Why Not Divide and Conquer? Targeted Bargaining and Violence in Civil War

Abstract

How do rebel groups maintain cohesion when faced with powerful and wealthy governments? Successful rebel groups are a paradox in international relations because they manage to solve collective action problems in wartime environments that lack third-party enforcement and monitoring opportunities. Governments ought to be able to split these rebel groups using a divide-and-conquer strategy of directed amnesty, rewards, and intimidation. I investigate three specific dimensions of the cohesion of a rebel movement: (1) the degree to which rebels can operate among the civilian population outside of their main base areas; (2) the rate at which rebel groups lose members through defection to the government; and (3) the degree to which civilians are willing to withhold information from the government to protect rebels from arrest and assassination even at risk to themselves.

Of all the tools of persuasion (e.g. ideological appeals, public goods, targeted rewards), I show that rebel cohesion is most clearly threatened by brute military force. Civilians will provide the tips needed for selective targeting when the alternative is indiscriminate targeting. Rebels will defect when the government militarily takes control of their communities. I explain the success of brute force over softer appeals by detailing the organizational and principal agent problems inherent to rebel organization in a war-zone. Large structural factors, particularly the ability to develop a monopoly of control over territory, provide far more leverage over civilians and fighters than punishment and reward strategies ever could.

I test and illustrate this argument using remarkably detailed microlevel evidence from the Vietnam War. Using textual and electronic archival records, I provide new data on Viet Cong defections, government assassinations of civilians, the division of territorial control, and other novel measures. New kinds of data require new techniques for analysis. I show that by disaggregating selective targeting by the source of information used to identify the suspect, the role of government coercion in generating civilian cooperation is revealed. I also show that even amorphous concepts like territorial control can be accurately measured on a large scale by expert surveys through application of models from item response theory.