Democratization, Sequencing, and State Failure in Africa: Lessons from Kenya

Abstract

In order to complement ongoing current empirical research, this article draws wider lessons from the crisis that grew out of the disputed Kenyan presidential election of December 2007. Looking beyond the immediate trigger for the subsequent violence - namely, the election itself - the paper instead locates the roots of the crisis within three historical trends: elite fragmentation, political liberalization, and state informalization.The origins of each can be traced to the style of rule employed by Daniel arap Moi. Even though his first government of 2002-5 perpetuated these trends, President Mwai Kibaki failed to recognize their implications for national unity and the exercise of power in 2007. The article then addresses the sequencing debate within the literature on democratization, identifying the lessons that can be taken from the Kenyan case for other states. Kenya has shown again that political liberalization is a high-risk activity that can produce unintended side-effects. Drawing on examples from other African states, we argue that the processes of democratization and reform can be undertaken simultaneously, but that this twin-tracked approach requires institutional reforms not yet undertaken by a large number of African polities.

Full text

African Affairs, 108/430, 1–26 doi: 10.1093/afraf/adn065
C
The Author [2008]. Published by Oxford University Press on behalf of Royal African Society. All rights reserved
Advance Access Publication 2 December 2008
DEMOCRATIZATION, SEQUENCING,
AND STATE FAILURE IN AFRICA:
LESSONS FROM KENYA
DANIEL BRANCH AND NIC CHEESEMAN
ABSTRACT
In order to complement ongoing current empirical research, this article
draws wider lessons from the crisis that grew out of the disputed Kenyan
presidential election of December 2007. Looking beyond the immediate
trigger for the subsequent violence – namely, the election itself – the paper
instead locates the roots of the crisis within three historical trends: elite frag-
mentation, political liberalization, and state informalization.The origins of
each can be traced to the style of rule employed by Daniel arap Moi. Even
though his first government of 2002–5 perpetuated these trends, President
Mwai Kibaki failed to recognize their implications for national unity and
the exercise of power in 2007. The article then addresses the sequencing
debate within the literature on democratization, identifying the lessons that
can be taken from the Kenyan case for other states. Kenya has shown again
that political liberalization is a high-risk activity that can produce unin-
tended side-effects. Drawing on examples from other African states, we
argue that the processes of democratization and reform can be undertaken
simultaneously, but that this twin-tracked approach requires institutional
reforms not yet undertaken by a large number of African polities.
LESS THAN A WEEK BEFORE KENYANS WENT TO THE POLLS IN
DECEMBER 2007, the chairman of the Electoral Commission of Kenya
(ECK), Samuel Kivuitu, addressed domestic and international election ob-
servers. Kivuitu, an experienced politician and electoral official, was confi-
dent that the general elections would proceed calmly and that any resulting
transition would occur smoothly. He unequivocally said of the electorate
that ‘This time they are the ones giving the warnings.’1Like many in the
audience, Kivuitu drew his confidence in Kenyan democratization from two
events. The first of these was the defeat by Mwai Kibaki and his National
Rainbow Coalition (NARC) of the then ruling party, the Kenya African
Daniel Branch (d.p.branch@warwick.ac.uk) is Assistant Professor of African History, Uni-
versity of Warwick. Nic Cheeseman (nicholas.cheeseman@africa.ox.ac.uk) is a Lecturer in
African Politics at the African Studies Centre, University of Oxford.
1. Samuel Kivuitu, ECK chairman, briefing to domestic and international observers,
Kenyatta International Conference Centre, Nairobi, 21 December 2007, from authors’ own
notes.
1
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

2AFFRICAN AFFAIRS
National Union (KANU), and its presidential candidate, Uhuru Kenyatta,
in the 2002 general election. The second was the rejection of the Kibaki
government’s controversial draft constitution in a referendum held three
years later. Kenyan politicians, so Kivuitu then believed, had at last learnt
to listen to the electorate. However, subsequent events revealed that the
democratic commitment of many political leaders was shallow at best.
The 2007 election was a two-horse race between the incumbent, Mwai
Kibaki, of the Party of National Unity (PNU), and his main rival, Raila
Odinga, of the Orange Democratic Movement (ODM).2While the parlia-
mentary vote was comfortably won by ODM, the presidential vote was,
as predicted, extremely close. Kibaki was controversially declared to have
won the latter poll by a slender margin, despite accusations of electoral
malpractice.3The announcement triggered widespread civil conflict and
political order subsequently broke down. There were two principal strands
of violence in operation. One strand involved the targeting of members of
ethnic communities broadly supportive of Kibaki and PNU by members of
other ethnic groups that had backed ODM and Odinga. This most com-
monly occurred in the Rift Valley Province as Kalenjin youths attacked
their Kikuyu neighbours. A second strand of violence witnessed across the
country but mostly within urban centres took the form of a series of violent
backlashes from state police and rival militias against those protesting the
result of the election. In total the violence claimed over 1,000 lives and
caused the displacement of over a hundred thousand more, many of whom
are still too frightened to return. The bloodshed was halted by the creation
of a ‘government of national unity’ in March 2008.4Kenyan ‘democracy’
was clearly neither as stable, nor as consolidated, as many had dared hope
just days before.
The 2007 polls and the subsequent violence have already generated sig-
nificant academic debate.5Doubtless, many articles and books are set to
follow in the coming months and years. Both the studies already published
and the precedent of the literature produced in the wake of past elections
suggest that forthcoming studies will follow a particular line of inquiry: what
the outcome of the election actually was (although this will never be known
for certain); how the rigging occurred; the role of donors; what inspired
the civil conflict; how different communities interpreted and responded to
the election; and who perpetrated the violence and why. While acknowl-
edging such empirical studies to be important, this article represents an
2. For the first analyses, see Daniel Branch and Nicholas Cheeseman (eds), Election Fever:
Kenya’s crisis, special edition of Journal of Eastern African Studies 2, 2 (2008).
3. Nic Cheeseman, ‘The Kenyan election of 2007: an introduction’, Journal of Eastern Afr ican
Studies 2, 2 (2008), pp. 166–84.
4. International Crisis Group, ‘Kenya in crisis’ (International Crisis Group Africa Repor t
No. 137, Brussels, 2008).
5. Branch and Cheeseman, Election Fever.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 3
attempt to consider the wider lessons and implications of the Kenya crisis
for other states undergoing political liberalization. The election of 2007 and
subsequent events have much to say to a more widespread malaise within
democratization, described by Diamond as a global crisis of democracy.6
In exploring such questions, we explicitly reject any misguided and in-
flammatory attempts to reduce the Kenya crisis to a story of ‘tribal warfare’.7
It is undeniable that competition between different ethnic groups for land
and resources came to be a defining feature of the Kenya crisis. Moreover,
it is clear that the violence that followed the elections owed much to the
evolution of pro- and anti-Kikuyu voting blocs, particularly as a result of
the divisive majimbo (regionalist) debate undertaken during the campaign.
However, more often than not, ethnic identities become salient because they
have come to embody other societal divisions, such as regional inequali-
ties, control over land, and access to political opportunities. The increased
salience of ethnicity is better understood as the outcome of changes in in-
stitutional context and the decision-making matrix facing political leaders,
rather than their cause.8Mueller argues persuasively that the origins of
these changes in the Kenyan context were in large part a creation of the
early 1980s. The one-party state was transformed during the first decade
of Daniel arap Moi’s presidency and then broke down entirely. As Mueller
argues, the subsequent return to multi-partyism in 1991 was accompanied
by two developments that boded ill for Kenya’s future stability: the disso-
lution of the state’s monopoly over the means of violence, and the rise of
‘winner takes all’ elections contested by parties dependent upon ethnic and
clientelist bases.9
In addition to the historical roots identified by Mueller, and the short-
term trigger provided by the contested election, we believe three other
prior interwoven processes contributed significantly to the Kenya crisis:
elite fragmentation, political liberalization, and state informalization. The
‘bureaucratic-executive’ state that dominated Kenyan political life after
independence established an all-powerful presidency, whose control was
grounded in the ability to direct political activity at the grassroots via the
prefectural structure of the provincial administration.10 The stability of this
bureaucratic-executive state rested on the collusion of a range of elites,
all determined to protect the highly inegalitarian post-colonial settlement
on which their wealth depended. Significantly, this elite alliance preceded
6. Larry Diamond, ‘The democratic rollback’, Fo re ign A ff air s 87, 2 (2008), pp. 36–48.
7. For example, Adrian Blomfield, ‘Inside Kenya: the tribal slaughter’, The Daily Telegraph,
4 January 2008.
8. Robert Bates, When Things Fell Apart: State failure in late century Africa (Cambridge Uni-
versity Press, Cambridge, 2008).
9. Susanne Mueller, ‘The political economy of Kenya’s crisis’, Journal of Eastern African
Studies 2, 2 (2008), pp. 185–210.
10. Branch and Cheeseman, Election Fever.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

