Scholarly article on topic 'Peace and Power Sharing in Africa: A Not So Obvious Relationship'

Peace and Power Sharing in Africa: A Not So Obvious Relationship Academic research paper on "Political Science"

Share paper
Academic journal
African Affairs
OECD Field of science

Academic research paper on topic "Peace and Power Sharing in Africa: A Not So Obvious Relationship"

African Affairs, 108/432, 453-473 doi: 10.1093/afrai7adp038

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

Advance Access Publication 27 May 2009


Andreas Mehler


Peace accords usually involve top politicians and military leaders, who °

negotiate, sign, and/or benefit from an agreement. What is conspicuously e

absent from such negotiations is broad-based participation by those who r

should benefit in the first place: citizens. More specifically, the local level |

of security provision and insecurity production is rarely taken into account. tr

The analysis of recent African peace agreements shows important variations in power-sharing devices and why it is important to ask who is sharing power with whom. Experiences with power sharing are mixed and far less positive than assumed by outside negotiators.

Power sharing looks like a logical approach to sustainable conflict .

management in multi-ethnic societies. In fact, it has been proposed time and /

again and often inscribed in peace agreements. As Jarstad notes, 'power g

sharing is attractive to peace negotiators'.1 It might be less so for (all) t

warring parties and the general population. Negotiators and signatories g

may ignore and sideline security concerns of ordinary citizens, doing little S.

to address the underlying causes of conflict, and thus reducing the usefulness §

of power sharing. It is also doubtful whether the power-sharing ingredients O

of peace agreements are conducive to peace. J

One problem of the current tendency to establish hasty power-sharing so- U

lutions from the outside is the widespread popular perception that democ- g

racy is thereby sacrificed in the name of peace. This problem is frequently g

overlooked in specialized academic debates. In this contribution I first sum- 1

marize some relevant arguments in the academic debate on power sharing and war termination. In a second step, I screen all recent major African peace agreements (1999-2007) for their power-sharing content. Finally, I analyse two country cases of peace agreements (Liberia and Cote d'Ivoire) and one post-election crisis settlement (Kenya), all ofwhich relied on powersharing ingredients. It will become obvious that power-sharing agreements

Andreas Mehler ( is Director of the Institute of African Affairs at the German Institute of Global and Area Studies, Hamburg.

1. Anna Jarstad, 'Power sharing for peace and democracy?' (Paper presented at the 47th annual meeting of the International Studies Association, San Diego, 22-25 March 2006), p. 9.

cannot provide sustainable solutions to all relevant aspects of complex crisis situations, particularly those beyond a confined elite focus. Only intensely debated and transitional power-sharing arrangements provide the expected beneficial effects.

Sustainable peace by power sharing?

There is a large and growing body of literature on the causes of successful conflict resolution. It would be impossible to give an account of all arguments put forward and all the approaches followed in this field. Only a selection of main works will be quoted for the sake of the argument: power sharing is not uniform and has no uniformly positive effects on peace and war, and that it ignores local security concerns.

Barbara Walter is a prominent supporter of power sharing. She posits that the more power sharing is built into a peace agreement, the less international commitment is needed to guarantee it.2 This belief in power sharing as a miracle formula is not rare, but should lead to further questions. What is meant by power sharing? Who has to share power with whom, and who can still be excluded from the sharing arrangement? Those questions are essential, for power sharing maybe limited to only a few areas of governance; or power maybe shared only between the government and hand-picked rebel groups; or radical movements, on the one hand, and civilian opposition parties, on the other, may be left out.

Hoddie and Hartzell test provisions in the peace agreement - whether implemented or not - as an independent variable for explaining the durability of peace, and also distinguish between different levels of power sharing: (1) central (political), (2) territorial (federalism/decentralization), (3) military, and (4) economic power sharing. Consecutive hypotheses are formulated and statistically tested. Their conclusion is positive while specific with regard to the level of conflict escalation: 'Power-sharing provisions in peace settlements have a demonstrated ability to provide a sense of security to former combatants facing the immediate prospect of working together peacefully after a severe conflict such as a civil war.'3 But how could they come to this conclusion without asking the combatants? They continue: 'In particular, our research indicates that both military and territorial power sharing have a positive role to play in fostering post-war peace.'4 Political

2. Barbara Walter, 'The critical barrier to civil war', International Organization 51, 3 (1997), pp. 335-64.

3. Matthew Hoddie and Caroline Hartzell, 'Power sharing in peace settlements: initiating the transition from civil war' in Philip G. Roeder and Donald Rothchild (eds), Sustainable Peace: Power and democracy after civil war (Cornell University Press, New York, NY, 2005), p. 103.

4. Ibid.

power sharing, by far the most frequent version in Africa, does not have the same value, according to Hoddie and Hartzell.

In a widely quoted article Stedman makes a case for focusing on so-called spoilers as the most important barrier to the implementation of peace agreements.5 While Stedman brings back in a number of issues of importance - such as actors' interests and intents in the discussion - and therefore is an important advocate of being less naive about peace processes, some critical aspects of his work have to be noted. First, spoilers are always portrayed as the 'bad guys', stepping out of something that is perceived as good Q - regardless of what harm it does to particular group interests. Second, we W have to remember that the spoilers of the peace process may have been sup- 3 porters of the previous democratization. And conversely, a rebel movement d that is accommodated by a peace process and is apparently sticking to the o provisions of the peace agreement may have been the spoiler during earlier ^ reforms. The spoiler perspective may hide more than it reveals. / Finally, Tull and Mehler argue that the institutionalization of the practice r of power sharing for the sake of 'peace' - that is, providing rebels with a 0

share of state power, °

has important demonstration effects across the continent. It creates an incentive struc- g

ture would-be leaders can seize upon by embarking on the insurgent path as well. As 3

a result, and irrespective of their effectiveness in any given case, power-sharing agree- 0

ments contribute to the reproduction of insurgent violence.6 /

Mediators may not even think in this long-term perspective. This may not >

be entirely their fault, since the challenges they face can be immense. From a §

diplomat's point of view a bad agreement may be better than no agreement. i

One key problem of the fire engine diplomacy that is so typical in Africa is choosing mediation partners on the rebel side. In a context devoid of peace, with perhaps a record of only manipulated elections over decades, it is very difficult to know who can command the loyalty and legitimacy of specific groups. A much more realistic assumption is that the self-declared leaders and representatives of a neglected group are politico-military entrepreneurs without a genuine interest in representing anything beyond themselves. They are usually involved in the negotiation because of their spoiling capacities. Implicitly, they are also perceived as representatives of

