate. Soon after 2002, the elites had already begun to wander off that default position. They had already acquired enough foresight by 2005 to return securely to their ethnic trenches before the 2007 elections.

Elite fragmentation and uncertainty could sometimes produce a recipe for stability. As demonstrated in the conceptual framework of this article, the flow of rents within the dominant coalition is facilitated by the synchronization of political, economic, and ethnic organizations. The stability of the dominant coalition is maintained through connecting the three sets of organizations like the gears of a conveyor belt. Ironically, the constructive adjustments that the dominant coalition in Kenya had undergone in 1978 and 2002 happened because some of the crucial political and ethnic organizations were fragmented to the extent of driving the whole coalition to extreme uncertainty. The fragmentation on both fronts was so severe that the elites could not regroup quickly in their ethnic enclaves, as they would have done under normal circumstances. The rule of law was observed because the elites could no longer afford to gamble by trying to rewrite the rules in the face of such fragmentation.

For many African countries, one of the challenges in ensuring political stability stems from the lack of organizational platforms that can lure elites out of their ethnic fortresses. Political and economic uncertainties can be instrumental in motivating elites to try out new alliances. In general, institutions are expected to help create some level of certainty about the rules of the game. However, given the low levels of institutional development prevailing in most African countries, the certainty of rules that institutions provide may afford ethnic entrepreneurs sufficient foresight to reformulate their ethnic agendas with respect to those institutions. Hence, institutions should be dynamic enough not to be bogged down in inter-ethnic competitions. Often, political stability in weakly institutionalized countries depends on whether the elites are more certain about the enforcement of inter-ethnic pacts than they are about intra-ethnic ties. Institutions are likely to promote political stability if they create greater certainty about inter-ethnic interactions than about intra-ethnic relations.

Mike Pflanz and Richard Holt, ‘Kenya on the brink amid “genocide” claims’, The Telegraph, 2 January 2008, <http://www.telegraph.co.uk/news/worldnews/1574358/Kenya-on-the-brink-amid-genocide-claims.html> (16 March 2015).
See, for instance, Susanne D. Mueller, ‘The political economy of Kenya's crisis’, Journal of Eastern African Studies 2, 2 (2008), pp. 185–210; Daniel Branch and Nic Cheeseman, ‘Democratization, sequencing, and state failure in Africa: lessons from Kenya’, African Affairs 108, 430 (2008), pp. 1–26.
For succinct analysis of the role of ethnicity in Kenyan politics since the early twentieth century, see Nicholas Nyangira, ‘Ethnicity, class, and politics in Kenya’, in Michael G. Schatzberg (ed.), The political economy of Kenya (Praeger, New York, NY, 1987), pp. 15–31; John Lonsdale, ‘Moral and political argument in Kenya’, in Bruce Berman, Dickson Eyoh, and Will Kymlicka (eds), Ethnicity and democracy in Africa (Ohio University Press, Athens, OH, 2004), pp. 73–95.
There were six changes made to the old constitution in 1982, the year Kenya became a de jure one-party state. The number of constitutional amendments peaked at 10 per year in the election years of 1992 and 1997.
Anyang Nyong'o, ‘State and society in Kenya: The disintegration of the nationalist coalitions and the rise of presidential authoritarianism 1963–78', African Affairs 88, 351 (1989), pp. 229–51.
Nicolas van de Walle, African economies and the politics of permanent crisis, 1979–1999 (Cambridge University Press, Cambridge, 2001).
See, for example, Patrick Chabal and Jean-Pascal Daloz, Africa works: Disorder as political instrument (James Currey, Oxford, 1999), and Jean-Francois Bayart, The state in Africa: The politics of the belly (Longman, London and New York, NY, 1993).
One of the few exceptions is the work of Gero Erdmann and Ulf Engel, who redefined neo-patrimonialism as ‘a mixture of two co-existing, partly interwoven, types of domination: namely, patrimonial and legal-rational bureaucratic domination’. See ‘Neopatrimonialism reconsidered: Critical review and elaboration of an elusive concept’, The Journal of Commonwealth and Comparative Politics 45, 1 (2007), pp. 95–119.
Angelique Haugerud, The culture of politics in modern Kenya (Cambridge University Press, Cambridge, 1993).
Robert H. Bates, Rui J. P. de Figuiredo, Jr, and Barry R. Weingast, ‘The politics of interpretation: Rationality, culture, and transition’, Politics and Society 26, 4 (1998), pp. 603–42.
Chabal and Daloz, Africa works.
Douglass C. North, John J. Wallis, and Barry Weingast, Violence and social orders: A conceptual framework for interpreting recorded human history (Cambridge University Press, Cambridge, 2009).
