bq. Civil war settlements create institutional arrangements that in turn shape postsettlement politics among the parties to the previous conflict. Following civil wars that involve competing nation-state projects, partition is more likely than alternative institutional arrangements—specifically, unitarism, de facto separation, and autonomy arrangements—to preserve the peace and facilitate democratization. A theory of domestic political institutions as a constraint on reescalation of conflict explains this unexpected relationship through four intermediate effects—specifically, the likelihood that each institutional arrangement will reinforce incompatible national identities, focus the pursuit of greed and grievance on a single zero-sum conflict over the allocation of decision rights, empower the parties to the previous conflict with multiple escalatory options, and foster incompatible expectations of victory. The theory’s predictions stand up under statistical tests that use four alternative datasets.
That is from a new article by Thomas Chapman and Philip Roeder. (Can anyone find an ungated version?) Various people have advocated partition as the best answer to the ongoing civil strife in Iraq. While Chapman and Roeder are cautious in drawing implications for Iraq, their evidence of partition’s benefits deserves consideration. See, however, the differing perspective of Nicholas Sambanis.
Addendum: see also this paper by James Fearon. [Hat tip to Holger Schmidt]