5. Stephan John Stedman, 'Spoiler problems in peace processes', International Security 22, 2 (1997), pp. 5-53.

6. Denis Tull and Andreas Mehler, 'The hidden costs of power sharing: reproducing insurgent violence in Africa', African Affairs 104, 416 (2005), p. 375. Tull and Mehler were criticized by Rene Lemarchand ('Consociationalism and power sharing in Africa: Rwanda, Burundi, and the Democratic Republic of the Congo', African Affairs 106,422 (2007), pp. 1-20) for allegedly not wanting to take into account rebel organizations' grievances, in particular those that were excluded and had spoiling capacities. This was not our argument, as we primarily wanted to show that peace agreements fail to take into account civilian opposition's grievances, voiced in peacetime, and end up sidelining those more constructive forces in peace negotiations.

ethnic groups, while those rarely form an undisputed entity. The outer limits of these groups are in fact frequently disputed and the internal homogeneity is a fiction.

This selective literature review shows that specification is imperative as power sharing can take very different forms. It further questions whether unintended negative or ambiguous effects are systematically accounted for in the academic literature. Over recent decades a number of peace settlements were agreed upon in the name of preventive diplomacy whose sustainability could have been doubted early on. Roeder and Rothchild count 22 comprehensive peace settlements in ethnically divided societies worldwide.7 In some of those cases severe escalations happened in their aftermath (Rwanda in 1993, Burundi in 1994, Sierra Leone in 1996, Afghanistan in 2001), in others additional agreements were necessary to more or less stabilize the situation (Burundi in 2001 and 2002, DRC in 2002 and 2003).8

Recent African peace agreements .

In this light, it makes sense to review briefly those African peace agree- &

ments established after Hoddie and Hartzell stop their calculation (1998) U

and take only recent African cases before moving to our particular cases 3

in more detail (Liberia, Cote d'lvoire, and Kenya). The following overview O (Table 1) is organized in a way intended to make the multi-dimensionality

of power-sharing dimensions visible. A

This overview shows (1) that power sharing was a prominent aspect of

most recent African peace settlements,9 and (2) that there are important g variations in the form power sharing can take, confirming Hoddie and e Hartzell's approach. Political and military power sharing is much more Infrequent than territorial or economic power sharing. Even more striking is g (3) the variation in the important question of who is sharing power with g

7. Roeder and Rothchild, Sustainable Peace, p. 3. These peace settlements include Azerbaijan (1991), Croatia (1991), Slovenia (1991), Eritrea (1993), South Africa (1993), Rwanda (1993), Burundi (1994), Bosnia (1995), Sierra Leone (1996), Northern Ireland (1998), Kosovo (1999), Afghanistan (2001), Burundi (2001), Macedonia (2001-2), Burundi (2002), Congo (2002), Democratic Republic of Congo (DRC) (2002), East Timor (2002), DRC (2003), Cote d'Ivoire (2003), Iraq (2003), and Sudan (2004). Some gaps may be noted here (Djibouti, 2000; Angola, 1991).

8. Caroline Hartzell, Matthew Hoddie , and Donald Rothchild, 'Stabilizing the peace after civil war: an investigation of some key variables', International Organization 55, 1 (2001), p. 195. They calculated that peace agreements on average lasted three and a half years before conflict re-escalated.

9. The big exception is obviously the Casamance conflict between separatist rebels and the central government of Senegal. Here, numerous peace agreements have failed to have any sustainable effects, but maybe not because of the absence of power-sharing devices. Failure is more probably linked to the amorphous nature of the MFDC rebel movement and the fact that their historical leader Abbe Senghor (1928-2007) could not sign on behalf of the different scattered wings.

Table 1. Essential power-sharing elements of recent African peace agreements (since 1999)10


Power-sharing details

No power sharing

Senegal, 30 December 2004 (Ziguinchor)

Angola, 4 April 2002 (Luanda)

Comoros, 17 February 2001 (Fomboni)

DRC, 19 April 2002 (Sun City)

Somalia, 29 January 2004 (Nairobi)

MFDC (Abbé Senghor), government

One-dimensional power sharing

Army representatives of government and UNITA

Nine key politicians of Moheli, Anjouan and Grande Comore/OAU, OIF, EU as guarantors

Government, MLC, RCD-ML, RCD-N, six different Mai-Mai groups, 19 political parties, 45 representatives of civil society

Eight out of 38 invited factions

Military: Integration of UNITA officers and soldiers into the national army; the rest to be demobilized. (Elements of political power sharing contained in earlier agreements.)

Political: Creation of a tripartite commission to decide on transition modalities. After the transition: government of national union, equal representation.

Political: In the transition phase Kabila remains President of the republic; Bemba (MLC) becomes Prime Minister; president of National Assembly goes to RDC; and president of the Senate to the unarmed opposition. Signatories to the agreement to present candidates for government positions. General dispositions to fill transitional National Assembly and Senate by signatories.

Political: Traditional elders of the four major clans will select 61 MPs and a coalition of small clans will select 31 MPs -transitional parliament to elect President.



