UC Berkeley

This dissertation studies how to improve the outcomes of Centralized Choice and Assignment Systems (CCAS) for schools, with a focus on both social objectives such as socioeconomic, cultural, or racial integration, as well as family welfare. CCAS systems are increasingly used worldwide for both student and teacher assignments, and their popularity is expected to rise further as digitalization continues to expand globally. Therefore, identifying cost-effective policies that can improve the functioning of CCAS systems can have significant implications, as they are likely to be applicable in many different contexts.

In the first chapter, co-authored with Isabel Jacas, we investigate an assignment rule policy: socioeconomic reserves (SR) for low-socioeconomic status (low-SES) families in the Chilean CCAS for PreK schools. Although the implementation of the CCAS system was expected to reduce school segregation by eliminating family background-based selection, similar levels of segregation persist. However, we find that optimizing the size of SR to local conditions can significantly improve outcomes. By setting SR at the schools' municipality low-SES applicants' share, educational segregation can be reduced by five times compared to the current flat 15% level when compared to a minimum segregation benchmark obtained by estimating family preferences and generating counterfactual applications that eliminate the drivers of differences in choice between socioeconomic groups. Additionally, leveraging demand-side policies as a complement to SR can further reduce segregation considerably, as SR effectively assigns additional low-SES applicants to congested schools where their share is underrepresented. In the second chapter, co-authored with Gregory Elacqua, Leidy Gómez, Luana Marotta, Carolina Méndez, and Christopher A. Neilson, we investigate an information demand-side policy that leverages the "smart" personalized feedback potential of digital application processes in CCAS. Specifically, we examine the causal impact of a personalized non-assignment risk warning, combined with a list of "achievable" teaching position recommendations, on teacher applications in the Ecuadorian "I Want to Become a Teacher" selection process. Our analysis reveals that treated teachers are significantly more likely to modify their application and secure an assignment, and we also provide evidence suggesting that the intervention resulted in an increase in selection scores used by the Ecuadorian Ministry of Education to evaluate teacher performance.

The third chapter of this dissertation, co-authored with Gregory Elacqua, Isabel Jacas, Carolina Méndez, and Christopher A. Neilson, takes a step back from improving the outcomes of CCAS to examine the broader comparison between these systems and alternatives. Specifically, we compare the welfare effects on families of a CCAS implementation in the city of Manta, Ecuador in 2021, incorporating household preferences for the first time as an assignment criterion (using the deferred acceptance algorithm) and compare it to the alternative assignment mechanism previously used, which was also centrally coordinated but based on minimizing residence-to-school distances. Our findings reveal that considering applicant preferences leads to significant welfare gains, suggesting that CCAS can have a substantial impact on welfare in developing country contexts, even without complete optimization using policies such as those studied in the first two chapters.