Thèse de doctorat en Sciences économiques
Sous la direction de Jacques Crémer.
Soutenue le 18-09-2018
Le résumé en français n'a pas été communiqué par l'auteur.
This paper studies firms’ disclosure decisions of product information in a duopoly setting, as well as the welfare implication of compulsory disclosure policy. I show that there is a problem of externality between the two firms: even if disclosure weakens price competition in the market and increases total industry profits, a firm could have incentive not to disclose product information because it decreases his market share. As a result, regulatory policy could increase total industry profits as it could rectify the problem of externality. Therefore, despite more information allows consumers to make a better choice between different alternatives, it might backfire as it could increase the average price in the market. I also present simple conditions on when providing more information could harm consumers, and when it will improve consumer welfare. This paper studies the information processing behavior of a decision maker (DM) who has limited information processing ability. More specifically, the DM can process only a subset of all available information. Before taking an action, he chooses whether to process or ignore signals about the state of the world which he receives sequentially. I show that at the optimum, the DM processes only signals which are strong enough, but will process a weaker signal if it confirms his existing strong belief or if it supports a much more desirable state of the world. This explains some phenomena which have been well documented in the psychology literature, such as preference for strong signals, confirmation bias for individuals with strong prior and wishful thinking. Moreover, I analyze how the Internet, and in general changes in information structures, affects the processing behavior of the DM. The results shed light on different issues in the information era, including polarization and media strategy. This paper studies experimentally whether confirmation bias arises when individuals are exposed to information overload, or equivalently have limited ability to perfectly update their belief with all available information. In our experiment, subjects have to form beliefs as they navigate a sequence of signals within a limited period of time. We compare belief formation under two settings, where the treatment setting imposes a larger information/cognitive load than the control setting. We find that subjects in the treatment setting exhibit a stronger confirmation bias than those in the control setting. Upon receiving a belief-challenging signal, subjects in the treatment group update their belief less than those in the control group. In contrast, upon receiving a belief-confirming signal, subjects update similarly in both settings. As a result, subjects in the treatment setting are also less likely to switch sides: once they believe that one state is more probable than another, they are less likely to switch even if they receive enough belief-challenging signals. Not only do these results show that the limited ability of information processing plays a role in the formation of confirmation bias, they also improve our understanding on the impact of information overload, for example on polarization.