Power-sharing details

Multi-dimensional power sharing

Burundi, 8 October 2003 (Pretoria)

Transitional government of Burundi and CNDD-FDD

Central African Republic, 21 June

Congo, 16 November 1999, 29 December 1999 (Pointe Noire/ Brazzaville)

Chad, 14 December 2003 (Ouagadougou)

Chad, 24 December 2006 (Tripolis)

APRD, UFDR, government, Gabon's Minister of Foreign Affairs

CNR, government

Abbo, MDJT faction leader exiled in Burkina Faso/Chad's Minister of Security (Moussa)

FUC, government/Libya

Political: Executive: CNDD-FDD gets four ministries. Legislative: CNDD-FDD gets second Vice-President and deputy secretary-general of the bureau, plus 15 seats. Military: Integrated general staff and officer corps, to be composed of 60% officers selected from the government army and 40% officers from the FDD. The allocation of command posts shall be on the basis of ethnic balance (50-50). (Similarly: police.)

Territorial: Provincial level: CNDD-FDD gets three governors' positions plus five advisers. Local level: CNDD-FDD gets 30 administrators. Economy: CNDD-FDD heads 20% of public enterprises.

Political: Participation of rebel representatives in 'management of state affairs' (vague).

Military: Integration of rebel fighters into the national army (vague).

Political: Creation of a follow-up committee to prepare national dialogue (weak).

Military: Limited integration of rebel combatants into the national army.

Political: Transformation of MDJT into a political party, participation in government (vague). Mlitary: Integration of rebels into the national army (vague).

Political: Participation of FUC in management of state affairs (vague). Joint follow-up committee to define details. Military: Reintegration of FUC combatants into the defence and security forces (vague).

Chad, 25 October 2007 (Sirte)

Comoros, 23 April 1999 (Antananarivo)

Côte d'Ivoire, 14 January 2003 (Linas-Marcoussis)

Côte d'Ivoire, 4 March 2007 (Ouagadougou)

RFC, CNT, UFDD, UFDDF, government/Sudan, Censad

Representatives of Moheli and Grande Comore/OAU, UN, and Arab League as guarantors

FPI, RDR, PDCI (major political parties); PIT, UDCY, UDPCI, MFA (smaller parties); MPCI, MPIGO, MJP (rebels)/France

Gbagbo (head of state), Soro (head of the rebellion)/ Compaore (President of Burkina Faso, mediator)

Political: Joint committee to determine share of rebels in executive domains on all state levels (vague). Military: Integration of rebel combatants into the defence and security forces (vague).

Political: Large prerogatives with the individual islands. Equitable representation of the islands in government, rotation of the position of President between the three islands (in office for three years each). National Assembly elected 50% in direct elections, 50% elected by the Island Assemblies. Territorial: Significant autonomy for the islands, change of the denomination of the republic (Union of Comoros Islands). Economic: Equitable sharing of resources between the islands and the central government.

Political: Government of National Reconciliation: equal representation of all participating parties. Eligibility: change of constitution to allow Ouattara (RDR) to stand for election. Military: Immediate demobilization of government soldiers recruited after the rebellion; amnesty for officers involved in the rebellion, but no clear power-sharing formula for creation of new army.

Political: Rebel leader Soro becomes Prime Minister (separate decree), a permanent framework of concertation is created with Ouattara and Bedie joining the signatories. Territorial: Redeployment of the administration to the north of the country; no particular measure of power sharing. Military: Creation of a joint command centre (equal shares) to prepare for a joint integrated army.



Power-sharing details

Djibouti, 12 May 2001 (Djibouti)

Liberia, 18 August 2003 (Accra)

Mali, 4 July 2006 (Algiers)

Sierra Leone, 7 July 1999 (Lomé)

Government, FRUD

Government, LURD, MODEL, and 18 political parties

Government, ADC / Algerian ambassador (Gheraieb)

Government and RUF/four heads of state (Liberia, Nigeria, Burkina, Togo), representatives of Côte d'Ivoire and Ghana, UN, OAU, ECOWAS, Commonwealth

Military: Integration of FRUD combatants into the security forces (or compensation).

Territorial: Creation of five (largely autonomous) regions. Political: Transitional government and legislature with warring parties (over)represented.

Military: New armed forces to be drawn from government forces, LURD, MODEL, and civilians. Economy: Major state corporations and agencies shared between warring parties (and the other civilian signatories).

Territorial: Creation of a provisional regional coordination council, acceleration of the decentralization process. Economy: Organization of a development forum for the Kidal region. Military: Mutineers to be included in new special security units.

Political: Government: RUF leader Sankoh becomes Vice-President, one senior ministry (finance, foreign affairs, or justice) plus three other cabinet posts and four deputy ministers to RUF.

Economy: Chairmanship of the Commission for the Management of Strategic Resources, National Reconstruction and Development to rebel leader Sankoh. Military: Ex-combatants of the RUF/SL, CDF, and SLA integrate new army according to some criteria. Recruitment into the armed forces shall proportionally reflect the geo-political structure.

Sudan, 9 January 2005 (Naivasha)

Government, SPLM/A

Political: Executive: Transitional National Executive with 52% of seats held by NCP, 28% SPLM, 14% other northern political forces, 6% other southern political forces. Head of state remains in place; First Vice-President is SPLA leader Garang. Legislative: Transitional Assembly with 52% of seats held by NCP, 28% SPLM, 14% other northern political forces, 6% other southern political forces. More favourable terms for NCP and SPLM on the state level (70%). Territorial: Southern Sudan becomes an autonomous region for 6 years; referendum about secession option to be held afterwards. Decentralization with significant devolution of powers to state and local level.

Economy: Sharing of oil incomes from oil produced in southern Sudan on a 50-50 basis.

Military: North and South with separate armies operating only in their territory. Joint force of 21,000 for three disputed territories.

Source: Compiled by the author based on the actual texts of the peace agreement (where available; see footnote 10) and media reports (where not).

