territorial power sharing and the regulation of conflict in africa
TRANSCRIPT
Civil Wars, 2013
Vol. 15, No. S1, 123–143, http://dx.doi.org/10.1080/13698249.2013.850891
Territorial Power Sharing and the Regulation ofConflict in Africa
CHRISTOF HARTMANN
The article argues that there is a relative lack of territorial power sharing in sub-Saharan Africa, but that this should not surprise us given the socio-ethnic context of most African countries which lack both majoritarian ethnic groups and clear culturally based distinctions between contending groups. A comprehensive analysis of the various federal systems and of some decentralising countries reveals a sobering record of conflict management and discusses some explanatory variables. In this context, forms of power sharing that include provisions on territorial autonomy, amounting to full-fledged federalism or restricted to some degree of decentralisation, should be regarded with caution. Regional autonomy as a solution to civil war will not work in most cases.
INTRODUCTION
Among students of politics in deeply divided societies, ‘territorial approaches’ have
always been considered as a major institutional strategy of conflict regulation.
Territorial strategies rely on the assumption that conflicts can be managed by
dividing power between central and regional elites. The possibility of self-rule over
political, economic or cultural issues within a part of the national territory should
thus minimise the potentially conflictive sharing of rule at the central level.1 At the
same time, the institutional features of such territorial conflict regulation strategies
vary enormously with regard to the degree of autonomy that one or all sub-national
units enjoy as well concerning the co-existing elements of shared rule built into the
institutional mechanism such as representation and decision-making rights for sub-
national units at the central level.2
Historically, institutional features of territorial autonomy and decentralisation
were introduced in many polities because previously independent states decided to
join in a larger polity while keeping own institutions and identity, or because a
decentralisation of the state structures was considered to increase the democratic
quality and/or the economic and developmental efficiency of the political system.
The capacity of territorial autonomy or self-governance to influence the dynamics of
existing or potential violent conflict did generally not feature among the main
concerns of many constitution-makers.
q 2013 Taylor & Francis
124 CIVIL WARS
Many violent conflicts, especially in non-western contexts, centre on the role of
the state in society and are directly related to its structure and organisation. It has
been argued that some of these conflicts could be prevented or mediated by a
conscious design of sub-national political institutions such as federalism, autonomy
or other types of decentralised governance. This should be particularly so for sub-
Saharan Africa with its majority of ethnically heterogeneous states and its many
protracted violent conflicts. The decision of Ethiopia to create the first ethno-federal
system on the continent in the early 1990s has attracted a lot of interest, although the
assessment on its management of conflicts is largely negative. The broader often
donor-driven trends towards more decentralisation in most African continents which
was initiated for developmental objectives might also have increased political
stability and reduced the likelihood of violent conflict.
The apparent advantages of territorial autonomy might explain why such
provisions also feature prominently in peace agreements after civil wars. In this
context, power-sharing formulas seem to have become the dominant mode of ending
violent conflict worldwide, but certainly so in sub-Saharan Africa.3 The empirical
verification of this trend is certainly rendered difficult by contrasting
operationalisations of power sharing itself. Following Hoddie and Hartzell, one
might distinguish between different dimensions of power sharing, i.e. the type of
powers to be shared: power sharing might thus refer to political power at the central
level, to economic or to military power, and finally to territorial issues (federalism/
decentralisation).4 Others have distinguished between horizontal and vertical power
sharing with the latter referring to the sharing of (political or economic) power
among multiple layers of government5 or between mandates (relatively hard
guarantees) and opportunities (soft guarantees).6
This debate has therefore slightly modified the contours of the concept of power
sharing.7 It has a rather wide understanding of power sharing which might relate to
any institution dividing or sharing any degree of political, economic, territorial and
military power. Hartzell and Hoddie stress that the likelihood of maintaining peace
depends on the number of dimensions that are included in the power-sharing
agreement.8 In his analysis of African power-sharing agreements, Mehler also
follows this logic by including in his analysis all cases where at least one type of
power is included (in Africa this is nearly always the military one).9 Political power
sharing within this research perspective covers very different situations, ranging
from roughly proportionate appointment of leaders of warring groups within
government (as in Chad) to a new constitution based on an explicit pact among all
main political-military actors (as in South Africa). From a conflict management
perspective, details and differences of political or territorial power sharing are less
important than the basic question of ‘who shares power with whom and who can still
be excluded from the sharing arrangement?’.10
Most debates on power sharing after violent conflict have so far put little
emphasis on the role of territorial self-governance or vertical power sharing as one
aspect of solving conflict and institutionalising democratic rule. Hartzell and Hoddie
argue that ‘recognising that the government is typically the most powerful
125 TERRITORIAL POWER SHARING IN AFRICA
organisation within a country, territorial power-sharing institutions call for the
devolution of powers to regionally concentrated groups in the hopes that this will
enhance their sense of security’.11 They define territorial power sharing as including
both federalism and regional autonomy, where ‘only one or a few country’s regions
are likely to have special powers devolved to them’.12
On the basis of their statistical test, they state that the inclusion of provisions for
establishing territorial power-sharing institutions ‘decrease the likelihood of
settlement failure by 99 per cent’. These provisions ‘have the demonstrated capacity
to set the stage for the period of transition by enhancing a sense of confidence among
former enemies that their interests will not be jeopardised in the context of the post
war state’.13 Lake and Rothchild, however, failed to find any evidence of successful
institutionalisation of territorial provisions in any post-war constitutional order,
although it has to be acknowledged that Hoddie and Hartzell were concerned with
the immediate post-war peace and not with long-term benefits in terms of stabilising
a polity.14 The recent Uppsala Peace Agreement Data Set proposes a typology with
six different types of ‘territorial provisions’ as included in peace agreements:
autonomy, federalism, local power sharing, regional development, cultural freedoms
and local government.15
Looking at all African power-sharing agreements for the period since 1990, we
see that the majority of these agreements do not contain any provisions with regard
to territorial power sharing. Such provisions are absent in Central African Republic
(2007), Chad (2003, 2006, 2007), Congo (1999), Democratic Republic of Congo
(2002, 2009), Cote d’Ivoire (2003, 2007), Guinea-Bissau (1999), Liberia (1993,
1996, 2003), Rwanda (1993), Sierra Leone (1996, 1999, 2000) and Somalia (1994,
1997, 2004, 2008). Table 1 shows those that have been classified as containing
territorial power sharing.
While we have got a better access to data over the last years, we still lack a more
systematic and comprehensive research about these specific institutional features of
peace agreements and their effects on violent conflict. Over the last two decades,
TABLE 1
TERRITORIAL POWER SHARING AFTER VIOLENT CONFLICTS IN SUB-SAHARAN AFRICA
1990 – 2011
Peace agreements Type of territorial power sharing
Mali 1992, Mozambique 1992, Niger 1993-4, Various types of decentralisation reforms Angola 1994, Djibouti 2001, Burundi 2003, Senegal 2004, Angola 2006
South Africa 1993, Comoros 2003, Sudan Different types of federal state reforms 2005 with constitutional rights for sub-national units
Sources: Own compilation based on information from the database on power-sharing agreements of Hartzell and Hoddie (data for the phase until 1999), as complemented by Andreas Mehler, Not Always in the People’s Interest: Power-Sharing Arrangements (Hamburg, German Institute of Global and Area Studies 2008) for the African continent, and the Uppsala Data Bank on Peace Agreements (covering the period until 2011, as updated by Stina Hogbladh). Coding of peace agreements differs among these sources, though.
