A Stochastic Signaling Model With Heterogeneous Risk Attitudes: Theory and Experiments

We developed and tested a stochastic signaling model where incomplete separation could be explained by sellers' heterogeneous risk attitudes. In this model, sellers of both high‐ and low‐quality products can subject their products to a noisy certification test. The high‐quality product has a lower chance of failing the test and getting the certificate given the same level of investment in cost. While the difference in the failure risks or signaling costs could induce complete separation in sellers' decisions if they are risk‐neutral, it might fail to do so when the high‐type sellers are too risk‐averse or low‐type sellers are too risk‐seeking. Under the rank‐dependent utility framework, we discuss the effects due to probability pessimism and aversion to income variations. Through laboratory experiments, we find more evidence on the effects of probability pessimism. We also find that, once we control for risk attitudes, sellers do not adopt the separating strategy when the pooling equilibrium cannot be refined and the separating equilibrium is not unique.