M V in

10Some agreements are based on earlier texts or are compilations of texts (in the case of Sudan); only the main agreements are listed. Text versions of peace agreements available for Angola, <>, Burundi, < pa/burundi/burundi_10082003.html>, CAR, <>, Chad, <>, <>, and < 20de%20Syrte.pdf>, Comoros, <> and <>, Cote d'lvoire, <> and <http://>, DR Congo, <>, Djibouti, chttp://>, Liberia, <>, Mali, <>, Sierra Leone, <>, and Sudan, <> (all 29 December 2008). No simple witnesses or facilitators (not signing the agreement) are recorded.

whom. Obviously, only those present at the negotiation table could really count on being included. In many peace agreements the government and individual rebel movements sign the final document, while civilian parties and other actors are left out. This is then more often than not translated into a power-sharing formula that excludes the non-signatories from the exercise of power or gives them a disproportionately small share. Warring parties are overrepresented in government. At times some coordinating positions (for example, Prime Minister) are given to civilian parties or independent

candidates, but usually without the means to pursue independent policies. Q

What makes peace agreements successful? Only a few of the listed agree- W

ments can be termed successes - the best bets are Burundi (2003), Liberia a

(2003), Sudan (2005), and Cote d'lvoire (2007). But does power sharing |

a prominent role in explaining these successes? It is not clear. An elaborate §

multidimensional power-sharing arrangement is not per se a success factor, ^

as Comoros (1999) and Cote d'lvoire (2003) show. After all, peace is when /

people think they are at peace. The perceptions of the population are the r

best indicator of successful and failing peace settlements. But these data are §

missing and are usually not available. One thing seems to be clear: superfi- §

cial negotiations a la Chad will not bring peace closer. In the case of Sierra §

Leone, it was not the peace agreement and its power-sharing content, but g

rather peace enforced by - mainly British - intervention forces that proved .

the key to a late success. Another prominent example: the relative weight /

for peace of the 2002 agreement in Angola is limited when compared to A

the military defeat of UNITA rebels. Africa's recent experience with power §§

sharing is far from a success story. U

The picture gets clearer if we look in some detail at a small selection of e

cases, which also permits us to see that power sharing is a dynamic process. St


The military stalemate in the first Liberian civil war formed the background of peace negotiations. After inconclusive agreements and meetings in Cotonou, Benin (1993), Akosombo, Ghana (1994),12 and Accra, Ghana (1995), a major breakthrough was recorded.13 In August 1995 the main

11. David Keen, Conflict and Collusion in Sierra Leone (James Currey, Oxford, 2005), p. 273; Michael Kargbo, British Foreign Policy and the Conflict in Sierra Leone, 1991-2001 (Lang, Bern, 2006), p. 320.

12. Akosombo included major power-sharing arrangements: Taylor, Kromah, and the commander of the Armed Forces of Liberia, Hezekiah Bowen, should have become members of a Council of State (five members). The text of the agreement can be accessed at <> (29 December 2008). The agreement was co-signed by Rawlings and a UN representative.

13. Dorina A. Bekoe, 'Toward a theory of peace agreement implementation: the case of Liberia' in Rose M. Kadende-Kaiser and Paul J. Kaiser (eds), Phases of Conflict in Africa

warring factions signed an agreement brokered by Ghana's President Jerry Rawlings in Abuja. Charles Taylor's National Patriotic Front of Liberia (NPFL), the strongest military organization, agreed to a ceasefire. A ruling council of six members was formed, including three main warlords -Charles Taylor, Alhaji Kromah of the Krahn faction of the United Liberation Movement of Liberians for Democracy (ULIMO-K) and George Boley of the Liberia Peace Council. Other factions were included in a government that proved unable to run the affairs of the country effectively.14 However,

hostilities continued and were only stopped after a further agreement in Q

Abuja, Nigeria (1996). g

Amos Sawyer, a prominent Liberian intellectual and former interim Pres- a

ident, criticizes the peace settlements negotiated to end the violence in d

Liberia (1996) and Sierra Leone (1998), particularly for their power-sharing §

content.15 Both agreements established governments that were 'substan- ^

tially, if not totally, controlled by armed groups whose leaders could hardly /

find in such arrangements sufficient incentive to blunt their greed and r

ambition'.16 Peace brokered this way was not sustainable. When Charles §

Taylor won elections in 1997, partly through intimidation, he could con- §

tinue his warlord politics as elected President. War broke out again in 1999 §

when a Guinea-backed rebel group, Liberians United for Reconciliation and g

Democracy (LURD) emerged. The rebellion was later joined by a second .

military group, Movement for Democracy in Liberia (MODEL), operating /

from Cote d'Ivoire. 1

The renewed civil war ended in 2003 with the LURD rebels close to §

a military victory and the negotiated abandonment of power by Charles U

Taylor (ceasefire agreement on 17 July 2003 in Accra). The three warring e

parties plus 18 civilian party representatives met in Akosombo and Ac- St

cra, from 4 June 2003 to 18 August 2003, within the framework of the 0

ECOWAS Peace Process for Liberia, under the auspices of Ghana's Pres- p

ident John Kufuor, chairman of ECOWAS, and under the mediation of g

General Abdulsalami Abubakar, former head of state of Nigeria. The Accra 1

(de Sitter Publications, Toronto, 2005), p. 113, counts 16 peace and ceasefire agreements between 1990 and 1997. She details all power-sharing arrangements and gives an inside view of motivations of different armed factions to accept or refuse arrangements. She concludes that the incremental steps taken by warring factions count most as they are seen as a concession by the other parties, who may then match them (p. 131). This may explain the relative success of the final Abuja arrangements.

14. Text available from <> (29 December 2008).

15. Amos Sawyer, 'Violent conflicts and governance challenges in West Africa: the case of the Mano River basin area', Journal of Modern African Studies 42, 3 (2004), p. 454.

16. Ibid., p. 451. Jarstad, 'Power sharing for peace and democracy?', p. 23, recalls that in the case of Sierra Leone the Lome accord of 1999 simply reflected the enhanced bargaining position of the RUF rebel group due to its military capacity. Consequently, the vice-presidency was allotted to its leader, Foday Sankoh. Only in 2002 could the war be declared over.

