The (in)effectiveness of torture for combating insurgency

It is commonly believed that torture is an effective tool for combating an insurgent threat. Yet while torture is practiced in nearly all counterinsurgency campaigns, the evidence documenting torture’s effects remains severely limited. This study provides the first micro-level statistical analysis of torture’s relation to subsequent killings committed by insurgent and counterinsurgent forces. The theoretical arguments contend that torture is ineffective for reducing killings perpetrated by insurgents both because it fails to reduce insurgent capacities for violence and because it can increase the incentives for insurgents to commit future killings. The theory also links torture to other forms of state violence. Specifically, engaging in torture is expected to be associated with increased killings perpetrated by counterinsurgents. Monthly municipal-level data on political violence are used to analyze torture committed by counterinsurgents during the Guatemalan civil war (1977–94). Using a matched-sample, difference-in-difference identification strategy and data compiled from 22 different press and NGO sources as well as thousands of interviews, the study estimates how torture is related to short-term changes in killings perpetrated by both insurgents and counterinsurgents. Killings by counterinsurgents are shown to increase significantly following torture. However, torture appears to have no robust correlation with subsequent killings by insurgents. Based on this evidence the study concludes that torture is ineffective for reducing insurgent perpetrated killings.