126 CIVIL WARS
there has been increasing research on the supposed theoretical nexus and the causal
mechanisms linking decentralisation with conflict16 and I will discuss some of this
research in more detail further. Students of federalism have also analysed whether it
matters for conflict management, but there are few federal systems worldwide and
comparative empirical analysis has been additionally hampered by the problematic
operationalisation of both decentralisation and federalism and of (violent) conflict.
The academic debate on territorial self-governance and conflict in sub-Saharan
Africa has been interested in the specific issue of secession,17 or mostly restricted to
the assessment of specific cases, especially South Sudan and the federal systems of
Nigeria, Ethiopia and South Africa (although there are some comparative analyses
linking Nigeria to India, for example).18 A few studies dealing with Africa are
noteworthy for their scepticism concerning federalism and territorial autonomy as
an adequate strategy of conflict management.19
The article will thus engage with these assumptions. The core argument is that the
social and political context in most African countries is not conducive for territorial
strategies to have an impact on conflict. For territorial power sharing to be relevant,
some pre-conditions have to be met and these pre-conditions are mostly absent in sub-
Saharan Africa. We will start by exploring the theoretical nexus between different
forms of territorial power sharing (federalism anddevolution) and conflict, highlighting
some critical context variables for territorial power sharing to be relevant.Wewill then
review Africa’s experience with federalism and decentralisation, including the
territorial power-sharing provisions of the recent peace agreements. The empirical
analysis reveals that there is indeed a relative lack of territorial power sharing in sub-
Saharan Africa, and that these few cases represent rather exceptional trajectories of
ethno-political contexts on the continent. I will finally argue that the track record of
territorial power sharing is not without success but not for the reasons advanced in the
literature, and largely because self-rule is complemented by some degree of power
sharing at the national level.
DECENTRALISATION, FEDERALISM AND CONFLICT
The decentralisation of central power and resources has been a constant element of
state reforms both in industrialised and developing countries. Over time, the
meaning of decentralisation has become rather blurred. It has been applied both to
the vertical reorganisation of the state apparatus and to the horizontal delegation of
state powers to non-state actors. Decentralisation was also discovered as a theme in
development policy discourse, and international donors urged many developing
countries to adopt decentralisation in order to enhance the effectiveness and
efficiency of their administration and the quality of services. External pressures to
decentralise led, not unsurprisingly, to rather superficial reforms of decentralisation
in many countries, especially in Africa.
Since the early 1980s and Rondinelli’s influential contribution, the conventional
distinction is between (administrative) deconcentration as a form of decentralisation
which mainly concerns the reorganisation of responsibilities within the government
127 TERRITORIAL POWER SHARING IN AFRICA
apparatus and (political) devolution as the type of decentralisation which shifts
power, resources and policymaking to sub-national levels of government with
elected councils thereby giving local citizens a say in decision-making.20 The
political practice of decentralisation policies in many developing countries tended to
blur Rondinelli’s basic distinction, because many sub-national governments were
more than implementing agencies and had some minimal degree of decision-making
while at the same time were lacking guaranteed fiscal transfers or significant own
revenue generating capacity. For decentralisation to be considered a form of power
sharing, some real power has obviously to be devolved to sub-national units.
A federation as a specific type of political devolution presupposes a division of
sovereignty, a constitutionally protected status for the devolved units (usually states
in their own right with parliaments, governments and a distinct political process
whose rights can only be revoked by constitutional amendment), a constitutional
procedure to solve conflicts between the central and the decentralised levels and an
organ at central level where the decentralised units have a veto-power over national
policy-decisions affecting their rights. A defining feature of federalism is thus that
self-rule and shared rule are constitutionally enshrined.21 A federation is certainly a
form of vertical power sharing as the representatives of the sub-national units do
influence policy decisions, but federalism itself says nothing about how much power
each level enjoys.
The provinces within a decentralised unitary state, may, however, have more
effective power than the member-states of a federation. This is one of the reasons
why recent research on federalism has proposed to replace the dichotomy between
federalism and unitary states with a more differentiated typology of variables such
as the degree of fiscal decentralisation.22 Attempts to actually compare types and
degrees of vertical power sharing have not yet produced any generally accepted
measurement or universally applicable criterion. Fiscal approaches to decentralisa
tion use the share of sub-national government expenditure and revenue as main
indicator of real decentralisation.23 Many political scientists insist, on the contrary,
on the categorical distinction between federalism and other forms of decentralisa
tion.24
There are other aspects of institutional design beyond the degree of autonomy
which also matter. There are many details to be considered in the constitutional
division of powers, the electoral system, and the modalities of representation in
central institutions, or how many subunits the federation has.25 The general
observation seems to be that federalism tends to be more stable with multiple
constitutional units rather than with two or three.26 Autonomy can be granted as a
specific representation right to a selected territory or the degree of sub-national
competencies is identical across the country. Federalism can accordingly be
symmetrical or asymmetrical.
The question whether regional political structures of self-rule are likely to
encourage or impede violent conflict or lead to secessionism has sparked a lively
debate over the last decade.27 The debate is mainly concerned with the concept of
ethno-federalism or multi-nation federalism, a type which was introduced in
128 CIVIL WARS
industrialised countries such as Belgium, Canada or Switzerland (but also India),
and which consciously grants federal (or quasi-federal) forms of territorial
autonomy to at least one sub-state national or ethnic group. In the ‘non-ethnic’ or
‘national’ federal systems, such as the USA, the Latin American federations or
Germany, sub-state units were not designed to enable ethno-national groups to
exercise self-government over a traditional territory.28 The debate is based on the
assumption that the conflicts to be managed are of an ethnic or linguistic nature, and
are rooted in a desire for increased autonomy from the central state, and in the belief
that the groups’ survival is threatened by the actions or inactions of the central state.
Scholars are interested in the capacity of federalism to avoid a civil war between
central state elites and ethno-regional elites. Interestingly, other types of territorial
power sharing such as political devolution are hardly considered in this debate.29 In a
similar way, the likelihood of federalism or territorial self-governance to influence
other types of conflict is rarely addressed.30
A number of scholars have considered territorial autonomy and more specifically
multi-nation federal arrangements as useful strategies for managing violent
intrastate conflict ‘by increasing political participation and allowing minorities a
degree of political autonomy sufficient to persuade them of the merits of continued
engagement in the existing state structures’.31 Bermeo stressed the ‘peace
preserving’ function of federalism.32 Others have argued that ethno-federalism helps
preserving ethnically divided states by satisfying demands for autonomy on key
issues, localising potential conflicts, promoting unifying identities and reducing
opportunities for the central government to exploit minority regions.33
According to the critics of ethno-federalism, the territorial recognition of
minorities perpetuates and strengthens the differences between groups and provides
minority nationalists with the institutional tools for eventual secession. ‘Group
recognition ensures the perpetuation of the differences and provides minority elites
with a vested interest in the continuation of the divided system’.34 Ethno-federalism
creates strong incentives for elites to mobilise mass support around ethnic themes
and would encourage secessionist movements by providing institutional bases and
political legitimacy for alternative national projects.35 For Roeder, ‘particularly after
civil wars, ethnofederalism and autonomy arrangements represent imprudent
institutional choices’.36
While the debate was apparently stirred by the (temporary) enthusiasm among
policymakers for federal solutions in internationally administered countries such
as Iraq or Afghanistan, many newer contributions rather discuss under which
conditions different types of federalism and territorial autonomy may have positive
effects.
