Improving Transparency and Verifiability in School Admissions: Theory and Experiment

Abstract

Students participating in centralized admissions procedures do not typically have access to the information used to determine their matched school, such as other students' preferences or school priorities. This can lead to doubts about whether their matched schools were computed correctly (the 'Verifiability Problem') or, at a deeper level, whether the promised admissions procedure was even used (the 'Transparency Problem'). In a general centralized admissions model that spans many popular applications, we show how these problems can be addressed by providing appropriate feedback to students, even without disclosing sensitive private information like other students' preferences or school priorities. In particular, we show that the Verifiability Problem can be solved by (1) publicly communicating the minimum scores required to be matched to a school ('cutoffs'); or (2) using `predictable' preference elicitation procedures that convey rich 'experiential' information. In our main result, we show that the Transparency Problem can be solved by using cutoffs and predictable procedures together. We find strong support for these solutions in a laboratory experiment, and show how they can be simply implemented for popular school admissions applications involving top trading cycles, and deferred and immediate acceptance.