4AFFRICAN AFFAIRS
the formation of the one-party state and underpinned the stability of the
post-colonial polity. Following the accession of Daniel arap Moi to the pres-
idency in 1978, the elite alliance began to fragment. Against a worsening
economic backdrop and enjoying far fewer political resources than his pre-
decessor, Moi adopted an increasingly exclusionary system of government,
especially after the failed coup attempt of 1982. Kenya thus fits the frame-
work recently set out by Robert Bates, in which the long road to disorder
begins with a decision by the executive to adopt an increasingly predatory
stance, itself triggered by some combination of falling public revenues and
increased political uncertainty.11 During the course of the 1980s, Moi’s
approach ruptured the fragile elite consensus. The subsequent process of
elite fragmentation undermined the ability of the regime to demobilize op-
position, leading to the end of the one-party state and a process of political
liberalization.
The introduction of multi-party politics created new opportunities for
political leaders to abandon the ruling party, and so contributed to the
ongoing process of elite fragmentation. The breakdown of the KANU al-
liance, in turn, rendered Moi ever more vulnerable.12 He responded to the
new challenges posed by political liberalization by pursuing a policy of state
informalization characterized by the looting of the Kenyan state, and by
transforming some of Kenya’s gangs into ethnic militias as a means to hold
onto power through force. Urban gangs had emerged initially as a result of
an autonomous set of processes relating to urbanization, unemployment,
and the vacuum of control in urban locales. However, the infusion of elite
patronage and funding emboldened gangs to become more ambitious and
more violent than ever before. The continued political relevance of gangs
was subsequently encouraged by the process of elite fragmentation; the de-
sire of a range of political actors to employ gangs for ‘protection’ during
election campaigns meant that for the first time protest movements ‘from
below’ were connected to a divided elite leadership that could provide them
with funding and direction. At the same time, the increasingly central role
played by militias both eroded trust among the elite and contributed to
the sense that Kenyan elections were ‘high stakes’ events, reinforcing elite
fragmentation. The erosion of trust between individuals and between oppo-
sition political actors in state institutions was demonstrated on numerous
occasions in the aftermath of the elections; ODM’s strategy of mass disobe-
dience was predicated on a lack of faith in key political institutions to deliver
impartial verdicts.
11. Bates, When Things Fell Apart.
12. By 1992, commentators believed that transfer of power was inevitable. See 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).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 5
Taken together, the mutually reinforcing processes of elite fragmenta-
tion, political liberalization, and state informalization radically altered the
balance of power between the centre and the periphery. It was in this con-
text that Kibaki – faced with an uncertain election outcome – sought to
wield the power of the bureaucratic-executive state as his predecessors had
done. In doing so, he took confidence from Kenya’s ‘top-heavy’ one-party
constitution. That constitution has remained largely unmodified despite the
transition to multi-party politics and continues to confer great power on the
executive. Kibaki was also well aware that the coercive power of many of
the key security institutions of today’s state is as strong as – or stronger
than – in earlier periods of post-colonial history. What Kibaki failed to
realize was that without the elite alliance that had given the Kenyan state
its initial strength, and against an opposition remarkable for its unity, the
physical appearance of the bureaucratic-executive state was no longer a true
reflection of its ability to demobilize opposition. The centre tried to hold as
it had done in the past. But this time Kibaki and his supporters found that
the centrifugal forces within the Kenyan political system could no longer
be contained. In other words, political disorder in Kenya was not the result
of a straightforward weakening of the state, as has been hitherto commonly
assumed. Instead, that disorder was the consequence of the Kibaki govern-
ment’s decision to employ the state in a manner that both Kenyatta and
Moi would have recognized, but in a far less favourable context.
The interconnected processes of fragmentation, informalization, and lib-
eralization described here are shared by many other states in a similar stage
of democratization: it is this that renders the events of early 2008 in Kenya
all the more alarming.13 The case of Kenya tells us much, therefore, about
the importance of the sequencing of democratic reforms as debated by
Carothers and his critics.14 Most significantly, it reveals the challenges fac-
ing a good many ongoing democratic experiments, from Uganda to Zambia,
Malawi to Burundi, and Cˆ
ote d’Ivoire to Nigeria, and the possible unin-
tended outcomes of such experiments when exposed to the stresses of an
election such as that witnessed last December.15
Elite fragmentation
As we have described elsewhere, the bureaucratic-executive state
that dominated Kenyan political life after independence established an
13. William Reno, ‘The politics of insurgency in collapsing states’, Development and Change
33, 5 (2002), pp. 837–58.
14. Thomas Carothers, ‘How democracies emerge: the sequencing fallacy’, Journal of Democ-
racy 18, 1 (2007), pp. 12–27.
15. Michael Bratton and Nicolas van de Walle, Democratic Experiments in Africa: Regime
transitions in comparative perspective (Cambridge University Press, Cambridge, 1998).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

6AFFRICAN AFFAIRS
all-powerful presidency. Executive control was grounded in the ability to
direct political activity at the grassroots via the prefectural structure of the
provincial administration.16 Except for the reintroduction of multi-party
politics, the institutional format of that bureaucratic-executive state has re-
mained largely unaltered since the late 1970s. The initial stability of this state
rested on the collusion of a range of elites, the emergence of which needs
to be understood in the distinctive cultural, ideological, and institutional
context of the late-colonial period. Of particular importance was the cre-
ation of a cleavage between supporters of the Mau Mau rebellion and those
loyal to the colonial regime. Membership or support of the latter group
was what bound together the interests of the executive, foreign capital, the
bureaucratic elite (top civil servants, government advisers, and the heads
of parastatals), and the administration (provincial commissioners, district
commissioners, and so on).
The shared interests of this alliance were deliberately engineered by the
British government in the dying days of the empire, as the colonial gov-
ernment sought to transfer power to a reliable and sympathetic elite. A
broad range of policies were employed to this end, most notably the use
of loyalty tests to determine voting rights, the manipulation of land reset-
tlement schemes to create a land-owning middle-class, and the granting of
preferential treatment to ‘loyalists’ in the course of the Africanization of
the administration.17 In so doing, the British effectively cultivated a bour-
geoisie in waiting, which shared the common interest implied by its dom-
inant socio-economic position. Through its control of the administration
and the government, members of the elite also enjoyed the capacity to re-
produce the system on which their privileged positions depended. It was the
common interest and outlook of this group that allowed KANU smoothly
to assimilate the opposition Kenya African Democratic Union (KADU)
shortly after independence, transforming Kenya into a de facto one-party
state. Of course, individual political leaders could exist outside of this elite
alliance, the most prominent example being Oginga Odinga, the father of
Raila Odinga. However, individuals such as Odinga remained on the fringes
of power and were unable to build up significant groundswells of support
outside their home provinces.
Members of these elites proved determined to protect the highly inegali-
tarian post-colonial settlement on which their wealth depended. Elite collu-
sion served to contain dissent from below, most notably enabling Kenyatta
to demobilize the more radical elements of Kenya’s diverse ‘nationalist’
movement. As a result, many of the key fault-lines in Kenyan politics and
16. Daniel Branch and Nic Cheeseman. ‘The politics of control in Kenya: understanding the
bureaucratic-executive state’, Review of African Political Economy 33, 107 (2006), pp. 11–31.
17. Ibid.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 7
society remained unactivated during this period, despite the fact that the his-
torical grievances that underpinned the Kenya crisis can be traced to three
dynamics established in the period 1940–70. First, land in the Rift Valley
was made available to Kikuyu settlers, to the great chagrin of the Kalenjin
who saw the Kikuyu as ‘outsiders’.18 Second, Luo leaders were system-
atically excluded from power: Tom Mboya through assassination, Oginga
Odinga by the refusal of first Kenyatta and then Moi to allow him to con-
test one-party elections.19 Third, most of the positions that comprise what
one might think of as the core executive – head of the army, president, fi-
nance minister, foreign minister – went to members of Kenyatta’s Kiambu
Kikuyu inner circle.20 Many groups, including the Kamba and the Luhya,
were not represented in proportion to their share of the voting population.
Finally, within the Kikuyu community ex-Mau Mau felt betrayed at the lack
of land redistribution, while non-Kiambu Kikuyu could also complain of
marginalization.21
The elite alliance that initially mitigated against the mobilization of these
grievances proved to be unsustainable in the long term. Elite fragmentation
was already visible by the time of Jomo Kenyatta’s death in 1978. Kenyatta’s
unquestioned status as Baba Taifa, the Father of the Nation, initially
provided sufficient cement to hold the elite alliance together. However, his
consolidation of power amongst a small Kiambu elite, and the failure of his
government to address land grievances, weakened the elite alliance, which
threatened to collapse altogether under the strain of the succession politics
of the first President’s final years. Once installed as President, Kenyatta’s
successor, Daniel arap Moi, briefly restored that alliance, and initially ap-
pealed for support on the basis of a policy of continuity. His slogan of ‘Nyayo’
(literally ‘footsteps’) suggested that he intended to rule in the image of his
predecessor whilst undertaking a series of populist reforms. However, Moi
faced numerous challenges to any simple continuation of Kenyatta’s style
of rule. The new President lacked his predecessor’s sizeable ethnic base and
vast personal wealth, and could not point to any great record of anti-colonial
resistance with which to boost his appeal across the country. Nor could Moi
benefit politically from the distribution of the fruits of independence in the
way Kenyatta had. The most notable of these sources of patronage with
which Kenyatta consolidated his support had been the millions of acres of
land vacated by European settlers which were distributed, albeit contro-
versially, to the land-hungry population through the 1960s. The absence
18. 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.
19. Oginga Odinga, Not Yet Uhuru (East African Educational Publishers, Nairobi, 1967).
20. David Throup, ‘The construction and destruction of the Kenyatta state’ in Michael
Schatzberg (ed.), The Political Economy of Kenya (Praeger, London, 1987), pp. 33–74.
21. Branch and Cheeseman, ‘The politics of control’.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

