Three Essays in Mechanism Design

Abstract

In the first and second chapters we study whether the current trend of using stronger solution concepts is justified for the optimal mechanism design. In the first chapter, we take a simple auction model and allow for type-dependent outside options. We argue that Bayesian foundation for dominant strategy mechanisms is valid when symmetry conditions are satisfied. This contrasts with monotonicity constraints used before in the literature. In the second chapter we develop the idea further by looking into the practical application of type-dependency of outside options in auctions - namely, a possibility of collusion between agents. We show that in this environment for a certain range of primitives no maxmin foundation for dominant strategy mechanisms will exist. Finally, in the last chapter we study a voting environment and non-transferable utility mechanism design. We argue that strategic voting as opposed to truthful voting may lead to higher total welfare through better realization of preference intensities in the risky environment. We also study optimal mechanisms rules, that are sufficiently close to the first best for the uniform distribution, and argue that strategic voting may be a proxy for information transmission if the opportunities to communicate preference intensities are scarce.