Following an influential contribution by Stepan, it is generally agreed that the
context of formation of federal systems matters.37 Stepan distinguishes between
three types of federations: those that came together voluntarily from distinct polities,
those that are created from unitary states in an attempt to hold the polity together and
those that are put together (or forced together) by a dominant group. While many
theorists of federalism stressed that the willingness to cooperate and accommodate
129 TERRITORIAL POWER SHARING IN AFRICA
must precede the creation of federal institutions,38 the actual context in which
federations emerge might include a civil war settlement or the desperate attempt to
keep a state together. Federations which are not coming together on a voluntary basis
will have a more difficult life in managing internal conflicts.39 Federal systems are
rarely agreed upon in peace times, but rather after ethno-political groups started to use
military force to defend their interests. Horowitz noted in his seminal study on ethnic
conflicts that ‘a more general constitutional change provides an auspicious setting in
which to consider new territorial arrangements to cope with ethnic problems’.40 If the
constitution is negotiated under a democratic dispensation, though, introducing a
federal system will not be easy, because the threat of constitutionally entrenched
autonomy rights will water down the independence of sub-national units or defeat the
option of federalism.
There are also economic and structural background factors. Many authors have
stressed the relevance of the level of economic development and the degree of
economic disparity.41 A prosperous state and society can provide for instruments
of fiscal equalisation and distributive fairness or specific policies designed to
accommodate the interests of marginalised ethnic groups. It might offer economic
benefits for members of an ethnic group outside their ethnic homeland. At the same
time, secession will also be facilitated by strong institutional and societal capacities
at the sub-national level.42
Since Horowitz’ book appeared in 1985, most of the debate has turned around
what kind of ethnic demography might be more conducive to resisting secessionist
pressures in a federal context. According to O’Leary, federalism relies on a
Staatsvolk, a national or ethnic people who are demographically or electorally
dominant,43 because a group with a decisive majority of the federal population has
no reason to fear federation. Hale, on the contrary, perceived a ‘core ethnic region’
as the major threat.44 There is thus a general disagreement whether the creation of
ethnically homogenous sub-national units should be engineered or avoided, with
McGarry and O’Leary arguing that historically mobilised ethno-national groups do
not take kindly efforts to disorganise them through the re-drawing of internal
political boundaries.45 The core puzzle seems to lie in the combination of the relative
size of ethnic groups with their concentration and dispersion in specific territories.
Horowitz had already discussed the different strategies in having heterogeneous and
homogeneous sub-national units and suggested, inter alia, that federations can
be partly designed to prevent ethnic minorities from becoming local provincial
majorities.46
A final critical variable is the regime type. According to Brancati,
decentralisation mitigates conflict in democratic contexts by increasing political
participation, but indirectly increases it by encouraging the growth of ethnic and
regional parties.47 Bermeo claimed that ‘no violent separatist movement has ever
succeeded in a federal democracy’.48 A democratic federalism could block the
systematic transgression of individual and group rights, and prevent minority
secessionist elites from claiming an exaggerated support for their preferences.
130 CIVIL WARS
An autocratic system will, on the contrary, strengthen the hegemony of ethnicised
elites and further increase polarisation.49
FEDERALISM AND CONFLICT IN AFRICA
Although there have been various attempts at creating federations in sub-Saharan
Africa following independence, many were quite short-lived and superficial. Failed
experiments include at the supra-national level the Federation of Mali and Senegal,
of Senegal and Gambia and the East African Federation, and at the national level
Cameroon (1962–72), Zaire (1960–65) and Uganda (1962–66).50
Among the still existing federations, Nigeria has been a federation since
independence in 1960. The four other cases are of a more recent nature, although
both Sudan and Ethiopia had been federations before and Apartheid’s creation of
bantustans had created a strong de facto asymmetrical autonomy regime (Table 2).
Even without taking a closer look at conflict databases, the conflict dynamics in
these cases represent quite different trajectories. We will not further consider the
case of the Comoros for lack of reliable data on the functioning of the federal system
in this small island archipelago.
South Africa
South Africa’s post-apartheid system has the best record in managing ethnic conflict.
Notwithstanding a strong legacy of violence and social unrest, and despite isolated
violent protests at local level and xenophobic attacks againstmigrant communities over
the last years, the political system has been quite effective in building a peaceful order.
South Africa’s quasi-federalismwas the result of protracted constitutional negotiations
in the transition fromApartheid. Vertical power sharing did not result from the wish to
‘come-together’, but from the need to ‘hold together’.51 Federalismwas not introduced
due to its functional advantages (such as effectiveness of policymaking), but as the
outcome of bargaining processes between contending groups who saw federalism as a
second-best solution but could not agree on something else.
It is doubtful that this weak federalism which was instituted in the mid-1990s has
a major responsibility in achieving political stability. Electoral victories of the
opposition in the Western Cape have strengthened accountability and the democratic
quality of the regime. Political stability and inter-ethnic cooperation have mainly
been produced by the formal and informal consociational elements of the political
system. Nearly 20 years after the introduction of the new political dispensation,
there is so little fear of ethno-federalism that the ruling African National Congress (ANC) was recently considering to return to the four old provinces of the Apartheid
times (ostensibly to weaken the Democratic Party’s grip on the Western Cape which
would be diluted in one bigger Cape Province).
Nigeria
In Nigeria, the federal option came with independence and in an attempt by the
British to ‘force together’ a vast and ethnic and religious heterogeneous territory.
131
TABLE 2
AFRIC
A’S
CURRENT FEDERALSYSTEMS
Since
Main structural features
No. of federal units
Nigeria
1960
Product of Independence; ‘holding-together’ federalism w
ithin a religious and ethnically plural country
36a
Tanzania
1964
Asymmetrical federalism w
ith Zanzibar retaining its own government
2
Sudan
1991
Part of post-w
ar power sharing; federalism as a transitory m
echanism
17b
Ethiopia
1992c
Product of Civil W
ar, ‘ethnic federalism’ with secession clause
11
South
Africa
1994
Partofpost-apartheidpower
sharing;strongdevolutionas
barrier
toANC dominance
9
Comoros
1999
Product of separatist tendencies
3
Sources:Own compilation.Three cases(South
Africa,Comoros andSudan)ofpost-conflictverticalpower
sharing werelackingfrom thelistofDavid
A.Lakeand
DonaldRothchild,‘TerritorialDecentralizationandCivilWar
Settlem
ents’,in
PhilipRoeder
andDonaldRothchild(eds) Sustainable
Peace: P
ower
and
Dem
ocracy
after
Civil
Wars
(Ithaca:
CornellUP 2005), because they did not consider
South A
frica as civilwar (following B
arbaraF. Walter,Committing
to
Peace: The
Successful Settlem
ent of Civil
Wars
(Princeton: PrincetonUP2002),andComoros andSudan notbeingcovered
by theirdata-set (to1999).
a The number of states stood as three at independence, and has been increased in several steps.
b Thefirstfederalconstitutionof1991provided
forninestates
(theform
erprovinces).In
1994,a
majorreorganisationcreateda totalof26states
andthe number
was
reducedto
25in
2005.Withtheindependence ofsouthern Sudan on9July
2011,the10southern statesdeparted.Since
January2012,thefederalrepublicofSudan
counts 17 federal states.
c Ethiopia
was
afederal
system
(withthetwo federal
unitsEthiopia
and Eritrea) alreadybetween 1952and1962.