8AFFRICAN AFFAIRS
of such comparative advantages was compounded by circumstances over
which Moi had no control. Most significantly, Kenya’s second President
took office just as the prevailing global economic climate took a turn for
the worse following the oil price shocks of the 1970s.22 As the framework
set out by Bates suggests, the combination of falling public funds and rising
political uncertainty set in motion a vicious cycle of predation, economic
decline, and uncertainty.23
With his reformist agenda stalled and fearful of hostile vested interests
that had been entrenched during the Kenyatta regime, Moi quickly moved
to exert control. Lacking the resources to co-opt likely opponents, the
President set out to marginalize and weaken them instead. First, ethnic
welfare organizations, such as the Gikuyu, Embu and Meru Association
(GEMA), were prohibited in 1982. Then, a de jure one-party state was in-
troduced in that same year.24 Moi’s exclusionary bent angered key interest
groups, including Luo intellectuals, Kamba military officers, and Kikuyu
businessmen, politicians, and civil servants. A confused coup attempt later
in 1982 narrowly failed, but had dramatic consequence nonetheless.25
By confirming Moi’s fears for his own position, the coup exacerbated his
paranoia and triggered a fresh round of predation. Most obviously, Moi
introduced Kenya’s most dramatic phase of elite rotation. Many Kikuyu
ministers, administrative officers, and senior figures within the upper ranks
of the military and police were replaced with loyal Kalenjin. In the 1983
general elections, the executive intervened to prevent the success of ‘subver-
sive elements’ to an unprecedented degree.26 Political exclusion went hand
in hand with a process of economic transformation, as government inter-
vention was used to confer greater advantages on Moi’s traditional support
base, in the process harming the interests of Kikuyu farmers.27
Despite this, a semblance of an elite alliance remained intact. In part, this
was because Moi followed the old adage of keeping his friends close but
his enemies even closer. While prominent Kikuyu figures such as Attorney
General Charles Njonjo were forced to depart, others, such as Mwai Kibaki,
22. David Throup, ‘The construction and destruction’.
23. Bates, When Things Fell Apart.
24. For more on the nature of the transition see Jennifer Widner, TheRiseofaParty-Statein
Kenya: From ‘harambee’ to ‘nyayo’ (University of California Press, Berkeley, CA, 2002).
25. There appear to have been a number of coup attempts, which partly failed because
different sections of the army were unaware that other sections were also planning coup
attempts. In the confusion, the GSU was able to reassert control in the name of the President.
See Throup, ‘The construction and destruction’.
26. Elections were held for constituency MPs. The President was indirectly elected via an
electoral college of MPs, and, as the executive controlled the vetting process for legislatures,
was always elected unopposed. See Nic Cheeseman, The Rise and Fall of Civil Authoritar ianism
in Africa: Patronage, participation in political parties in Kenya and Zambia (University of Oxford,
unpublished DPhil manuscript, 2006).
27. Robert Bates, Beyond the Miracle of the Market: The political economy of agrarian development
in Kenya (Cambridge University Press, Cambridge, 1989).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 9
were retained. The balance between inclusion and exclusion was shifting,
but it would not become untenable until 1988. Even in the final days of
the one-party state, KANU still boasted a remarkable collection of ethno-
regional patrons. It is an oft-forgotten fact that all three main presidential
candidates in the 2007 presidential elections had previously served as min-
isters under Moi. However, the success of this strategy now depended on
the high costs of exclusion. Moi used the state’s strong coercive capacity to
increase the regime’s control over political and economic life. Oppression
and intimidation were not unknown under Kenyatta.28 Yet the brutality
of Moi’s rule hardened opposition to the regime, forcing opposition deeper
underground.29 Coercion and fear had replaced elite consensus as the mech-
anism holding the system together. Elite figures no longer invested in the
regime because they believed in it; they tolerated it because they had no
alternative.
As economic conditions undermined Moi’s ability to provide the patron-
age necessary to maintain key ‘clients’ from public revenue, he responded
with ever more grandiose schemes designed to transform state resources
into private slush funds. Although the process of state informalization be-
gan in earnest in the late 1980s, its roots can be traced back to the attempt
of the Moi clique to establish a firm economic base in the late 1970s. Si-
multaneously, the executive’s control over political space was enhanced by
the introduction of queue voting for party and general elections, which
allowed the regime to identify and punish those brave enough to line up
behind opposition candidates. Initially, this empowered Moi to eliminate
opposition during the KANU elections of 1986 and to reinvent the party as
a body wholly beholden to the President.30 However, in the 1988 general
elections the public nature of queue voting meant that the magnitude of
the vote inflation undertaken by government-appointed returning officers
was clear for all to see. The elections marked a turning point. On the one
hand, the ‘defeat’ of many of the most senior non-Kalenjin figures within
the Parliament tilted the balance in favour of outright exclusion, and repre-
sented the final nail in the coffin in KANU’s elite alliance. Shortly after the
polls political heavyweights including Martin Shikuku, Charles Rubia, and
Kenneth Matiba publicly announced their opposition to Moi and the one-
party state. On the other hand, the open rigging of the general elections
served to undermine public faith in one of the only institutions left that
conferred any legitimacy on the regime. In short, executive predation led
to a process of elite fragmentation. In turn, fragmentation provided the
28. Nic Cheeseman, ‘Political linkage in the Kenya post-colony: assessing the structure of
colonial legacy’, Africa Today 51, 1 (2006), pp. 3–24.
29. Angelique Haugerud, The Culture of Politics in Modern Kenya (Cambridge University
Press, Cambridge, 1995).
30. Widner, TheRiseofaParty-State.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

