Introduction
The political economy origins of Africa’s development challenges have been linked to the adverse influences of unconstrained elites (Ndulu and O’Connell 1999) and ethnic fragmentation (Easterly and Levine 1997). The interlocking nature of the two factors – that is, elites rely on ethnic allegiances at the same time as ethnic allegiances are sustained by elite manipulation – has complicated the task of identifying the primary source of the problem. This challenge seems to have rendered many authors undertaking quantitative research to resort to treating ethnicity as a predetermined factor. In the realm of the more qualitative Africanist literature, however, there are plenty of insightful studies that have dealt with the historical and institutional origins of political ethnicity in Africa (see, for example, Berman 1998; Lonsdale 2004; Lynch 2006). These authors have provided a nuanced analysis of ethnicity from a historical point of view, employing a mix of instrumentalist and moral economy explanations. However, there is still a dearth of systematic explanation – especially at a theoretical level – as to what makes political ethnicity a perennial feature of an otherwise changing society.
This article seeks to examine why ethnic allegiances may persist as the most dominant medium for organising collective action in a society where competing forms of political identity exist. Specifically, the article attempts to explain why trans-ethnic forms of organisation fail to take root among elites in certain polities in spite of economic and political changes that are supposed to transform intra-elite relationships. The contribution of the article is twofold. First, on a theoretical level, the article situates ethnicity in a broader framework of social orders and elite bargaining, highlighting its role as an organisational tool. Second, on an empirical level, it applies the theoretical framework to explain the persistence of political tribalism in Kenya. In this regard, the article sheds light on an existing empirical debate from a new theoretical angle.
Historically, Kenya was a regional centre in the colonial system of production. Therefore, it had already developed a relatively robust modern sector before independence. The Mau Mau rebellion in the decade before independence forced the colonial administration to nurture a loyalist African elite with the goal of driving a wedge of class differentiation among the native population. Economic growth performance was strong, albeit fluctuating, in the first two decades after independence. However, patron–client networks continued to be organised along ethnic lines throughout the postcolonial history of the nation. The persistence of high income inequality did not give rise to a Marxian class society. Instead, working-class Kenyans remained allied to wealthy elites from their respective ethnic groups as ethnicity continued to be the most viable means of collective action.
I use the Violence and Social Orders framework developed by North, Wallis, and Weingast (2009, NWW hereafter) to characterise the organisational role of ethnicity in facilitating intra-elite pacts. I also consult a few other theories of political identity to formulate a coherent framework to explain the persistence of political tribalism in Kenya. I attempt to highlight the instrumental role of ethnicity without forfeiting the historical dimension of group identity. The NWW framework posits that political order emerges in weakly institutionalised societies when powerful actors with the capacity for violence opt for cooperation out of mutual self-interest. The elites may choose to cooperate in the interest of reaping the peace dividend generated through the mutual third-party enforcement provided by the resulting coalition. Such states in which order is maintained by the consent of the elites instead of impersonalised institutions are called natural states. The overarching organisation of the elites – who are referred to as ‘specialists of violence’ in NWW – is called the dominant coalition. In the case of Kenya, the dominant coalition may be construed as a loose alliance between political leaders, business elites, senior bureaucrats and traditional leaders. These actors need not belong to the same party or organisation as long as they implicitly agree to the existing social order and hold a stake in it.
The elites who normally generate political bargaining power from their client organisations come together to build a dominant coalition buttressed by the selective distribution of rents. Rent, in this context, refers to economic value created thanks to exclusive access to regulatory discretion or scarce resources. The dominant coalition in natural states is an adherent organisation built on incentive-compatible allocation of rents among elites and sustained without outside third-party enforcement. Unlike intra-elite relations, the relationship between the elites and their clients is facilitated through contractual organisations such as patron–client ties enforced by the dominant coalition. Building on key insights from the NWW framework, I portray ethnic groups as clientelist organisations headed by ethnic patrons. Then I draw further on insights from other frameworks in the literature to characterise the interplay between ethnic allegiances and historical institutions in a postcolonial society such as Kenya’s.
In the first half of the empirical section, it is shown that the process of capitalist class formation in Kenya was not robust enough to replace ethnic allegiances with class consciousness, although income inequality continued to run high for several decades. In the presence of significant inter-regional inequalities underlain by agro-ecological disparities, economic growth in the first decade and a half after independence failed to weaken ethnic sentiments because the gain accrued to some ethnic groups more than to others. In a latter part of the empirical section, I take a close-up look at the role of elite bargaining in making political ethnicity a key building block of the postcolonial social order in Kenya. As long as the elites see their leadership of ethnic organisations as their most important bargaining chips, they will strive to preclude other forms of mass mobilisation that could potentially undermine the political salience of ethnicity. Kenya’s history is replete with political assassinations of powerful actors who had threatened to build viable trans-ethnic alliances. In the 1990s, Kenya’s elites tried to pre-empt the liberalising effect of multiparty competition by fortifying their tribal enclaves with the help of ethnically homogenous political parties. More fundamentally, the nature of socioeconomic change in Kenya has been such that the rents – such as white-collar employment and agricultural land – that the elites use to stabilise the dominant coalition remain tied to group leadership.
