This thesis considers mechanism design problems with dual objectives of efficiency and fairness in three different problems.

The first one is fair mechanism design for competition sequencing problem. We design an auction system to determine a fair number of credits one may sacrifice or request for a position she or he desires or dislikes, respectively. It is modeled as a sequencing problem that demands incentive compatibility, budget-balance and egalitarian conditions. We characterize such deterministic protocols. We then find a unique randomized protocol with respect to these three conditions.

The second application is the resource sharing over the network. In such a system, participants act as both suppliers and consumers. We consider tit-for-tat popular proportional sharing...[ Read more ]

This thesis considers mechanism design problems with dual objectives of efficiency and fairness in three different problems.

The first one is fair mechanism design for competition sequencing problem. We design an auction system to determine a fair number of credits one may sacrifice or request for a position she or he desires or dislikes, respectively. It is modeled as a sequencing problem that demands incentive compatibility, budget-balance and egalitarian conditions. We characterize such deterministic protocols. We then find a unique randomized protocol with respect to these three conditions.

The second application is the resource sharing over the network. In such a system, participants act as both suppliers and consumers. We consider tit-for-tat popular proportional sharing mechanism which has been proved to be both fair and efficient. However, agents may play strategically to improve their utilities by influencing the allocation. Examples show that proportional mechanism is not incentive-compatible against this strategy, therefore the concept incentive ratio is employed to represent the proportion to which a player’s utility is increased. We prove that: the incentive ratio is 2 on trees and √2 on complete graphs; it is between 2 and 4 on circle graphs and is exactly 2 if the number of nodes in the circle is even.

The third work is efficient and fair mechanism design in vehicle licenses allocation. Due to traffic and air quality concerns in urban cities, many big cities have begun to adopt the vehicle licenses control policies. The current pricing and allocation mechanisms differ from city to city, including auction, lottery, lottery with reserved price, and the simultaneous auction and lottery. We target to design the optimal mechanism to balance efficiency and fairness. We first propose a unified mechanism framework that either includes or outperforms all the existing mechanisms. Under this framework, we prove that the optimal mechanism is always first auction then lottery. Besides, the optimal allocation rule is distribution-free and easy to implement in practice due to its truthfulness and simple structure. The experimental results also show that first auction then lottery is the best mechanism to use in practice.