10 AFFRICAN AFFAIRS
alternative elite leadership required to transform Kenya’s ‘diffuse’ oppo-
sition into a mass movement for change, paving the way for political
liberalization.31
Moi mistakenly believed that the coercive capacity of the state would
enable him to contain the inevitable groundswell of opposition following
the rigged polls. But these institutions had proved so effective in the past
partly because half of the job of containing dissent had already been done
courtesy of the elite alliance and the demobilization of popular forces it
implied. Until the late 1980s, the Kenyan state had never faced a sustained
and widespread challenge.32 Following Rubia’s and Matiba’s public call for
the reintroduction of multi-partyism, and the formation of the Forum for
the Restoration of Democracy (FORD), the threat that elite fragmentation
posed to the one-party state became clear.33
The excluded elite sought to reinstate multi-party politics as a way to
unlock Moi’s monopoly over political and economic opportunities. Their
new-found willingness to mobilize grassroots protest against the regime un-
dermined KANU’s ability to demobilize opposition to the one-party state.
KANU initially responded to mass protest with brutal repression. But the
outpouring of popular frustration following a long period of economic and
political stagnation meant that the protests could not be contained easily.
Elite fragmentation had shifted the balance of power away from the centre,
undermining KANU’s ability to maintain political order. KANU’s brutal
repression of pro-democracy activists, including the detention and torture
of opposition leader Kenneth Matiba, shocked the international community
into belated condemnation of the Moi regime. More significantly, the distri-
bution of foreign aid and financial support was halted pending political and
economic reforms.34 Facing intense pressure from above and from below,
Moi was finally persuaded to announce a return to multi-party politics in
1991.
Political liberalization
Moi’s decision to move quickly to multi-party elections was premised on
two assumptions: first, that multi-partyism represented KANU’s best op-
portunity to retain power; and, second, that such a gesture would resume
31. Ibid.
32. A brief challenge was offered by the Kenya People’s Union (KPU), but the combination
of administrative capacity, executive legitimacy, and broad elite support enabled KANU to
contain the KPU threat with little difficulty. See Branch and Cheeseman, ‘The politics of
control’.
33. Throup and Hornsby, Multi-Party Politics, pp. 94–102.
34. Ibid., pp. 72–4.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 11
the critically required flow of international finance.35 However, the process
of political liberalization also brought with it new challenges. The release
of political prisoners, the legalization of opposition parties, and the partial
opening of political space created new opportunities for elite fragmentation.
Although some KANU MPs were unable to defect from the party because of
financial debts to senior party leaders and institutions,36 many did leave to
join FORD, while Mwai Kibaki subsequently established his own organiza-
tion, the Democratic Party (DP). The intensification of elite fragmentation
that ultimately split FORD into its two rival parties, FORD-Kenya and
FORD-Asili, reduced KANU to a minority government. Yet this new level
of fragmentation also created new opportunities for Moi to retain power
through the careful deployment of divide and rule policies. By encouraging
the further fragmentation of rival parties and exacerbating ethnic tensions,
Moi successfully promoted opposition disunity. This disunity was a major
contributory factor to the failure of opposition leaders to create a viable al-
liance until 2002; although the level of vote rigging was significant, Throup
and Hornsby estimate that without it Moi would probably still have won
the 1992 election.37
As well as feeding the ongoing process of elite fragmentation, political
liberalization provided incentives for Moi to pursue the informalization of
the state. Consequently, the story of Kenya in the 1990s is not one of
democratic consolidation and institutional reform. Rather, it is a tale of cor-
ruption, increasing elite polarization, the rise of militias, and what Mueller
has called the diffusion of violence.38 State informalization gave Moi the
ability to disrupt moves to weaken his hold on power, thus enhancing the
power he already derived through control over the remaining coercive ca-
pacity of the provincial administration and other organs of internal security.
Critically, his government consistently demonstrated its ability to obscure
and defer the debate over constitutional review. This left the ‘top-heavy’
constitution inherited from the one-party era in place. Although a range
of non-governmental actors bravely strove to open up political space, ulti-
mate control over the judiciary, provincial administration, and the media
remained vested in the executive. Kenya may have routinely held multi-
party elections, but saw little in the way of democratic consolidation in the
1990s. If anything the main organs of the state became less, rather than
more, institutionalized. The only significant exception was the security ap-
paratus. Moi was careful to maintain the capacity and professionalism of the
35. Stephen Brown, ‘Authoritarian leaders and multiparty elections in Africa: how foreign
donors help to keep Kenya’s Daniel arap Moi in power’, Third World Quarterly 22, 5 (2001),
pp. 725–39.
36. Throup and Hornsby, Multi-Party Politics, pp. 186–7.
37. Ibid., pp. 450–1.
38. Mueller, ‘The political economy of Kenya’s crisis’.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

12 AFFRICAN AFFAIRS
provincial administration and the paramilitary General Service Unit (GSU),
both directly accountable to the Office of the President.
Perhaps the most obvious example of this process of state informalization
was the spiralling corruption perpetrated by the Moi regime from the late
1980s onwards. The personalization of public wealth was (of course) noth-
ing new to Kenyan politics. It was the scale of the theft of public funds that
changed dramatically following the return to multi-party politics. In part,
this was a direct response to the shortened time horizons of the Moi govern-
ment. The increasing likelihood of electoral defeat resulted in senior figures
within the Moi government seeking to accumulate wealth at an accelerated
rate in order to secure their ‘retirement’. Practices such as land grabbing
proliferated.39 At the same time, democracy proved expensive. The need to
fund costly election campaigns provided an additional – and just as signif-
icant – motivation for spiralling corruption. Kenyan elections have always
seen vast amounts of official and unofficial expenditure. In 1992 it is esti-
mated that, despite the economic difficulties facing the country, the ruling
party spent over $100 million on the campaign.40 By the 2007 contest, this
figure had escalated.41
In a straightforward sense, the corruption of the Moi regime impover-
ished the state, as millions of dollars were looted through the infamous
Goldenberg scandal.42 However, an equally significant outcome was that
institutions of oversight and accountability had to be weakened in order for
the scams to remain undiscovered, or at least unreported. This meant that
central budgeting measures from the Finance Ministry to the Central Bank
of Kenya (CBK) were undermined. To this day, Kenya is unable to receive
international budgetary support because successive presidents have refused
to address this flaw. Greater corruption, then, created additional reasons for
the executive to encourage a process of state informalization that continues
to constrain anti-corruption efforts.
Of course, political liberalization actually increased political uncertainty
and shortened Moi’s time horizon. In this way, political liberalization pro-
vided incentives for the government to pursue state informalization as part
of a desperate attempt to ensure its short-term survival. But this does not
mean that there is an inevitable relationship between these two processes,
or that civil conflict in Kenya could not have been averted. Nevertheless,
similar but not identical developments can be observed in other African
39. Jaqueline Klopp, ‘Pilfering the public: the problem of land grabbing in contemporary
Kenya’, Africa Today 47, 1 (2000), pp.7–26.
40. Throup and Hornsby, Multi-Party Politics, pp. 357–9.
41. The Standard, ‘Storm over Sh3.5 billion poll campaign spending’, 24 April 2008,
<http://allafrica.com/stories/200804231185.html>(16 October 2008).
42. Tom Wolf, ‘Immunity or accountability? Daniel Toroitich arap Moi: Kenya’s first retired
president’ in Roger Southall and Henning Melber (eds), Legacies of Power: Leadership change
and former presidents in African politics (Nordic Africa Institute, Uppsala, 2006), pp. 197–233.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 13
countries that have undergone partial political liberalization. For example,
Zambia and Nigeria have experienced an increase in corruption; the respect
for basic human rights has declined in Burundi and Rwanda; and liberaliza-
tion has not produced a change of regime in Cameroon or Tanzania. Indeed,
Bratton and van de Walle find that the increase in the political rights en-
joyed by Africans in the early 1990s went hand-in-hand with a fall in the
quality of civil liberties.43 Consequently, it is important to consider whether
nascent processes of political liberalization undertaken in the absence of an
institutionalized and stable state may prove to be ultimately self-defeating.
We return to this theme in the final section of the article.
State informalization
As pernicious as the corruption and deliberate deinstitutionalization of
some state organs were to the Kenyan political system, perhaps the most
damaging dimension of state informalization during this period was the de-
centralization of control over violence. Bates’s paradigm of state collapse
provides an interesting starting point for thinking about the privatization
of violence during the 1990s. He is surely right that political liberalization
and the uncertainty it brings is the key back-story to the heightened use of
political violence in many African countries. However, it is also important
to recognize that the attraction of co-opting militias stems from their clear
comparative advantage over existing institutions of violence. Most obvi-
ously, militias deflect culpability from their sponsors. When the state police
commit human rights abuses the executive is clearly culpable. When a se-
cretive and poorly understood militia commits human rights abuses, it is
far less clear who should be held to account, and frequently impossible to
prove culpability. By using militias, political leaders can carry out tasks they
could never have ordered the police or the army to undertake during the
one-party era. But this short-term gain brings with it dangerous long-term
consequences, because the very advantage of militia groups is also their
key flaw: transferring the capacity for violence to ambiguous and complex
structures necessarily decentralizes control over the use of force and reduces
the ability of the centre to control conflict.44
Despite this, the KANU elite was prepared to devolve control over vio-
lence to local militias in the 1990s. The aim was two-fold. First, this would
surround the conduct of the elections with an atmosphere of violence and
suspicion likely to deflate the opposition vote. This effect was heightened by
the killing of Foreign Minister Robert Ouko, the detention of both Matiba
43. Bratton and van de Walle, Democratic Experiments in Africa.
44. Stathis Kalyvas, The Logic of Violence in Civil War (Cambridge University Press,
Cambridge, 2006), p. 108.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