It should be acknowledged that the Africanist literature has addressed many of the above issues in one way or another. Berman (1998) provides a rigorous analysis of the historical origins of political ethnicity in Africa. Lonsdale (2004) draws a useful distinction between moral ethnicity and political tribalism. In doing so, he has shown that the role of ethnicity in public life is more nuanced than the simplistic view it receives in popular discourse. Lynch (2006) offers more insights by highlighting the process of negotiating ethnicity in an environment of political and social fluidity. Using specific cases from two provinces in Kenya, she demonstrates that ethnic identity can be reformulated as a response to external factors. There has been a renewed interest in exploring the political salience of ethnic identities in Kenya following the post-election violence in 2007–2008. Bratton and Kimenyi (2008) present statistical evidence showing that Kenyans vote along ethnic lines partly because they have mistrust of ethnic ‘others’ whose behaviour, they think, is influenced primarily by ethnicity. Jenkins (2012) shows how grassroots consciousness of ethnic territoriality leads to the persistence of the native–immigrant distinction which has contributed to the salience of ethnic identities at the community level. In a sense, both Bratton and Kimenyi, and Jenkins emphasise the role of bottom-up ethnic exclusionism. However, this does not preclude the elites’ ability to introduce alternative narratives at times when their coalition stands to benefit from a new narrative. Lynch (2014) shows how the winners of the 2013 elections framed their alliance of convenience which would normally go against bottom-up understanding of ethnic allegiances in the context of protecting Kenya’s sovereignty from a threat by the International Criminal Court.
In this article, I hope to shed a new light on the existing debate from the point of view of intra-elite bargaining and dynamic social orders. In so doing, I will attempt to locate the source of ethnic territoriality emphasised in Jenkins (2012) in the nature and distribution of economic rents. I will also characterise the fear of ‘others’, a phenomenon that has been highlighted in recent literature, as an equilibrium response supported by the past actions of elites as part of the bargaining process towards a stable social order. Finally, I will place the discussion firmly in the context of historical grievances and economic and political marginalisation that are acknowledged by the Truth, Justice and Reconciliation Commission as the root causes of ethnic tensions in Kenya.1 In both the theoretical and empirical sections of the article, I build on the wealth of insights available in the literature to weave a coherent body of explanations centred around the nexus between structural constraints and strategic bargaining.
The rest of the article is organised as follows. Section two provides a theoretical framework. Section three reviews the historical development of political ethnicity and class as competing identities in Kenya. Section four applies the theoretical framework to explain the persistence of ethnic allegiances. Section five concludes.
Ethnicity, modernisation and nation-building: a theoretical framework
Ethnic groups as patron–client organisations
Organisations are the backbone of natural states. The stability of natural state social orders hinges on the capacity of elites to organise their clients. In ethnically stratified societies, ethnic grouping can be an appealing medium to organise a network of patrons and clients. In many societies, there are multiple tiers of sub-ethnic divisions such as clans, sub-clans and families. The boundaries of ethnic delineation could be fluid to the extent that adjoining communities might identify themselves as the same or different ethnic group at different points in time. Sub-ethnic divisions may also morph into solid boundaries depending on the cultural and socioeconomic distance between the sub-groups. Such dynamics could be endogenous to external changes of broader scale as well as to the preferences of powerful actors.
In an autarkic setting, ethnic groups are self-sustaining socioeconomic structures fitted with their own set of patron–client networks. In that case, ethnic boundaries define the adherent organisations of sub-ethnic patrons, ensuring the third-party enforcement of intra-ethnic contractual ties. Nevertheless, different ethnic groups often converge on the same political-economic space, mostly in the form of a nation state. Such amalgamation could potentially put multiple ethnic leaders with viable capacity for violence on a collision course. In order to maintain the stability of the union, a dominant coalition of ethnic leaders needs to be formed. In the process, the co-ethnic leaders compete for a more prominent place in the larger dominant coalition. They also try to secure a bigger share out of the ‘national’ pool of resources for their clients in exchange for the clients’ support. Berman (1998) argues that the political mobilisation of ethnicity can be an instrument of demands from below for internal redistribution and external protection. The dominant coalition at a national level emerges as a mechanism to support the sub-ethnic organisations and to stave off potential violence among the ethnic leaders themselves.
Not all ethnic groups are the same in terms of their internal hierarchy and in how such cultural structures are aligned with corresponding socioeconomic contracts. In tight-knit communities with elaborate power structures enforced by traditional sanctions, patron–client ties could be self-enforcing. This is because rights and obligations in tight-knit communities are well balanced in that they often approximate impersonal institutions. On the contrary, in communities where the vertical boundaries of traditional power and the horizontal boundaries of group membership are more fluid, the dominant coalition could be instrumental in stabilising intra-ethnic contractual organisations.
As long as patrons realise that they could enforce contractual relationships within their own cliental network by invoking ethnic allegiances, they are likely to have weaker incentives to stabilise the dominant coalition than in the absence of strong ethnic allegiances. This, of course, depends on how distinct the ethnic groups are and how easy it would be for the clients to pass for another ethnic identity whenever they find it more beneficial to regroup under a different patron.2 If the dominant coalition is already stable, strong ethnic allegiances might not hinder the emergence of new interests and the corresponding reconfiguration of patron–client ties. But, generally, ethnic allegiances may lead to frequent violence by strengthening the fallback position of the elites in case of dominant coalition breakdown.
