Implementation of the matching mechanism for the new school admissions system and modeling of the school choice for chilean families

Abstract

Matching mechanisms for school assignment have been adapted on a global scale since the popular application in New York City in 2004, which used the Deferred Acceptance algorithm to assign thousands of students. Implementation of these systems is not a trivial task because of the large scale of the problem and educational regulations particular to each country. One of the goals of this thesis was to develop the implementation of the matching algorithm for the Chilean case and to show the nuances in the design of this mechanism. Then, taking advantage of the strategy-proofness of this system and the centralized data it produces, this work develops structural models for discrete choice to study families' preferences for schools, speci cally, a Bayesian multivariate ordered probit with a hierarchical Bayesian structure to model heterogeneity in the school choice. A methodology was developed to obtain choice sets for families using unsupervised learning techniques in the Coquimbo region where these structural models could be applied. This way, meta-analysis was conducted to evaluate what characteristics in school choice are consistently signi cant, where results indicate that, for example, while schools with poor academic performance are not preferred on average, schools with superior results on standardized test are only preferred by students with good academic records. The estimation of these preferences allows for a series of counterfactual analyses that can aid in the design and implementation of public policies and support the decision making of schools. A no-pricing policy was simulated for the Coquimbo region---which is actually in the process of being adopted by subsidized private schools in Chile---, where it was found that it would improve the social welfare of the assignment of the region, specially for families with children with disabilities, but would impact negatively students that are not in the lowest socioeconomic family groups.