Four essays on strategic communication

This dissertation studies patterns of strategic communication in cases, in which the involved parties disagree in their preferences or opinions. In Chapter 1, I study a model of strategic communication in networks, in which the players diverge in their preferences and information can be communicated either through a costly verifiable information (hard) channel or through a low-cost cheap talk (soft) channel. I find that the availability of hard links allows each agent to get a weakly greater number of truthful messages compared to the pure cheap talk setting. If only one party bears the cost of a hard link, then introducing hard links increases the total expected welfare. In contrast, if the cost of a hard link is shared by both parties, then allowing for verifiable communication can decrease the total welfare. In Chapter 2, I consider a model of strategic cheap talk communication in networks, in which the players can disagree in their preferences or their opinions. I find that the information transmission pattern crucially depends on the nature of the disagreement. If the agents diverge in their preferences, then information transmission exhibits a negative externality effect: greater information obtained by some agent discourages further information accumulation by harming the credibility of other agents. In contrast, information transmission displays a positive externality effect when the agents have divergent opinions: greater information obtained by some agent encourages further information accumulation by improving the credibility of other agents. Chapter 3 studies a benevolent authority's decision to constrain or inform a population of individuals. It demonstrates that the authority's decision to regulate an activity depends on whether she deems it a matter of preference or opinion. In the former case, the benevolent authority is libertarian: she gives truthful advice and safeguards liberty. In the latter case, the benevolent authority is paternalistic: believing that she acts in the individuals' best interest, the authority forces another action than the individuals would choose for themselves. In Chapter 4, I consider communication between an informed Sender and an uninformed Receiver. The Sender has a preference bias and is guilt averse to letting down the Receiver's payoff expectations. I show that no separating equilibrium exists; rather, in case of uniform state of the world and quadratic utilities, I demonstrate that there exist partition equilibria (as in Crawford and Sobel (1982)). An increase in the guilt aversion intensity is akin a decrease in the preference divergence: higher guilt aversion intensity allows for more intervals in the equilibrium partition; and holding the number of elements in the partition fixed, greater guilt aversion intensity results in more balanced intervals.