Abstract

This thesis consists of three essays on the Economics of Information, focusing on strategic information acquisition and on the design of disclosure rules.

In Chapter 2, a decision-maker relies on the information reported by a panel of experts to take an action. These experts may observe relevant information and have preferences over the decision-maker’s actions. Each expert possesses three qualities: (i) the probability that he acquires information, (ii) the probability that this information is inaccurate in favor of his agenda, and (iii) the level of this error. Conscious of these qualities, the decision-maker forms her optimal panel to make the most informed decision. We show that it is generally better for the decision-maker to have experts with identical agendas if these experts cannot misreport their information, even if it is inaccurate.

In Chapter 3, we consider a variation of the model presented in Chapter 2 in which experts’ information is always accurate. Moreover, each expert can now incur a cost to increase his probability of obtaining information. The objective of the decision-maker is to form a panel to make the most informed decision, while experts have their own preferences. We show that there exist levels of cost of effort such that experts in a homogeneous panel will choose to exert no effort, while experts in a diverse panel will exert some. In addition, we find that for a sufficiently high return on effort, a diverse panel is optimal.

In Chapter 4, one agent – the sender – benefits from two other agents – the receivers – taking some actions. Each receiver possesses private information referred to as his type, and can disclose it truthfully or not. The sender has a private type which she does not observe at first but can condition a communication device, observed by every agent, on its different realizations and on the reported types. In turn, this provides the receivers with action recommendations. We show that depending on the alignment of the preferences, the optimal rule ranges from fully revealing the sender’s type to no revelation at all, and that most of the times only partial information is revealed.