Estimating Crisis-Signaling Games with Multiple Equilibria

• In Event: Poster Session: Political Methodology

Abstract

We develop new methods to structurally estimate the canonical crisis-signaling game using the dynamic games literature. Unlike previously employed techniques, the estimators are particularly well suited for signaling models in political science because they accommodate the possibility of multiple equilibria. Monte Carlo experiments illustrate that they outperform current approaches in terms of bias and variance and can dramatically reduce computational time. We apply the estimators using two data sets on interstate wars and economic sanctions and compute comparative statics even in the presence of multiple equilibria.

Authors

• Casey Crisman-Cox, Washington University in St. Louis

• Michael Gibilisco, California Institute of Technology