African Affairs, 108/433, 581–598


doi: 10.1093/afraf/adp043 Advance Access Publication 8 July 2009

The Author [2009]. Published by Oxford University Press on behalf of Royal African Society. All rights reserved


ABSTRACT This article investigates the relationship between national politics and local violence in the aftermath of Kenya’s 2007 election. Focusing on the Kibera slum, the article shows that while the area’s ‘big man’ Raila Odinga at times appeared to have a strong hold over his constituents at the grassroots, patrimonialism and big man politics cannot provide a full explanation of the post-election violence. Instead, local socio-economic factors played a key role and lent the conflict its own specific dynamics. The article also shows that while Raila’s strong patron–client relationship with Kibera residents has empowered him as a national politician, in his current role as Prime Minister this relationship restricts his political room for manoeuvre. Thus the focus on Raila’s ‘big man’ status in Kibera illustrates the pressures faced by Kenyan politicians in mediating between their public roles and the demands of their constituents.

WHEN ON 30 DECEMBER 2007 THE INCUMBENT PRESIDENT, MWAI KIBAKI of the Party of National Unity (PNU), was suddenly declared the winner of the presidential elections – turning his main opponent, Raila Odinga of the Orange Democratic Movement (ODM), into the loser – violence erupted in many parts of Kenya. Previous multi-party elections in the 1990s had seen similar clashes, but this time the scale and intensity were unprecedented, creating the worst political crisis since independence. The initial demonstrations rapidly turned into revenge killings, targeting ethnic groups linked to the ‘other political camp’. Violence continued until a power-sharing agreement between PNU and ODM was signed on 28 February 2008. The two months of violence caused over 1,100 deaths and up to 350,000 internally

Johan de Smedt ( is a social anthropologist with extensive experience in humanitarian aid, most recently in southern Sudan and Somalia. The author would like to thank Jan Kees van Donge and Rita Abrahamsen for their advice.




displaced persons (IDPs), while the economic loss was estimated to run to billions of dollars.1 The 2007 elections and subsequent violence have been extensively discussed,2 and there is general consensus that it is an oversimplification to see the violence as an ethnic, or tribal, problem. Instead most analyses emphasize ‘underlying precipitating factors’, while ‘the elections provided the spark that ignited them’.3 Some of these underlying causes are identified as historical grievances over resources (mainly access to land), deliberate weakening of government institutions (like the judiciary), and the gradual loss of the state’s monopoly of legitimate force, allowing the large-scale proliferation of militias and gangs, which were in turn used and mobilized by politicians in their pursuit of electoral victory. Other factors include economic and political exclusion, as well as the strongly ethnicized discourse of Kenyan politics.4 While these analyses point to the complexity and historical roots of the post-election violence, they all, to varying degrees, emphasize the importance of ethnicity and patron–client relations in Kenyan politics. This article follows this lead, and seeks to investigate the relationship between national politics and local violence in more detail. Focusing on Kibera, the stronghold of Raila Odinga, the article explores the relationship between ‘big man’ politics and violence at the grassroots. Kibera was one of the areas most affected by the post-election violence – shops and houses were looted or destroyed, people killed, and for several weeks angry crowds engaged the police in running battles.5 Raila Odinga has been the area’s MP since 1992, and Kibera is his main support base. As the article shows, Raila
1. Human Rights Watch, ‘Ballots to bullets: organized political violence and Kenya’s crisis of governance’ (HRW Report, March 2008); International Crisis Group, ‘Kenya in crisis’ (Africa Report No. 137, 21 February 2008). See also the so-called ‘Waki Report’ (2007) of the Commission of Inquiry into Post-Election Violence, headed by Judge Waki, <> (17 April 2009). 2. See, for example, the special issue of the Journal of Eastern African Studies, 2, 2 (2008). 3. Susanne Mueller, ‘The political economy of Kenya’s crisis’, Journal of Eastern African Studies 2, 2 (2008), pp. 185–210. 4. Daniel Branch and Nic Cheeseman, ‘Democratization, sequencing and state failure in Africa: lessons from Kenya’, African Affairs 108, 430 (2008), pp. 1–26; Nic Cheeseman, ‘The Kenyan elections of 2007: an introduction’, Journal of Eastern African Studies 2, 2 (2008); Mueller, ‘The political economy’; Michael Bratton and Mwangi Kimenyi, ‘Voting in Kenya: putting ethnicity in perspective’, Journal of Eastern African Studies 2, 2 (2008), pp. 272–89; Mwangi wa Githinji and Frank Holmquist, ‘Kenya’s hopes and impediments: the anatomy of a crisis of exclusion’, Journal of Eastern African Studies 2, 2 (2008), pp. 344–58; Gabrielle Lynch, ‘The fruits of perception: “ethnic politics” and the case of Kenya’s constitutional referendum’, African Studies 65, 2 (2006), pp. 233–70. 5. The author worked in Kibera between 1998 and 2002, and has since 2003 undertaken research there, mainly on the Nubi community and Kibera history. Research for this article was conducted from January to July 2008, and is based on over a hundred informal, semistructured interviews and oral testimonies, as well as newspaper articles. Informants were of all ages, sexes, ‘classes’, and ‘tribes’, and many were interviewed several times. To protect them, no names or other personal details are given.



