To attend the conference, please contact the secretariat Christelle Fotso Tatchum
- Andrew Rhodes : Professor Université Toulouse Capitole Supervisor
- Bruno Jullien : Senior Researcher CNRS-TSE-R Co-Supervisor
- Jacques Cremer : Senior Researcher CNRS-TSE-R Examinateur
- Catherine Bobtcheff : researcher CNRS-PSE-Ecole d'économie de Paris Rapporteure
- Sinem Hidir : Professor, University of Warwick Rapporteure
- Sandro Shelegia : Professor, BSE-UPF Rapporteur
This thesis comprises three chapters on information economics. The first chapter focuses on sequential screening through trial versions. I develop a model where firms use trial versions to convey information, whereby trial versions serve as both the first-stage consumption and the source of information. I find that when consumers acquire knowledge of a product's valuation through good-news belief updates, it is more profitable to offer a rich menu of trial versions. Conversely, in the case of bad-news belief updates, a single trial version (referred to as bunching) proves to be the optimal strategy.
The second chapter studies price signaling and the recommender's ranking mechanism. Consumers sometimes evaluate a product based on its price, resulting in bad products outperforming good ones. I find that if the recommender always provides recommendations, even ones with high credibility (indicating a high probability of being a good product), it does little to deter consumers from relying on price judgments. To address this issue, the recommender must make commitments to not recommend any products occasionally, even if the probability of doing so is infinitesimally small. Through this commitment, consumers will no longer rely solely on price and instead pay greater attention to the recommender's suggestions. Consequently, the probability of selling high-quality products increases, leading to enhanced social welfare.
The third chapter investigates the impact of information entropy on coordination. I find that to achieve better coordination among different players, it is beneficial to provide them with precise but not accurate information. This involves offering information with a refined type space, such as increased significant figures or a highly detailed classification, while allowing for a relatively large margin of error. Despite the seeming redundancy of precise information reporting when accompanied by large errors, it plays a crucial role in refining the state space, augmenting entropy, and ultimately improving coordination.