It is widely accepted that terrorists attack civilians to coerce their governments into making political concessions (Crenshaw 1990; DeNardo 1985; McCormick 2003). The field is deeply divided, however, on the efficacy of this tactic. In the 1980s and 1990s, the dominant view held that terrorism seldom induces government compliance (Cordes et al. 1984; Crenshaw 1986; Freedman and Hill 1986; Lomasky 1991; Schelling 1991; Wilkinson 1986). Political scientists gravitated to the opposite stance around the turn of the century, maintaining that terrorism is a productive instrument of coercion (Kydd and Walter 2006; Lake 2002; Pape 2003, 2005; Sprinzak 2000). The pendulum has recently swung back, with several studies claiming that terrorism does not pay politically (Abrahms 2006a, 2006b, 2008; Cronin 2009a; Rose and Murphy 2007). The policy relevance of this debate requires little elaboration: the question of whether terrorism will advance the political agendas of its practitioners is understandably a paramount concern to the international community (see Blair 2005; Bush 2002).

In this study, I contribute to the debate in four ways. In the first section, I summarize the relevant scholarship and assess its methodological basis. In the second section, I build on this scholarship to establish my own research design for empirically testing terrorism's utility in compelling political concessions. In the third section, I present my findings from an original dataset of 125 violent substate campaigns. My principal finding is that terrorist campaigns are an inherently unprofitable coercive tactic because governments resist politically complying when their civilians are the focus of substate attack.