Wallis (2011) argues that agents may develop preference for impersonal rules over relation-based organisations when the balance between organisation-specific rents and individual-specific rents shifts away from organisation-specific rents. Organisation-specific rents are the type of privileged benefits that are tied to membership of a particular organisation. For instance, a monopoly right negotiated by a farmers’ cooperative provides members with organisation-specific rents. On the contrary, individual-specific rents are rents that are tied to the personal quality and skills of individual elites, accruing across the dominant coalition regardless of specific organisations. In the context of the previous example, the individual-specific rents of a farmer lie in their farming skills as opposed to their membership of a specific cooperative. One of the most appealing organisation-specific rents that ethnic leaders could generate from their ethnic network comes in the form of lower downside risk when resources are scarce and violence looms. The relevance of this class of rents diminishes when times are good. Hence the elites could capitalise on their individual-specific rents to start a virtuous cycle of reconfigurations that might eventually lead to a more liberal arrangement. The prominence of organisation-specific rents results in what Émile Durkheim refers to as ‘mechanical solidarity’, whereas the prominence of individual-specific rents contributes to the development of ‘organic solidarity’. Mechanical solidarity arises from homogeneity of individuals whereas organic solidarity requires specialisation and complementarities between individuals.
The following subsection presents theoretical perspectives on how economic growth and class stratifications within ethnic groups could change forms of allegiances. I also discuss under what conditions modernisation and growing inequality might fail to alter forms of allegiances.
Identity and modernisation
The foregoing discussion suggests that, in a growing economy, elites might see opportunities to exploit individual-specific rents. Hence, they seek more stable rules that would guarantee those rents irrespective of their own group affiliation. In the short run, this could lead to regrouping of cliental networks according to complementary individual-specific rents. That way, the old individual-specific rents will be instrumental in creating new and bigger organisation-specific rents. The new organisations, however, are less likely to be based on ethnic allegiances since complementary individual-specific skills are more likely to be distributed across ethnic cleavages. A boost in trading activities, for example, could lead a tribal leader with good commercial acumen to identify more with members of the Chamber of Commerce than with his co-tribesmen (at least as far as public policy issues are concerned). This argument echoes the early modernisation theories which claim technological and economic development will assist in the birth of a new set of socioeconomic roles and group identities (for example, Fried 1967; Fox 1971).
One of the most common types of group identification that come along with capitalist development is class differentiation. The supremacy of ethnic grouping does normally imply the compartmentalisation of the working class and peasants in their own ethnic enclaves. The Marxian literature predicts that urbanisation and industrial development bring the working class together spatially (McAll 1990). The new spatial proximity results in class awareness, which in turn, according to Max Weber, leads to class struggle. This will solidify the new set of identities along class lines. If intra-ethnic wealth inequality is rising across the board, it provides a fertile ground for the development of class divisions that could eventually dominate ethnic cleavages.
So why did the advent of capitalism in many multi-ethnic societies fail to relegate ethnic identity to the backseat? Basically, affective ties, once they have formed, do not easily morph into other forms of association based only on the rational choice of individual group members. Horowitz (1985) summarises plenty of experimental evidence that shows individuals with some sense of loyalty to their group are willing to incur costs to maximise intergroup differentials. Melson and Wolpe (1970) claim that modernisation could further intensify communal tensions instead of reshaping group allegiances. The process of social mobilisation that normally accompanies modernisation creates new competition for scarce resources. This competition renders elites to stick to organisation-specific rents to ensure support from their clients while maintaining their place in the dominant coalition. Moreover, not all elites have the same potential to capitalise on individual-specific rents. Bates (1974) highlights the fact that both the distributions of modernity and ethnic groups are governed by the factor of space: ‘Where modernisation takes place often largely determines who gets modernised’ (464, emphasis in original).
Growing intra-ethnic inequality also may fail to weaken communal allegiances. The class consciousness that might emerge within ethnic groups thanks to capitalist development could be pre-empted by existing customary channels of redistribution. Apart from the moral economy of customs, the political imperative of forging a stable dominant coalition also requires the sharing of rents between patrons and clients. No matter how unequal the incidence of benefits may be among co-ethnics, the collective advantage of extracting goods and services from the modern sector as a cohesive group contributes to the persistence of ethnic identification (Bates 1974). Drawing on evidence from African societies, Miller (1974) argues that kinship plays a vital social welfare function in communally stratified societies as most of them do not have the resources necessary to emulate the modern welfare state.
Presumably, a multi-ethnic society should first develop a viable national identity before it becomes feasible for its members to regroup along trans-ethnic lines such as class or professional affiliation. Penn (2008) pins the viability of the nation state on the balance between two factors: the relative homogeneity of ethnic groups and the level of horizontal inequality among groups.
Individuals will be more likely to identify with their ethnic group as it becomes more homogenous and as its rival groups become less homogenous … When groups are not too dissimilar in wealth then a uniform increase in national wealth decreases the likelihood that members of all groups will identify ethnically. (Penn 2008, 962–963)
In summary, elites may use communal bonds as a bargaining chip in the process of nation-building until the rents they get from trans-ethnic organisations outstrip the rents from ethnic organisations. However, the development process that is expected to promote trans-ethnic sources of rents produces further ethnic antagonisms that may perpetually eclipse the prominence of other forms of allegiances.
The historical development of class and political ethnicity in Kenya
In pre-colonial Kenya, where population pressure was not an issue and the village homestead was characterised by a self-contained economy, ‘people lived ethnicity rather than rationalised it, having little need to defend its ways against outside oppression’ (Lonsdale 2004, 77). Nevertheless, there was already a discernible degree of variation in cross-communal relations across different geographic regions. For example, the scramble for land and consequent tribal raids that used to happen between the Kikuyu and the Massai ethnic groups near the centre of the country even before the British arrived indicate that competitive tribalism is not necessarily of a colonial creation (Miller and Yeager 1994).
