The Compromise Game: Two-sided Adverse Selection in the Laboratory

We analyze a game of two-sided private information characterized by extreme adverse selection, and study a special case in the laboratory. Each player has a privately known "strength" and can decide to fight or compromise. If either chooses to fight, there is a conflict; the stronger player receives a high payoff and the weaker player receives a low payoff. If both choose to compromise, conflict is avoided and each receives an intermediate payoff. The only equilibrium in both the sequential and simultaneous versions of the game is for players to always fight, independent of their own strength. In our experiment, we observe among other things (i) frequent compromise, (ii) little evidence of learning, and (iii) different behaviour between first, second and simultaneous movers. We explore several models in an attempt to understand the reasons underlying these anomalous choices, including quantal response equilibrium, cognitive hierarchy, and cursed equilibrium.