at times appears to have an almost total control over his constituents in Kibera, and in the weeks after the elections the slogan ‘No Raila, no peace!’ was frequently heard on the streets and in demonstrations.6 At the same time, a closer examination of Kibera shows that patrimonialism and big man politics cannot provide a full explanation of the post-election violence. Instead, local, socio-economic factors played a key role and lent the conflict its own specific dynamics. Moreover, the article shows that while Raila’s strong patron–client relationship with Kibera residents has in many ways empowered him as a national politician, in his current role as Prime Minister this relationship also restricts his political room for manoeuvre. Thus the focus on Raila’s ‘big man’ status in Kibera illustrates the pressures faced by Kenyan politicians in mediating between their public roles and the demands of their constituents. Kenya’s ‘big man’ politics The politics of patronage occupies a central place in discussions of Kenya’s political history, past and present. Patronage was one of the characteristics of pre-colonial ethnic communities: the local ‘big men’ exercised authority by sharing out their wealth – the recipients of this redistribution, the poor, ‘inevitably owed obedience’.7 During the colonial period, the British introduced ethnically defined administrative units – ‘tribes’ – and as a result ‘ethnic groups became political tribes’.8 The ‘big men’ stayed. In independent Kenya, President Kenyatta’s politics to a certain extent reflected the old ‘moral economy’ of redistribution and obligation, but the encouragement and gradual emergence of ethnic big men, tied to Kenyatta through patron–client relationships, widened nascent ethnic divisions. As argued by Daniel Branch and Nic Cheeseman, this (new) elite colluded with Kenyatta to get and maintain access to wealth and privileges, without much concern or sympathy for the plight of the poor. In return they contained dissent from below.9 A system of ethnic patronage thus became entrenched in Kenyan politics, whereby eventually all the ties of local, ethnic big men led to the all-powerful President, the ultimate ‘big man’. Invoking the analogy of a royal establishment, Colin Leys describes this as ‘Jomo Kenyatta’s court’,

6. See also Lynch, ‘The fruits of perception’, p. 255. 7. John Lonsdale, ‘Moral ethnicity and political tribalism’ in Preben Kaarsholm and Jan Hultin (eds), ‘Inventions and boundaries: historical and anthropological approaches to the study of ethnicity and nationalism’ (IDS Occasional Papers No. 11, Roskilde University, 1994). 8. John Lonsdale, ‘The political culture of Kenya’ (Occasional Papers No. 37, Centre of African Studies, Edinburgh University, 1992); Lynch ‘The fruits’, p. 236. 9. Branch and Cheeseman, ‘Democratization, sequencing’; Lonsdale, ‘The political culture’, pp. 7–8.



where the President was surrounded by close Kikuyu allies.10 Politicians who became too powerful, independent, or popular, and appeared to be a threat, were removed. In 1966 Oginga Odinga, the Luo Vice-President, was manoeuvred out of his post and sidelined; in 1969 the leading Luo politician Tom Mboya was killed, allegedly on the orders of high-level Kikuyus. Later that year, Oginga was imprisoned and the Luo party, the Kenya People’s Union (KPU), was banned, effectively introducing the single-party state.11 When Kenyatta was succeeded by Moi, the patronage system remained in place, but the political elite became more fragmented. Moi had fewer resources available for distribution and favoured his own Kalenjin, excluding other ethnic groups. Unable to maintain elite cohesion, Moi increasingly resorted to oppression, further alienating the excluded elite and effectively undermining their willingness to contain dissent; instead, the excluded leaders called for the re-introduction of multi-partyism.12 When re-introduced in 1992, multi-party competition and elections created new opportunities for ethnic ‘big men’, who could now profile themselves as the defenders of their ethnic group on the national platform, promising a piece of the national economic cake, including jobs, favours, and hard cash, in return for votes. Though the presidency remains the biggest prize to be won in elections and ‘the ethnic identity of a new President defines patterns of favouritism’,13 it is no longer the only centre of power and even the incumbent leadership has to (re)build and maintain its support base.14 With ethnicity seen as central in determining the distribution of national resources, many ethnic big men resort to what John Lonsdale calls ‘political tribalism’, the deliberate use and manipulation of ethnic identity in political competition with other groups. At the risk of over-simplifying, this amounts to pitting one community against the other, exploiting latent (ethnic) grievances about scarce resources (often ‘land’), and triggering ethnic violence through incitement, as a means of securing political power and facilitating elite accumulation of wealth. Violence thus becomes a normal part of political culture, while political ideology or a clear political programme is often absent as political parties function primarily as vehicles for getting into
10. Colin Leys, Underdevelopment in Kenya: The neo-political economy of neo-capitalism (Heinemann, London, 1976 [1975]), pp. 246–51; John Lonsdale, ‘The political culture’. 11. R. M. Maxon, East Africa: An introductory history (West Virginia University Press, Morgantown, 1986), pp. 264–9; W. R. Ochieng’, ‘Independent Kenya, 1963–1986’ in W. R. Ochieng’ (ed.), A Modern History of Kenya 1895–1980 (Evans Brothers Ltd, London, 1989), pp. 215–17; W. R. Ochieng’, ‘Structural and political changes’ in B. A. Ogot and W. R. Ochieng’ (eds), Decolonization and Independence in Kenya 1940–93 (James Currey, London, 1995), pp. 101–2. 12. Branch and Cheeseman, ‘Democratization, sequencing’. 13. Angelique Haugerud, The Culture of Politics in Modern Kenya (Cambridge University Press, Cambridge, 1995), p. 42. 14. David Throup and Charles Hornsby, Multi-Party Politics in Kenya: the Kenyatta and Moi states and the triumph of the system in the 1992 election (James Currey, Oxford, 1998) p. 584.