Agreement17 contained some obvious power-sharing arrangements. While Taylor's Vice-President, Moses Blah, was allowed to run government affairs for a short period until October, the transitional institutions were clearly designed according to power-sharing prerequisites. No representative of a warring faction could hold the position of chairman or vice-chairman in the National Transitional Government of Liberia (NTGL). In the end the independent businessman Gyude Bryant became head of the executive branch of government. However, 15 out of 21 cabinet posts were allocated between the two rebel movements and the former Taylor government - again, a Q

strong majority for the warring factions. The Taylor side retained internal W

affairs, defence, planning and economic affairs, health and social welfare, a

and post and telecommunications; LURD got transport, justice, labour and d

finance, and the Ministry of State; MODEL would have agriculture, com- §

merce, foreign affairs, public works and land, mines and energy. Taylor's ^

Defence Minister was thereby allowed to continue in his job. In the legis- /

lature some similar problems were encountered. The National Transitional rg

Legislative Assembly was composed of the three warring parties (12 seats §

each), political parties (18), civil society and special interest groups (7), and §

one representative each for the different counties (15). A disputed LURD §

leader was made the interim speaker of the Legislative Assembly. This was g

meant to be an interim arrangement, but it showed what the agreement .

was all about: 'That the persistent focus at Accra on jobs, cars and money /

rather than the challenges confronting Liberia gives a clue to the character A

of the transitional government.'18 The heavy involvement of peacekeepers §§

later permitted the rushing of national elections, to be held by the end of U

2005, under a new constitution. Those elections permitted several key fig- e

ures of Taylor's regime and warlords like Prince Johnson to become elected St

legislators - and thereby gain immunity. Taylor, by contrast, was indicted 0

by the UN-sponsored Special Court for Sierra Leone, and finally ended up p

behind bars in The Hague. g

On the military level the National Commission for Disarmament, Demo- 1

bilization, Rehabilitation, and Reintegration was formed, with all warring 2

parties contributing. The Accra Agreement determined that forces could be 4 drawn from the ranks of the then Government of Liberia forces, the LURD and the MODEL, 'as well as from civilians with appropriate background and experience'. Economic power sharing, on the surface, was not the main purpose of the Accra Agreement, but Article 25 noted:

Allocation of ministerial positions, deputy and assistant ministerial positions, headship of autonomous agencies, commissions, public corporations and state-owned

17. <> (29 December 2008).

18. International Crisis Group, 'Liberia: security challenges' (International Crisis Group Africa Report No. 71, Brussels, 3 November 2003), p. 4.

enterprises shall be made to the Parties to this Agreement through a process of negotiation. The allocations as agreed to by the Parties are contained in Annex 4 attached to the Agreement. Annex 4 is an integral part of this Agreement.

Even more worrying than the distribution of political positions was the decision to give each warring faction key public corporations, which would allow them to siphon off state resources and thereby fund whatever they pleased. This annex shared the spoils between the warring parties at a ratio of four state corporations and two national agencies to each warring party.

The power sharing involved in the Accra Agreement has been heavily o

criticized. In the view of Amos Sawyer, 'fixing the central state is important n

but insufficient.... Authority must be constitutionally shared at other levels g

of government and local people must become empowered participants.'19 S

The security architecture proposed by Sawyer would be organized across o

borders and involve, where appropriate, religious bodies and community tr

militia units. This seems plausible as numerous local-based ethnic disputes /

have fuelled the civil war at the national level. r

Liberia's history of externally brokered agreements has plenty of power- o

sharing devices involving the top level of warring parties. They may have r

been partly conducive to peace at a secondary level, but the negative expe- o'

rience of the settlement of the first civil war raises doubts. Only the strong |

and active presence of international peacekeepers - and not power sharing r

during the transition period - achieved a respectable degree of stability and a

security.20 >

Cote d'Ivoire

The list of negotiations and negotiators engaged to end Cote d'Ivoire's civil war is long. After a first round under the aegis of ECOWAS in 2002,21 a complex attempt was made by the country's former colonial power, France, leading to the much-cited agreement of Linas Marcoussis (23 January 2003). Participants in the negotiation were all political parties represented in Parliament or government, plus the three rebel organizations Mouvement Patriotique de la Cote d'Ivoire (MPCI), Mouvement Populaire Ivoirien du Grand Ouest (MPIGO), and Mouvement pour la Justice et la Paix (MJP).22 The agreement is a very interesting document as it addresses most of the

19. Sawyer, 'Violent conflicts and governance challenges', p. 454.

20. Festus Aboagye and Alhaji Bah, 'Liberia at a crossroads: a preliminary look at the United Nations Mission in Liberia (UNMIL) and the protection of civilians' (ISS Paper 95, Institute for Security Studies, Pretoria, 2004).

21. For an account of the ECOWAS negotiation attempts see 'Report of the Secretary-General on Cote d'Ivoire' (UN Security Council, S/2003/374, 26 March 2003) p. 3ff.

22. MPIGO and MJP surfaced in late 2002 in the western part of the country; in 2003 they merged with MPCI to become the so-called Forces Nouvelles.

salient political problems of the country, including the question of citizenship with its ramifications for electoral eligibility, human and civil rights, and land ownership.23 There is no doubt that some of those solutions were not wholeheartedly supported by all those who signed the agreement. But the agreement failed because of (1) the role of France as a mediator and (2) the power-sharing formula imposed shortly afterwards at the so-called Kleber meeting in the French capital. It was simply unacceptable to Abidjan (and particularly the army leadership) that the rebels should get the Defence