TERRITORIAL POWER SHARING IN AFRICA
132 CIVIL WARS
Suberu claims that at independence, leaders of the three major ethnic groups
‘appeared to come together’ in a multinational federation and that subsequently
there has ever since the failed Biafra secession in the late 1960s been a large elite
consensus on federalism as building principle of the Nigerian state.52 ‘Nigeria is
arguably the sole African country in which federalism is deeply entrenched and its
constituent sub-federal governments are among the most powerful sub-national
units on the continent’.53
Nigeria’s record in managing violent conflict has been more mixed. It is true that
following the Biafra civil war with its massive loss of lives, the federal reforms have
been successful in preventing a recurrence of large-scale ethno-secessionist conflict. At
the same time, the country is plagued by high levels of ethnic protest, violent communal
conflicts and political and economic discrimination against the so-called non-indigenes
at the sub-state level. Many of these violent conflicts have occurred not in spite of
federalism, but exactly because of the specific dynamics of Nigerian federalism,
especially the creation of ethnically heterogeneous states. At the same time, many
students of Nigerian politics claim that the formal and informal institutions of power
sharing at the centre have been as important as federalism in creating stability. The so-
called ‘federal character’ is actually a geographic distribution rule which stipulates that
the composition of public institutions should reflect the federal character of the country
at each level of government.54
Ethiopia
Ethiopia comes closest to Bermeo’s category of ‘forced together’ federalism. Its
constitution did not result from a participatory constitution-making process, but was
imposed after the military victory of the Ethiopian People’s Revolutionary
Democratic Front (EPRDF) and was apparently inspired by the Soviet-Stalinist
model.55 While Ethiopia’s federalism has avoided state disintegration and large-
scale civil war, it is held together by the rule of a highly integrated bureaucratic party
structure, and actually governed very much like ‘a centralised unitary state with
most powers residing at the center’.56 Despite the international attention accorded to
the Ethiopian constitution, the political practice is certainly not providing for any
vertical power sharing. The government effectively rules by force.57 The success (or
not) of Ethiopia’s regime in creating some degree of political stability can thus not
be attributed to the decentralised features of its constitutional system.
Using the criteria that we deduced from the comparative literature, Ethiopia is also
the least likely case for a functioningmulti-nation federal system. It is a poor country,
has an authoritarian regime, and federalism was forced upon a certainly multiethnic
society but without any aspiration of creating a federation of nations. It has not passed
the test of peaceful departure from power of the government that created it.58
Tanzania
Tanzania has actually not a federal system, but a Union Government which was
created in 1964 by the merger of two pre-existing territories, Tanganjika and
Zanzibar. Both governments created a new centralised union government with
133 TERRITORIAL POWER SHARING IN AFRICA
representatives from both previously independent countries, but Zanzibar being so
small in comparison to Tanganjika retained a certain degree of autonomy with a
separate government and parliament. Although various constitutional reform
commissions proposed the transformation of this peculiar structure into a fully
fledged federal system, the government preferred to retain the status quowhichmeans
that mainland Tanzania (Tanganjika) has no separate tier of government below the
union structures.59
The autonomy regime awarded to the Zanzibar islands grants thus more political
discretion to sub-national elites than is the case in most of the other federal systems.
The system did not prevent the break out of violence especially in the aftermath of
the re-introduction of multi-party politics in 1995 as a fierce competition between
the former single party Chama Cha Mapinduzi (CCM, result of a merger of the then
dominating parties of Tanganjika and Zanzibar in 1977) and the main opposition
party Civic United Front (CUF, a major player only in Zanzibar, but not on the
mainland) emerged. Peace agreements in 1999 and 2001 eventually led to a local
power-sharing agreement in 2009. Further claims for a secession have thus been
quelled, especially within CUF, and the functioning of the broader asymmetrical
quasi-federal system is not really challenged. A more informal rotating principle at
the Presidency induced Nyerere to hand-pick Ali Hassan Mwinyi, a Zanzibari
Muslim, as his successor as Tanzanian President (1985 – 95).
Sudan
Sudan went through a first quasi-federal experiment (1972 –83) following years of
violent conflict, but the central government gradually undermined the federal
institutions until it was effectively ended with the break-out of another civil war. The
discovery of huge oil in southern Sudan 1978 and the direct interference of Ethiopia
in southern Sudan largely explain the failure of the federal system. In the middle of
the civil war, Sudan returned to a federal constitution transforming the existing
provinces to states. In order to weaken the southern states, the system was further
fragmented by increasing the number of federal units to 26 in 1994.
The federal system was never effectively working and certainly not instrumental
in ending the civil war(s). Federalism featured prominently in the 2005 Machakos
agreements ending the war between Sudan and South Sudan, but the federation was
‘seen as a compromise on the road towards secession rather than a permanent
solution to the problems of national accommodation in Sudan’.60
With independence of South Sudan on 9 July 2011, the southern states left in
order to build a new non-federal state, while the remaining states continued to build
a federation whose weakness was further exposed when President al-Bashir created
two further states by decree in January 2012.
Institutional Design and Ethnic Demography
The five cases represent contrasting institutional options. Ethiopia is a clear-cut case
of ethno-federalism with the whole state divided into supposedly ethnically
homogeneous homelands although practically none of these states has any tradition
134 CIVIL WARS
of self-rule. The risk of two core groups (Oromiya, Amhara both roughly 39 per cent
of the population) dominating the federation is so far counterbalanced by the
dominance of the ruling party.61 South Africa with its Apartheid history of
‘homelands’ and strong provinces decided to create new provinces which do not
coincide with any ethnic homeland. The Horowitzian ‘drawing through’ principle
was thus applied, although it was made possible through a major transformation of
the political system. South Africa’s weak federalism is decisively non-ethnic.
Although in different formats, in South Africa, Sudan, Tanzania and Ethiopia,
federalism was introduced to protect the interests of ethno-regional minorities.
Ethiopia’s ethno-federalism with secession clause served as exit-option for the
ruling minority, while South Africa’s constitution rather represented a guarantee for
political, cultural and economic rights of white and black minorities. Zanzibar’s
autonomy regime was introduced in the 1960s to guarantee the ethno-religious
interests of the small islands in a much larger polity, but the elites of both constituent
and independent units clearly ‘came together’ in the wake of the 1964 Zanzibari
revolution. Sudan’s federalism in its post-2005 version was a compromise solution
to keep southern Sudan in the Sudanese state while protecting the economic,
political and military interests of the southern elites.
Nigeria, finally, started with a multi-nation federation, but has over time moved
towards a centralised and non-ethnic federalism.62 The incremental increase of sub-
state units (from 3 to 36) diluted hegemonic or secessionist tendencies of the three
major ethnic groups. Suberu called this ‘a conscious and creative, albeit somewhat
controversial shift since the civil war towards a US-style national federal model in
Nigeria’s otherwise multiethnic or multi-national federal society’.63
Level of Economic Development
The specific economic and social contexts also matter in this comparison. Ethnic and
racial tensions in South Africa have been significantly absorbed by social policies
and black economic empowerment programs while (the otherwise completely
different) types of federal systems in Sudan and Nigeria were largely stabilised by
the redistribution of oil rents through federalism. Although Ethiopia has seen a
strong economic growth over the last decade, it continues to belong to the category
of least developed countries. Sub-national units heavily rely on grants and subsidies
from central government and the most lucrative income sources such as foreign
trade taxes are given to the central government.64 Such a federal structure is
therefore more likely to increase conflict both within the poorer sub-national states
and between the states and the central government.
