Power-Sharing Agreement

A power-sharing agreement allocates governing authority (executive seats, legislative quotas, territorial autonomy, fiscal control, and security-sector composition) across previously warring parties or recognized constituencies, as a deliberate substitute for winner-take-all politics during and after a transition.

Context

A power-sharing agreement appears when a conflict ends without a clear victor and the parties have to govern together, or at least govern in parallel, before they trust one another to lose an election. The settlement cannot run on simple majority rule, because the side that would lose the first vote has the guns, the territory, or the constituency to refuse the result, and it won’t accept a count it can overturn by force. So the agreement writes shared authority into the transition itself.

The working typology comes from the comparative record. The PA-X coding of peace agreements treats power-sharing as four distinct axes, and the distinction matters because each axis fails differently. Political power-sharing distributes executive offices, cabinet seats, and legislative quotas. Territorial power-sharing devolves authority to regions, sets up federal or autonomous arrangements, or draws administrative boundaries around a group. Economic power-sharing allocates control over revenue, natural resources, central-bank or budget authority, and reconstruction funds. Military power-sharing composes the post-war security sector: who commands which forces, how former combatants integrate, and whether parallel chains of command survive the settlement.

Underneath the four axes sit two competing design philosophies the field has argued about for forty years. The consociational school, associated with Arend Lijphart, builds the settlement out of a grand-coalition executive, a mutual veto for each major group, segmental autonomy, and proportional representation. Authority follows group identity, and the design protects each group from being outvoted. The integrative school, associated with Donald Horowitz, distrusts identity-locking and engineers incentives for cross-group coalition instead: alternative-vote or single-transferable-vote electoral systems, vote-pooling, and rules that reward a politician for reaching across the cleavage rather than mobilizing within it. Most real agreements are hybrids, and a working drafter needs to know which instinct a given clause is serving.

Problem

The parties need to share authority enough that no side fears annihilation by ballot, but not so much that the settlement freezes the conflict’s cleavages into the permanent structure of the state.

That is genuinely hard, because the two failure modes point in opposite directions. Share too little, and the losing side reads the transition as a slow-motion defeat and walks, or rearms, or boycotts the institutions until they collapse. Share too much, handing every group a veto, ring-fencing every ministry, guaranteeing every quota in perpetuity, and the state acquires so many blocking points that ordinary governance stalls. Civic identities that cut across the war’s cleavages find no electoral path, and the arrangement that was meant to end the conflict becomes the conflict’s monument. Lebanon’s confessional system and Bosnia’s tripartite presidency are the standard cautionary references: settlements that bought peace by institutionalizing the very divisions they were supposed to transcend.

Forces

• Reassurance competes with governability. A mutual veto reassures a minority that it cannot be steamrolled; the same veto, multiplied across groups, can make the cabinet unable to pass a budget.
• Identity-protection competes with civic re-alignment. Ethnic or sectarian quotas guarantee a seat at the table, but they also lock the electorate into the war’s categories and starve any cross-cutting, issue-based politics of oxygen.
• The transitional bargain competes with the permanent constitution. Clauses written to get fighters out of the field are often inherited, unexamined, into the standing constitutional order, because no one writes the sunset and no group volunteers to surrender a guarantee.
• Inclusion competes with capture. The case for adding one more group to the grand coalition is always concrete and immediate; the case against the cumulative veto load is diffuse and shows up only later.
• Military integration competes with security guarantees. Merging former combatants into one army reassures civilians and ends parallel command, but it also asks each commander to dissolve the force that is their insurance against betrayal.

Solution

Treat the four PA-X axes (political, territorial, economic, military) as separate design surfaces, allocate authority on each one deliberately, and write into each clause the answer to a single question: what does this give a party the power to do, and what does it give that party the power to stop?

The first discipline is axis separation. Political seats, territorial autonomy, fiscal control, and security-sector composition are different goods with different timetables and different failure modes, and they should not be bargained as one undifferentiated basket. Bundling them hides the trades. A drafter who keeps the axes separate can see that a group is conceding on the cabinet while quietly capturing the resource-revenue axis, which is the trade Lomé made and the trade the next process should be able to read at a glance.

The second discipline is to distinguish voice from veto on every clause. Proportional representation, a reserved ministry, and a seat on an oversight commission give a group voice and standing. A mutual-veto clause, a concurrent-majority requirement, or a community-of-interest blocking rule gives a group the power to stop the state. Both can be legitimate, but they’re not interchangeable, and a settlement that hands out vetoes wherever it meant to grant voice will discover its blocking points only when it tries to govern. The discipline is to grant the minimum stopping power that delivers the reassurance, and no more.

The third discipline is to name the time horizon of each provision. A power-sharing clause is either transitional (scaffolding meant to come down once elections, security integration, or a constitutional convention can carry the load) or permanent, a standing feature of the constitutional order. The agreement should say which, and for transitional clauses it should name the trigger that retires them. Unmarked, a transitional clause defaults to permanence, because the group it protects will treat any attempt to retire it as a renewed threat.

The fourth discipline is to design the military axis against parallel command. Power-sharing that distributes army or police authority without a real integration timetable, a vetting standard, and a single chain of command institutionalizes exactly the divided coercive capacity the war ran on. A seat in the cabinet that comes with a private brigade is not a share of the state; it is a license to bargain with the state by other means.

How It Plays Out

A government and a rebel movement that fought to a stalemate agree to a transitional grand coalition. The mediation team resists writing “the parties shall share power” as a single principle and instead builds the bargain axis by axis. The political axis gives the movement a proportional cabinet with two reserved security-related portfolios. The territorial axis grants regional autonomy for the two provinces it controls. The economic axis sets a revenue-sharing formula for the mineral districts with an independent audit body. The military axis phases the movement’s fighters into the national army under a joint command with international monitors. Because the axes are separate, the parties can see that the movement is trading deep territorial autonomy for shallow cabinet representation, a defensible trade they couldn’t have evaluated if it had been buried in one omnibus clause.