Nyong'o, ‘State and society in Kenya’.
Branch and Cheeseman, ‘Democratization, sequencing, and state failure in Africa’, pp. 3.
Economic rent, in the current context, is defined as economic value created due to exclusive access to regulatory discretion or scarce resources. In other words, it is a value accruing to the owner of a resource over and above the price that would be fetched in a perfectly competitive market.
See, for example, Bayart, The state in Africa, and Erdmann and Engel, ‘Neopatrimonialism reconsidered’.
North, Wallis, and Weingast, Violence and social orders.
Chabal and Daloz, Africa works.
Robert H. Bates, ‘Ethnic competition and modernization in contemporary Africa’, Comparative Political Studies 6, 4 (1974), pp. 454–84.
Bayart, The state in Africa.
Nicola Swainson, ‘State and economy in post-Colonial Kenya, 1963–1978’, Canadian Journal of African Studies 12, 3 (1978), pp. 357–81.
Nic Cheeseman, ‘Political linkage and political space in the era of decolonization’, Africa Today 53, 3 (2006), pp. 3–24.
Bayart, The state in Africa.
Colin Leys, Underdevelopment in Kenya: The political economy of neo-colonialism (East African Educational Publishers, Nairobi, 1975).
Rose W. Ngugi, ‘Development of the Nairobi stock exchange: A historical perspective’ (Nairobi, KIPPRA Discussion Paper 27, 2003).
David Himbara, ‘The failed Africanization of commerce and industry in Kenya’, World Development 22, 3 (1994), pp. 469–82.
In a telling declaration of the elites' reliance on patronage networks, Moi once spoke to a crowd at a fundraiser: ‘Some of you are wondering where I get so much money to donate. You think Moi has one or five friends? I have many friends.’ See Weekly Review, ‘Favourite sons: Moi campaigns for friends’, 24 August 1979.
Republic of Kenya, Development plan 1979–83 (Government Printers, Nairobi, 1979).
Mueller, ‘The political economy of Kenya’s crisis’.
Jomo Kenyatta, Suffering without bitterness. The founding of the Kenya nation (East African Publishing House, 1968).
Weekly Review, ‘Ngwataniro at crossroads as internal problems surface’, 12 December 1977.
Peter Gibbon, ‘Markets, civil society, and democracy in Kenya’, in Peter Gibbon (ed.), Markets, civil society, and democracy in Kenya (Nordiske Afrika Institutet, Uppsala, 1995).
Ben Knighton, ‘Going for cai at Gatundu: reversion to a Kikuyu ethnic past or building a Kenyan national future’, in Daniel Branch, Nic Cheeseman, and Leigh Gardner (eds), Our turn to eat: Politics in Kenya since 1950 (LIT Verlag, Berlin, 2010), pp. 107–28.
Rick Lyman, ‘In Kenya, a path toward multiparty politics’, The Inquirer, 3 May 1992.
Human Rights Watch, Divide and rule: State-sponsored ethnic violence in Kenya (Human Rights Watch, New York, NY,1993).
Joel D. Barkan, ‘Divergence and convergence in Kenya and Tanzania: pressure for reform’, in Joel. D. Barkan (ed.), Beyond capitalism vs. socialism in Kenya and Tanzania (East African Educational Publishers, Nairobi, 1994).
Weekly Review, ‘GEMA speaks on politics’, 19 May 1975.
John R. Nellis, ‘Is the Kenyan bureaucracy developmental? Political considerations in development’, African Studies Review 14, 3 (1971), pp. 389–401.
Weekly Review, ‘Please cool it’, 10 November 1975.
Wainainah Kiganya (ed.), Kenyatta cabinets: Drama, intrigue, triumph (Kenya Yearbook Editorial Board, Nairobi, 2012).
Charles Njonjo, ‘Interview with Jeff Koinange’, K24 TV, 10 July 2012.
Jay E. Hakes, The parliamentary party of the Kenya African National Union: Cleavage and cohesion in the ruling party of a new nation (Duke University, Unpublished PhD dissertation, 1970).
Kenya Box Office, ‘Who killed J. M. Kariuki?’ [Documentary film], 2013, <http://www.kenyaboxoffice.com/component/movies/?view=movie&mid=121> (24 October 2014).
US Embassy Nairobi, ‘Unclassified telegram’, The US National Archives and Records Administration, 12 March 1975, <http://www.archives.gov/> (1 July 2014).
Kenya Box Office, ‘Who killed J. M. Kariuki?’.
Joseph Karimi and Philip Ochieng, The Kenyatta succession (Transafrica, Nairobi, 1980).