14 AFFRICAN AFFAIRS
and Rubia, and the death of the outspoken Bishop Muge in 1990 alone.
Second, a ‘KANU zone’ would be created through a process of ‘ethnic
cleansing’ that saw members of ethnic groups likely to support opposition
parties, most notably the Kikuyu, driven from Moi’s heartlands in the Rift
Valley. More than 1,500 people lost their lives and over 300,000 were dis-
placed in the clashes.45 Even in the 1990s the violence was not limited
to Rift Valley, as ‘a private vigilante group known as Jeshi la Mzee (The
Elder’s Army) was unleashed on the public’ during key debates over consti-
tutional change.46 The violence of 2008 may have been tragic, then, but it is
important to remember that it was not unprecedented. The large-scale in-
strumental use of political violence, structured along ethnic lines and fuelled
by notions of majimbo, was introduced by Moi as a direct response to the
processes of elite fragmentation and political liberalization that threatened
to remove him from power. This tactic re-energized historical tensions that
had previously been effectively suppressed by the elite alliance, empowering
the very militia groups that would ultimately play a central role in the recent
crisis.
Although Bates’s treatment of the overall process of state failure is
helpful – primarily by modelling the militarization of society as a dynamic
process in which the executive and the population become engaged in an
‘arms race to the bottom’ – he fails to allow sufficient space for the con-
nections between the executive, the wider political elite, and the militia
groups that emerge through the process of predation.47 When elite divi-
sions are low and the centre holds, there are good reasons for the executive
to demobilize and disarm popular movements in general.48 When the cen-
tre fragments, the ‘instrumentalization of disorder’ may come into play
as militias are directed by members of the government against suspected
opposition supporters.49 In short, once elite fragmentation has occurred,
the militarization of society, and the taking up of arms by the wider popu-
lation, become significantly more likely.
Bates’s analysis also underplays the autonomous processes that work to
create the gangs that often evolve into militias over time. Urban gangs typi-
cally evolve in response to a vacuum left by a disinterested or incompetent
state. Their numbers are fuelled by a combination of urbanization and high
levels of youth unemployment, while they achieve local legitimacy through
45. Brown, ‘Authoritarian leaders’.
46. Shadrack Nasong’o, ‘Constitutional reform and the crisis of democratization in Kenya’
in Daniel Branch and Nic Cheeseman (eds), Our Turn to Eat: Politics in Kenya since 1950 (Lit
Verlag, Berlin, forthcoming).
47. Bates, When Things Fell Apart.
48. As indeed happened in many post-colonial African states immediately after indepen-
dence.
49. Patrick Chabal and Jean-Pascal Daloz, Africa Works: Disorder as political instrument (James
Currey, Oxford, 1999).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 15
their ability to provide ‘protection’; the infamous Mungiki gang emerged
and consolidated itself in precisely this way.50 Although the initial emer-
gence of gangs was in some cases facilitated by a prior decline in the scope
of authoritarian control, it is important to recognize that it was also rooted
in a set of factors independent of elite fragmentation and political liberal-
ization. The emergence of gangs throughout Kenya in the 1990s, largely in
response to economic rather than political stimuli, laid the foundations for
the privatization of violence. It was when the two trends of elite fragmenta-
tion and the growth of vigilante groups became fused that the Kenyan state
finally lost its monopoly over the legitimate use of force.
Given that the majority of militias in Kenya evolved first because of the
absence of the state, and later because of the direct sponsorship of state
actors, it is misleading to speak, as Bates does, of society militarizing against
the executive. Rather, we must recognize that following a process of elite
fragmentation, as occurred in Kenya, the instrumentalization of violence
that occurs is a process in which there are no firm dividing lines. Militia
groups that arise through this process are often co-opted by the executive,
engaged in conflict with rival communities, and then dropped. Such militia
groups may then fade into the background, side with opposition leaders,
or become ‘enemies of the state’ in general. In either of the latter two
eventualities it seems to be more accurate to speak of a society militarizing
through, rather than in opposition to, the executive.
Mueller captures this ambiguity well when she writes about the gangs in-
volved in KANU’s ethnic crusade: ‘In most cases these gangs were formed,
aided, or abetted by the state’s security apparatus and the provincial admin-
istration. Gangs of youth were organized by key KANU politicians who were
identified by name in both human rights reports and those produced by a
government commission.’51 But despite the violence being state-directed, it
was not wholly state-controlled: ‘Politics by other means had taken root all
over the country while various gangs both appeared and disappeared. All
of this was a further indicator that the state’s monopoly of legitimate force
was being challenged and diminished.’52 In part, this reflected the ‘gang
for hire’ nature of many of the militias that sprang up. These gangs were
not the permanent armed wing of a political party, but semi-autonomous
bodies employed by political leaders at election time. However, it also re-
flected the fact that alliances between different factions of the political elite
and Kenya’s urban militias quickly led to rival politicians fighting proxy
wars through groups such as the Baghdad Boys, the Taliban and Mungiki.
The large social base of some of these groups, coupled with their rapid
50. David Anderson, ‘Vigilantes, violence, and the politics of public order in Kenya’, African
Affairs 101, 405 (2002), pp. 531–55.
51. Mueller, ‘Political economy’, p. 190.
52. Ibid., p. 194.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

16 AFFRICAN AFFAIRS
expansion, meant that during the 1990s an increasing number of Kenyans
were being integrated into a world of political violence and mob rule.
It was the new-found willingness of members of the Kenyan elite to
challenge the status quo that led to the airing of once-submerged historical
grievances, and the refusal of the elite to demobilize forces from below which
rendered militia groups such a potent political threat. The dual processes
of the privatization of violence and elite fragmentation thus created the
foundation for widespread civil conflict. The interconnected nature of these
processes means that Bates’s depiction of the executive as a ‘specialist in
violence’ as distinct from the wider population is misleading. Even though
the coercive capacity of the Kenyan state has remained stable, struggles
within the elite have facilitated the incorporation of militias into the internal
logic of the political system. In other words, militia groups have become
embedded in the state itself (in some cases it may be more accurate of say
that the state is reduced to a constellation of militias). In this way militias
may be used to replace, and in extreme cases may confront, those elements
of the state over which the executive’s control is less secure.53 When po-
lice officers enter into deadly clashes with gangs sponsored by government
ministers, as occurred in Kenya in December 2007 and March 2008, the
consequence is not simply a battle between the ‘specialist in violence’ and
pretenders to that crown, but is more accurately described as a state at war
with itself. The picture is further complicated when police officers conduct
operations alongside militia members or themselves become virtually indis-
tinguishable from militia members due to rampant corruption and reliance
upon extra-judicial killings within the force.
We do not believe Kenya is a failed state. However, it is chastening to
think that the situation over the first months of 2008 closely resembled the
fluid and potentially uncontrollable situation typical of collapsing states.54
While the Kenya crisis revealed a residual capacity of the part of the state
to withstand protest from below, it also demonstrated how thin the dividing
line between order and disorder had become.
Things fall apart
Kibaki’s NARC government inherited a precarious state. Under Moi, po-
litical uncertainty and falling public revenue inspired an increasingly preda-
tory executive, which in turn undermined the elite alliance that had under-
pinned the bureaucratic-executive state. Elite fragmentation necessitated
a period of political liberalization, which in turn reinforced the process
of fragmentation. Together, these interconnected processes provided the
53. Chabal and Daloz, Africa Works.
54. Reno, ‘The politics of insurgency’.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 17
conditions that encouraged the steady informalization of the state. During
this latter process, institutions intended to protect the political rights of the
population, such as the judiciary, were weakened or simply bypassed. The
loosening of the state’s monopoly over the use of force was a second, per-
haps more significant, consequence of KANU’s desperate attempt to retain
power. Although allowing Moi to stay in office through the 1990s, these
same developments made any attempt by Kibaki to retain power after his
election victory in 2002 more difficult.
That is not to claim that Kibaki could only vainly struggle against the
legacies of Moi’s misrule. Instead, decisions taken following the NARC
electoral victory served only to intensify the looming political storm. In
particular, the refusal of Kibaki to sanction any genuine process of con-
stitutional reform was of great significance. The deliberate sabotage of the
constitutional review process destroyed a renewed bout of elite consensus
that had emerged in the second half of 2002. Then, with few exceptions,
the key regional powerbrokers coalesced around agreement on the need to
remove KANU from office. Shortly after achieving that aim, the failure to
create the post of Prime Minister in accordance with pre-election agree-
ments with Odinga led directly to the split of the NARC government and
the formation of ODM.55 In other words, the best opportunity to reverse
the trend of elite fragmentation was jettisoned by an executive fearful of
decentralizing power.
Moreover, the collapse of the constitutional review process left Kenya’s
existing constitution in place, and hence provided Kibaki with executive
control over coercive institutions. He made good use of that power. In the
months leading up to the polls, Kibaki refused to ‘gazette’ (and hence ac-
tivate) legislation covering the behaviour of political parties, demonstrating
the impotence of Parliament in the face of the President’s veto. Kibaki also
took the opportunity to appoint 19 of the 22 electoral commissioners, violat-
ing the spirit of an Inter-Party Parliamentary Group agreement and eroding
opposition confidence in the electoral process.56 Just two days before the
elections, the President appointed five new High Court judges to an already
partisan bench.57 The appointments, combined with the fact that the gov-
ernment had been able to delay previous electoral petitions for an entire
parliamentary term, made it clear that the judiciary could not be relied
upon to act as an independent arbiter. Opposition leaders and supporters
unsurprisingly lacked faith in key democratic institutions in the run-up to
the 2007 polls.
55. Godwin Murunga and Shadrack Nasong’o, ‘Bent on self-destruction: the Kibaki regime
in Kenya’, Journal of Contemporary African Studies 24, 1 (2006), pp. 1–28.
56. International Crisis Group, ‘Kenya in crisis’.
57. Jillo Kadida, ‘Kibaki appoints six judges’, The Nation, 25 December 2007,
<http://allafrica.com/stories/200712270383.html>(15 October 2008).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

