Signalling in Deterministic and Stochastic Settings : A Theoretical and Experimental Investigation

We contrast a standard deterministic signaling game with a variant where the signal-generating mechanism is subject to stochastic perturbations. In the theoretical part, we obtain a unique equilibrium with stochastic signals. This equilibrium is separating and has intuitive comparative-static properties in that the degree of signaling is a function of the prior distribution of types. With deterministic signals, both pooling and separating configurations occur. We report laboratory data which support the theory remarkably well. In the stochastic variant, there is more signalling behavior than with deterministic signals, and less frequent types distort their signals relatively more. Moreover, the degree of congruence between equilibrium and subject behavior is greater in stochastic settings compared to the deterministic variant