The formation of African ‘reserves’ in 1915 is an important landmark in the installation of ethnicity as the primary basis of political organisation in Kenya. The colonial administration sliced the native reserves into a series of tribally homogenous units. Given this administrative structure, ethnic identity emerged as the only means of mobilising collective action. Accordingly, the first two African organisations formed in 1922 to lobby for native interests, namely the Young Kikuyu Association and the Young Kavirondo Association, were based on the exclusive membership of the Kikuyu and the (formerly) Kavirondo ethnic groups, respectively (Nyangira 1987). However, the two organisations could not be considered to have arisen out of a sense of competitive tribalism. Rather, they were motivated by the common cause of native empowerment against colonial subjugation. They emerged as two separate organisations owing to the administrative constraint imposed by the colonial government. As a matter of fact, it was Harry Thuku, the very person who founded the Young Kikuyu Association, who travelled to Kisumu to persuade ‘a small number of mission educated Luos to form the Young Kavirondo Association’ (Miller and Yeager 1994, 20).
The colonial administration itself was engaged in some form of ethnic re-engineering as in the case of the Luhya ethnic group of Western Province, whose collective identity was minted out of nine linguistically close tribes. That was done mainly in the interest of administrative convenience. The continued confinement of political mobilisation to ethnic associations gave rise to an endogenous process in which new forms of ethnic identity emerged through amalgamation. A prime example in this regard is the creation of the ‘Kalenjin’ ethnic group out of numerous Nandi-speaking subgroups towards the mid twentieth century. The Kalenjin identity is a result of a conscious attempt by local notables to garner greater bargaining power in an increasingly ethnicised surrounding (Lynch 2011). Once the dominant coalition was strongly rooted on ethnic grounds, the Nandi-speaking elites were convinced that their position in the national political space would be enhanced by a larger ethnic base.
The first native organisation in colonial Kenya that could be described as pan-ethnic was the Kenya African Union (KAU), established in 1942. The organisation’s founding leaders were dispassionate liberals with little ethnic-based following. However, as the political reality demanded more communally oriented leaders with populist appeal, the KAU grew into an overwhelmingly Kikuyu-dominated organisation with little mass presence outside the Central Province (Hornsby 2012). The key rallying point of the KAU was the return of ‘stolen lands’, an issue that was not as pressing for other groups farther from the ‘White highlands’ as it was for the Kikuyu. The White highlands represent parts of the country that were settled by, among other groups, the Kikuyu before colonisation and that were later reserved for settlers of European origin by the colonial administration. The geographic variation in colonial economic practices, particularly with respect to land, had effectively differentiated the causes of grievances of the native population in different parts of the country. This trend led different ethnic groups to rise up with divergent key demands, ultimately pushing them to their respective enclaves for political mobilisation.
By the time independence was on the horizon, the salience of ethnicity in Kenyan politics was fairly evident. In the lead-up to the pre-independence elections in the late 1950s and the Lancaster House Conferences that were organised by the British government to negotiate eventual self-governance in the early 1960s, two national parties emerged representing an alliance of politically and economically dominant tribes, i.e. Kenya African National Union (KANU), and that of other tribes who feared being dominated by the former, i.e. Kenya Africa Democratic Union (KADU). ‘KADU, in reality, was a defensive coalition of regional party bosses who felt threatened by KANU and had to take a defensive posture as a bargaining point in sharing power after independence’ (Nyong’o 1989, 234).
Parallel to ethnic clustering, Kenya’s society had also begun undergoing some level of differentiation along class lines before independence. To be sure, colonial policies were used to actively discourage African accumulation by reserving export crop production to white settlers. Nevertheless, the effect of the Great Depression necessitated the sourcing of African export crops as well as native economic mobilisation in an effort to increase tax collection. This development opened the door to the rise of an African middle class situated inside the native reserves. Increased demand for agricultural produce during the Second World War created further opportunities for some African traders to share in the already rising surplus brought about by intensification (Kitching 1980). Furthermore, the settler demand for more secure property rights in exchange for increasing supply in the war years created an institutional structure that would later be taken over by the African middle class (Bates 1989).
Independent Kenya emerged as one of the most unequal countries in sub-Saharan Africa, as shown in Table 1. Two decades after independence, Kenya’s Gini coefficient remained 41% higher than the sub-Saharan Africa average. The way accumulation was fostered in the first decade after independence had effectively hardwired the reproduction of inequality in the young nation. The two major resources that the new African elites had at their disposal were land and salaried employment. After acquiring large tracts of land, one would need to mobilise financial capital through loans in order to develop large-scale commercial farms. In most cases, wage employment was the only viable collateral and indicator of credit worthiness (Chege 1987, 98). The practice of providing credit based on landownership reinforced existing inequality and contributed to the predominance of the same group of elites in the political power structure for a long time (Leys 1975, 99).
| Period | Real GDP growtha | Gini coefficientb | Labour’s share of incomec |
|---|---|---|---|
| Kenya | |||
| 1961–1969 | 7.4 | 63.9 | 0.82 |
| 1970–1979 | 7.1 | 61.5 | 0.73 |
| 1980–1989 | 4.2 | 61.8 | 0.68 |
| Sub-Saharan Africa | |||
| 1961–1969 | 4.9 | 44.7 | 0.55 |
| 1970–1979 | 4.1 | 47.9 | 0.56 |
| 1980–1989 | 1.7 | 43.8 | 0.56 |
Sources:
aWorld development indicators.
bThe standardised world income inequality database.
cHeston, Summers, and Aten (2011).