18 AFFRICAN AFFAIRS
Once on the campaign trail, ODM lost no time in announcing its determi-
nation to introduce regionalist constitutional arrangement and to investigate
cases of grand corruption if elected. ODM thus appeared to threaten the
status quo for both the Kibaki regime and the latter’s predominant support
base from the Mount Kenya region, which would lose most in any attempt
to redistribute wealth and influence to other areas of the country.58 The
refusal of the opposition to play by the established rules of the game, itself
a direct consequence of the process of elite fragmentation discussed above,
has not been given sufficient emphasis in discussions of the Kenya crisis.
As chequered as the relationship between Moi and Kibaki had been, the
transfer of power in 2002 was underpinned by a common historical experi-
ence that generated a genuine trust. Moi was able to walk away from power
because he understood that Kibaki would never seek to investigate him, or
his closest supporters, for the corruption perpetrated under his regime.59
Indeed, Moi’s political rehabilitation began almost as soon as his tenure
as President ended. Kibaki was quick to appoint his predecessor as an en-
voy to Sudan during the peace talks there, before the two brokered a pact
that brought much of KANU into the government after the referendum
defeat. The rapprochement between Moi and Kibaki culminated in the two
sharing a stage during PNU campaign rallies in the Rift Valley. The last ves-
tiges of the elite collusion of the 1960s and 1970s underpinned the smooth
transition in 2002, but Kenya was to have no such luck in 2007.
Odinga’s candidature, and the broad anti-Kikuyu alliance incorporated
within ODM, rendered 2007 a ‘high-stakes’ election. On the one hand,
ODM contained some of Kenya’s most controversial political figures. The
presence of individuals such as William Ruto in a party campaigning on the
issue of majimbo (regionalism) naturally inspired fear among those members
of the Kikuyu community residing in non-Kikuyu areas. The focus on re-
gionalism played directly on the set of political grievances that had been so
carefully managed during the one-party era. Understood variously to mean
a form of respectable federalism, a reconsideration of land rights, and a li-
cence to commit ethnic cleansing, regionalism became the key divisive issue
of the campaign.60 In the run-up to the election, Kikuyus were significantly
more likely to report a fear of election-related violence than any other ethnic
group.61
58. Cheeseman, ‘The Kenyan election of 2007’.
59. Wolf, ‘Immunity or accountability?’
60. For the roots of this see David Anderson ‘“Yours in the struggle for majimbo.” Nationalism
and the party politics of decolonization in Kenya, 1955–64’, Journal of Contemporary Histor y
40, 1 (2005), pp. 547–64; John Lonsdale, ‘Soil, work, civilisation and citizenship in Kenya’,
Journal of Eastern African Studies 2, 2 (2008), pp. 205–314.
61. Stefan Dercon and Tessa Bold, ‘Ethnicity and the elections’ (paper presented at ‘Kenya
Elections 2007’ workshop, Somerville College, University of Oxford, 17 January 2008).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 19
On the other hand, Odinga was a different generation of political leader,
and senior PNU figures doubted that he could be trusted. PNU propaganda
ensured that this fear was translated to the grassroots, comparing Odinga to
Idi Amin, Stalin, and Hitler.62 This perception was reinforced by Odinga’s
willingness to reveal his personal role in the 1982 coup plot and to openly
talk about a plan to bring John Githongo back as anti-corruption Tsar.63
The mention of Githongo, who had resigned as Anti-Corruption Minis-
ter under NARC citing political interference in his work, was particularly
significant. It suggested that ODM might use Githongo’s knowledge of the
Anglo-Leasing scandal to prosecute senior PNU figures. By the end of 2007,
elite fragmentation and the privatization of violence had eroded trust at all
levels of the political system, a context in which Kibaki’s decision to retain
power at any cost becomes explicable. That decision was underpinned by
the assumption that the coercive apparatus of the state was sufficient to cope
with protests in the aftermath. The PNU leadership knew exactly what to ex-
pect once the ECK announced the final results. In acting on this assumption
Kibaki was simply following in the footsteps of Kenyatta and Moi in trusting
in the bureaucratic-executive state to protect the holder of power against all-
comers. However, Kibaki’s decision proved to have even higher costs than
the PNU leadership expected. The processes of elite fragmentation and
state informalization had undermined the capacity of the centre to hold.
Most significantly, the decentralization of control over violence meant
that the wave of attacks triggered by the declaration was broader and harder
to manage than the government had envisaged. In the Rift Valley, militia
leaders who had learnt their trade in the ethnic clashes of the 1990s came
to the fore, chasing Kikuyu from their homes and attacking those suspected
of supporting the ruling party, including the Moi family. The extent to
which the violence was planned, and how much of it occurred under the
direction of political leaders, will never be known. Nonetheless, the speed
and coordinated nature of the early attacks strongly suggests that it was not
spontaneous. Many senior ODM figures failed to do all they could to rein in
the violence. Najib Balala, one of ODM’s coterie of regional powerbrokers
and now Minister of Tourism, told the authors that the response of ODM
supporters was an inevitable reaction to the blatant injustices perpetrated
by the Kibaki regime.64 However, it would be misleading to suggest that
62. Flyers given out at PNU’s final rally at Uhuru Park depicted Odinga as Idi Amin. For
more on the images used in the campaign see Justin Willis, ‘What has Kibaki got up his sleeve?
Advertising the Kenyan presidential candidates in 2007’, Journal of Eastern Afr ican Studies 2,
2 (2008), pp. 264–71.
63. East African Standard, ‘What else does Githongo know?’, 17 December 2007, <http://
allafrica.com/stories/200712170561.html>(16 October 2008).
64. Conversation with the authors at Panafric Hotel, Nairobi, 31 December 2007, with
Charity Ngilu and Musalia Mudavadi in attendance. Both William Ruto and Raila Odinga
denied some of the worst atrocities in interviews with the BBC World Service.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

