Chile's Sistema de Admisión Escolar — A Nationwide Deferred-Acceptance School-Matching Algorithm

Chile's 2015 School Inclusion Law banned publicly subsidised schools from selecting students by social, religious, economic or academic criteria and eliminated co-payments, forcing a redesign of how students are matched to schools nationwide. Chile's Ministry of Education (MINEDUC) partnered with a team of Universidad de Chile researchers to design and implement the Sistema de Admisión Escolar (SAE).

SAE adapts the Deferred Acceptance algorithm (the Gale–Shapley mechanism, for which Lloyd Shapley and Alvin Roth shared the 2012 Nobel Memorial Prize in Economic Sciences) to a simultaneous, nationwide, all-grade matching problem spanning pre-K to 12th grade, piloted from 2016 in the Magallanes region before national rollout. Because the new law removed walk-zone priority and there is no guaranteed public school transport, MINEDUC asked the research team to also help keep siblings together, though this is not a legal requirement; the researchers designed a two-lottery, family-level tie-breaking mechanism for this purpose.

A peer-reviewed paper in Operations Research reports that, in the studied application cycle, over 274,000 students applied to more than 6,400 schools simultaneously. The researchers' own simulations, run on real family preference data, show the two-lottery mechanism raises the share of siblings placed in the same school by 13 percentage points compared with standard student-level tie-breaking. Simulated outcomes under different socio-economic quota orderings show 51–68% of students receiving their first-preference school, with 5.6–14.6% left unassigned in the main round depending on quota design and disadvantaged status; unassigned students are covered through a complementary round and ultimately guaranteed a no-co-payment seat at the closest school with space.

Chile's Ministry of Education publicly praised the algorithm's transparency and strategy-proofness in 2022, though this is a ministry statement rather than an independent audit, and the cited outcome figures come from the system's own designers rather than a third-party evaluator.