power. According to Mueller, political parties in Kenya, like ‘big men’, are driven by ‘ethnic clientism, with a winner-takes-all view of political power and its associated economic by-products’.15 As a result, it is not uncommon for political leaders to move from party to party, and take their ethnic following with them. By the same token, alliances with other parties, in or outside government, are made and unmade with seemingly little regard for political ideology or agendas. For example, in the 1992 elections Raila Odinga was in FORD-Kenya, in 1997 in NDP, in 2002 in NARC/LDP, and in 2007 in ODM. He was imprisoned by President Moi several times in the 1980s, but nevertheless joined his government in 2001, before again returning to the opposition in 2002. For ‘political tribalism’ to be successful, politicians will have to suppress local debates expressing ‘moral ethnicity’, that is, ethnicity and more complex forms of identification springing from ‘below’, as an ordinary aspect of daily life and social intercourse. As Lonsdale describes it, in such everyday engagements, ethnicity is imbued with moral values, civic virtue, and responsibility, and is not necessarily disruptive within communities.16 At this moment in time, however, it seems that Jacqueline Klopp might be correct in wondering whether moral ethnicity can present a serious challenge to political tribalism. Her case study of Nandi nationalism, like this study of Kibera, shows that there is resistance at the grassroots level to the politics of ethnic hatred. However, given the success of this political strategy, the current political elite may well prove reluctant to abandon political tribalism and its accompanying violence.17 Today, Raila Odinga is one of Kenya’s leading big men. As ODM’s charismatic Luo leader, he is the MP of Langata Constituency in Nairobi, a position he has held since 1992 when the first multi-party elections since 1966 were held. Gabrielle Lynch aptly describes him as ‘a man who stirs up the strongest of emotions – be it Railamania or Railaphobia. Yet, love him or hate him, Raila holds considerable political clout, because of his fiercely loyal support base.’18 There was never any doubt that Raila would retain his parliamentary seat; he has a large Luo following in the Kibera slum, which is only a small part of Langata constituency (probably less than 2 percent of the area, but with more than 60 percent of its population). Raila’s support base followed him from party to party, and has kept him in Parliament since 1992, in what appears to be an almost paradigmatic illustration of the patron–client relationship in Kenyan politics. Raila appears at times to
15. Mueller, ‘The political economy’, p. 186. 16. Lonsdale, ‘Moral ethnicity’, pp. 132, 141; Gabrielle Lynch, ‘Courting the Kalenjin: the failure of dynasticism and the strength of the ODM wave in Kenya’s Rift Valley Province’, African Affairs 107, 429 (2008), pp. 541–68. 17. Jacqueline Klopp, ‘Can moral ethnicity trump political tribalism? The struggle for land and nation in Kenya’, African Studies 61, 2 (2002), pp. 269–94. 18. See also Lynch, ‘The fruits’, p. 255.



have almost total control over his constituents, drawing large crowds when visiting Kibera. But the reality on the ground is not quite so straightforward. The following sections analyse how Raila’s ‘big man’ machinations in Kibera (may) have contributed to the post-election violence. It provides an account of the flow of events in Kibera, before analysing the underlying causes of the violence. A key factor here is how, during the violent episodes of the last 15 years, the Luos have managed to turn large parts of Kibera into ‘Luo territory’, chasing away the (mainly) Kikuyu landlords. It also shows how Raila’s recent half-hearted attempts, as Prime Minister, to correct the situation have not endeared him to his Luo voters in Kibera – he has to walk a tightrope between being Raila the Prime Minister and Raila the Luo ‘big man’. As such, Raila faces considerable pressures in mediating between his ethnic Luo status and his public role as Prime Minister. Kibera Kibera is an informal settlement within the city of Nairobi, started more than 100 years ago as a settlement for retired Sudanese soldiers of the British colonial army in East Africa. It is now (allegedly) the largest slum in subSaharan Africa, with an estimated population of 600,000.19 The Nubis, descendants of those Sudanese soldiers, still live in Kibera as a distinct group. They are a small minority, but they do wield some power as they own an estimated 15 percent of the rental houses. Much of the rest of the rental accommodation is owned by Kikuyus. Kibera is Kenya in microcosm, and is home to members of all Kenyan (African) ethnic groups. The settlement is divided into a number of ‘villages’, each with its own characteristic ethnic make-up, and although most villages have people of all ethnic groups, often one group is dominant. For example, Gatwikira is dominated by Luo tenants, while Laini Saba is a Kikuyu stronghold – but both areas also have residents of other ethnic groups. As is clear from this brief description, Kibera’s demography is more complex than can be captured here, and for the purpose of this article, the focus is on the Kikuyus and Luos, the main ‘actors’ in the post-election violence. Because of the villages’ different ethnic characteristics and locations relative to the main road, each village experienced different levels of violence; some were seriously affected, others hardly at all. To understand these issues, a brief look at the historical context of Kikuyu and Luo presence in Kibera is necessary. The first Kikuyus came to Kibera as early as the 1920s, and more followed during the next decades as Kikuyus lost most of their land to the white settlers and were desperately looking for places to settle. From the early
19. Estimates vary between 250,000 and 1.2 million people, but there are no reliable statistics.



