Rational Disagreement

We propose rational disagreement as a formal framework for analyzing seemingly irrational behavior that can persist despite the wide availability of objective information in a steady-state. Agents are rational in that they correctly anticipate the distribution of aggregate outcomes, yet disagree about which specific individuals perform better than others. Notably, the subjective belief of any individual may be objectively correct. We illustrate the key concepts with a simple entry game. We show how unordered individual outcome distributions can be identified solely from aggregate statistics. We then characterize the resulting game, define its Nash equilibria, and develop a statistical test for the null hypothesis of agreement. Finally, we situate our framework within the broader literature on Bayesian games, behavioral biases, and the rational expectations hypothesis.