How and when armed conflicts end: Introducing the UCDP Conflict Termination dataset

Abstract

Abstract This article presents new data on the start and end dates and the means of termination for armed conflicts, 1946—2005. These data contribute to quantitative research on conflict resolution and recurrence in three important respects: the data cover both interstate and intrastate armed conflicts, the data cover low-intensity conflicts, and the data provide information on a broad range of termination outcomes. In order to disaggregate the UCDP-PRIO Armed Conflict dataset into multiple analytical units, this dataset introduces the concept of conflict episodes, defined as years of continuous use of armed force in a conflict. Using these data, general trends and patterns are presented, showing that conflicts do not exclusively end with decisive outcomes such as victory or peace agreement but more often under unclear circumstances where fighting simply ceases. This pattern is consistent across different types of conflict, as is the finding that victories are more common in conflicts with short duration. The article then examines some factors that have been found to predict civil war recurrence and explores whether using the new dataset produces similar results. This exercise offers a number of interesting new insights and finds that the determinants for civil war recurrence identified in previous research are sensitive to alternate formulations of conflict termination data. The findings suggest that intrastate conflicts are less likely to recur after government victories or after the deployment of peacekeepers. If the previous conflict is fought with rebels aiming for total control over government or if the belligerents mobilized along ethnic lines, the risk of recurrence increases. The discrepancy in findings with previous research indicates the need for further study of conflict resolution and recurrence, for which this dataset will be useful.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 674 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.