1970s, Kikuyus were favoured by the Kenyatta administration and were accordingly given the first opportunity to build rooms to rent in Kibera. Some invested their pension in the construction of a small house and a few rental rooms, while others with more money (civil servants, businessmen, politicians) constructed hundreds of rooms. The Nubis had to follow suit in order not to be left out. Today, Kikuyus possibly own about 50 percent of the rooms in Kibera.20 The big landlords do not usually live in Kibera, and these ‘absentee landlords’ often use agents to collect rent. The small resident landlords (most with less than 20 rooms) often live in the same compound as their tenants, and basically in the same conditions. The Luos too arrived early; in 1948 they constituted 27 percent of the non-Nubi population.21 Over the years their number increased fast, like the overall Kibera population which grew from around 9,000 at independence to 60,000 in 1980.22 Before the 1992 elections many more Luos moved to Kibera; according to the residents themselves, many were brought by Raila in trucks from Nyanza to ensure that he would be elected to Parliament. In subsequent years, the flow of Luos into Kibera was maintained, in part to ensure the retention of Raila’s seat.23 Unlike the Kikuyus, the Luos did not become much involved in the construction of houses or rental rooms, but are instead known to prioritize investment in the education of their children, and to buy land or build a house in Nyanza with a view to returning at retirement.24 The Luos thus became the tenants of the Nubis and the Kikuyus. As Luos arrived from Nyanza, they usually stayed with or close to relatives who could take care of them, pay school fees, or assist in finding a job. The majority of Luos thus settled together, mainly in Gatwikira (including Kisumu Ndogo). Over the years, Gatwikira/Kisumu Ndogo became a hardcore Luo area. Today, a large part of the population consists of young Luo men, who, constrained by economic circumstances, remain unemployed and poor. They hang around the streets, looking for jobs, for anything or for anyone that pays them. Currently the Luos are believed to be the largest ethnic group in Kibera, making up the majority of tenants not only in Gatwikira and Kisumu Ndogo, but also in large parts of Shilanga, Lindi and Kianda (see Figure 1).25
20. There are no real statistics on Kibera landlords; this figure comes from the chairman of the Kibera Landlord and Housing Cooperative, a (Kikuyu) landlord organization. 21. ‘Census of August 23rd, 1948’ (13 September 1948, Kenya National Archives RCA (MAA) – 2/1/3 ii). 22. Philip Amis, A Shanty Town of Tenants: The commercialisation of unauthorised housing in Nairobi, 1960–1980 (University of Kent, unpublished dissertation, 1983), pp. 166–7. ´ e ` 23. Marie-Ange Goux, ‘Les ev´ nements de Novembre–D´ cembre 2001 a Kibera: crise e d’origine interne ou externe?’, L’Afrique Orientale (Annuaire 2002), p. 328; Michelle Osborn, From Forest to Jungle: Tracing the evolution of politics in Kibera (Oxford University, unpublished MS thesis, 2006), p. 38. 24. The claim here is not that there were no Luo landlords, but that there were relatively few. 25. Again, it must be stressed that no reliable statistics are available.



Figure 1. Kibera.

Previous conflicts in Kibera Over the years, Kibera has seen a number of violent conflicts about rent, resulting in death, destruction, and displacement. The dynamics of big man politics at grassroots level become evident here. The Luos are mainly tenants, and therefore interested in low rents. By supporting his Luo voters in their violent demands for lower rent (and in their refusal to pay rent), ‘big man’ Raila secured votes and his re-election. In 1992, some months before the general elections, there were major clashes between the Nubis and Luos, caused by the latter’s refusal to pay rent, which, they claimed, was too high. Though the Kikuyus stayed out of the fighting, they suffered most, as they owned most of the houses in the affected areas. Rents were eventually reduced, though some tenants refused to pay any rent at all to Kikuyus, chasing away some of the landlords. More clashes between the Nubis and Luos occurred in 1995, and again the rent levels played a role. The November 2001 clashes between Nubis and Luos came at a time when Raila’s National Development Party (NDP) was about to merge with President Moi’s KANU and join the government. In a public meeting in Kibera, Raila asked Moi to reduce rent levels, and the President, needing the NDP’s support, agreed immediately, as lowering rent was a way of securing the Kibera voters’ support for KANU/NDP, and for Raila in particular. However, landlords did not agree, leading many tenants to refuse to pay



any rent at all. In the ensuing clashes 15 people were killed, houses went up in flames, and thousands fled Kibera.26 As a result, in the Luo-dominated areas of Gatwikira and Kisumu Ndogo, hardly any rents, or only substantially lower rents were paid. The absentee landlords, almost all Kikuyu, but also some Nubi, lost control over their rooms, as most gave up and stopped coming for the rent. Their tenants thus became the de facto owners of the rooms. The small Kikuyu resident landlords, who had not been actively involved in the actual fighting, again suffered more, as tenants basically refused to pay rent to them (as Kikuyu). A number of Kikuyus (an estimated 20 percent)27 left Gatwikira and Kisumu Ndogo, though the majority stayed and tried to make do with what the tenants gave them; sometimes half the rent, sometimes nothing. In other areas with a high percentage of Luo tenants, like Shilanga (and even Kianda), many Luos and even people of other tribes also temporarily stopped paying rent. However, after negotiations between landlords and tenants, rent levels went down in most parts of Kibera, and most tenants resumed payment. In the main Luo-dominated areas, however, it was both difficult and risky for landlords to take action against non-paying tenants, and the situation of non-payment of rent persisted up to the elections of 2007, and formed an important part of the local dynamics of the postelection violence. The 2007 post-election violence in Kibera The months before the elections were marked by political tension and sporadic violence, while the political rallies of Raila’s opponents in Kibera were disrupted, sometimes allegedly after incitement by ODM politicians.28 As the votes were counted, people in Kibera keenly watched the live broadcast of the election results, expecting Raila to win. Ready to celebrate, people were also tense, expecting that the results might be rigged. Immediately after Kibaki was declared the winner, the violence erupted. As one interviewee described it, ‘people went on a rampage, destroying things. Kikuyu started moving out immediately. People took to the streets, shouting “Kikuyu have to go”. People also started looting, especially Kikuyu houses and businesses.’29 Violence had also erupted in other parts of Kenya, mainly in Rift Valley, where the Kalenjin community was destroying Kikuyu property, killing people, and chasing away Kikuyus from villages where they had
26. PeaceNet-Kenya, The Quest for Human Dignity: Kibera violence (Acken Media Services, Nairobi, 2001); Osborn, ‘From forest’, p. 34; Goux, ‘Les evenements’. 27. Based on estimates of several informants in Gatwikira. 28. ‘ODM in new poll rigging plot claim’, East African Standard, 18 September 2007; ‘Chaos at Livondo’s Kibera meeting’, Sunday Nation, 30 September 2007, p. 52. See also Cheeseman, ‘The Kenyan elections’, p. 170. 29. Interview, Luo lady, small landlord, Kianda, 18 March 2008.