Regime Type
The five cases represent contrasting regime trajectories. South Africa’s federal
constitution was both the result of a democratic negotiation and the basis for a
further democratic process. Nigeria’s federalism has been a constant feature of post
independence history, independent of the alternating political regimes. It has been
strengthened under democratic rule, notably with regard to the activism of the
135 TERRITORIAL POWER SHARING IN AFRICA
Supreme Court in arbitrating intergovernmental conflicts, although it increased the
pressures for more decentralisation and arguably increased the level of corruption.
Tanzania, Ethiopia and Sudan have introduced federalism in clearly non-democratic
circumstances, and it would be difficult to claim that especially the Ethiopian and
Sudanese regimes have evolved with their federal constitutions in a more democratic
direction since then. Kymlicka claimed that for a multinational democracy to have a
positive impact on conflict, it has to be democratic and there has to be a cross-ethnic
consensus on human rights and liberal constitutionalism.65 ‘The ultimate stability of
post-conflict multi-nation federalisms, however, will probably depend on
overcoming the security and human rights fears that prevented their peaceful
adoption in the first place. In Ethiopia, for example, it is difficult to see how genuine
autonomy will be possible for groups like the Somalis, who are viewed by the
government as secessionists, backed by foreign enemies’.66
DECENTRALISATION IN AFRICA
Political decentralisation within a non-federal polity is hard to measure, and it is thus
more difficult to give an empirical overview of devolution in the remaining non-
federal states of sub-Saharan Africa. Schneider measured political decentralisation
by the holding of elections at different sub-national levels,67 but this is certainly not
very convincing (and he included few African countries in his analysis). Brancati
used the constitutional division of competencies as main indicator and considered
countries to have vertical power sharing if they have regional legislatures with
independent decision-making power over at least one issue area.68 Brancati further
measured decentralisation with a four-point index on whether or not regional
legislatures are elected and the types of issues over which regional legislatures have
control – tax authority, education and public order or police.
According to such criteria, none of the non-federal cases in Africa implements a
strong form of devolution. Central governments in Africa have tended to keep
decentralised authorities under strict political surveillance and with little financial
discretion over how to spend the largely centrally collected resources.69 Qualitative
research on individual processes of decentralisation in Africa highlights the many
differences among the countries and the growing assertiveness of local and regional
councils in many places. It should thus not be excluded that even weak forms of
decentralisation might be relevant for conflict management.
For lack of data, it is impossible to give a comprehensive assessment of how the
very different decentralisation reforms have affected the dynamics of violent
conflict in African states. It is possible that within a decentralised unit, different
groups with contrasting political views or structural disparities in economic power
or different ethnic or religious affiliations may conflict with each other, or different
local government units may clash with each other.70 The scenario of conflict between
one (or several) decentralised unit and the central government over the extent of
political, cultural, financial and economic autonomy and authority is thus just one
possible outcome of decentralisation reforms.
136 CIVIL WARS
Let us have a closer look at the cases of decentralisation/territorial power sharing
with post-conflict peace agreements. We can distinguish three types:
In Angola, Mozambique or Djibouti, decentralisation reforms were promised
without any guarantee that meaningful powers are devolved to sub-national level.
While the power sharing might have a territorial component, this was not the real
issue in any of these agreements, and hardly amounts even to some sort of autonomy,
let alone to power sharing. In Angola and Djibouti, decentralisation was actually
never implemented; in Mozambique, the law was passed nearly 7 years after the
ending of conflict, and in a way which minimises the influence of the rural-based
opposition party Resistencia Nacional Moc�ambicana (RENAMO) by effectively
adopting an asymmetrical type of decentralisation granting local self-government to
selected municipalities only (Table 3).71
In Burundi, the very comprehensive power sharing within government also
included the different territorial layers of government. Again, this provision does not
share power, but simply attributes posts to the rebels who are appointed by the
central government.72
In a final group, decentralisation was consciously used by state elites to counter
violent regionalist movements. Seely has analysed the decentralisation policies
enacted in Mali since 1992 as a deliberate attempt of the new democratic regime to
co-opt the Tuareg rebels into the political system.73 The National Pact provided for
integration of Touareg fighters in the security forces and reserved four seats in the
National Assembly to delegates from the northern peoples of Mali. The agreement
amended the political structure of Mali through a decentralisation of authority to
regional and sub-regional assemblies of an executive character. Special
administrative structures were agreed upon for the northern region. These new
assemblies mainly had powers in socio-economic and cultural sectors of
governance. It might be questioned, however, whether the territorial component
had a particular importance for managing the conflict, especially in comparison to
TABLE 3
DECENTRALISATION AFTER VIOLENT CONFLICTS IN SUB-SAHARAN AFRICA 1990 – 2011
Peace agreements Type of decentralisation
Mozambique 1992, Angola 1994, Djibouti 2001
Decentralisation reforms were promised without any guarantee that meaningful powers are devolved to sub-national level. While the power sharing has a territorial component, this was not the real issue in any of these agreements
Burundi 2003 Comprehensive power sharing within government also included the different territorial layers of government. Attributes posts to the rebels that are appointed by the central government
Mali 1992, Niger 1993/94, Senegal 2004, Angola 2006
Asymmetrical devolution/regional autonomy
137 TERRITORIAL POWER SHARING IN AFRICA
the integration of rebel fighters into the army. In the Cabinda Peace Agreement of
2006, the Angolan government offered the rebels some degree of local autonomy
which represented a progress with respect to the status quo ante, but did not really
allow the population of this small oil-rich exclave to effectively control the polity or
economy on their territory.
Although not introduced through a peace agreement, the decentralisation
reforms in Senegal and Uganda fit into this same pattern. During the early 1980s, a
violent conflict erupted in the southern province of Casamance, virtually cut from
the rest of Senegal by the enclave of the Gambia. The mostly Dioula and Christian
population, living in the western (or lower) part of Casamance, resented a growing
economic and social marginalisation and a large influx of ‘Northerners’ within the
administration and business elites of the Casamance.74 The local populations
eventually started to give widespread support to a rebel movement (Mouvement des Forces Democratiques de Casamance, MFDC) fighting for the separation of
Casamance from Senegal. Large-scale violence was brought to an end in the early
1990s, but violence re-erupted again and again and no sustainable political solution
has been found yet.75 Decentralisation was mainly introduced in Senegal as a
developmental strategy, but included political subtexts in dealing with Casamance.
Although President Abdoulaye Wade promised a more substantial autonomy in the
latest peace agreements of 2004, nothing substantial followed, and in the run-up to
the last presidential elections in 2012, new violence erupted.
The Ugandan case has attracted much interest because decentralisation was
designed in the mid-1990s as a strategy to appease the proponents of federalism
(largely concentrated in wealthy Buganda). Interestingly, it was not designed to deal
with the ethnic grievances of the northern populations and the brutal civil war in
northern Uganda. As the survival of the ruling coalition depended on Buganda’s
support of the new Constitution in the Constituent assembly, and decentralisation
was welcomed by donors, some concessions to autonomy demands had to be taken.76
In the further decentralisation process, minorities emerged to demand the creation of
new districts thus fragmenting the existing sub-national units and denying the bigger
ethno-nationalist movements any possibility to use decentralised units as
springboard for secession while shifting the problem of ethnic conflict to local
levels where it was supposedly less threatening to state integrity.77
On balance, these political hidden agendas explain why decentralisation is rarely
amounting to territorial self-governance or territorial power sharing. If
decentralisation is introduced to placate donors or to appease localised rebel
movements, it is very unlikely to lead to a durable sharing of power between the
central state and local or regional state elites. These cases thus illustrate to what
extent decentralised institutions can effectively function as an obstacle to regional
autonomy. Regime type might matter here again as decentralisation with its non-
constitutional basis is certainly more likely to be manipulated in the presidentialist
and hybrid or authoritarian regimes. The core intervening variable, however, is the
ethnic geography or nature of ethnic identity in the African context.