and the Interior portfolios, ministries that would give them preponderance Q

on all security issues. The preceding step, making the rebels presentable by W

inviting them all to negotiations in Marcoussis, was already difficult to bear a

for some regime hardliners, and indeed gave the rebels a political weight d

they previously did not have. The presidential camp distanced itself im- §

mediately from the agreement upon the delegation's return, declaring the ^

text a simple draft needing refinement. Therefore, the whole agreement - /

despite its interesting ingredients - had from the outset very poor prospects rg

of implementation. Some concrete details of the formation of a government §

of national unity were left to a new summit in Accra (Ghana) on 7 March §

2003, when the Forces Nouvelles got two senior ministries (Territorial Ad- §

ministration and Communication). Additionally, a 15-strong National Se- g

curity Council was established with all parties represented.24 This could §

be interpreted as a second layer of power sharing (military establishment), /

the first being the share of government positions. The consensual Prime A

Minister Diarra was able to build a grand coalition including ten ministries §

for the FPI, seven each for RDR and PDCI, while nine went to the rebel U

union Forces Nouvelles plus six to smaller parties. A compromise was found e

regarding the Interior and Defence ministries, which went to technocrats. St

However, neither the National Security Council, nor the independent Prime 0

Minister, together with the handful of technocrats in government, were able p

to influence government policy in key respects. g

More popular concerns were not addressed. The country remained di- 1

vided and this division was frozen by the presence of UN, ECOWAS, and 2

French peacekeepers, although they were not able to prevent several erup- 4 tions of violence in the west of the country, the so-called confidence zone, and in Abidjan over the following years.25 The territorial division was in no

23. All contained in an annex devoted to the programme of the Government of National Reconciliation. A good analysis of the agreement can be found in Jessica Kohler, 'From miraculous to disastrous: the crisis in the Cote d'Ivoire' (Centre for Applied Studies in International Negotiations, Geneva, August 2003), <> (29 December 2008).

24. <> (29 December 2008).

25. Peacekeepers can claim a far less positive balance sheet than in Liberia; see Andreas Mehler, 'Positive, ambiguous or negative? Peacekeeping in the local security fabric', Critical Currents 5 (2008), pp. 41-65.

way the result of a power-sharing agreement, but obviously mimicked some effects that might have been gained from regional autonomy, as the northern part was no longer under central administration. Even essential commercial flows were controlled by a couple of rebel leaders for at least four years. In the following period one camp did not agree to disarm, while the other would not agree to change the constitution to allow for fair elections.

The interim agreements of Accra, on 30 July 2004 (in the name of ECOWAS), as well as those mediated by South Africa's President Thabo

Mbeki in the name of the African Union (Pretoria I + II, 6 April and 29 °

June 2005) could not resolve the deadlock. The signing of the Agreement W

of Ouagadougou (4 March 2007) apparently created a much more viable a

power-sharing arrangement.26 However, with all the previous failures, it g

appears premature to be optimistic about long-term effects. What are the §

main differences between the agreements of Linas-Marcoussis and Oua- ^

gadougou? The first peace plan was put on the table and pushed through in /

about eight days. As good as the intentions may have been, it was rushed. r

It lists important conditions for sustainable peace, such as the revision of §

eligibility rules and the revision of land ownership, but the 'implementa- §

tion rules', that is, the concrete power-sharing formula for a government §

of national reconciliation in the following meeting, were not realistic. The g

Agreement of Ouagadougou was signed by Gbagbo and Soro after a month .

of intense negotiations (and still longer preparation) in what was called /

'direct dialogue'.27 The mediator, Blaise Compaore, President of Burkina A

Faso, himself the third signatory, is an 'insider' in contrast to the mediators §§

of all preceding agreements. Soro was appointed Prime Minister shortly U

after the agreement and presented a new government. e

The rapid ascension of a political entrepreneur from student leader to tit-

rebel leader and now head of government therefore reinforced - in the 0

name of power sharing - the signal to other aspirant youth leaders that p

'violence pays'. Important questions remained unresolved: what to do with g

local rebel commanders and militia leaders close to the President?28 How 1

to dismantle the entrenched war economy in both north and south? What 2

about the local theatres of violence over the last couple of years? And how 4 to deal with the deep mutual distrust inside the population? While Linas-Marcoussis singled out more root causes of the armed conflict, it looked less realistic than the Ouagadougou agreement that more clearly paid off

26. French text of the agreement from <> 12 May 2009.

27. Background reports can be found in Jeune Afrique, 11-17 March 2007, 25-31 March 2007. See also Le Monde, 13 March 2007.

28. See Jeune Afrique, 20-26 May 2007; Africa Research Bulletin, 1-30 June 2007, pp. 17-122; Marches Tropicaux et Mediterraneens, 6 July 2007, p. 48.

for rebel leader Soro. More people believed that this agreement could bring peace nearer. Here the process mattered more than the content.

Kenya might be termed an atypical case, as the agreement reached was not called a peace agreement (and therefore is not included in Table 1) and the deadly clashes, with over a thousand deaths, were not described

as a civil war. On 27 December 2007, the second presidential and parlia- o

mentary elections in the post-Moi era were held. Raila Odinga's Orange g

Democratic Movement (ODM) won 99 out of 210 parliamentary seats, g"

while the Party of National Unity (PNU), led by President Mwai Kibaki, g

won only 43 seats. However, Kibaki was declared winner of the presidential 0

elections. According to the official figures he held an advantage of 3 percent tr

over his opponent Odinga.29 The ODM, in line with international election /

observers, claimed fraud. Violent riots swept across Kenya after the an- gg

nouncement of the election results.30 Only two out of eight provinces were X

spared, but the local conflict scenarios varied strongly. Rift Valley, Nyanza, &

Western Province and Nairobi were most seriously affected. Around 1.200 o'

people died and officially 350,000 were displaced before a power-sharing g

agreement was signed between Kibaki and Odinga on 28 February 2008. 0

Several attempts at mediation were made after the outbreak of hostilities a

in early January 2008. The most prominent was by John Kufuor, Ghana's A

President and outgoing Chairperson of the African Union. However, o

Kufuor failed to arrange a meeting between Kibaki and Odinga. On 25 o

January, former UN Secretary-General Kofi Annan brokered face-to-face &

negotiations between the rivals. Annan mediated in the name of the African y

Union's Panel of Eminent Africans. Benjamin Mkapa (former President g

of Tanzania) and Graca Machel (wife of Nelson Mandela) had oversight g

roles, while the Kibaki camp rejected - for alleged links to Odinga - the J

participation as a chief negotiator of South African business tycoon Cyril 8

Ramaphosa, who had played a lead role in negotiating the end of apartheid 0

in South Africa.31 Tanzanian President and new AU chairman Jakaya 4

29. <> (29 December 2008). ODM officially recorded 47.1 percent of the votes at the legislative elections (PNU 20.5), but Odinga just 44 percent at the presidential contest (Kibaki 47, the third candidate Kalonzo 9).

