Interpersonal trust is a one-sided social dilemma. Building on the binary trust game, we ask how trust and trustworthiness can evolve in a population where partners are matched randomly and agents sometimes act as trustors and sometimes as trustees. Trustors have the option to costly check a trustee's last action and to condition their behavior on the signal they receive. We show that the resulting population game admits two components of Nash equilibria. Nevertheless, the long-run outcome of an evolutionary social learning process modeled by the best response dynamics is unique. Even if unconditional distrust initially abounds, the trustors' checking option leads trustees to build a reputation for trustworthiness by honoring trust. This invites free-riders among the trustors who save the costs of checking and trust blindly, until it does no longer pay for trustees to behave in a trustworthy manner. This results in cyclical convergence to a mixed equilibrium with behavioral heterogeneity where suspicious checking and blind trusting coexist while unconditional distrust vanishes. (author's abstract) / Series: Department of Economics Working Paper Series

Experimental tests of choice predictions in one-shot games show only little support for Nash equilibrium (NE). Poisson Cognitive Hierarchy (PCH) and level-k (LK) are behavioral models of the thinkingsteps variety where subjects differ in the number of levels of iterated reasoning they perform. Camerer et al. (2004) claim that substituting the Poisson parameter = 1:5 yields a parameter-free PCH model (pfPCH) which predicts experimental data considerably better than NE. We design a new multi-person game, the Minimizer Game, as a testbed to compare initial choice predictions of NE, pfPCH and LK. Data obtained from two large-scale online experiments strongly reject NE and LK, but are well in line with the point-prediction of pfPCH.

Experimental tests of choice predictions in one-shot games show only little support for Nash equilibrium (NE). Poisson Cognitive Hierarchy (PCH) and level-k (LK) are behavioral models of the thinking-steps variety where subjects differ in the number of levels of iterated reasoning they perform. Camerer et al. (2004) claim that substituting the Poisson parameter tau = 1.5 yields a parameter-free PCH model (pfPCH) which predicts experimental data considerably better than NE. We design a new multi-person game, the Minimizer Game, as a testbed to compare initial choice predictions of NE, pfPCH and LK. Data obtained from two large-scale online experiments strongly reject NE and LK, but are well in line with the point prediction of pfPCH. / Series: Department of Economics Working Paper Series