Lavish returns on cheap talk: Two-way communication in trust games

We conduct trust game experiments in which subjects can sometimes exchange proposals either in numerical (tabular) form, or using chat messages followed by exchange of numerical proposals. Numerical communication significantly increases trusting and trustworthiness; inclusion of 1-min verbal communication in a chat room generates an even larger and more robust effect. On average, trustors send $9.21 of their $10 endowment as compared to $7.66 in the standard trust game, and trustees return 56% vs. 45%. Chat enhances the likelihood that trustors and trustees will adhere to non-binding agreements they make—an additional interpretation of trusting and trustworthiness—and increases the probability that subjects will propose, accept, and abide by equal-division agreements. Analysis of the content of subjects’ verbal communication shows that what is said, and not only the fact that things are said, significantly affects outcomes.

Research highlightsWe conduct trust game experiments in which subjects can sometimes exchange proposals either in numerical (tabular) form, or using chat messages followed by exchange of numerical proposals. Numerical communication significantly increases trusting and trustworthiness; inclusion of 1-min verbal communication in a chat room generates an even larger and more robust effect. On average, trustors send $9.21 of their $10 endowment as compared to $7.66 in the standard trust game, and trustees return 56% vs. 45%. Chat enhances the likelihood that trustors and trustees will adhere to non-binding agreements they make—an additional interpretation of trusting and trustworthiness—and increases the probability that subjects will propose, accept, and abide by equal-division agreements. Analysis of the content of subjects’ verbal communication shows that what is said, and not only the fact that things are said, significantly affects outcomes.