lived for generations. Kikuyus took revenge and attacked Luos and Luhyas in Kikuyu-dominated areas like Limuru and Kawangware. Shops and houses were looted all over Kibera, mainly those belonging to Kikuyus and known PNU supporters. Toi Market, an informal market mainly controlled by Kikuyu traders, went up in flames. Burning of property was not widespread, however, as the risk of burning down hundreds of houses (including those of the perpetrators themselves) is high in a place like Kibera, and fire was therefore generally avoided. During the violence, people were beaten up and robbed, some were killed, and women and girls were raped. Much of the violence was simply looting; though the main actors were male youths, there were also many adults, girls and children involved. Looting groups were often randomly formed; youths from all parts of Kibera, and even from outside, took part. Looters were of all ethnic groups, including Kikuyus. According to most informants, the violence was mainly instigated by youths from Gatwikira, the Luo area: ‘Groups of youths came looking for Kikuyus, especially youths from Gatwikira. Many people followed those groups for their own safety – if you don’t join them, you are seen as against them.’30 These Gatwikira groups were said to have been organized by local (youth) leaders, often linked to ODM. They went all over Kibera; after the initial ‘shopping spree’ at Kikuyu shops and houses, they turned to the property of people of other ethnic groups. Police did not do much to stop the looting, often just standing by and watching – but sometimes they became actively involved, opening the padlocks with their guns and going in first.31 The areas mainly affected were Kianda and Olympic, the richer parts of Kibera. In other areas looting was limited, sometimes because they were protected by neighbourhood vigilantes, sometimes because most shops were owned by non-Kikuyu residents. The upper part of Kikuyu-dominated Laini Saba was guarded by Kikuyu vigilantes and escaped most of the looting. In revenge, Luos living there were told to leave; sometimes they were beaten, while their homes were looted and occupied by Kikuyus, often those fleeing other parts of Kibera. Luos fleeing Laini Saba often found refuge in other villages, in rooms deserted by Kikuyus. Notably, there was not much killing in Kibera, as the aim of the violence was to chase away the Kikuyus and loot. Most Kikuyus left almost as soon as the violence started, certainly in areas with a high Luo or Luhya population, leaving all their belongings behind. Their shops and rooms were then looted and immediately occupied by others. Very few Kikuyus stayed, mainly young men who were born in Kibera: ‘I had no problem, the guys that came to the house, I know all of
30. Interview, Luo man, tenant, Kianda, 19 March 2008. 31. See the Waki Report for descriptions of violence and police failure to do their duty.



them, they are my friends, we grew up together. . . .’32 It is important to stress, in the light of this young man’s perception, that a trace of ‘moral ethnicity’ was at work. For example, during my fieldwork I heard of quite a few cases where fleeing Kikuyus were protected by their Luo neighbours (and vice versa) – they were hidden and escorted out of Kibera, and their belongings and rooms were kept safe. Although it is impossible to quantify the extent of such actions, it is clear that some people were able to defend moral values, and to resist succumbing to ‘political tribalism’ towards their neighbours and friends. John Lonsdale’s ‘moral ethnicity’ was thus not entirely absent within the (imagined) ‘Kibera community’, and served to resist and challenge ‘political tribalism’, albeit on a small scale. The violence, in other words, was not caused by ethnic difference per se, but rather by its politicization.33 Kibera was rife with rumours during the post-election period: one suggested a plot to assassinate Raila, another that the army would take power – and there were many more. These rumours spread fast and wide, using modern technology like text/SMS messaging, while local radio stations broadcasting in vernacular (mainly Kalenjin and Kikuyu, but also Luo) also played a role in further inciting people, fuelling fear and panic, and exacerbating violence, displacement, and the destruction of property.34 As is well known, rumour plays an important role in Kenyan politics; politicians may start rumours themselves, exploiting them to their own benefit. In this case, Michelle Osborn shows how the rumour of Raila’s arrest brought many ODM supporters onto the streets to vent their frustration. The ODM leadership did nothing to quash that baseless rumour, instead using misinformation in an obvious attempt to keep their supporters on the streets – and thus directly contributing to the violence.35 Another rumour was that the Mungiki, a violent, pseudo-religious, predominantly Kikuyu gang, was coming to Kibera to take revenge.36 In response, Kibera vigilante groups were organized from the first days of January, and in many areas such groups would stay in the streets at night, to be ready for defence against Mungiki, but also to prevent more violence and looting. Importantly, there is little evidence to suggest that the Mungiki were actually in Kibera, but the rumours nevertheless had an impact on the course of events.

32. Interview, young Kikuyu man, Gatwikira, 21 April 2008. 33. See Lonsdale, ‘Moral ethnicity’; Klopp, ‘Can moral ethnicity’. 34. Michelle Osborn, ‘Fuelling the flames: rumour and politics in Kibera’, Journal of Eastern African Studies 2, 2 (2008), pp. 325–6; Waki Report, pp. 295–303. 35. Osborn, ‘Fuelling the flames’, p. 323. 36. On the Mungiki, see David Anderson, ‘Vigilantes, violence and the politics of public order in Kenya’, African Affairs 101, 405 (2002), pp. 531–55, Peter Kagwanja, ‘Facing Mount Kenya or facing Mecca? The Mungiki, ethnic violence and the politics of the Moi succession in Kenya, 1987–2002’, African Affairs 102, 406 (2003), pp. 25–49.



