Essays on Strategic Information Transmission

Abstract

This dissertation analyzes strategic information transmission between informed and uninformed economic agents. Chapter 1 overviews the contents of the dissertation. Chapter 2 and chapter 3 analyze strategic communication under the assumption that the uninformed party has reference-dependent preferences and is loss averse a là Köszegi and Rabin. In particular, chapter 2 studies the link between reference dependence, loss aversion and credible communication in an environment where an uninformed agent (B) has to decide whether to participate or not in a risky project, whose probability of success is known to an informed agent (A) only. We show that reference dependence and loss aversion may give rise to credible information transmission. This happens because inaccurate information has two effects: it leads B to choose the action A prefers in the short run, but it also generates unrealistic expectations that, modifying his reference point, may induce B to take actions in the long run that hurt A. This phenomenon is not possible in a model where B is an expected utility maximizer. In Chapter 3, we use a similar insight to analyze a model of electoral competition in which two parties compete to get the support of a mass of voters. Each party is represented by a politician whose valence is unobservable and can take one of two values: high or low. All voters prefer politicians with high valence, but ideological biases may lead them to vote according to party affiliation. Candidates can make statements concerning their valence; however, if voters are expected utility maximizers, politicians' statements lack any credibility and no information transmission takes place. By introducing reference dependence and loss aversion, information transmission becomes possible. This happens because reference dependence introduces an implicit cost of lying: lies may raise voters' expectations about the candidates' valence and, if detected, may lead them to vote against ideological biases in order to avoid the psychological loss associated with supporting a candidate worse than expected. In this context, we show that the set of parameters under which the fully informative equilibrium exists expands with increases in loss aversion. In chapter 4 we consider a mechanism design problem and we characterize the extent to which an uninformed agent can use hard evidence to induce informed agents to truthfully reveal their, possibly exclusive, information. More precisely, we study the problem of full implementation in Bayes-Nash equilibrium in environments with incomplete information and hard evidence. We provide a full characterization of the set of implementable social choice functions in economic environments with at least 3 agents, while, in general environments with at least 3 agents, we provide separate necessary and sufficient conditions.