In a different process, a settlement grants each of three communities a mutual veto over “matters of vital interest,” undefined. Within two years the phrase has been invoked to block a census, a budget, an army-reform bill, and a war-crimes prosecution. Each community reads the others’ vetoes as proof that shared authority is working as designed. Governance does not advance because every contested act is a matter of vital interest to someone. The repair isn’t a new ceremony but a renegotiated, narrowly enumerated list of veto-eligible subjects and a tie-break mechanism, and the renegotiation is far harder than the original drafting, because every group now treats its open-ended veto as an acquired right.

A third agreement writes a transitional power-sharing executive explicitly meant to govern only until a constitutional convention can design permanent institutions. The convention is scheduled, then postponed, then quietly abandoned, and the “transitional” coalition is still allocating ministries by the original wartime formula a decade later. The scaffolding didn’t come down. The clause that was scaffolding has become the building. Practitioners reviewing the case conclude that the failure was not in the transitional design but in the absence of a self-executing trigger: nothing made the convention happen, so the path of least resistance was inheritance.

Consequences

Benefits

• It gives a losing side a reason to accept the result of a war it did not win outright, by guaranteeing that the transition will not be used to eliminate it politically.
• It separates the four allocation axes so drafters, guarantors, and later reviewers can see which part of the settlement is being captured, conceded, or contested.
• It supplies a structured alternative to winner-take-all politics in exactly the settings where majority rule is least survivable: deep cleavage, recent atrocity, no decisive victor.
• The quantitative record is encouraging on the immediate margin: comparative evaluations find that organized violence tends to fall sharply in the period right after a power-sharing agreement, and that the effect can compound where the arrangement holds.

Liabilities

• It can institutionalize the conflict’s cleavages, locking the electorate into wartime identity categories and foreclosing the cross-cutting, civic politics that would let the society move past them.
• Veto-conferring clauses written for reassurance can accumulate into a state that cannot govern, where any group can stop any act it dislikes.
• Transitional clauses get inherited as permanent ones when no exit trigger is written, so the emergency scaffolding becomes the constitutional order by default.
• Military power-sharing without genuine integration preserves parallel command, leaving each party a coercive option that the political settlement was supposed to retire.
• It can become a Spoiler Empowerment vehicle: the same seats, quotas, and resource controls that distribute authority among good-faith parties hand a process-breaker durable purchase on the transition.

Variants

Consociational settlement builds the grand-coalition executive, mutual veto, segmental autonomy, and proportional representation Lijphart described. It is the most common form for deep identity cleavages and the one most exposed to veto-accumulation and identity lock-in.

Integrative design engineers cross-group incentives instead of group guarantees: vote-pooling, alternative-vote or single-transferable-vote systems, and rules that reward coalition across the cleavage. It asks more of the electoral system and the parties’ willingness to compete for the other side’s votes, and it is harder to sustain where the cleavage is freshly violent.

Territorial settlement resolves the conflict mainly through devolution, federalism, or autonomy, sharing authority across space rather than across seats in one central government. It suits conflicts organized around a region or a concentrated minority, and it carries its own secession-versus-integration tension.

Hybrid arrangement mixes consociational guarantees with integrative incentives and territorial devolution, which is what most real agreements actually are. The risk is incoherence: a mutual veto and a vote-pooling incentive can pull the same actor in opposite directions.

Transitional-only power-sharing is scaffolding with an explicit sunset, meant to hold until elections, security integration, or a constitutional convention can carry the settlement. It is the variant most often undone by the absence of a self-executing trigger.

When Not to Use

The pattern also misfires when guarantors treat it as a turnkey template. Power-sharing arrangements are highly sensitive to the specific cleavage, the balance of forces, and the credibility of the integration and exit mechanisms. A consociational design lifted from one settlement and dropped into another can entrench divisions the second society was not actually organized around, producing the lock-in costs without the reassurance benefit.

Related Articles

Sources

• PA-X Peace Agreements Database, PA-X Codebook, Political Settlements Research Programme, University of Edinburgh. The PA-X coding scheme supplies the four-axis typology (political, territorial, economic, military) that lets a drafter keep the allocation surfaces separate and auditable rather than bargaining them as one basket.
• Arend Lijphart, Democracy in Plural Societies: A Comparative Exploration, Yale University Press, 1977. Lijphart’s account of grand coalition, mutual veto, segmental autonomy, and proportionality is the canonical statement of the consociational design philosophy the pattern’s first variant draws on.
• Donald L. Horowitz, Ethnic Groups in Conflict, University of California Press, 1985. Horowitz’s case for incentive-based, integrative electoral engineering is the counterweight to consociationalism and the source for the pattern’s integrative variant.
• Christine Bell, On the Law of Peace: Peace Agreements and the Lex Pacificatoria, Oxford University Press, 2008. Bell’s analysis of how power-sharing clauses travel across processes anchors the warning that a design lifted from one settlement carries its lock-in costs into the next.
• Caroline A. Hartzell and Matthew Hoddie, Crafting Peace: Power-Sharing Institutions and the Negotiated Settlement of Civil Wars, Pennsylvania State University Press, 2007. Hartzell and Hoddie’s comparative study of how multiple dimensions of power-sharing affect the durability of negotiated settlements supplies the empirical case that the four axes operate, and fail, on different logics.
• United Nations, Guidance for Effective Mediation, 2012. The UN guidance frames inclusion and substantive settlement together, and supplies the doctrinal substrate within which a mediator distinguishes a power-sharing arrangement that reassures from one that ransoms the transition.