The official policy of Africanisation was also used to create a new class of African bureaucratic elites. The Africanisation of the civil service was prioritised as the core of postcolonial nation-building. The civil service structure, however, was kept intact with its privileged ‘European’ and lower-level ‘African’ positions retaining their colonial characteristics. Only the race of the occupants of the ‘European’ positions had changed. In about five years after independence, the enhanced upward mobility of African personnel came to an end with the consolidation of a distinct bureaucratic elite class (Leonard 1984). The reliance of the elites on white-collar employment to get ahead economically in the formative decades could be part of the reason that Kenya featured a substantially higher share of labour in national income than the sub-Saharan Africa average (see Table 1).
The persistence of inequality in postcolonial Kenya did not happen in a vacuum. It was accompanied by commendable, albeit erratic, economic growth performance in the first decade and a half after independence. The uncertainties that had prevailed around independence in 1963 were defused with reassurance from the new African elites. As shown in Table 1, real GDP growth remained significantly higher than the sub-Saharan Africa average even in the politically rocky decade of the 1980s. The encouraging performance of the country in the first decade after independence had a lot to owe to the partial land reform which had redistributed settler land to more efficient smallholders, and the extension of the area cultivated in high-value cash crops (Ndulu and Mwega 1994).
Taking stock of the foregoing narrative on ethnicity, inequality and economic growth in pre- and early post-independence Kenya, one is able to discern the coexistence of political ethnicity and stubborn inequality. In summary, three important patterns are worth noting:
ethnicity came to define the political agenda on the eve of independence owing to the need for strengthening bargaining power in a society that is already balkanised by an existing administrative structure;
economic growth occurred hand in hand with persistent inequality throughout the formative years although most of the benefit seems to have accrued to labour;
after independence, most of the class formation was driven by the mere Kenyanisation of colonial privileges without changing the distribution significantly.
Persistence of ethnic allegiances: analytic narratives
I shall move next to investigate why ethnicity persisted as the prime basis of political mobilisation in the face of high inequality and expanding economic space. The potential explanations are organised in four claims motivated by the theoretical discussion in section two. The claims will subsequently be fleshed out with empirical evidence.
Income inequality remained high in the formative years of postcolonial Kenya. However, the development of the capitalist class has not been complete enough to create widespread class consciousness. The prominence of gentrified agriculture and white-collar employment, instead of capitalist industrialisation, as the mainstays of the postcolonial elite enhanced the use of ethnic allegiances for neutralising the impact of high inequality.
The growth of the fledgling African capitalist class that was propelled by increased aggregate demand during the Second World War was curtailed by the Mau Mau rebellion in the first half of the 1950s. To be sure, the colonial administration worked toward intensifying the process of class formation in an attempt to dismantle the broad social foundations of the Mau Mau. However, the state’s reaction to the rebellion shifted the locus of class formation from capitalist accumulation to rent-seeking by loyalist elites propped up with direct patronage from the state. During the state of emergency that was declared as a result of the rebellion, ‘the number of registered African companies dropped from 15% of Kenya’s total in 1952 to 2% in 1953 and then to zero in 1954 and 1955’ (Miller and Yeager 1994, 27–28). The insurgency forced the administration to recruit a loyalist self-defence force in Central Province, the Kikuyu ‘Home Guard’, numbering close to 25,000 men by 1954 (Hornsby 2012).3 The effect of the state’s reaction to the Mau Mau rebellion was not limited to weakening the basis of class formation. It also dismantled the channels of class action by emasculating the nascent union movement (Stichter 1975).
Table 1 shows that inequality continued to run high throughout the first three decades after independence. However, the markedly high share of labour in national income suggests that much of the inequality must have been a result of high wage inequality. The effect of salaried employment in sustaining inequality is signified by its role as a driver of agricultural investment and innovation (Collier and Lal 1984). However, the private sector had become less important in generating employment. The contribution of the public sector to recorded employment outside agriculture had grown from 43% in 1952 to nearly 55% in 1975 (Kitching 1980). Hence, the enduring link between urban employment and rural investment stunted the process of capitalist class formation.
What was emerging from the combination of an agrarian economic base and a lineage-based social context was not a nascent bourgeoisie but rather a group more akin to an incipient gentry – managers not of individually owned, industrial corporations but rather of family-owned, agricultural estates. (Bates 1989, 39)
Piecing all the above narratives together, we find a picture of how ethnic allegiance trumped class cleavages. First, wealth creation in Kenya depends on access to well-paid salaried employment; second, the economy was highly dependent on the public sector for generating salaried employment; finally, the public sector is controlled by a political class which draws its legitimacy from ethnic allegiances. The official policy of the post-independence administration states that ‘the vigorous implementation of traditional political democracy in the modern setting will eliminate, therefore, one of the critical factors promoting class divisions’ (Republic of Kenya 1965, 13). This amounts to admitting the role of ethnic allegiances in neutralising class consciousness since ‘traditional political democracy’ in postcolonial Kenya presumes the presence of civic-republican citizenship which in turn is embedded in ethnic identity (Ndegwa 1997).
Economic growth widened existing income gaps between regions with differential access to markets and agricultural opportunities. Since geographic income disparities coincided with ethnic cleavages, economic growth ended up enhancing inter-ethnic wealth inequality. Therefore the potentially liberalising effect of the increase in national wealth was offset by the feeling of collective alienation owing to increasing group inequality.
