The value of information in centralized school choice systems

Strategy-proofness and assurance of a fair matching are desirable qualities for school choice mechanisms. Although these qualities are theoretical properties of the unrestricted-list deferred acceptance mechanism (DA), it has proven to be hard to go without list-size restrictions in practice. This paper shows how a simple modification to the restricted-list DA, in which students are provided more information about vacancies and offered higher-value options in the event of rejection, can mitigate uncertainty and yield matches very close to what would be obtained under the unrestricted-list DA. I estimate an application-portfolio choice model using administrative data from Tunisia, where a sequential implementation of the DA is used to assign high-school graduates to universities. This sequential implementation creates quasi-experimental variation that allows to separate the identification of students’ preferences for programs from their expectations about their admission probabilities. Counterfactual simulations show that the average student’s expected ex post utility is significantly lower when assignments are made using the standard restricted-list DA than what it would be in the unrestricted-list DA match. Using a sequential implementation, as is done in Tunisia, instead of the standard restricted-list DA, can reduce this welfare loss by 90%. ∗Email address: mluflade@sas.upenn.edu †I am extremely grateful to Meryam Zaiem whose insights about the context, help with obtaining the data, and encouragement have been critical to the realization of this project. I am grateful to Arnaud Maurel, Peter Arcidiacono, Joe Hotz, and Matt Masten for help and suggestions, as well as continuous support and encouragement. I thank Atila Abdulkadiroğlu, Eli Liebman, Pedro Mira, Derek Neal, Aviv Nevo, Seth Sanders, Modibo Sidibé, Petra Todd, and the participants at the Duke Labor and Microeconometrics workshops and at many other seminars, for very helpful comments and discussions. I thank the Ministry of Higher Education of Tunisia for providing the data. All errors are mine.