138 CIVIL WARS
THE AFRICAN CONTEXT
What follows from this previous discussion for the prospects of federalism and
territorial power sharing in Africa and its role in managing violent conflicts? We
have seen that there are few cases of substantive territorial power sharing in Africa
(although in contrast to many other world regions during the last two decades, two
new federations emerged on the continent, Ethiopia and South Africa) and that the
existing cases of vertical power sharing have contributed only to a limited extent to
conflict management.
We assume that the degree to which federal institutions can contribute to
preserving peace depends on how these institutions reflect the characteristics of the
societies they govern. The idea of multi-nation or ethno-federalism relies on the idea
of ‘competing nationalisms’ within a single state. According to this view, a
dominant core group forming a clear majority of the population used nationbuilding
policies to impose its language and culture on the rest of the population. Minorities
had to constitute themselves as nations and typically tried to consolidate this
nationhood in institutions at sub-national level.
Most African states, however, do not have such a core majoritarian ethnic group.
They are rather composed of a multitude of ethno-cultural groups. Compared to
countries in other regions of the world, African countries contain large numbers of
small groups, none with sufficient numerical strength to constitute a majority in a
country.78 ‘Comparatively few ethnic groups in black Africa are candidates for
nationhood, and few of them have such aspirations’.79 In most cases, it is not the
language of a proto-nation to monopolise public space and public institutions, but
the language of the former coloniser. There is consequently less need for ethno
federalism to protect minorities from the injustices that arise as a result of majority
nationbuilding, as pointed out by Mozaffar and Scarritt:
Because of the combination of modal characteristics of ethnopolitical groups in
Africa – many small and relatively similar groups that do not differ greatly from one
another by global standards, the heterogeneity of these differences, multiple levels of
ethnopolitical identity, the absence of long histories of autonomous rule, and significant
changes in the ethnopolitical landscape – most African countries are not deeply divided
societies. They are more accurately characterised as multiethnic societies comprised of
large numbers of relatively evenly balanced, or dispersed, communal groups.80
The ethnic conflicts we observe in Africa rather ‘take the form of struggles for a
share of state power at the central level, rather than ethno-nationalist struggles for
self-government and autonomy at a regional level, (. . .) the problem is that while the
state may be neutral (that is, language, culture and symbols are not tied to any
particular ethnic group), access to the state remains ethnicized’.81 This requires
groups to maintain a strong group identity, but does not necessarily provide
incentives to exaggerate cultural differences.
In the context of a ‘gatekeeper-state’ which sits astride the interface between a
territory and the rest of the world collecting and distributing resources that derive
from the gate itself82 (such as customs revenue and foreign aid, permits to do
139 TERRITORIAL POWER SHARING IN AFRICA
business in the territory, entry and exit visas), fighting for political and economic
autonomy within peripheral regions of this territory is less obvious than aspiring to
be included in some form of horizontal power sharing at the central level, i.e. the
gate.83 The few centre-periphery conflicts which exist in Africa thus do typically not
lead to ethno-nationalist struggles. Even in places such as Casamance or the recent
Tuareg uprising in Mali where the dominant discourse frames these rebellions as
nationalist struggles, these movements rely much more on transnational criminal or
terrorist networks than on any substantial domestic mobilisation.
A small number of African states do, however, not correspond to this model of
primarily horizontal (or unranked) ethnic relationships and the ensuing competition
among non-hierarchical ethnic groups.84 In Ethiopia, Sudan, Liberia, Rwanda,
Burundi and the former Apartheid states of southern Africa (Namibia, South Africa),
statebuilding was not neutral amongst ethnic groups, but one racial or ethnic group
managed to shape the state which reflected its language, history and culture and in
which social status corresponded with ethnic affiliation. This is most evident in the
fact that non-European languages spoken by the core group such as Amharic or
Afrikaans became national languages. These societies can indeed be characterised as
deeply divided societies in which ethnic or other particularistic groups are strongly
polarised by overlapping political, social and economic cleavages.
Some minorities in these countries with a strong territorial base and a history of
autonomy have responded with ethno-nationalism. What we find there ‘is something
akin to the problem of competing nationalisms’.85 Ethno-federalism could thus be
the institutional answer and has been chosen in Ethiopia (and to an extent in southern
Sudan). Liberia, Rwanda and Burundi are both too small as countries and have also
mixed settlement patterns which excluded a consideration of federalism. For very
obvious historical reasons, ethno-federalism was no longer an option in South Africa
and Namibia (not even for the White hardliners during the constitutional
negotiations).
CONCLUSION
The article has argued that there is a relative lack of vertical power sharing in
sub-Saharan Africa, but that this should not surprise us given the socio-ethnic
context of most African societies which lack both majoritarian ethnic groups and
clear culturally based distinctions between contending ethnic groups. In this context,
forms of power sharing that include provision on territorial autonomy should be
regarded with caution. Regional autonomy as a solution to civil war will not work
in most cases. By interregional comparison, many African states are small in
population (a third of all states having less than 3 million inhabitants), and
federalism certainly makes sense only in states of a certain size and population. Most
African peace agreements which included territorial power sharing might have been
essential in sending signals of commitment86 but did nothing to create a viable and
institutionalised form of vertical power sharing.
140 CIVIL WARS
Critics of ethno-federalism mainly discuss the spectre of disintegration. The
African experience shows that decentralisation might fuel or mitigate violent
conflicts below the level of full-scale partition, and future research should thus be
more interested in this multitude of conflict patterns and how these conflicts are
influenced by territorial power sharing.
We have shown that territorial power sharing might be a solution in a couple
of countries with rather hierarchical inter-ethnic relationships. Our analysis has
revealed, however, that even in cases such as Nigeria and South Africa, conflict
management effectively relies on non-ethnic federalism or a combination of
different modes of power sharing, in that territorial self-rule is complemented by at
least some elements of consociationalism. Where it lacks, such as in Ethiopia,
federalism does not work well, and positive effects on inter-ethnic conflict can
be excluded. Further research should thus be primarily interested in a better
understanding of the precise interaction patterns of vertical and horizontal forms of
power sharing in Africa and beyond.
ACKNOWLEDGEMENTS
A first version of the paper was presented at the project workshop on Institutions for Sustainable Peace at the International Studies Association Conference in San Diego in April 2012. I thank all participants and the two anonymous reviewers for their helpful feedback.
NOTES
1. Arend Lijphart, Democracy in Plural Societies (New Haven, CT: Yale UP 1977). 2. John McGarry and Brendan O’Leary, ‘Territorial Approaches to Ethnic Conflict Settlement’ in Karl
Cordell and Stefan Wolff (eds) Routledge Handbook of Ethnic Conflict (London: Routledge 2011) pp.249 – 65.
3. Anna Jarstad and Timothy Sisk (eds) From War to Democracy: Dilemmas of Peacebuilding, Cambridge (Cambridge: Cambridge UP 2008); Andreas Mehler, ‘Not Always in the People’s Interest: Power-Sharing Arrangements in African Peace Agreements’ (Hamburg, German Institute of Global and Area Studies 2008); Philip Roeder and Donald Rothchild (eds) Sustainable Peace: Power and Democracy After Civil Wars (Ithaca: Cornell UP 2005).
