Terrorist Factions ∗

I study how a variety of structural and strategic factors affect terroristmobilization, the likelihood of a splinter faction forming, and the positions adopted by terrorist leaders. The factors considered include the state of the economy, the viability of institutions for the nonviolent expression of grievance, the ability of the factional leaders to provide nonideological benefits, and the risks associated with splintering. The model highlights that, for strategic reasons, changes in the structural environment often entail trade-offs between decreasing terrorist mobilization and increasing extremism. For instance, strengthening the economy or institutions for the nonviolent expression of grievance is found to decrease terrorist mobilization, increase the extremism of terrorist factions, and decrease the likelihood of a splinter faction forming. These results suggest competing micro-level effects of such changes on the expected level of violence that, because they are offsetting, might not be observed in macro-level data analyses, which have been the mainstay of empirical studies of terrorism. Terrorist organizations are not monolithic nor is their structure stable. Rather, they are made up of heterogeneous factions that frequently splinter from one another as the political and economic landscape shifts (Bell 1998, Zirakzadeh 2002). Consider a few examples. ∗ I am indebted to Scott Ashworth, Bruce Bueno deMesquita, Steve Callander, Amanda Friedenberg, Catherine Hafer, David Lake, Dimitri Landa, Maggie Penn, Bob Powell, Cyrus Samii, Matthew Stephenson, the editors of the QJPS, and seminar participants at Bar Ilan University, Berkeley, Caltech, Chicago, Columbia, HebrewUniversity, Michigan, NYU, UCSD, UCLA, and the CESifo conference on “Guns and Butter: The Economic Causes and Consequences of Conflict.” I thank the Lady Davis Fellowship, the Department of Political Science, and the Center for the Study of Rationality, all at the Hebrew University of Jerusalem, for support. MS submitted 21 January 2008; final version received 6 October 2008 ISSN 1554-0626; DOI 10.1561/100.00008006 © 2008 E. Bueno de Mesquita 400 Ethan Bueno de Mesquita Republicanmilitants inNorthern Ireland have experienced a variety of splinterings. In the late 1960s, the Provisional Irish Republican Army (IRA) split from the Original IRA due to disagreements over military policy. In the mid-1980s, the extremist Continuity IRA splintered from the Provisionals when the Provisionals abandoned their policy of refusing to participate in parliament. Another radical splinter group, the Real IRA, broke from the Provisionals in the 1990s over the peace process that led to the Good Friday Agreement. Militant Palestinian nationalism has been represented by a variety of terrorist groups that have also splintered a number of times. In the 1970s a group of radical, secular nationalist factions split from the Palestine Liberation Organization over the value of compromise. Similarly, twomilitant Islamic terrorist groups—Palestinian Islamic Jihad (in the 1970s) and Hamas (in the 1980s) — split from the Muslim Brotherhood over the value of violent versus nonviolent resistance. In recent years further divisions have occurred between factions in both secular nationalist and Islamic organizations as these various groups vie for power. Basque separatists have divided into several factions throughout the history of the terrorist group Euskadi Ta Askatasuna (ETA). For instance, in the late-1970s ETA divided into the extremist ETA-militar and the more moderate ETA-politico militar over whether Basque separatists should participate in regular politics following the death of Franco, Spanish democratization, and the grant of partial autonomy to the Basque Country. Such internal divisions within terrorist organizations have important affects on both patterns of terrorist violence and counterterrorism strategies.1 For instance, factions often disagree over the relative value of negotiated settlement versus continued violence ( Jaeger and Paserman 2006). As a result, when one faction accepts government concessions violence can increase, both because the remaining faction is more extreme than the faction that accepted concessions and because the extremists use violence to undercut peace negotiations (Stedman 1997, Kydd and Walter 2002, Bueno de Mesquita 2005a). I model a variety of determinants of mobilization, extremism, and factionalization. The model explores how the risks of factionalization affect and are affected by the level of extremism of the original terrorist group. It also allows me to examine how the extremism of factions and the likelihood of factionalization are affected by the economy, institutions for the nonviolent expression of grievance, factional leaders’ abilities to provide nonideological benefits, and the risks associated with forming a splinter faction. A key theme is that many policies that are expected to decrease mobilization — e.g., economic aid or building institutions for the nonviolent expression of grievance — will also lead terrorist factions to become more extreme. These effects may be offsetting in their impact on the level of terrorist violence, implying a trade-off for governments. Moreover, this same observation suggests a challenge for empirical studies that focus on the relationship between structural features of a society and macro-level measures of the 1 For models of terrorism and/or counterterrorism with internally divided terrorist organizations, see, among others, Berrebi and Klor (2006), Bloom (2004, 2005), Bueno de Mesquita (2005a), de Figueiredo and Weingast (2001), Kydd and Walter (2002), and Siqueira (2005). Terrorist Factions 401 level of terrorism. Even if such studies find no statistical relationship between the level of terrorism and nonviolent institutions or the economy, one need not conclude that these features of the political-economic environment do not matter for understanding terrorism. Rather, changes in the political-economic environment may have important micro-level effects that, because they are offsetting, are not observed in the type of macro-level data that are the mainstay of empirical studies of terrorism.