The vigilantes continued their nightly vigils for a few weeks. In some ‘tribal hardcore’ areas like Laini Saba and Gatwikira they also checked people during the day, asking for ID: if you were of the wrong ethnic group, you could be in serious trouble, and some people were killed. There were also clashes between different vigilante groups, notably the groups from Gatwikira and the ‘defence force’ of Laini Saba, in which a number of people are reported to have been killed. Youths who initially had made money from looting now – as ‘vigilantes’ – started levying taxes on the inhabitants for ‘protection’ when out in the streets at night. Some put up roadblocks on Kibera Drive to collect money from passing traffic. Others offered their services to Kikuyus who needed to transport their belongings out of Kibera or to newly arrived Luos and Luhyas,37 chased away from Kikuyu-dominated areas, who were shown houses vacated by the Kikuyus. During negotiations between PNU and ODM to solve the crisis, ODM called several times for nationwide demonstrations to put pressure on Kibaki. One of the most popular slogans during this period was ‘No Raila, no peace!’, a slogan that resonated particularly strongly in Kibera. Demonstrations were invariably stopped by security forces using water cannon and teargas, resulting in more clashes all over the country. In Nairobi, the security forces tried to stop people from Kibera participating in demonstrations – they were forced to turn back into Kibera, where they would go on a rampage, looting and fighting the police. The main road, Kibera Drive, was subsequently barricaded by youths to prevent the police from entering Kibera. As the violence continued, life in Kibera became more difficult: in most areas the shops were either empty after looting, or had closed as a security measure. Fresh food, if available, was very expensive. Supermarkets near Kibera could not always be reached, and even if most of the loot was for sale in the streets of Gatwikira, it was at highly inflated prices that few could afford. Some food aid was given by some churches and the Red Cross, but this was insufficient to cover the needs of the population. The first wave of violence lasted for about a week. After that Kibera calmed down, and some shops opened again here and there. The situation nevertheless remained very tense, pending negotiations between ODM and PNU. After former UN Secretary-General Kofi Annan arrived on 22 January to mediate, the situation improved and most vigilante groups stopped their nightly vigils. However, the assassination of the ODM MP Melitus Mugabe Were, seen as politically motivated, triggered renewed attacks on Kikuyus, with groups dominated by youths from Gatwikira going from door to door to flush out all remaining Kikuyus. This time more Kikuyus were killed, causing revenge killing of Luos and Luhyas in Laini
37. The Luhyas were generally (considered to be) ODM voters.



Saba. In total around 60 people may have been killed in Kibera, about half of the total of those killed in Nairobi.38 Throughout February, sporadic fighting continued in Rift Valley and Western Kenya, but Kibera remained relatively calm, though tense, as people were anxiously waiting for an agreement between the parties. On 28 February, when negotiations seemed to have ended in failure and tension was rising, a power-sharing agreement was suddenly announced – Kibaki remained President, Raila became Prime Minister. Demonstrations planned for that day were called off, and calm was restored instantly almost everywhere. People had been waiting for the agreement, expecting that everything would go back to normal. This is not what happened, however, as few Kikuyus returned to Kibera. In relatively quiet parts of Kibera, where Kikuyus and Luos had lived together peacefully for many years, a considerable number of Kikuyus came back: not to live there, but to re-open their shops. Its high population density makes Kibera a good place for business, and most burnt-down shops have been rebuilt, and many re-opened. But, as one shopkeeper expressed the common sentiment: ‘Even if we do business here, we do it in fear.’39 In most other parts of Kibera, the areas with a majority of Luo or Luhya tenants, very few Kikuyus returned. Those who did experienced aggression and threats, so that some were even too scared to talk; they were harassed, and some had to run for their lives again.40 Importantly, Kikuyu landlords received little assistance from the local authorities (the District Officer and Chiefs) to recover their rooms, as even the authorities were afraid to go into these areas. It would be risky for the landlord as well, because ‘the police is not going to guard me day and night; the Luos will just kill me the next day’.41 Most Kikuyus, afraid to come back to Kibera, preferred to wait for things to cool down. The announcement on 13 April 2008 of a new Cabinet with Raila as Prime Minister did not change the situation: Kikuyus remained unwelcome. As this discussion demonstrates, one effect of the post-election violence was to bolster the process of growing Luo dominance in Kibera. In particular, the Kikuyu landlords have been almost completely removed from the Luo hardcore areas of Gatwikira and Kisumu Ndogo, where the Luos have become the de facto owners of the houses. In the areas around Gatwikira, the same process is now in its initial stages, as many Kikuyu landlords have lost control over all or a large proportion of their rooms. In most cases rent
38. Interview, Luo youth involved in data collection for local NGO, Gatwikira, 21 April 2008. The Waki Report estimates 125 deaths in Nairobi – see table on p. 308. 39. Interview, Kikuyu businessman, Kianda, 20 March 2008. 40. Interviews, Kikuyus, mainly small landlords (male and female) from Gatwikira, in IDP camp, April 2008. 41. Interview, Kikuyu landlord, IDP camp, April 2008.