30. Recent analyses have attributed the post-electoral violence to a large extent to elite fragmentation and elite struggles; see 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; and Daniel Branch and Nic Cheeseman,'Democratization, sequencing, and state failure in Africa: lessons from Kenya', African Affairs 108, 430 (2009), pp. 1-26.

31. The Guardian, 'Kenya government forces out key mediator in political talks', 4 February 2008, <> (29 December 2008).

Kikwete joined the mediation team at a critical moment on 26 February.32 Negotiations between the Kibaki and Odinga camps lasted for nearly 40 days. Each of the contestants appointed a four-person negotiating team of high-ranking party officials. Outside pressure to come to a quick solution was high. Donors threatened to withdraw their assistance, while the powerful Kenyan tourism industry and Kenya Association of Manufacturers pushed for a solution to the crisis. A number of intermediate agreements were signed on 1, 4, and 14 February.

The final agreement signed by Kibaki and Odinga is a short document, °

which contains the essentials of the power-sharing deal: The PNU and the W

ODM should form a government of national unity, a grand coalition. The P

composition of the Cabinet should reflect the parliamentary strength of g

the coalition members. The post of Prime Minister was newly created. And §

the leader of the largest party or the governing coalition in the National ^

Assembly was to be named Prime Minister and as such authorized to coor- /

dinate and supervise the execution of the functions and affairs of the govern- g

ment. Odinga thus became Prime Minister. Concomitantly, there would be §

two deputy prime ministers, one from each member of the coalition. Fur- §

thermore, the contestants signed several side agreements to establish sepa- §

rate commissions:33 a Constitutional Review Commission; a Commission g

of Inquiry on Post-Election Violence (CIPEV) to investigate the violence .

that took place after the poll; a Truth, Justice and Reconciliation Commis- /

sion (TJRC); and an Independent Review Commission (IREC) to examine A

long-standing sources of grievance. On 18 March 2008 the Kenyan Parlia- §

ment unanimously passed the key pieces of the settlement: The National U

Accord and Reconciliation Bill (which outlines the concept of power shar- e

ing) and a Constitutional Amendment Bill (which creates the post of the it

Prime Minister and of the two deputy prime ministers). 0

The core document had several flaws. Most importantly, it fell short of p

specifying the functions of the Prime Minister. Thus it was no surprise that g

neither camp could agree on their roles in the following months. In several 1

instances, Kibaki proved reluctant to cede government authority to Odinga. 2

The concrete formation of a government was not included in the text of the 4 agreement. As a consequence, on 8 April the announcement of the suspension of the talks over the coalition Cabinet's composition resulted in violence in Kibera, a Nairobi slum. Further sporadic use of violence by armed militias was recorded after the signing.34 The character of the agreement was an

32. Pambazuka News, 'How Annan magic worked to seal Kibaki-Raila deal' <http://www.> (29 December 2008).

33. For the text of the agreements, see <> (29 December 2008).

34. United States Institute of Peace, 'Kenya: setting the stage for durable peace?' (United States Institute of Peace, April 2008), < 2008/0415_kenya.html> (29 December 2008).

elite pact. The AU was in fact criticized for not 'reaching out to civil society organizations' 35 during the mediation of the agreement. This was however remedied to some extent by the serious work of the CIPEV, calling upon non-governmental organizations (among others) to 'provide information to the Commission of Inquiry related to post-election violence'.36 The agreement itself focused on institutional engineering, and failed to address root causes of conflict such as land tenure and human rights abuses. Some of this work was left to the different commissions created. The CIPEV report was published on 15 October 2008 and detailed responsibilities for the violence on both the government and the ODM side - without naming names.37 It may prove to be a basis for more meaningful peace arrangements at the local level.

Conclusion j|

Power-sharing agreements offer no miraculous solutions to complex crisis &

situations. Further evidence is provided by the Zimbabwean case, where an 0 uneasy process continued at the time of writing. A power-sharing agreement & was signed on 15 September 2008 between the ruling Zimbabwe African 0 National Union/Patriotic Front (ZANU-PF) and the two Movement for g Democratic Change (MDC) formations. The agreement was meant to end o a severe crisis after the contested first round of presidential elections in a March 2008. The key element of the deal, mediated by South Africa's former President Thabo Mbeki, was a government of national unity. According 0 to the agreement the executive powers would be shared between the two g main protagonists: Robert Mugabe, winner of a meaningless second round e of elections, would remain President of Zimbabwe and head the Cabinet, whilst Morgan Tsvangirai, leader of the main opposition faction who had g refused to participate in the second round citing security concerns, was des- g ignated as Prime Minister and would chair a council of ministers supervising J the Cabinet. Arthur Mutambara, leader of the second MDC faction, would 1 be appointed Deputy Prime Minister.38 The government would comprise 0 thirty-one ministers, 15 nominated by ZANU-PF, 13 by MDC-T and 3 by 4 MDC-M. Quarrels started anew immediately after the signing of the deal on the concrete distribution of key ministries, not least in the security sphere. Although this is a crucial aspect, it was striking that the contenders were

35. Ibid.

36. <>, (29 December 2008), p. 2.

37. <> (29 December 2008).