Kenya is a creation of agriculturally oriented settler colonialism. Therefore, the distribution of economic opportunities is shaped by suitability for cultivation of cash crops. As Hazelwood (1979) points out,
the greatest regional inequalities are the work of nature. One need look no further than the 307 mm rainfall in North Eastern Province and the 1084 mm in Central Province to find a major cause of inequalities between the two. (Hazelwood 1979, 175)
Therefore, the incidence of economic opportunities across different ethnic groups was bound to favour the residents of the Central and Western Highlands which were suitable for cash crop farming. The effect of agro-ecological differences was exacerbated by the privilege afforded to some Kikuyu districts during the Mau Mau emergency in terms of fast-track land consolidation and therefore opportunities to cultivate export crops (Kitching 1980).
The horizontal inequalities that were rooted in agro-ecological differences and Western influences such as modern education were not static. They gave rise to further inequalities through path dependence in the administrative and political processes. Hence, groups in the well-endowed regions had built productive capacity that would give them a head start once the yolk of political and economic repression was removed. As of 1974, Collier and Lal (1984) attribute only 4% of the variance in household income to inter-regional differences. This may seem to attenuate the importance of regional disparities. However, it is rather an indication that the effect of inter-regional differences manifests itself through other factors such as access to innovation and improved inputs.
Recall that salaried employment is a key driver of accumulation in early postcolonial Kenya. In 1972, 41.4% of top-level government officials in the country hailed from the Kikuyu ethnic group, which constituted only half as much of the total population (Nellis 1972).4 This can be contrasted with the shares of the Luo, who accounted for 8.6% of government positions in spite of 13.9% population share, and the Luhya, who accounted for 10.3% of government positions whereas they constituted 13.2% of the population. But it is difficult to attribute these disparities to conscious political bias. After all, the Kikuyu were the most educated of all ethnic groups in Kenya. ‘If all promotions and appointments could be controlled strictly by an objective, computerised test of merit, the Kikuyus probably would still hold almost all the positions they have today’ (Mesiler 1970, 117). However, the political dominance of the Kikuyu must have contributed to better access to complementary resources for the Kikuyu. Such resources are likely to have helped the Kikuyu utilise their human capital more effectively. By the mid 1970s the regional inequalities in welfare and development had become fairly visible (see Table 2). Central Province was clearly ahead of every other province, having registered substantial gains in education and access to welfare-enhancing services compared with the national average. When one contrasts that with the figures for Rift Valley Province, which attained far less than the average provision of education and welfare services despite being endowed with most of the country’s high-potential land, it becomes clear that development in Kenya of the 1970s was lopsided.
| Province | Central | Coast | Eastern | Nyanza | Rift Valley | Western |
|---|---|---|---|---|---|---|
| Major ethnic group (% of the most populous group, out of the national total, living in the province) | Kikuyu (73%) | Coastal people (98.4%) | Kamba (87%) | Luo (87%) | Kalenjin (95%) | Luhya (80%) |
| Population | 15.3% | 8.6% | 17.4% | 19.4% | 20.4% | 12.3% |
| Land | −13.0 | −6.0 | +10.0 | −17.2 | +9.3 | −10.9 |
| Land of high potential | −1.9 | −3.1 | −10.0 | −1.4 | +24.2 | −1.4 |
| Coffee smallholders (number) | +21.8 | −8.4 | +21.2 | +0.6 | −19.3 | −9.3 |
| Tea smallholders (number) | +18.5 | −8.6 | +19.0 | −3.4 | −12.7 | −6.1 |
| Wage employment | +0.3 | +3.8 | −9.6 | −12.0 | +4.9 | −8.1 |
| School enrolment, primary | +4.4 | −3.1 | +1.4 | −0.4 | −2.1 | +3.1 |
| School enrolment, secondary | +8.5 | −2.7 | −1.3 | −3.0 | −8.1 | +3.0 |
| Access to drinking water (dry season)** | +15.1 | −39.9 | −15.7 | −3.7 | −3.2 | +20.0 |
| Access to health centre*** | +24.1 | −11.3 | −38.2 | +17.7 | −15.2 | +3.3 |
| Access to market*** | +8.1 | −41.2 | −24.3 | +21.6 | +0.1 | +1.8 |
*The positive or negative scores of provinces show the advantage or disadvantages, respectively, that the provinces have in that particular aspect with respect to the national average. The national average is normalised to zero. The numbers may not add up to zero because the provinces of Nairobi and North West are omitted from the table.
**Access within 1 mile distance. ***Access within 4 mile distance.
Source: Hazelwood (1979).
It remains to be shown how these disparities are linked to persistence of ethnic allegiances. The first instance when the political salience of ethnicity was tested in an open contest came with the elections in 1963. The results showed the clear alignment of the supposedly less privileged and largely rural ethnic groups of the Rift Valley and Coast with the agenda of majimboism (federalism). The same pattern re-emerged after 30 years when multiparty competition was reintroduced in 1992. Four out of five general elections since the reintroduction of a multiparty system – the exception being the 2002 elections – had two frontrunners hailing from different ethnic groups. In those elections, the average gap between the percentage of votes the winner garnered from his home province and from that of his opponent’s home province was 73.5%. The polarisation of ethnic voting patterns increased in the last two elections regardless of decent economic growth performance compared with the 1990s.