is not paid, certainly not by Luo tenants. These areas are now in a similar situation to Gatwikira before the 2008 violence. The underlying causes of violence in Kibera Reviewing the violence in Kibera raises the question of the relationship between Kenya’s ‘big man’ politics and local socio-economic dynamics. Phrased differently, to what extent was the Kibera violence ‘designed’ to boost the electoral position of Raila and the ODM, and how much of it relates to local experiences and struggles over livelihoods? While it is clearly too simplistic to suggest a strict dualism of interpretation, as the two factors overlap and intersect, posing the question in these terms helps draw attention to some of the ambivalences of Kenyan patrimonial politics. It is evident that there was a lot of anger in Kibera at the alleged rigging of the elections, which resulted in rioting, which then transformed into ‘looting for economic benefit’. Poverty plays a role here, but so does resentment and jealousy of the ‘rich’ landlords, as the looters were mainly tenants. This was certainly not a politically motivated ‘rebellion against the existing social order’, against a ‘landlord class’, as Luos simply took over rooms and became landlords themselves. However, there is an ethnic dimension to this: to many Luos, the Kikuyus are the representatives of an oppressive order. As one interviewee perceptively put it, the Kikuyus ‘seem to control everything in Kibera, and the Luos resent the fact that they have to pay them for everything: for the rooms, for water, for electricity, and most shops in Kibera were run by Kikuyus as well. The anti-Kikuyu feeling was there before the election; people feel they are being exploited by the Kikuyus.’42 As discussed above, there are also historical reasons for the Luos’ antiKikuyu sentiments, like the killing of Mboya and the economic and political marginalization of Nyanza in Kenyatta’s time. Kibaki, like Kenyatta and Moi before him, surrounded himself with his own ‘tribesmen’, excluding other ethnic groups, and this contributed to a growing anti-Kikuyu sentiment in many parts of Kenya.43 Moreover, in 2003 there was a strong sense of betrayal among Luos when Kibaki (and by extension, the Kikuyus) did not honour the agreement to reform the constitution and make Raila Prime Minister.44 These underlying anti-Kikuyu and anti-landlord sentiments have been used by Raila for his own benefit, in a classic illustration of political tribalism to mobilize political support and voters: ‘Politicians see Kibera not as a large slum, but as a strategic reservoir for votes.’45
42. Interview, non-Luo man, Gatwikira, 12 April 2008. 43. HRW report, ‘Ballots to bullets’, p. 4; Lynch, ‘Courting the Kalenjin’, pp. 556–9. 44. Lynch ‘The fruits’, p. 258; Jeffrey Steeves, ‘Beyond democratic consolidation in Kenya: ethnicity, leadership and “unbounded politics”’, African Identities 4, 2 (2006), pp. 195–211. 45. PeaceNet, ‘The quest’, p. 30.



Using ethnic discourse during election campaigns is not enough; the reciprocal nature of patronage demands ‘redistribution’ in exchange for votes. However, it is not only the votes that count. Additional ‘support’ is just as important – for example coming to political rallies, and creating mayhem when required. Already in the 1960s KANU used its Youth Wing to intimidate political opponents, while in the 1980s it became commonplace for politicians to have their own (violent) gangs of supporters, generally of the same ethnic group. In Gatwikira, ODM has an unofficial office, a base for the local party mobilizers. From here they organize the (mainly Luo) youths that hang around, ready for a ‘job’ – they collect protection fees from shops, or ‘repair tax’ if someone wants to repair a house. It is these young men that form and join gangs like the Taliban, or, in the case of the Kikuyu, the Mungiki. This phenomenon is by no means unique to Gatwikira – most villages of Kibera and other slums have similar groups of youths, gangs that easily adjust to the circumstances. Sometimes they are ‘vigilantes’, sometimes they ‘support’ landlords to kick out non-paying tenants, sometimes they are plain criminals.46 These youths can also be mobilized easily for political interests – they make up the ODM youth wing, supporting Raila by doing anything necessary and required, whenever, to make him win the elections.47 ‘The Gatwikira Luos are fanatics, they follow Raila, who controls them completely; they worship him like a God.’48 Many informants (including Luos) claimed that it was mainly these youths that were responsible for inciting (forcing) others to loot and destroy Kikuyu shops and houses, and remove Kikuyus from Kibera. It is this kind of ‘political’ support that requires a ‘reward for a job well done’. This reward can be simple and direct: the fieldwork revealed that the youths get paid for the job they do. For example, for uprooting the railway in Kibera at the height of the mass demonstrations in early January, they received between 100 and 200 shillings each.49 Rewards can also be indirect, more substantial, or long-term. There are for example persistent rumours that Raila, for many years, has paid rent for many of his supporters, and monthly allowances to his main ‘mobilizers’, although such information is notoriously difficult to verify. Furthermore, in 2001 Raila asked President Moi to lower rents in Kibera, thus creating a rent conflict that he subsequently made no effort to solve. In Kibera it is frequently claimed that, during rallies, Raila would ‘talk peace’ in Kiswahili but then throw in a few Luo words to the effect of ‘Don’t pay rent!’; as a result, many Luo
46. Interviews, Luo and non-Luo informants in Gatwikira, March 2008. See also Anderson, ‘Vigilantes’, and Mueller, ‘The political economy’. 47. See also Anderson, ‘Vigilantes’, p. 551; Mueller, ‘The political economy’, pp. 189–94; PeaceNet, ‘The quest’, pp. 27–30. 48. Interview, Luo man, Kibera, 14 April 2008. 49. 200 Kenyan shillings is the daily rate for unskilled casual construction work.



