Are Costly Signals More Credible? Evidence from Three Experiments

Abstract

The idea that costly signals are more credible is a long-standing hypothesis in international politics. However, little is known on how costly signaling actually works. Causal evidence is elusive because the effect of a costly signal is almost always confounded with the effects of other previous or simultaneous information. I design three controlled experiments to study how the logic of sinking costs operates. I find that signalers randomly assigned with high resolve are more likely to sink costs, but receivers do not acquiesce in line with signaler expectations, despite the sunk costs suffered. The logic of sunk-cost signaling is strong at the signaler’s end but not at the receiver’s end. There is a sender-receiver gap in how the same deterrence interaction is perceived at the two ends of the signaling mechanism, contrary to what the theory of costly signaling automatically assumes.