Abstract

A rich body of theory in political economy suggests that civilian support is central to the success of counterinsurgent campaigns. Civilian collaboration can significantly improve military operations, enhance soldier efficiency, and avoid disruption of costly security infrastructure. Yet there have been few direct tests of the claim that harm to civilians, and who harms them, influences when and with whom noncombatants collaborate. We provide such a test, drawing on newly declassified military records and large-scale survey data. We demonstrate that civilians responded to harm suffered in insurgent-initiated attacks by providing intelligence to security forces in Afghanistan. Moreover, we show that these tips improved the success of subsequent counterinsurgent operations. These results clarify the conditions under which civilian casualties can shape the course of internal wars, with implications for future research on political violence.