| dc.description.abstract | Information design in an incomplete information game involves a designer that influences players’ actions through signals generated from a designed probability distribution to optimize its objective function. For quadratic objective functions, if players have quadratic payoffs that depend on players’ actions and an unknown payoff-relevant state, and signals on the state that follow a Gaussian distribution conditional on the state realization (LQG game), the information design problem is a semi-definite program (SDP). The doctoral research is pursued in three thrusts: analytical and numerical characterization of optimal information design to maximize social welfare and the agreement among players’ action in LQG games, analysis of individual information preferences of agents in LQG network games, and robust optimal information design in LQG games under perturbed utilities.
Firstly, it is shown that full information disclosure maximizes social welfare under common payoff state, under homogeneous payoff dependencies, or under public signals. When the objective is to maximize agreement among players’ actions, no information disclosure is optimal. Under joint objective, full information optimality condition on weight of agreement is determined for public information structures and common payoffs. In the second thrust, conditions for which rational agents are expected to benefit from full information are characterized in general network games. In star networks, the central agent benefits more than a peripheral agent from full information unless the competition is strong and the number of peripheral agents is small enough. Numerical simulations of ex-post preferences showed that a risk-averse central agent may prefer no information under strong competition. In the third thrust, we consider the setting where the designer has partial information about players’ payoffs. We develop robust convex programs considering the worst-case realization of players’ payoffs under general perturbations. We obtain optimality conditions of no and full information disclosure based on uncertainty set radius and perturbation shifts under ellipsoid uncertainty. Numerical experiments show that the designer is inclined to reveal less informative signals as its uncertainty about the game increases. | |