4. Matthew Hoddie and Caroline Hartzell, ‘Power Sharing in Peace Settlements: Initiating the Transition from Civil War’ in Roeder and Rothchild (note 3) pp.83 – 106.
5. Pippa Norris, Driving Democracy. Do Power-Sharing Institutions Work? (Cambridge: Cambridge UP 2008).
6. Donald Rothchild and Philip G. Roeder, ‘Power Sharing as an Impediment to Peace and Democracy’ in Roeder and Rothchild (note 3) pp.29 – 50.
7. Anna Jarstad, ‘Power Sharing: Former Enemies in Joint Government’ in Anna Jarstad and Timothy Sisk (eds) From War to Democracy: Dilemmas of Peacebuilding (Cambridge: Cambridge UP 2008) pp.105 – 33.
8. Caroline Hartzell and Matthew Hoddie, ‘Institutionalizing Peace: Power-Sharing and Post-Civil War Conflict Management’, American Journal of Political Science 47/2 (2003) pp.318 –32.
9. Mehler (note 3). 10. Ibid. p.7; Jarstad (note 7). 11. Hoddie and Hartzell (note 4). 12. Ibid. 13. Ibid. p.103. 14. David A. Lake and Donald Rothchild, ‘Territorial Decentralization and Civil War Settlements’, in
Philip Roeder and Donald Rothchild (eds) Sustainable Peace: Power and Democracy After Civil Wars (Ithaca: Cornell UP 2005) pp.109 –32.
141 TERRITORIAL POWER SHARING IN AFRICA
15. Uppsala Conflict Data Program/StinaHogbladh, Peace Agreement Dataset Codebook Version 2.0 (Uppsala: Department of Peace and Conflict Research, 2012). This definition of peace agreement clearly extends beyond the idea of power-sharing.
16. Donald L. Horowitz, Ethnic Groups in Conflict (Berkeley: University of California Press 1985); Kristin M. Bakke and Erik Wibbels, ‘Diversity, Disparity, and Civil Conflict in Federal States’, World Politics 59/1 (2006) pp.1 –50; Jan Erk and Lawrence Anderson, ‘The Paradox of Federalism: Does Self-Rule Accommodate or Exacerbate Ethnic Divisions?’ Regional and Federal Studies 19/2 (2009) pp.191 – 202; Dawn Brancati, ‘Decentralization: Fueling the Fire or Dampening the Flames of Ethnic Conflict and Secessionism?’, International Organization 60/3 (2006) pp.651 –85 and Stefan Wolff, ‘Conflict Management in Divided Societies: The Many Uses of Territorial Self-Governance’, International Journal on Minority and Group Rights 20/1 (2013) pp.27 – 50.
17. Pierre Englebert and Rebecca Hummel, ‘Let’s Stick Together: Understanding Africa’s Secessionist Deficit’, African Affairs 104/416 (2005) pp.399 – 427.
18. Andrea Iff, Peace-Preserving Federalism. Making Sense of India and Nigeria (Saarbrucken: Sudwestdt. VerlagfurHochschulschr. 2010).
19. Eghosa Osaghae, ‘Federalism and the Management of Diversity in Africa’, Identity, Culture and Politics 5/1–2 (2004) pp.162 –78 and Shaheen Mozaffar and James R. Scarritt, ‘Why Territorial Autonomy is Not a Viable Option for Managing Ethnic Conflict in African Plural Societies’, Nationalism and Ethnic Politics 5/3–4 (1999) pp.230 – 53.
20. Dennis A. Rondinelli, John R. Nellis, and G. Shabbir Cheema, Decentralisation in Developing Countries: A Review of Recent Experience (Washington, DC: World Bank 1983).
21. William Riker, Federalism: Origin, Operation, Significance (Boston, MA: Little, Brown 1964) and Daniel J. Elazar, Exploring Federalism (Tuscaloosa: University of Alabama Press 1987).
22. Ugo M. Amoretti, and Nancy Bermeo (eds), Federalism and Territorial Cleavages (Baltimore, MD: John Hopkins University Press 2003) and Jonathan Rodden, ‘Comparative federalism and decentralization: On Meaning and Measurement’, Comparative Politics 36/4 (2004) pp.481 – 500.
23. Ronald Watts, ‘Models of Federal Power Sharing’, International Social Science Journal 167 (2001) pp.23 –32; Rodden (note 22).
24. Norris (note 5). 25. Yash P. Ghai, ‘Constitutional Asymmetries: Communal Representation, Federalism, and Cultural
Autonomy’ in Andrew Reynolds (ed.), The Architecture of Democracy. Constitutional Design, Conflict Management and Democracy (Oxford: Oxford UP 2002) pp.141 –70 and Watts (note 23).
26. Brendan O’Leary, ‘An Iron Law of Nationalism and Federation?’ Nations and Nationalism 7/3 (2001) pp.273 –96.
27. See references in note 16. 28. Alfred Stepan, ‘Federalism and Democracy: Beyond the U.S. Model’, Journal of Democracy 10/4
(1999) pp.19 – 34. 29. But see Kent Eaton and Ed Connerly, ‘Democracy, Development, and Security as Objectives of
Decentralization’, in Ed Connerly, Kent Eaton, and Paul Smoke (eds), Making Decentralisation Work (Boulder: Lynne Rienner 2010) pp.1 –24 and Joseph Siegle and Patrick O’Mahony, ‘Decentralization and Internal Conflict’, in E. Connerly, K. Eaton, and P. Smoke (eds) Making Decentralisation Work (Boulder: Lynne Rienner 2010) pp.135 –66.
30. For Africa, compare Andreas Mehler, ‘Decentralization, Division of Power and Crisis Prevention: A Theoretical Exploration with Reference to Africa, in Tobias Debiel (ed.), Fragile Peace: State Failure, Violence, and Development in Crisis Regions (London: Zed Books 2002) pp.72 –86 and Christof Hartmann, ‘Decentralisation and the Legacy of Protracted Conflict – Mauritius, Namibia and South Africa’ in Gordon Crawford and Christof Hartmann (eds) Decentralisation in Africa: A Pathway Out of Poverty and Conflict? (Amsterdam: Amsterdam UP 2008) pp.169–90.
31. Graham K. Brown, ‘Decentralisation and Conflict: Introduction and Overview’, Conflict, Security and Development 8/4 (2008) pp.387 –92, 389.
32. Nancy Bermeo, ‘Conclusion: The Merits of Federalism’, in Ugo M. Amoretti and Nancy Bermeo (eds) Federalism and Territorial Cleavages (Baltimore, CA: John Hopkins UP 2003) pp.457 – 82.
33. Will Kymlicka, ‘ Emerging Western Models of Multination Federalism: Are They Relevant for Africa?’, in David Turton (ed.) Ethnic Federalism. The Ethiopian Experience in Comparative Perspective (London: James Currey 2006) pp.32 –64 and John McGarry and Brendan O’Leary, ‘Federation as a Method of Ethnic Conflict Regulation’ in J.R.S. Noel (ed.) From Power-Sharing to Democracy: Post-Conflict Institutions in Ethnically Divided Societies (Toronto: McGill-Queens UP 2005) pp.263 – 96.
