Information Preferences of Individual Agents in Linear-Quadratic-Gaussian Network Games

We consider linear-quadratic-Gaussian (LQG) network games in which agents have quadratic payoffs that depend on their individual and neighbors’ actions, an unknown payoff-relevant state. An information designer determines the fidelity of revealed to about payoff state maximize social welfare. Prior results show full disclosure is optimal under certain assumptions payoffs, i.e., it beneficial for average individual. In this paper, we provide conditions general structures based strength dependence competition, a rational agent expected benefit, receive higher from disclosure. find all benefit star structure when game homogeneous. also identify central benefits more than peripheral unless competition strong number small enough. Despite fact expect ex-ante, can be worse-off many realizations indicating risk-averse prefer uninformative signals ex-ante.