20 AFFRICAN AFFAIRS
the incriminating links to organized violence were the sole preserve of the
opposition: the government responded to the challenge to its authority by
unleashing Mungiki – or at least armed gangs of Kikuyu youth prepared
to march under this banner – to combat pro-ODM actors in Kibera and,
later, in the Rift Valley. The diffusion of violence, along with the willingness
of rival politicians to give arms and leadership to the mob, contributed
direction to the breakdown of political order.
The extent to which the process of elite fragmentation also constrained
the response of the state has escaped comment in many accounts of the
crisis. That the armed forces were reluctant to become embroiled in the
suppression of opposition protests should not be considered too significant.
After all, the army and air force had been entirely absent from the domes-
tic political sphere since the coup of 1982. However, the unwillingness of
police to confront protesters in their own home areas was of far greater
importance. Such an outcome would have been unthinkable in the heyday
of the one-party state, when the strength of the elite alliance ensured that
the state’s institutions pulled in the same direction. The actual physical ca-
pacity of the police had not fallen by 2007. If anything, police officers were
more numerous and better armed than at any time in Kenya’s post-colonial
history. What had changed was the ability of the executive to trust that this
capacity would be exercised in its favour, and the balance of power between
the state and militia groups. Ultimately, the PNU was forced to rely on
Moi’s favoured instrument of control, the General Service Unit, to hold
Nairobi, while deploying as many police units as possible outside of their
home areas. This proved to be an effective holding strategy, as the protesters
progressively ran out of funds, energy, and options, but the government was
unable to reassert control in the absence of some form of elite compromise.
Political disorder thus occurred in a state that hitherto had been governed
by what Atieno Odhiambo has labelled ‘an ideology of order.’65
The inability to find an elite compromise until the arrival of international
mediators and the creation of the government of national unity in March
2008 reflected a widespread lack of trust in the ability of key democratic
institutions such as courts and electoral commissions to deliver fair political
outcomes. That lack of faith suggests that the longevity of the Kenya crisis
was rooted in the prior failure to carry out a genuine process of consti-
tutional review. The stalled process of democratization in Kenya neither
decentralized power nor created any truly independent democratic institu-
tions. The absence of viable institutions through which conflict could be re-
solved encouraged activists and leaders to take their protests to the streets.
From the very beginnings of the crisis, Odinga publicly stated that the
65. E. S. Atieno Odhiambo, ‘Democracy and the ideology of order in Kenya’ in Michael
Schatzberg (ed.), The Political Economy of Kenya (Praeger, New York, 1987), pp. 177–201.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 21
opposition would not submit themselves to the justice of ‘Kibaki’s courts’.66
For the government’s part, Martha Karua demonstrated a similar lack of
faith in the Kofi Annan-led mediation process. Karua, the most vocal sup-
porter of Kibaki during the crisis, accused the former Secretary-General
of the United Nations of attempting to lead ‘a civilian coup’ on behalf of
ODM.67 Although the government of national unity that emerged from
those talks may indicate a rebuilding of trust between elite political actors,
to date the trust between its main figures has yet to be rigorously tested.
Democracy, disorder and the sequencing debate
Political liberalization has been perhaps the least discussed ingredient of
the Kenya crisis. Few of the already numerous studies of the violence have
asked the uncomfortable question of whether or not the scenes witnessed
earlier this year were a product of the reintroduction of multi-party elec-
tions. Of course, multi-party elections do not cause violence in any mean-
ingful sense, but they can create incentives for leaders to adopt increasingly
antagonistic strategies. Moreover, when such elections are held within a
first-past-the-post electoral system, they may encourage a ‘winner-takes-all’
struggle for control of the state. Both tendencies pose real challenges for
political stability in countries where a sense of the greater good has not
yet subsumed ethnic or regional loyalties, and where the coercive strength
of social forces is greater than that of the state itself. The dangers these
challenges pose to the success of the democratic project, and to the very
integrity of the state, have given rise to a heated debate over the significance
of democratic sequencing. Sceptics such as Mansfield and Snyder contend
that opening up states which lack the effective rule of law, agreement on the
identity of ‘the people’, and a monopoly over the legitimate use of force,
to greater levels of political competition invites disaster.68 Because such
political systems lack both inter-communal trust (‘social capital’) and have
limited administrative control (‘rational-legal’ legitimacy), they are unlikely
to be able to manage effectively the additional social tension that multi-party
elections necessarily bring.69
Attempts to isolate the key factors underpinning the success of demo-
cratic transitions in Africa using aggregate indicators support this broad
66. Maseme Machuka, ‘Seek court action, PNU leaders tell Raila’, The Standard, 3 January
2008, <http://allafrica.com/stories/200801021298.html>(15 October 2008).
67. Roger Cohen, ‘How Kofi Annan rescued Kenya’, New York Review of Books 55, 13 (2008),
<http://www.nybooks.com/articles/21719>(2 October 2008).
68. Edward Mansfield and Jack Snyder, ‘The sequencing “fallacy”‘, Journal of Democracy 18,
3 (2007), pp. 5–10.
69. Max Weber, Politics as a Vocation (Fortress Press, Philadelphia, PA, 1965); Robert
Putnam, Making Democracy Work (Princeton University Press, Princeton, NJ, 1993).
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

22 AFFRICAN AFFAIRS
conclusion. Although Bratton and Chang do not find that state infrastruc-
ture and the delivery of welfare services have a significant impact on process
of democratization, and caution against adopting an overly state-centred un-
derstanding of the democratization process, their analysis shows that ‘what
matters most for democratization is whether the state has the capacity to
fulfil its prime function: creating a legitimate political order or, differently
stated, a rule of law’.70 However, even among those that accept Rose and
Shin’s argument that third-wave democracies are hampered by attempt-
ing ‘democratization backwards’,71 it is unclear what the correct advice for
policy makers should be. Fukuyama and McFaul have argued in defence
of democracy promotion, suggesting that ‘Western democracy promoters’
should take steps to safeguard fragile democracies, recommending that they
‘develop international institutions that enhance mutually beneficial cooper-
ation’ and ‘draft a code of conduct for democratic interventions’ in order
to consolidate the ‘right to free and fair elections’.72 At the other end of
the continuum, sceptics like Amy Chua go as far as to suggest that elec-
tions should be postponed until a suitable socio-institutional context can be
developed.73
It is easy to depict the Kenya crisis as another piece of supporting evi-
dence for the necessity of democratic sequencing of some form. The need
to fund election campaigns has clearly exacerbated the extent of corruption
and further entrenched the patron–client relationships that connect Kenyan
politicians to local clients. At the same time, the institutionalization of com-
petition among the political elite has been instrumental in connecting militia
groups to political leaders, through their use as private armies during elec-
tion campaigns. In turn, the increasing interplay between formal politics
and gang warfare has accelerated the diffusion of violence. Finally, elec-
tions themselves have come to be moments of severe national stress, around
which social and political relations may be thrown into a state of flux. The
widely reported refusal of Luo tenants to pay rents to Kikuyu landlords in
the run-up to the 2008 polls because ‘soon we, and not you, will be in State
House’, is a classic illustration of this phenomenon.74 Clearly, focusing the
nation’s hopes and fears on 48 hours of balloting has the potential to cause
social and political ruptures which the state may be unable to manage.
70. Michael Bratton and Eric Chang, ‘State building and democratization in sub-Saharan
Africa: forwards, backwards, or together?’, Comparative Political Studies 39, 1059 (2006), p.
1080. Emphasis in original.
71. Richard Rose and Doh Chull Shin, ‘Democratization backwards: the third-wave democ-
racies’, British Journal of Political Science 31, 2 (2001), pp. 331–54.
72. Francis Fukuyama and Michael McFaul, ‘Should democracy be promoted or demoted?’,
Washington Quarterly 31, 1 (2007/8), pp. 43–4.
73. Amy Chua, World on Fire: How exporting free market democracy breeds ethnic hatred and
global instability (Knopf, New York, NY, 2004).
74. For more on the election in Kibera, see Michelle Osborn, ‘Fuelling the flames: rumour
and politics in Kibera’, Journal of Eastern African Studies 2, 2 (2008), pp. 315–27.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 23
Most significantly, the events of late 2007 and early 2008 closely resem-
bled the nightmare scenario outlined by one of sequencing’s best-known
supporters, Amy Chua. In her Wo r l d o n F i r e , Chua identifies the presence of
‘market-dominant minorities’ as being the critical variable that pulls democ-
ratizing states into the maelstrom of ethnic violence. According to Chua,
democratization results in violence as a result of two factors, sometimes
working in tandem: first, a market-dominant minority resorts to extra-legal
action to protect its privileged position; or, second, a marginalized majority
resorts to violence in order to dislodge the minority from its entrenched
position. Kenya’s dominant Kikuyu and the other Mount Kenya peoples
fit Chua’s taxonomy of a market-dominant minority. In an effort to protect
the economic interests of a Kikuyu elite threatened by the prospect of a
Kibaki defeat, the state and its non-state partners undertook ‘a backlash
against democracy by forces favorable to the market-dominant minority’.
Furthermore, the actions of ODM and its supporters could be understood
as falling into the category of ‘majority-supported ethnic violence’ aimed at
members of the Kikuyu market-dominant minority.75 We therefore recog-
nize the potential relevance of some of Chua’s observations with reference
to Kenya, albeit a relevance that she herself mitigates against by her iden-
tification of Indians as the main market-dominant minority in East Africa.
We do not, however, agree that this evidence builds a case for a roll-back of
international advocacy for democratization.76 Rather, we suggest that the
forms of violence and instability that Chua observes are more likely to be
mitigated by a significant element of constitutional reform being built into
the process of democratization in its infancy. Moreover, Chua’s critics ar-
gue that constitutional reform is most likely to be stable and effective once
a process of democratization is under way.
Although the potential for democratization to produce unintended nega-
tive side-effects is high, it is also clear that sequencing is unfeasible in many
cases. For one thing, most African nations are in the process of holding
elections, and the international community can hardly advocate a reversal
of the liberalization programme. For another, it is a fallacy that benevolent
authoritarian regimes are more likely to create the necessary preconditions
for a successful transition.77 In the universe of authoritarian states, the
successful ‘developmental’ states of Asia are the exception, not the norm.
By contrast, non-democratic rule in Africa and Latin America has more
typically been brutal, exploitative, and arbitrary. As Kohli has argued, the
conditions needed for the Taiwanese or Korean model to be successful –
bureaucratic autonomy, broad economic equality and a pressing security
75. Chua, World on Fire, pp.123–5.
76. Ibid., pp. 259–64, 273–8.
77. Ibid.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

