Julien Grenet is a researcher at the CNRS, an Associate Professor at Paris School of Economics, and one of the founders of the Institut des Politiques Publiques. He is specialised in education economics, public economics and market design. He is known by the general public for his participation in the public debate and the vulgarisation of economic concepts in some media such as France Culture.
He agreed to talk to the magazine about his work as a researcher, the importance for economists to be involved in the public debate and about modern issues that the french educational system is facing today.
Why did you create l’Institut des Politiques Publiques? What are its specificities?
We created l’Institut des Politiques Publiques – IPP – with Antoine Bozio in 2011. It followed a six-year period that Antoine spent in London working for the Institute for Fiscal Studies – IFS, which is our main inspiration for IPP. What was lacking in France was an institute that evaluates public policy, tries to put together the insights of academic research and translates them into policy brief reports targeting a broader audience such as policymakers, journalists and citizens. We felt that there was a very good academic research in public policies existing in France, but most of the results were not really conveyed to the general debate, which is, in my opinion, quite unfortunate. IFS was a good model to import in France. We started small but we have grown up ever since, trying to cover a broad range of topics that are interesting for the public debate, such as tax policies, education, housing, pension, and environment. We also work on health issues.
What is your opinion, as a researcher, on the role of economists in the public debate?
I do not want to be judgmental on what we should do or not do. There are different ways to contribute to the public debate. From my point of view, you do so through the academic output you produce that then spills over onto the public debate. You should also try to meet policymakers. The important thing is to participate in the debate on topics that you know, and only on them. Unfortunately, it is not always the case, and that sort of attitude may damage the reputation of economists. I am personally trying to restrict my interventions to questions on education or housing, since I have worked on it.
Why did you choose to study education, and more specifically social segregation and selection processes, as your main topic?
I started to study education because it was the topic of my Master thesis. What drove me to this is that I come from a family of teachers whose social mobility upwards was entirely due to school. I was shocked in a way by the fact that through the education system, my family managed to climb up the social ladder. Today, we sometimes have the impression that it does not play this role anymore, and we wonder what is wrong with our educational system. I think the tools of economists have a lot to say. What we can learn with economics is improving the efficiency of the educational system.
I went into it for personal reasons; afterwards, the topics that I have addressed are more random. I started working on the return of education, which is a very classic question. Then, since I was working in the same office as Gabrielle Fack, who was working on housing, we thought about working on something in between those two fields of interest. We started working on the effect of school zoning (“la carte scolaire” in French) on housing prices. We thought that this system was one way to assign students to schools, but we actually found out there were many others. We started reading about the school choice mechanism and got interested in that. It is a very dynamic field in economics: how to assign students to schools? How to assign teachers to schools? How to drive students to higher education programs?
In France, there has been a lot going on on the subject lately, and this is important for the public debate. We heard a lot about Admission Post Bac and Parcoursup; those are, in my opinion, important technical tools for policy implications or policy effects. We empirically know quite little about how their effect in the real world. I think this is where we, as economists, can contribute: by improving these tools.
According to the OECD, France is one of the most unequal countries in terms of climbing up the social ladder. What is your analysis?
I think that there are many reasons to it; yet, we can hardly identify them. What the OECD has shown is that at the age of 15, your performance is more determined by your social background in France than in any other country. France is typically in the top three countries where social determinism is the strongest at school.
One reason is that our educational system, especially the middle school system – between 11 and 15 years old – is highly segregated. From research, we know that ghetto schools harm students who are studying there beyond the effect of social background. This segregation in the school system increases inequalities. This might be due to different things: the level of residential segregation is very high in France, and the way we assign students to schools is far from being optimal. As we are assigning students to their local school, if the neighborhood is segregated, then the school is going to be segregated too.
There are many other ways to assign students that we could use. For instance, there is what we call “control school choice” that tries to achieve a balance in the social composition. We could also redesign the school boundaries, or “school catchment”, so that they would be more diverse in their intake students. That is one important topic to be addressed: can we reduce segregation in school by using different methods of assignment?
There is also a problem with how teachers are assigned to school. Typically, young teachers, who are inexperienced, are assigned to the most deprived schools in France, which is obviously a problem. We know that teachers have their biggest efficiency improvement during their first few years of teaching. Hence, students from deprived schools have less probability to benefit from the most efficient teaching.
