Strategic Mass Killings

Abstract

Since World War II there have been about …fty episodes of large-scale mass killings of civilians and massive forced displacements. They were usually meticulously planned and independent of military goals. We provide a model where con‡ict onset, con‡ict intensity and the decision to commit mass killings are all endogenous, with two main goals: (1) to identify the key variables and situations that make mass killings more likely to occur; and (2) to distinguish conditions under which mass killings and military con‡ict intensity reinforce each other from situations where they are substitute modes of strategic violence. We predict that mass killings are most likely in societies with large natural resources, signi…cant proportionality constraints for rent sharing, low productivity and low state capacity. Further, massacres are more likely in a civil than in an interstate war, as in the latter group sizes matter less for future rents. In non polarized societies there are asymmetric equilibria with only the larger group wanting to engage in massacres. In such settings the smaller � We are grateful to CEPR for sponsoring the con‡ict conference in Gerzensee in July 2009, where we started this project. We also thank Borek Vasicek and Peter Van der Windt for research assistance. Helpful comments from Autumn Payton Lockwood, Kalle Moene, Debraj Ray, Alexander Wolitzky, Fabrizio Zilibotti and seminar and conference participants in Princeton, EUI, IMT-Lucca, Oslo, Dehli, Barcelona and Madrid are gratefully acknowledged. y Institut d’Analisis Economica, CSIC, Barcelona. Email: joan.esteban@iae.csic.es. Joan Esteban is a member of the Barcelona GSE Research Network funded by the Government of Catalonia. He gratefully acknowledges …nancial support from the AXA Research Fund and from the Spanish Government CICYT project n. SEJ2006-00369.