The effect of collective feelings of historical inequality has been apparent in the election-related violence which engulfed many parts of the country in the 1990s and 2007–2008. Particularly, in the highly violent elections of 2007, the electoral agenda was driven by the failure of the incumbent to effectively share executive power and review the constitution in a way that would rectify historical inequalities. A quick analysis of the Afrobarometer data on public opinion in Kenya sheds light on the link between feelings of historical injustice and ethnic identification at an individual level. I have computed the correlation between the two variables for three data points – in 2005, 2008 and 2011 – controlling for the effect of belonging to one of the four major ethnic groups in the country. The results show that the feeling of historical injustice against one’s own ethnic group increases the probability of identifying more with the ethnic group than with the nation by an average of 11%.
Adding on to the historical and structural constraints identified in the above claims, the following two claims discuss the strategic factors that might have contributed to the persistence of political ethnicity in Kenya.
The dominant coalition elites in Kenya have shut off organisational avenues for the non-elite so that the latter do not mobilise themselves. This is done with a mutual understanding among the elites that they would revert to playing the ethnic card in case intra-elite bargains fail to be enforced. This equilibrium strategy is part of what has held the dominant coalition together.
Kenya’s first generation of organised elites were composed of liberal, pan-ethnic technocrats. These elites constituted the short-lived leadership of the KAU of the early 1950s. The liberal platform of these pioneering elites was met with Mau Mau aggression in a way that is reminiscent of typical class struggle. As a consequence, the ‘dispirited tennis-court liberals’ left the scene, giving way to sterner ethnic conservatives (Lonsdale 2004). This ‘new’ generation of leaders (including Jomo Kenyatta), however, had learned a valuable lesson from their predecessors not to ever let their guards down in the face of simmering class consciousness. The elites had not only recognised the threat of class consciousness as they prepared to take over the new nation; they had also resolved to consolidate their authority around traditional moral values by ostracising revivalist religious groups and ‘troublemaking women’ (Peterson 2012).
The sporadic sprouts of populist resistance during the formative years spearheaded by politicians such as Oginga Odinga could be mistaken for organised class struggle. In almost all populist movements in Kenya, however, the major protagonists have been members of the petty bourgeoisie. ‘Although this populism has attracted substantial peasant and worker support, it is clearly a petty bourgeoisie ideology for it advocates the interest of that class in small business and promotes kulak over capitalist agriculture’ (Leonard 1984, 150–151). Figure 1 presents a model juxtaposing the class and ethnic dimensions of a society controlled by a dominant coalition of elites. As Figure 1 shows, the bi-circular band that represents direct access to the organisational privileges of the dominant coalition is mostly confined to the tip of the pyramid in each ethnic group occupied by the political and economic elite proper. There are at least three types of equilibrium scenarios that may characterise the potential outcomes of the strategic interaction between the elites of different ethnic groups: (1) middle-class cooperation, (2) retreating to ethnic enclaves and (3) all-out ethnic violence.
At times of middle-class cooperation, the dominant coalition band in Figure 1 expands further down to reach the petty bourgeoisie rungs of the pyramids. Such enlargement gives the petty bourgeoisie of ‘ethnic group A’ access to the petty bourgeoisie of ‘ethnic group B’. It is possible that such common access to petty bourgeoisie organisations might be used to coordinate on class interest. However, the elite proper will always make sure that the petty bourgeoisie are effectively separated from the non-elites (i.e. the bottom rung) to limit the contagion of class consciousness. In case the petty bourgeoisie of the two ethnic groups attempt to launch more coordinated action relying on the support of the non-elite, the elite proper will resort to the second scenario: ‘retreating to ethnic enclaves’. This strategy severs the link between the two petty bourgeoisie classes, limiting access to the dominant coalition to the tip of each pyramid and confining the rest of the communities to their ethnic enclaves.
The continued salience of ethnicity in political and economic life is crucial for the elite to be able to play the strategy of retreating to ethnic enclaves. Kenya’s elites have struggled to reserve ethno-regionalism as an option that would ensure the maximum possible return in the event of simmering class consciousness. The elites in Kenya were willing to use extra-legal means to punish defectors who threatened to jeopardise the availability of this option. Each of the first three instances of political crisis in the history of independent Kenya was triggered by a threat of trans-ethnic mobilisation. These were marked by the assassinations of Pio Gama Pinto in 1965, Tom Mboya in 1969 and J. M. Kariuki in 1975. Those leaders were promoting either class consciousness or cosmopolitan meritocracy in a way that would undermine ethnic allegiances. Hence, the dominant coalition elites scrambled to save the vitality of ethnic mobilisation as a fallback option by eliminating the defectors. Following the assassinations, most of the communities retreated to their own ethnic enclaves whereas the dominant coalition was saved with little scathe. Those personalities represented the exception rather than the rule because, as Ndegwa (1997) puts it, they ‘are contradicted by their own lives, which are embedded in ethnic communities, and, alas, by their deaths, which are even more intimately enmeshed in the reproduction of ethnic communities’ (608).
Therefore, as long as the strategy of retreating to ethnic enclaves without destabilising the dominant coalition is feasible, the third equilibrium scenario – all-out ethnic violence – remains what is referred to in game theory as an ‘off the equilibrium path’ strategy.5 But this is possible only so far as the elite proper are able to defuse class consciousness that is mostly radiated from the petty bourgeoisie in Kenya. In other words, the elite proper should be able to compartmentalise the political organisation of the petty bourgeoisie and the non-elites in separate rungs of the ethnic pyramid. This is precisely what the reintroduction of multiparty politics in the early 1990s took away from the elites. The cross-cutting political mobilisation that is normally inbuilt in multipartyism would break the compartmentalisation of the lower rungs of the ethnic pyramid. The petty bourgeoisie who had always tried to infuse class politics in Kenya could have a chance to mobilise the non-elites in a political party. As populist politics became imminent with the reintroduction of a multiparty system, the role of the dominant coalition to enforce intra-ethnic compartmentalisation seemed to be waning. Anticipating that the old protocols of the single-party dominant coalition would no longer be viable, Moi resorted to ethnic violence, which would have otherwise been ‘off the equilibrium path’ deterrence. Moi had openly argued that multiparty democracy would inevitably degenerate into tribal bloodshed.6 In light of the ethnic violence that was perpetrated after the 1992 elections mainly by Moi’s Kalenjin allies, his statement was a self-fulfilling prophecy (Republic of Kenya 1999).
