Experimentelle Untersuchungen zur Rolle von Higher-Order Beliefs über die kognitiven Fähigkeiten Anderer in Koordinationsspielen
Zusammenfassung der Projektergebnisse
We are interested in the possibly reasons for coordination failure in economic and social settings. Consider for instance a bank that can run smoothly and effectively if most people believe this to be the case, but can also go bankrupt if there is a "bank run" in which too many people all demand their money back at the same instance. How can such a bank run be prevented? Or consider the case of the unfortunate yet widespread practice of female genital mutilation (FGM). Bicchieri [2005] argues that despite efforts of NGOs and governments to educate families about the adverse consequences of FGM and despite a measurable change in families' beliefs about the usefulness of FGM there was essentially no measurable change in behavior. Why not? Both situations can be argued to be problems of coordination, with a bank run and FGM both examples of coordination on a bad equilibrium. If all people demand their money back, then I better do so too before there is no money left in the bank. If daughters without FGM are not accepted as brides I have to subject my daughter to FGM as well. So how can one get out of these bad equilibria? In an effort to get closer to answering this question we empirically study people's motifs in a coordination game. We are crucially interested in whether people employ "higher order belief" reasoning in determining their choices. This means do people worry about what others think about them when they make their choices? To do so, we take up the experimental results and setup of Blume and Gneezy [2010], in which there is an issue of cognitive difficulties, to analyze the effects of higher-order beliefs. We introduce a new treatment in which participants guess what other participants play against themselves. This allows us to identify first-order beliefs and therefore separate first- from higher-order beliefs. Using the data from these treatments, we find that players are frequently able to coordinate in the absence of common knowledge, that, however can coordination fail because players' first-order beliefs matter, but that also coordination can fail because players' higher-order beliefs matter. For the FGM example, for instance, our results therefore suggest that educating families individually about the adverse effects of FGM might not be enough to overcome the inherent coordination problem. Instead one should probably pursue a public campaign as only this would also change people's beliefs about others' beliefs.
Projektbezogene Publikationen (Auswahl)
- Identifying the Reasons for Coordination Failure in a Laboratory Experiment. Center for Mathematical Economics, Working Paper No. 567, September 2016 (letzte Revision November 2018)
Philipp Külpmann und Davit Khantadze
