The Logic of Threat and Assurance: An Experimental Investigation

• In Event: Poster Session: International Security II

Abstract

Why threats do not always work in international politics? Military threats are widely studied in international relations and regularly adopted as a strategy by policymakers. But history and recent experimental evidence have shown that threats can be counterproductive (e.g. Jervis 2017[1976], chap. 3; Dafoe and Weiss 2018; Quek and Johnston 2018). Building on Schelling’s (1966) seminal work, we argue that the lack of assurance can explain the ineffectiveness of threat. While the existing literature has studied threat effectiveness empirically, none has investigated the full strategy set of (1) threat, (2) assurance, and (3) threat and assurance combined. Using the case of North Korea’s nuclear weapon programs, and a parallel replication based on a hypothetical country, we manipulate these strategies in a survey experiment to measure their effects on U.S. attitudes toward removing nuclear sanctions. We find that threat is effective *only* when used with assurance. Assurance *alone* is also effective, but it operates through a different mechanism. Threat and assurance combined increase only the respondents' expected cost of noncompliance, whereas assurance alone increases both their expected benefit of compliance and the expected cost of noncompliance.

Authors

• Zenobia Chan, Princeton University

• Kai Quek, University of Hong Kong