Escalation: Competing Perspectives and Empirical Evidence

Abstract

Since the early days of the Cold War, “escalation” has been used as shorthand to refer to rising hostility in interstate disputes. The escalator metaphor reminds us that the phenomenon is a dynamic process as opposed to a static condition, but in some respects the analogy can be misleading. Unlike escalators, militarized disputes do not always start at the same level. Nearly half of them (49 percent) begin with the use of force by one side (Jones, Bremer, and Singer 1996:195). Moreover, unlike the smooth upward process of an escalator, interstate disputes are likely to exhibit jagged patterns as they move through several phases of escalation, de-escalation, and renewed escalation before terminating (Leng 1993). What is less clear, however, is the extent to which the escalation process is self-reinforcing, as the escalator metaphor suggests, as opposed to remaining under the control of the participants. The survey of empirical findings that follows considers that question as part of a broader examination of what we now know about dispute escalation and its consequences. To provide theoretical focus for the survey, I begin by contrasting the works of those conflict strategists within the realist tradition, who view escalation as manageable and informative, with those political scientists and social psychologists who focus on what they see as escalation's self-reinforcing properties. I then appraise the relative validity of the two perspectives through a survey of empirical findings drawn from three data sets: the Correlates of War Militarized Dispute data, Michael Brecher and Jonathan Wilkenfeld's International Conflict Behavior data, and Russell Leng's Behavioral Correlates of War data. The results are mixed. The findings are consistent with the realist view of the informative properties of escalation; more often than not, states are capable of controlling the escalatory process. But there also is a positive association between escalation …