Introducing strategy-proof secondary school choice in Amsterdam
Scholars were asked to devise a new system for assigning first-year students to secondary schools in Amsterdam. Hessel Oosterbeek, a professor at the University of Amsterdam (UvA), was one of the project’s collaborators.
It started with a TV show. In 2012, in De Wereld Draait Door, VU University Amsterdam Professor of Economics Pieter Gautier comments on the work of 2012 Nobel Prize in Economics winners Alvin Roth and Lloyd Shapley. They have developed a matching theory which analyses allocation mechanisms that can be used on markets where the use of the price mechanism is impossible or undesirable. An example of the use of their insights in the US is the matching of kidneys to patients in need of a transplant.
In the TV show, Gautier says that the assignment of students to high schools in Amsterdam is an example of where the insights of Roth and Shapley could work well. Gautier’s TV appearance encourages the city of Amsterdam and the Association of School Boards in Amsterdam (OSVO) to seriously consider an overhaul of their assignment system. At OSVO’s invitation, Gautier and VU Professor Bas van der Klaauw, University of Oslo Professor Monique de Haan and UvA Professor Hessel Oosterbeek get to work on analysing the functioning of the system and suggest alternatives. Three years later, in 2015, Amsterdam introduces a new assignment system, based, in essence, on the work of the two Nobel Prize winners.
A few turbulent years later, Oosterbeek looks back on the project. In the first year after the introduction of the new matching system, a group of parents came close to undoing what had been achieved. They went to court to demand that students be allowed to change schools with each other after the completion of the assignment process. However, the court understood that the right to change schools would completely undermine the new matching system. Since then, the dust seems to have settled. Oosterbeek: ‘It has been decided to run this system for at least three years. The system cannot prevent that some parents and students will be dissatisfied with the outcome of the assignment process, but I do believe that, overall, the new system performs better than the previous one.’
The Boston system
In Amsterdam, some 8,000 children transfer from primary school to secondary school each year. The Amsterdam school community as a whole has enough capacity to place all prospective students, but problems arise when individual schools cannot accept all students that prefer those schools. In Amsterdam, such ‘popular’ schools are Hyperion, a secondary school offering pre-university education, Spinoza Lyceum for general secondary education and IJburg College for pre-vocational secondary education. If all students were to be accepted by their most-preferred school, the popular schools would have to expand their capacity by some 1,000 seats.
Until 2015 Amsterdam utilised a system – referred to as the Boston system – where students could only express their preference for one school. If, in the assignment process it turned out this school couldn’t place them because it was full, the unfortunate students ended up at one of the schools that had seats left after the first assignment round. The system gave rise to frustration and provoked strategic behaviour. ‘People knew that it was not always in their own best interest to reveal one’s true preference, and that ranking another school first might pay off in the end’, says Oosterbeek. By ranking a second or third school first, a student could avoid ending up at a school where he or she did not want to go. ‘It was difficult to make strategic choices, the process came with a lot of uncertainty, and often resulted in feelings of regret when it turned out that it would have been better to not make a strategic choice. And not making a strategic choice could turn out to be just as painful.’
The four researchers compared the Boston system with two other matching systems, Deferred Acceptance Multiple Tie Breaking (DA-MTB) and Deferred Acceptance Single Tie Breaking (DA-STB). Their analysis needed to be based on information about students’ true school preferences, which was collected by questionnaires that were filled in by a great number of students. It allowed the researchers to run multiple simulations and compare the outcome of the assignment process under the various matching systems.
The simulations showed that applying the DA-MTB system results in less children being placed in their most-preferred school than when the Boston system was applied (82.5% versus 86.4%). On the other hand, the number of students placed in a school that had not been in their top 3 halved. ‘There will always be a trade-off when choosing how to spread the pain caused by some schools’ capacity constraints’, says Oosterbeek. ‘We have quantified this trade-off in our research and have left the choice between the various systems up to the city of Amsterdam.’
According to Oosterbeek, the system not only has quantifiable effects, but other types of advantages as well: ‘What makes this system so beautiful is that it is in the student’s best interest to reveal their true preferences. It avoids strategic behaviour and regret afterwards, and it provides information about the real popularity of schools.’ This information may help schools to make correct decisions about their capacity.
Oosterbeek regrets that the percentage of students who are assigned to their most-preferred schools is lower under the DA system, but points out that the chance that students are placed in a school that ranks among their three preferred choices has increased. As Amsterdam has a large number of excellent schools, even the school that ranked third on anyone’s list should be a good school.
A disadvantage of the new system is that students need to visit a great many schools to draw up their personal top 3. ‘That is a challenging task, but the fact that it leaves both parents and new students better informed can be seen as an advantage’, says Oosterbeek.
More information? Email H.Oosterbeek@uva.nl.