24 AFFRICAN AFFAIRS
threat – were not present in Africa in the 1970s.78 Nor are they now. It is
unlikely that African states will be blessed with significantly better leaders
or more favourable conditions in the near future. Democratization before
reform will therefore remain a risky action, but a necessary one.
Furthermore, Thomas Carothers correctly asserts that for many coun-
tries, including Kenya, there is simply no option but to attempt the tran-
sition to democracy and the transition to effective rule of law at the same
time.79 On the one hand, authoritarian leaders rarely create the necessary
conditions for the successful transition to democracy of their own accord.
On the other, the majority of African states have already introduced multi-
party elections, and it is seems both unfeasible and heartless to suggest
they should abandon them. As a result, the key challenge becomes one of
creating institutional mechanisms that can reduce the potential for inter-
group conflict as the process of democratization unfolds, a challenge which
necessitates far-reaching constitutional review.
Fortunately, Carothers is also right that in many cases simultaneous de-
mocratization and reform have proved possible. In Ghana, where political
liberalization began with a series of flawed elections as in Kenya, slow re-
form and repeated plays of the electoral game have supported a remarkable
process of democratic consolidation.80 Ghana is not an isolated case, with
Senegal, Botswana and Mauritius all able to point to substantial gains made
through simultaneous processes of democratization and institutional reform
far in excess of the authoritarian regimes in the region. Indeed, there are
good reasons to think that a process of democratization can confer legit-
imacy on a state, and in doing so may aid the process of building state
capacity. As Bratton and Chang conclude, ‘the state is unlikely to provide a
durable order unless it is legitimated by democracy. Thus, the best way for-
ward, for scholars and reformers alike, is to reconnect the study of Africa’s
states and regimes, to acknowledge the interaction of state structures and
democratic procedures, and to promote state building and democratization
together.’81
For all its shortcomings, more progress was made in terms of the freedom
of the media, the independence of the private sector from political control,
and the capacity of the state to collect tax revenue in the five years of the
NARC government than in the previous two decades under Moi. How-
ever, Kibaki failed to address the most pressing priority for a democratizing
multi-ethnic nation: institutional safety-nets to counteract the potentially
78. Atul Kohli, State Directed Development: Political power and industrialization in the global
periphery (Cambridge University Press, Cambridge, 2005).
79. Carothers, ‘How democracies emerge’.
80. Godwin Murunga and Shadrack Nasong’o (eds), Kenya: The struggle for democracy (Zed
Books, London, 2007).
81. Bratton and Chang, ‘State building’, p. 1081.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

DEMOCRATIZATION, SEQUENCING: LESSONS FROM KENYA 25
damaging impact of inter-communal tension and elite fragmentation. In
the 1970s the elite alliance made such institutions unnecessary. In 2008
these institutions were essential but they did not exist, having been utterly
subverted by Kibaki’s machinations in the build-up to the elections. Simi-
lar institutional backdrops currently hold in countries as diverse as Zambia,
Malawi, Uganda, the Democratic Republic of the Congo, Nigeria, Rwanda,
Burundi, the Central African Republic, and Mozambique. Given this, the
lack of attention given by international actors to constitutional change is
remarkable. Donors may have limited purchase in some of these countries,
but even when they do push for reform, the judiciary and the legislature
have been well down on the list of priorities.82 Of course, strengthening
such institutions would only create the potential for deadlock to be broken;
successful conflict resolution also requires political will that is often in short
supply. Nevertheless, even a small possibility of avoiding the breakdown of
political order is better than none.
In the Kenyan context, divorcing appointments to the electoral commis-
sion and the judiciary from executive control would give the opposition
greater faith in these institutions. On the one hand, the complete collapse
of the ECK would not have occurred if it had not been packed with govern-
ment stooges. On the other, evidence of independent action by the courts
might have persuaded opposition leaders to pursue legal redress rather than
street protests. The removal of the President’s ability to delay indefinitely
legislation passed by Parliament would also render the prospects of tak-
ing up parliamentary seats while continuing to challenge the outcome of
the presidential contest more attractive; especially in cases such as Kenya
where the opposition actually secured a parliamentary majority.
Of course, a more radical and far-reaching set of constitutional revisions
to finally dismantle the bureaucratic-executive state is desirable, if signifi-
cantly less realistic. The impressive stability of Mali demonstrates that gen-
uine and carefully considered decentralization can help to co-opt disgrun-
tled minorities into the system and to reduce the intensity of competition
for control of the political centre.83 Successful democratic consolidation in
Mauritius shows that electoral systems designed to encourage all parties to
field candidates of all ethnicities – in that case the use of two-member elec-
toral constituencies that encourage parties to run parliamentary candidates
from minority and majority ethnic groups simultaneously – can help to pre-
vent the emergence of mono-ethnic parties, even in a diverse and divided
82. Stephen Brown, ‘From demiurge to midwife: changing donor roles in Kenya’s democrati-
sation process’ in Godwin Murunga and Shadrack Nasong’o (eds), Kenya: The struggle for
democracy (Zed Books, London, 2007), pp. 301–29.
83. Richard Vengroff, ‘Governance and the transition to democracy: political parties and the
party system in Mali’, Journal of Moder n African Studies 31, 4 (1993), pp. 541–62; Jennifer
Seely, ‘A political analysis of decentralization’, Journal of Modern African Studies 39, 3 (2001),
pp. 499–524.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from

26 AFFRICAN AFFAIRS
society.84 Remarkable economic growth in Botswana also reveals the posi-
tive impact that a professional and comparatively meritocratic bureaucracy
can have on state building and economic growth by insulating economic
decision making from political whims.85 Institutional reform in all three
cases was particularly successful because it was underpinned by a support-
ive political culture of tolerance and respect for the greater good which has
been noticeably absent in Kenya. However, political will notwithstanding,
there is nothing to suggest that such institutional designs would not have
similar effects if implemented in Kenya.
The most significant lesson of the Kenya crisis is that it reveals how frag-
ile Africa’s new multi-party systems may be when political liberalization,
elite fragmentation, state informalization, and historical grievances come
together in the absence of effective institutional safeguards, forming what
Mwangi and Holmquist have called a ‘perfect storm’.86 At present this les-
son is not being heeded. As a result, countries such as Nigeria, Uganda, and
Kenya are repeatedly walking the tightrope of intensely contested elections
in the absence of an effective institutional safety-net. Without certain pre-
requisites such as basic state capacity, the effective rule of law, and an agreed
national identity the reintroduction of multi-partyism may exacerbate un-
derlying tensions which the state is powerless to manage.87 Given such a
context, effective institutional reform is essential to prevent the transition
to multi-partyism being self-defeating and forever incomplete.
84. Larry Bowman, ‘Mauritius: Democracy and development in the Indian Ocean’
(Westview Press, Boulder, CO, 1991); Deborah Br¨
autigam, ‘Institutions, economic reform,
and democratic consolidation in Mauritius’, Comparative Politics 30, 1 (1997), pp. 45–62.
85. Patrick Molutsi and John Holm, ‘Developing democracy when civil society is weak: the
case of Botswana’, African Affairs 89, 356 (1990), pp. 323–40; James Leith, Why Botswana
Prospered (McGill, Montreal, 2005).
86. 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.
87. Mansfield and Snyder, ‘The sequencing “fallacy”’.
at Bodleian Library on September 8, 2014http://afraf.oxfordjournals.org/Downloaded from