Designing Mechanisms to Focalize Welfare-Improving Strategies

Many institutions use matching algorithms to allocate resources to individuals. Examples include the assignment of doctors, students and military cadets to hospitals, schools and branches, respectively. Oftentimes, agents' ordinal preferences are highly correlated, motivating the use of mechanisms that provide agents with channels through which they can express some cardinal preference information. This paper studies two such mechanisms, one from the field and one we design. In each of the games induced by these algorithms, we identify the strategies that constitute the unique symmetric ex-post equilibrium. Interestingly, when we test the mechanisms in the lab, these equilibrium predictions fail. Subjects nevertheless behave largely in concordance with the mechanisms' intended strategies; the focalization of such strategies lead to greater welfare in relation to a popular existing mechanism