Political Influence on Homicide Reports under Civil Conflict

Abstract

This paper tests the link between political ideology and the reliability of homicide data in countries with local conflicts. It examines this question empirically using a regression discontinuity design approach to close-run elections in the context of the Colombian conflict. It finds that homicide reporting in conflict countries is highly political and the potential for manipulation is considerable. It shows that the number of homicides perpetrated by rebel troops increased substantially following close-run elections in municipalities where mayors adopted the incumbent ideology, whereas no impact is observed on the number of homicides perpetrated by any other group. Empirical evidence suggests that the violence reflects either retaliation by the rebels or the alteration of records by official bodies for political purposes.