Partition as a Solution to Wars of Nationalism: The Importance of Institutions

Is it better to partition a state after it underwent a civil war instead of keeping hostile parties in the same political entity? So far scholars have come up with differing answers to this question. This article is in favor of total partition of a state is the most promising solution to a nationalist civil war. They argue that the institutional arrangements after civil war ends have significant impact on resumption of violence, escalation of conflict as well as democracy level of the new state. They emphasize that the partition has to be implemented fully, half-separation, incomplete population exchanges etc. will worsen the situation. 
The authors talk about four known institutional arrangements. Partition leads to creation of a new sovereign state and separation of parties physically. De facto separations usually occur after truces or ceasefire agreements, and leaves the secessionist group in control of their region. Autonomy is another institutional arrangement that grants self-rule to a region while preserving the original boundaries of the state. Unitarism is an institutional arrangement that brings fighting parties together under a government. According to the authors, former studies often call the latter three arrangements also separation. 
Chapman and Roeder focus on nationalist civil wars meaning the rebelling party believes that they have the right to have their own state. Civil wars of this kind are caused by incompatibility of identities. They argue that “partition is more likely than autonomy or de facto separation to reduce the incompatibility of national identities”. They believe that after partition, incompatibility will fade away over time. But in case of de facto separation or autonomy, incompatible identities will remain and even be reinforced. Under unitarism, they believe, incompatibility of identities is likely to remain in the early stages of post-conflict period but will decline over time. 
They defend partition for several reasons. First, partition reduces the number of issues that has to be made jointly by the former fighters. New state will make its own decision about taxation, education, allocation of resources etc. Second, Partition will eliminate low-cost options to escalate conflict. Therefore, parties will be left with high-cost options to project their resolve, which will be evaded most of the time due to high cost. Third, unless full partition does not take place, sharing coercive capabilities can rise tension under alternative institutional arrangements. 
Their results support their contention. To be cautious about the findings, they test their argument with other data sets. The subsequent tests also reveal that “states created by partition were 1) more likely to be born democratic, 2) even when born nondemocratic were more likely to democratize, 3) were no more likely to experience postindependence ethnic violence, 4) the reemergence of intra-state conflicts as interstate conflicts did not make the relations among successor states more violent than relations among other states”. They also find that even if the new state born out of partition is nondemocratic, it is more likely to democratize faster. On average, it takes about 27 years for a state to democratize if partitioned and 40 years if born out of other arrangements.