So when does it become in the elites’ interest to cross ethnic boundaries and restructure the pyramids in Figure 1 differently? The next claim deals with that question.
Economic growth and social development in early postcolonial Kenya failed to make non-ethnic forms of political and economic organisation as sufficiently resistant to political uncertainties as ethnic organisations. This happened because most of the rents of the dominant coalition elites remained tied to their leadership of ethnic organisations. For various reasons, the development process did little to enhance the individual-specific rents of the elites.
The most common source of individual-specific rent is human capital obtained in the form of natural talent, education or productive experience. One can see the effect of a distinctively educated elite class on the development of a trans-ethnic liberal platform in the experience of the pre-independence first-generation leadership of the KAU as discussed above. Actually, the educational advantage of the ruling elites persisted until independence and beyond as the founding politicians were sufficiently educated. Besides, there were more young educated elites coming through the mass airlifts of ‘O level’ students to US universities between 1958 and 1961. However, the expansion of education opportunities to the broader population was catching up at a faster pace. Secondary school enrolment increased fourfold within a matter of seven years after independence. With free tertiary education, enrolment to University College Nairobi quadrupled to over two thousand between 1963–1964 and 1969–1970 (Hornsby 2012). Inevitably, this expansion diluted the individual-specific rent the elite could have reaped thanks to their head start in human capital.
While the elites’ advantage in individual-specific rents diminished, developments in the formative period of the post-independence nation entrenched the relevance of organisation-specific rents. The importance of land as a source of rent that is often accessed through group bargaining reinforced what Jenkins (2012, 580) refers to as ‘ethnicised understanding of belonging to and ownership of space’. Additionally, the significant expansion of salaried employment in the public sector, instead of the private sector where merit would be more important, implies organisation-specific rents continued to trump individual-specific rents in the decades after independence.
Even if there was any growth in the individual-specific rents of the dominant coalition elites, it would have been far from uniformly distributed among elites from different ethnic groups. Obviously, the Kikuyu elites had overwhelming advantage in terms of education and formal sector experience. Moreover, the elites of the formative years were split into two along the rural–urban divide. This split saw the rural-stationary segment juxtaposed with the bureaucratic-mobile segment (Holmquist 1984). One can deduce from the very characterisation of the two segments that the bureaucratic segment would have higher individual-specific rent than the rural segment. Such heterogeneity among powerful elites in terms of the potential gain from reconfiguration would keep patron–client networks confined to ethnic enclaves.
Conclusion
Ethnic tensions continue to threaten political stability and social cohesion in Kenya mostly because the elites could not shift the economic foundation of the dominant coalition to rents that are not necessarily tied to ethnic bargaining. The structure of the Kenyan economy and the presence of institutional path dependence have played a role in keeping the dominant coalition dependent on organisation-specific rents. This article has attempted to weave a body of explanations for the persistence of political ethnicity in Kenya using structural factors steeped in history and geography as well as strategic factors dictated by current institutions. I have employed key insights from the Violence and Social Orders framework by North, Wallis, and Weingast to put the organisational benefits of ethnicity in the context of negotiating political order in a weakly institutionalised polity.
The postcolonial elites in Kenya preserved the state-capitalist mode of economic organisation that they inherited from the colonial system. In making the economic regime compatible with the political settlement, they also stuck to ethnicity as the most viable form of political organisation. This happened in spite of the stubbornly high income inequality and the expansion of economic space during the formative years of the postcolonial nation. In the presence of substantial intergroup inequality such as the one that prevails in Kenya, it is shown to be difficult to promote other forms of collective organisation over ethnicity. The persistence of ethnicity has strategic origins on top of structural reasons. Once the elites have started using ethnic allegiances as bargaining chips to secure their position in the dominant coalition, it becomes hard to override tribal partitions even when the structural conditions are favourable for a more liberal organisation of society. Middle-class voters could be trapped in their ethnic enclaves even when they really prefer to vote based on policy differences. This is mainly because they do not trust that the middle class is large enough to make a difference and they fear others from different ethnic groups will vote along ethnic lines (Burbidge 2014).
In light of the analysis presented in this article, events such as the 2007–2008 post-election violence in Kenya should be understood as the complex outcomes of historical grievances and elite posturing. Accordingly, the potential solutions to the problem of ethnic strife in Africa should include strategic interventions that would eventually make the elites walk away from group-specific sources of rent without necessarily destabilising the dominant coalition in the short run. The shift from reliance on organisation-specific to individual-specific rents could be facilitated through structural transformation of the economy as well as the development of elite human capital.7 Industrial development reduces the need for location- and group-specific rents, eventually promoting trans-ethnic elite platforms. The development of elite human capital, especially in science and technology, creates strong incentives for the children of the rich and the powerful to transform the organisation-specific rents of their parents’ generation to the individual-specific rents of their own generation.