Jennifer A. Widner, The rise of a party-state in Kenya: From ‘Harambee!’ to ‘Nyayo!’ (University of California Press, Berkeley and Los Angeles, CA, 1992).
US Embassy Nairobi, ‘Unclassified telegram’, The US National Archives and Records Administration, 10 March 1977, <http://www.archives.gov/> (1 July 2014).
Karimi and Ocheng, The Kenyatta succession.
Weekly Review, ‘Strange tales’, 3 November 1978.
David W. Throup, ‘The construction and destruction of the Kenyatta state’, in Michael G. Schatzberg (ed.), The political economy of Kenya (Praeger, Westport, CT, 1987).
David W. Cohen and E. S. Atieno Odhiambo, Siaya: The historical anthropology of an African landscape (Heinemann Kenya Ltd, Nairobi, 1989).
David Goldsworthy, Tom Mboya: The man Kenya wanted to forget (Heinemann Educational Books Ltd, Nairobi, 1982).
Karimi and Ocheng, The Kenyatta succession.
Mueller, ‘The political economy of Kenya's crisis’.
Patrick Mutahi, ‘Political violence in the elections’, in Henry Maupeu, Musambayi Katumanga, and Winnie Mitullah (eds), The Moi succession: The 2002 elections in Kenya (Transafrica Press, Nairobi, 2005).
Over a quarter of the persons adversely mentioned in the Akiwumi Commission report are MPs. See Republic of Kenya, Report of the judicial commission appointed to inquire into tribal clashes in Kenya (Government Printers, Nairobi, 1999).
Leonardo R. Ariolla, Multiethnic coalitions in Africa: Business financing of opposition election campaigns (Cambridge University Press, Cambridge, 2013).
‘Moi's last lap’, Africa Confidential 39, 4 (20 February 1998).
Makau Mutua, Kenya's quest for democracy: Taming Leviathan (Lynne Rienner Publishers, Boulder, CO, 2008).
‘Close shave’, Africa Confidential 39, 1 (9 January 1998).
Raila Odinga, The flame of freedom (Mountain Top Publishers Ltd, Nairobi, 2013).
‘Unconstitutional’, Africa Confidential 41, 2 (21 January 2000).
Parliament was barred from making piecemeal amendments to the constitution until the completion of the review process. See Ahmad Khalif, ‘This country's future depends on law review’, Daily Nation, 26 April 2001.
Paul Muite, ‘Opposition plight in impending elections’, Daily Nation, 6 January 2002.
Odinga narrates the frantic and tentative nature of the negotiation process on the morning of 14 October 2002, only a couple of hours before the rally that saw the formation of NARC. See Odinga, The flame of freedom.
Mutahi, ‘Political violence in the elections’.
Njenga Karume, the wealthy chairman of the then-defunct GEMA, was candid enough to acknowledge the tribal calculation behind his last-minute switch to support Kenyatta: ‘This time round we must ensure our people join the winning team so that they are back in the government to improve their lot.’ See Njeri Rugene and Mburu Mwangi, ‘Karume crosses over to Uhuru’, Daily Nation, 2 September 2002, <http://allafrica.com/stories/200209020015.html> (10 November 2014). Others must have made the same calculation, but on the other side of the electoral divide.
‘At the end of the rainbow’, Africa Confidential 43, 21 (25 October 2002).
The Standard, ‘Kanu now bars Mungiki aspirant’, 28 November 2002, <http://allafrica.com/stories/200211280213.html> (16 March 2015).
Jacinta Ochieng, ‘KBC won't air Narc adverts’, Daily Nation, 24 December 2002, <http://allafrica.com/stories/200212230708.html> (10 November 2014).
The KAMATUSA are the dominant Rift Valley tribes – Kalenjin, Maasai, Turkana and Samburu – that formed the core of Moi's constituency.
Eliud Chisika and Aggrey Ouma, ‘Constitution uncertainty hurting economy, says NSE’, The Standard, 11 July 2002, <http://allafrica.com/stories/200207110085.html> (10 November 2014).
Uhuru Kenyatta, the man who undermined the Kikuyu hegemony by running against Kibaki in 2002, was already on his way back to his ethnic enclave several months before the elections. See Cabral Pinto, ‘Uhuru is headed straight to the Kibaki-Moi political arms’, Daily Nation, 11 July 2007, <http://allafrica.com/stories/200707110060.html> (16 March 2015).
Francis M. Mwega and Njuguna S. Ndung'u, ‘Explaining African economic growth performance: The case of Kenya’ (Nairobi, AERC Working Paper No. 3, 2002), pp. 1–44.