Cycles of Distrust: An Economic Model
We propose a model of cycles of distrust and conflict. Overlapping generations of agents from two groups sequentially play coordination games under incomplete information about whether the other side consists of "extremists" who will never take the good/trusting action. Good actions may be mistakenly perceived as bad/distrusting actions. We also assume that there is limited information about the history of past actions, so that an agent is unable to ascertain exactly when and how a sequence of bad actions originated. Assuming that both sides are not extremists, spirals of distrust and conflict get started as a result of a misperception, and continue because the other side interprets the bad action as evidence that it is facing extremists. However, such spirals contain the seeds of their own dissolution: after a while, Bayesian agents correctly conclude that the probability of a spiral having started by mistake is sufficiently high, and bad actions are no longer interpreted as evidence of extremism. At this point, one party experiments with a good action, and the cycle restarts. We show how this mechanism can be useful in interpreting cycles of ethnic conflict and international war, and how it also emerges in models of political participation, dynamic inter-group trade, and communication - leading to cycles of political polarization, breakdown of trade, and breakdown of communication.
We thank Sandeep Baliga, Sylvain Chassang, Edward Glaeser, James Fearon, Jon Levin, Qingmin Liu, and seminar participants at the Canadian Institute for Advanced Research, Microsoft Research, the NBER Political Economy Conference, the Northwestern Conflict and Cooperation Conference, and Stanford for useful comments. Acemoglu gratefully acknowledges financial support from the Canadian Institute for Advanced Research and the ARO. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.