142 CIVIL WARS
34. Erk and Anderson (note 16) p.192. 35. Valerie Bunce, Subversive Institutions. The Design and Destruction of Socialism and the State
(Cambridge: Cambridge UP 1999); Svante E. Cornell, ‘Autonomy as a Source of Conflict: Caucasian Conflicts in Theoretical Perspective’, World Politics 54/2 (2002) pp.245 –76 and Jack Snyder, From Voting to Violence: Democratization and Nationalist Conflict (New York, NY: W.W. Norton 2000).
36. Philip G. Roeder, ‘Ethnofederalism and the Mismanagement of Conflicting Nationalisms’, Regional and Federal Studies 19/2 (2009) pp.203 –19, 206.
37. Stepan (note 28); Bermeo (note 32) and McGarry and O’Leary (note 2). 38. Daniel J. Elazar, ‘Federalism and Consociational Regimes‘, Publius: Journal of Federalism 15/2
(1985) pp.17 –34. 39. Stepan (note 28). 40. Horowitz (note 16) p.623. 41. E.g. Bakke and Wibbels (note 16). 42. Erk and Anderson (note 16). 43. O’Leary (note 26). 44. Henry Hale, ‘Divided We Stand: International Sources of Ethnofederal State Survival and Collapse’,
World Politics 56/4 (2004) pp.165 –93. 45. McGarry and O’Leary (note 2). 46. Horowitz (note 16). 47. Brancati (note 16). 48. Nancy Bermeo, The Import of Institutions, Journal of Democracy 13/2 (2002) pp.96 – 110, 108. 49. Roeder (note 36). 50. Rotimi Suberu, ‘Federalism in Africa: The Nigerian Experience in Comparative Perspective’,
Ethnopolitics 8/1 (2009) pp.68f. 51. Stepan (note 28). 52. Suberu (note 50). 53. Ibid. p.67. 54. Daniel Bach, ’Indigeneity, Ethnicity and Federalism’, in Larry Diamond, Anthony Kirk-Greene, and
OyeleyeOyediran (eds) Transition Without End: Nigerian Politics and Civil Society Under Babangida (Boulder: Lynne Rienner 1997) pp.333 –49; Osaghae (note 15); Anthony Smith, ‘Fractured Federalism: Nigeria’s Lesson for Today’s Nation Builder in Iraq’, The Round Table 94/1 (2005) pp.129 –44 and Suberu (note 50).
55. Christopher Clapham, ‘Afterword’, in D. Turton (ed.) Ethnic Federalism. The Ethiopian Experience in Comparative Perspective (London: James Currey 2006).
56. Edmond J. Keller, ‘Ethnic Federalism, Fiscal Reform, Development and Democracy in Ethiopia’, African Journal of Political Science 7/1 (2002) pp.21 –50, 46.
57. International Crisis Group 2009, ‘Ethiopia: Ethnic Federalism and its Discontents’, Africa Report 153 (September 2009).
58. Clapham (note 55) p.237 and David Turton (ed.) Ethnic Federalism. The Ethiopian Experience in Comparative Perspective (London: James Currey 2006).
59. For details see Romuald R. Haule, ‘Torturing the Union? An Examination of the Union of Tanzania and its Constitutionality’ Zeitschrift fur auslandisches offentliches Recht und Volkerrecht 66 (2006) pp.215 – 33.
60. Rohan Edrisinha, Lee Seymour, and Ann Griffiths, ‘Adopting Federalism: Sri Lanka and Sudan’ in Ann Griffiths (ed.) Handbook of Federal Countries, 2005 (Montreal: Mc Gill-Queen’s UP 2005), p.445.
61. Edmond J. Keller, and Lahra Smith, ‘Obstacles to Implementing Territorial Decentralization: The First Decade of Ethiopian Federalism’, in Philip Roeder and Donald Rothchild (eds) Sustainable Peace: Power and Democracy after Civil Wars (Ithaca: Cornell UP 2005).
62. Horowitz (note 16); Rotimi Suberu and Larry Diamond, ‘Institutional Design, Ethnic Conflict-Management and Democracy in Nigeria’, in Andrew Reynolds (ed.) The Architecture of Democracy (Oxford: Oxford UP 2002) pp.400 –28.
63. Suberu (note 50) p.73. 64. On Ethiopian federalism see particularly Lovise Aalen, Ethnic Federalism in a Dominant Party State:
The Ethiopian Experience 1991–2000 (Bergen: Christian Michelsen Institute 2002). 65. Kymlicka (note 33). 66. Ibid. p.52. 67. Aaron Schneider, ‘Decentralization: Conceptualization and Measurement’, Studies in Comparative
International Development 38/3 (2003) pp.32 –56.
143 TERRITORIAL POWER SHARING IN AFRICA
68. Brancati (note 16). 69. Gordon Crawford and Christof Hartmann (eds) Decentralisation in Africa. A Pathway out of Poverty
and Conflict? (Amsterdam: Amsterdam UP 2008). 70. Mehler (note 30). 71. P.S. Reddy, Local Government Democratisation and Decentralisation. A Review of the Southern
African Region (Kenwyn: Juta 1999). 72. On Burundi Daniel P. Sullivan, ‘The Missing Pillars: A Look at the Failure of Peace in Burundi
Through the Lens of Arend Lijphart’s Theory of Consociational Democracy’ Journal of Modern African Studies 43/1 (2005) pp.75 –95 and Rene Lemarchand, ‘Consociationalism and Power Sharing in Africa: Rwanda, Burundi, and the Democratic Republic of the Congo’, African Affairs 106/422 (2006) pp.1 – 20.
73. Jennifer C. Seely, ‘A Political Analysis of Decentralisation: Coopting the Tuareg Threat in Mali’, Journal of Modern African Studies 39/3 (2001) pp.499 – 524.
74. For a more detailed historical account and political-economic explanation see Catherine Boone, Political Topographies of the African State (Cambridge: Cambridge UP 2003) pp.94ff.
75. On Senegal compare Vincent Foucher, ‘Pas d’alternance en Casamance? Le nouveau pouvoir senegalais face a la revendication separatiste casamanc�aise’, Politique Africaine 91 (2003) pp.101–19.
76. Pierre Englebert, ‘Born-Again Buganda or the Limits of Traditional Resurgence in Africa’, Journal of Modern African Studies 40/3 (2002) pp.345–68.
77. Anna Katharina Schelnberger, ‘Decentralization and Conflict in Kibaale, Uganda’, in Gordon Crawford and Christof Hartmann (eds) Decentralisation in Africa. A Pathway out of Poverty and Conflict? (Amsterdam: Amsterdam UP 2008) pp.191 – 212; Elliot Green, ‘Decentralization and Conflict in Uganda’, Conflict, Security & Development 8/4 (2008) pp.427 – 50.
78. Mozaffar and Scarritt (note 19). 79. John Markakis, ‘Nationalism and Ethnicity in the Horn of Africa’, in Paris Yeros (ed.) Ethnicity and
Nationalism in Africa: Constructivist Reflections and Contemporary Politics (Basingstoke: Macmillan1999) pp.65 –80, 77.
80. Mozaffar and Scarritt (note 15) p.241. 81. Kymlicka (note 33) p.48. 82. Frederick Cooper, Africa Since 1940: The Past of the Present (Cambridge: Cambridge UP 2002). 83. For a similar argument Englebert and Hummel (note 17). 84. The terminology goes back to Horowitz (note 16). 85. Kymlicka (note 33) p.48. 86. The idea of signaling, and ‘costly signals’ is based here on the discussion by Matthew Hoddie and
Caroline Hartzell, ‘Civil War Settlements and the Implementation of Military Power-Sharing Arrangements’, Journal of Peace Research 40/3 (2003) pp.303 – 20.