Implementing the right student assignment and school choice model in Chile
By Henrieta Isufllari
One of the Government institutions that saw some of the biggest reforms during the 80’s was, without a doubt, public education. These reforms established that public schools would no longer be under the direct control of the Ministry of Education, but were instead to be run by the municipalities. In addition, tuition-free private schools were to receive equal funding with public schools. In theory, the free market nature of the reforms would allow students (and their parents) free choice over education, but as the national school model in Chile shows, they inadvertently allowed the educational institutions to act like private sector entities. Private schools tried to maximize profit and minimize their costs by fixing school prices, and being very selective on their admissions. Students from well-educated and wealthy households require fewer resources to school, and give better results; therefore this model inherently rewarded parents and families with better economic resources, at the expense of those less fortunate.
Given the socio-economic composition of the Chilean society and the low degree of social and economic mobility, a perhaps unwanted result of these reforms was the deepening of the educational gap. According to The Economist, one of the OECD’s PISA tests a few years ago ranked Chile in the 64th place out of 65 participating countries with regards to the variance of the results according to social class. Results like these, combined with strong ongoing public pressure, are one of the factors at the root of the new educational reforms in the high school system implemented by the Bachelet Government starting this year.
A key change introduced in the new reforms, as MIPP researcher Juan Escobar points out, is assigning the Ministry of Education the role of public coordinator, expecting it to act as an unbiased regulator when matching students with schools, based on revealed preferences by the parents. Professor Escobar emphasizes that in order for the reforms to work, it is important that Chile implements a model that will ensure an optimal match of students with schools, all while maintaining transparency and fairness of the process. Current models already in place in other countries need to be thoroughly analyzed and understood, in order to avoid the long term negative effects that could result from implementing a less optimal model.
Finding the right school model that will optimize the student assignment process was the topic of a recent MIPP seminar by economics professor Tayfun Sönmez, based on his seminal work with colleague Atila Abdulkadiroğlu, titled "School Choice: A Mechanism Design Approach"[i]. During the presentation, Sönmez gave a detailed summary of the history of school choice models in Boston, starting with the early 1970’s until present day. Originally the assignment of students to schools in Boston (and other places within and outside the United States) was done based on geographic vicinity to the school, restricting student enrolment only to schools in their district. This model did not necessarily affect wealthy students, which can usually afford the cost of relocation to a district with a better school (or join a private institution instead), but it put a great burden on poorer families and their children. In the past few decades, as the negative effects of this type of school choice became more apparent, policy makers started to acknowledge the urgent need for a better mechanism: one that prioritizes each student’s needs and preferences, and the preferences of each school.
What followed was the (Old) Boston Mechanism - known as The Direct Mechanism - which required students to rank schools according to their preferences. For each school they had listed, students were then assigned a score based on a predetermined set of hierarchies involving vicinity to school and other important factors. Students with the same ranking were further ordered by a lottery. In the last step, The Direct Mechanism required that students be grouped together based on school preference, and it continued to assign students to their first choice school until all empty spots were filled. The remaining students in each group were then assigned to empty seats in schools listed as their second choice, and it went down the list of school preference until each student was placed.
But as the results of the paper indicate, the Boston Mechanism, although still being used in some places, has one major failure: it is not strategy-proof. Given that schools have capacity constraints, it is inevitable that some students will not get into their top choice, and this drives some parents into trying to manipulate the system by not being truthful with their preferences. The authors rightfully point out that this method inadvertently incentivises parents to “cheat the system” by not revealing their top school choices in hopes of getting their children to a higher ranked school. It especially affects students from low income, poorly educated families, which don’t necessarily have the tools for strategy game manipulation. An extension of this model, The Columbus Student Assignment Mechanism, although it attempts to somewhat correct some of the problems inherent in the Boston model, also fails to provide an optimal application strategy for students and it creates inefficient matching.
The Boston experience shows that assigning students to schools is by no means an easy task. What the authors propose as a solution to the persistent problem of school choice is a Mechanism Design approach, based on the Nobel Prize winning Gale-Shapley student optimal stable mechanism, already in place for college admission procedures. Under the Mechanism Design approach, a central authority (the Ministry of Education in Chile’s case), implements an algorithm that considers student preferences and school priority and capacity, in order to come up with the best matching outcome. This model uses the stability concept (elimination of justifiable envy) to guarantee that a student with the highest priority always gets a spot, unless they get an offer from a better school. The paper proves that this assignment criterion ensures that every student is better off, while strictly honouring the planner’s priorities. The design of a strategy-proof model eliminates the incentive to manipulate the admissions process, since revealing her true preferences is every parent’s optimal choice, and there is no need to game the system. By removing the possibility of “cheating the system” in the hopes for a better outcome, this mechanism significantly simplifies the otherwise exhaustive process of matching students with their preferred schools. It also ensures that students from low income families, and those not as proficient in strategy games, get a fair chance at an education from the school of their choice.
The Mechanism Design approach, based on the Gale-Shapely algorithm has been implemented and widely used in large cities in the US, and in municipalities in countries such as Britain and Spain. The first to adopt this model, (also known as the DA – the student-proposing deferred acceptance algorithm) was the city of Boston in 2005, replacing the previous Boston Mechanism after a lengthy community debate triggered by the rigorous critique that Sönmez and Abdulkadiroğlu outline in their paper.
Education reforms are not only a costly burden to Government budgets, but also a determining factor in the welfare of nations, and as such it is imperative that they are well designed and properly implemented. As MIPP researcher Juan Escobar concludes, “These education reforms provide a great opportunity to use state-of-the-art methodologies to design a school system that effectively gives families equal access to education and, at the end, to better lives.”
[i] American Economic Review, 2003