38. <> (29 December 2008).

focusing again on predominantly elite concerns. From the beginning this point had been criticized in the public debate. Among others, the Congress of South African Trade Unions (COSATU), through its spokesman Patrick Craven, termed the agreement an 'elite deal negotiated between a few individuals, with no attempt to involve civil society'.39 The Zimbabwe Congress of Trade Union (ZCTU) noted in its deliberations that 'the process used in coming up with the deal was not all-inclusive as the civic society was not given an opportunity to participate'.40 Wafulu Okumu, head of the African Security Analysis Programme at the Pretoria-based Institute for Security °

Studies, pointed out that in the case of Zimbabwe, as well as in Kenya, a W

debate has to be started about whether 'democracy is being sacrificed for P

peace, just like it has been pointed out that justice has been sacrificed for S

peace in Uganda and Sudan'.41 §

Power sharing remains in vogue despite such unresolved issues. Examin- ^

ing a full sample of recent African cases (1999-2008) shows that at least /

some power-sharing devices are present in most peace agreements in Africa, g

successful or not, while the specific aspects (political, military, territorial, §

economic) can vary. The lion's share of top positions in power-sharing §

governments routinely goes to the groups represented at the negotiation §

table - and those are mostly the warring parties. g

More specific lessons can be drawn from individual cases examined. In the §

Ivorian case, the Linas-Marcoussis (2003) and Ouagadougou agreements /

(2007) gave very different signals, although the latter agreement is formally A

based on the former. While Linas-Marcoussis singled out root causes of the §§

armed conflict, it looked less realistic than the Ouagadougou agreement, U

that more clearly paid off for rebel leader Soro. More people believed that e

this agreement could bring peace nearer. The negotiation process and its it

participants mattered more than the content. Liberia's history of externally 0

brokered agreements, including the 2003 Accra agreement, is studded with p

power-sharing devices involving the top level of warring parties.42 Some g

of the provisions are very detailed and handed sinecures to former rebel 1

leaders, much to the anger of civilian leaders. However, it was the strong and 2

active presence of international peacekeepers - and not the power sharing 4 formula enacted during the transition period - that achieved a respectable degree of stability and security. The Kenyan power-sharing deal was never said to be part of a peace agreement. The text itself and its interpretation show all the difficulties inherent in vague formulations. The major difference

39. <> (29 December 2008).

40. Links, 'Zimbabwe Congress of Trade Unions: power-sharing deal "a far cry" from expectations', <> (29 December 2008).

41. < ZIMSEMREPSEP08.PDF?link_id=14&slink_id=6608&link_type=12&slink_type=13&tmpl _id=3> (29 December 2008).

42. Sawyer, 'Violent conflicts and governance challenges'.

between this and many other negotiation processes lies in the potential effectiveness of side aspects of such a deal, in this case the creation of diverse commissions who would dig deeper over a substantially longer period of time and propose more differentiated solutions.

Although cases differ in the form conflict and civil war has taken, there is one striking commonality in most recent African peace agreements: the readiness of the international community to advocate or facilitate a strong power-sharing ingredient, which is to the benefit of armed rebel movements (or their leaders). This seems not to be a function of the strength of interna- d

tional engagement, as power sharing could be a cheap solution (for example, w

in the Central African Republic) or a rather expensive and transitional one a

(in Liberia). In Cote d'Ivoire and many other places (DRC, CAR, Chad) d

civilian political parties have suffered a setback not only because of war, but o

also because of the way peace was negotiated - from outside and above. h

In examining more closely the realities in the three cases of Liberia, Cote /

d'Ivoire, and Kenya, it becomes clear that there are even more aspects of rr

power sharing that need attention and which would profit from a 'large-N' o

examination - beyond the African cases. But the argument of this article °

also invites researchers to provide more in-depth case studies, as variations o

are so important. First we need to investigate who is defining the constitu- n

tive parts of the polity. Then, what are the 'ethnic' groups which supposedly need accommodation?43 Who is entitled to speak for such a community? And, above all, who is invited to the negotiation table? After all, this analysis shows the carelessness of outside actors when deciding upon those issues, either by inviting pre-war political formations with unclear popular credentials or, even more frequently, by bringing in vociferous political-military entrepreneurs and rebel organizations with one main quality: spoiling capacities. An effort to ask communities about their sense of the problems and the legitimacy of those sitting at the negotiation table, sometimes explicitly in their name, cannot be detected. This is only different when contested elections are the triggers of conflict escalation, as in Kenya (and Zimbabwe). The second problematic aspect of the settlement practice is the systematic assumption that the conflict is ethno-political in nature (particularly when it comes to Africa). While Burundi's civil war has an obvious ethnic colouration and significance, this is much more difficult to assert in cases such as Comoros or the Central African Republic. Third, the local level not only of (1) conflict generation and (2) escalation, but also of (3) conflict management and (4) security production is more often than not completely neglected in peace negotiations. In such cases, it cannot be denied that national power sharing will fail to bring a solution at the local level and may

43. Walker Connor, 'A few cautionary notes on the history and future of ethnonational conflicts' in Andreas Wimmer et al. (eds), Facing Ethnic Conflicts: Toward a new realism (Rowman and Littlefield, Lanham, MD, 2004), pp. 23-33.

even impede local solutions to problems that were at the root of an armed conflict. Fourth, the perceptions of the population as to whether there is/was a violent conflict and whether it was interrupted, terminated, or settled (or whether it continues) are obviously important, not least for the decision of whether a peace settlement was successful. But this issue is rarely analysed, so we are devoid of reliable data.

Therefore, the entire discussion about power sharing and sustainable peace looks inherently flawed when it does not focus on what peace is all about: security for the people. Local sustainability might also be the real test Q

of one of the most recent major power-sharing agreement on the continent: W

In Kenya, no single provision touches on the local level of security provision, a

for example, in the hardest-hit zones of violence in Rift Valley, Kisumu and d

some poor neighbourhoods of the capital, Nairobi.44 This may turn out to o

be problematic. After all, only tailor-made, intensely debated, and maybe ^

only transitional power-sharing arrangements can have beneficial effects. /

Such effects will by no means automatically arrive when a peace agreement r

has been signed; good intentions alone do not provide for peace. 0

44. See the text of the final parliament-endorsed agreement < Downloads/Bills/THE%20NATIONAL%20ACCORD%20AND%20RECONCILIATION %20BILl,%202008.pdf> (29 December 2008).