Abstract

Recognizing the need to find a solution to the European refugee crisis, given the political deadlock preventing countries from participating in responsibility-sharing, this thesis addresses questions on international refugee placement from a market design perspective. Beyond refugee settlement, our theoretical findings also apply to immigration, centralized university admissions, public school choice systems, and other settings.

Following the Introduction in Chapter 1, in Chapter 2 we advocate a centralized global refugee matching system and design new matching algorithms with explicitly modelled UNHCR-mandated hierarchical priority classes of refugee families. Combined with the preferences of both sides, this set-up allows us to capture and analyse the impact of the mandated priority classes on the stability and fairness of the resulting refugee matching. We conduct an axiomatic analysis to further support the new matching designs.

Chapter 3 is motivated by the objective of designing a priority policy that effectively helps the prioritized agents. To benefit refugees in emergency zones, for example, a policy designed to create representation is not appealing, since strong candidates who would qualify without relying on their priority status may take up some or most of the reserved positions. Hence, we propose an alternative policy with targeted priority reserves, the DA-TPR mechanism, which targets those agents in the priority group who are in need of a reserved position. We also study a general class of matching mechanisms with priority reserve policies, which includes both the DA-TPR and the primary representation rule, Hafalir et al.'s (2013) DA-MiR mechanism. The DA-TPR is the most targeted policy in this class of mechanism, which we characterize by a priority reserve stability axiom and a strong incentive property.

Chapter 4 introduces two new classes of matching mechanisms with explicitly specified partially targeted priority reserves. Both of these classes include the DA-TPR and DA-MiR mechanisms as extreme members. We identify one of these as a subclass of the characterized class of mechanisms from the previous chapter. Both studied classes of mechanisms are transparent and offer a range of policies between the DA-TPR and the DA-MiR policies, providing the designer with flexibility and clarity when choosing a priority reserve policy.