tenants refused to pay rent because, as they claimed, ‘Raila told us not to pay rent.’50 While the veracity of these statements is difficult to ascertain, what matters in this context is the perception that political support results in material reward. In the recent clashes, these material rewards have been substantial: the proceeds of the looting; the burnt-down Toi Market, first Kikuyu-controlled, but now rebuilt and in the hands of Raila supporters; and, most importantly, the expansion of ‘Luo territory’ in Kibera where no rent is paid, and where Luos have effectively taken over from the Kikuyus as the new landlords. The question is whether this was part of a ‘Grand Plan’, a reward for Raila’s supporters. Understandably, many Kikuyus do believe that the violence was part of a plan to rid Kibera of Kikuyus and take over their rooms and businesses. Several people mentioned that before the elections Kikuyus received threats like ‘after the elections, this shop will belong to me’, and that Luos were already ‘booking’ houses and shops that had been marked for takeover.51 All this seems to indicate a certain amount of advance planning. Kikuyu informants frequently claimed that it was probably not planned at a top ODM level, but at a lower, local, level: the ‘smaller’ politicians, like councillors, the party people on the ground, the youth leaders (the people based in those unofficial party offices), that must be seen to champion the Luo cause to climb the ranks. Evidence from the Rift Valley also suggests that local leaders have considerable autonomy and that they organize the youths and pay them part of what they receive from ‘above’.52 There was a general belief amongst Kikuyu informants that if Raila had won the elections, it would have been worse for them: they would have been removed from Kibera with even more conviction, violence and impunity, since the Luos would expect to be protected by the new President himself. This may be true, considering that the local Kibera administration hardly intervenes, and the District Officer and the Chiefs are said to be scared of the violent and fanatical Luos in Gatwikira, and of Raila’s connection there. Raila’s dilemma – the complications of the patronage system On 24 April President Kibaki and Prime Minister Raila Odinga started their so-called ‘Peace Tour’ around Kenya to ensure an end to violence and a safe return home for displaced people.53 On 29 May Raila came to Kibera (without Kibaki) for a peace meeting in which he mentioned the problem of people illegally squatting in other people’s houses. He added that legal
50. Information from at least 6 non-Luo people who understand the Luo language. 51. Different informants, April/May 2008. See also Mueller, ‘The political economy’, p. 203. 52. See the HRW report, ‘Bullets to ballots’, pp. 35–9. 53. Quite unsuccessfully in the Rift Valley; see ‘Don’t return, refugees warned’, Daily Nation, 1 April 2008, p. 34.



action would have to be taken by the government against them, but that he himself was not in charge of that. With this diplomatic statement, ‘Raila the Prime Minister’ confirmed that the occupation of rooms is illegal, but at the same time, ‘Raila the Luo politician’ condoned it by not telling his people to get out of the rooms or pay rent. By not taking a position, his actions show that patron–client relationships can be more complicated than ‘one leads, the others follow’, with the patron in total control. Raila’s Luo constituents expect resources, protection, and other advantages at national level from their ‘big man’ Raila. His earlier call for blanket amnesty for all post-election violence offenders can be seen in that light: as most people arrested nationwide were ODM supporters, Raila was expected, maintaining the patron–client relationship, to try and protect them.54 Being in such a powerful position, his supporters would expect even more from Raila – helping them to stay in their newly occupied houses, thus keeping their reward for supporting him. This puts him in a difficult position, caught between being the Prime Minister and being the Luo politician. He is, to a certain extent, a prisoner of the political game he plays, and his choice is in fact very limited: as Prime Minister he must be seen to follow the rule of law. However, this risks alienating many of his supporters in Kibera, and thus might ultimately affect his prospects of re-election and his political career. According to unofficial reports, Raila had some informal meetings in Kibera in which he was told by his supporters that any attempt to get them out of their new houses would be at the peril of losing their support. Nevertheless, in October 2008 Raila’s personal assistant Edward Ketta went, with the local administration, to Gatwikira for a meeting with the ‘support base’, and announced that people should leave the occupied houses. The crowd reacted violently and Ketta had to leave in a rush, his car damaged. This seems to indicate that Raila is not able to exert complete control over his support base, and that key constituents in Kibera do not automatically accept his authority. After this Raila’s discourse changed; at his ‘homecoming party’ on 29 November 2008, while PNU minister Martha Karua was jeered when she asked Raila to help out in the ‘house issue’,55 Raila defended his people by saying that many Luos were chased out of Kikuyu-dominated areas, and had also lost property. Nevertheless, recent developments have indicated that Raila’s support in Kibera is being undermined by his failure to deliver, as Prime Minister and as Luo leader. His popularity in Kibera (and nation-wide) is decreasing because of rising prices of basic commodities,

54. ‘Raila: free youths without conditions’, Daily Nation, 30 May 2008, p. 1. 55. ‘Raila opens up’, East African Standard, 30 November 2008, <http://www.eastandard. net/archives/InsidePage.php?id=1144000484&catid=4&a=1> (30 May 2009).



especially maize, and government corruption.56 There are rumours that in the next elections Raila will not stand in Kibera, since he is no longer sure to win there again after too many broken election promises.57 This article has shown that a full understanding of the post-election violence cannot be centred at the national level only, but requires attention to local dynamics. Focusing on the local dynamics in Kibera shows that historical and socio-economic factors play an important role in explaining the violence, while also pointing to patronage and big man politics as key factors. While the ‘underlying factors’ are linked to dynamics at national level, that relationship is complex and cannot be read in a straight logic of patrimonialism. Despite his popularity in Kibera as a ‘big man’ Luo leader, Raila must reward his supporters; redistribution is a basic element of the patronage system. Being Prime Minister, he has not been able to do that satisfactorily, and his waning popularity is an indication that he is not able to live up to the expectations of his voters. This illustrates the difficult situation and the pressures ‘big men’ may face in mediating between their public roles and the demands of supporters.

56. See also ‘Dozens dash for maize flour for the poor’, Daily Nation, 10 December 2008, <> (30 May 2009); ‘The coalition has failed and should resign’, East African Standard, 1 March 2009, <> (30 May 2009). 57. In the 2007 parliamentary elections he had only 70 percent of the votes against his nearest PNU rival Stanley Livondo’s 28 percent. < default.asp> (17 April 2009). Reliable sources claim that Raila had been afraid he would lose these elections.

Sign up to vote on this title
UsefulNot useful