There is also an issue with the educational system. The French system is very good at selecting an elite and the whole system is created to detect these students who go all their way up to “classes préparatoires”, “grandes écoles” and so on. However, it is not so good to have as many students as possible to succeed. We have a very strong elite but, in the meantime, we are losing a lot of students along the way. France has a high drop-out rate: many students quit school with no certification. Another problem with the system is that vocational courses are seen as a personal failure, unlike many other countries. Therefore, a lot of students who follow this path feel like they failed their studies.
Your research focuses on assignment algorithms. What consequences did you find of such algorithms on students’ choices?
France is a very centralised country; hence, it is more inclined to use these algorithms to assign students and teachers than other countries. There has been very little involvement of researchers and economists to design these algorithms. In fact, a lot of research on this assignment mechanism comes from the U.S.. It is a branch of design mechanism theory which received a lot of visibility thanks to the Nobel prize of Alvin Roth and Lloyd Shapley in 2012. They really transformed the landscape in many dimensions: for example, the assignment of students to school in the U.S. has been completely redesigned in many cities using these algorithms. Kidney exchanges now rely on these algorithms, and there are many new applications, such as social housing allocation.
In France, in my opinion, the main problem is the fact that there is not enough transparency about these algorithms. They exist in order to produce the best possible matching between students and schools, to try to maximize satisfaction while respecting several priority rules. The problem is that, the way the algorithms and the priority rules work are not well known. This has led many people to reject the whole idea of selecting people with algorithms because they feel that there is a black box, like a lottery, when in fact, an algorithm is just a tool.
What really matters is the way you design priorities. If you have two students who apply to a school and there is only one seat left, which student has the priority over the other is a political decision depending on which criteria you promote – students with better grades, students who live closer to the school, students with a lower social background, … This is not sufficiently explained and democratically decided. The issue today is to bring research into these algorithms, so that there are more discussions and a better understanding of the way they work.
You are currently working on a project on social mix. Why it is a topic of interest? What are your preliminary results and your analysis?
We have already said that the lack of social mobility is one of the reasons why there is so little mobility upward in France. The question is how to address this problem. We have several potential ways of doing it. We could use the , we could redesign the school catchment area, we could also close some schools and send some students away from their original choice, like in the city center rather than in a suburban area.
We do not have many empirical results telling us in which case we should use this or that tool nor do we know the actual effect of some tools on segregation. Moreover, these effects are mitigated by the behavior of the parents: if they decide to send their child to a private school, we might not get as much social mix as we initially wanted. Therefore, we are trying to evaluate different ways to assign students to school in order to create social mix and evaluate their effect. To do so, we are using several experiments that were launched across the country, and we try to compare the effect of these experiments on social mix.
The reason why we want to increase social mix is because we believe it is going to reduce inequalities. We are interested in the effect of social mixing on both students’ performance and their non cognitive aptitudes: their self-confidence, their social fatalism and the way they perceive others – the perception of difference. What we are trying to use here is the fact that, in some experiments, even if we found a large effect on social mix,
We try to evaluate this through surveys that are conducted in schools. We are now proceeding in the second wave; two other waves are coming. What we try to evaluate is how does the change of the school social composition individually affect the students through their performance in school and their non-cognitive outcomes. If we look at the literature, there is no evidence of this, especially on the non-cognitive aptitudes, because we cannot really measure it with administrative data. We need to go to the schools and directly ask students some questions. That is our contribution to the literature: trying to answer one of these questions.
Finally, what results in your research were you surprised of?
I did not anticipate the fact that this students’ assignment mechanism would have such a big impact on the composition of schools. I started to work on these assignment mechanisms looking at several high schools in Paris. In 2013, the educational authority of Paris adopted an algorithm to replace the manual procedure. As a part of the algorithm, they created a bonus for low-income students. This bonus would increase their priority, and as a result, the social segregation in high schools in Paris went down by 30 % in only two years, which is huge. This had not been anticipated by the local education authority because they did not think that the way the bonus had been created would make that bonus so large. They did not realise that they gave almost automatically their first choice to low-income students. This completely changed the landscape of Paris, which was the most segregated area in France. This is no longer the case.
By working on this data, I realised that these tools are in fact even more powerful than any reform. For instance, the “assouplissement de la carte scolaire” was relaxing these schools’ catchment areas, so that students could apply to schools that are away from their homes. In reality, this had very little effect on the social composition, whereas these school choice algorithms, like the one implemented in Paris, had a huge impact with very little coverage in the media. The numbers shown in the graph are explanatory: the low-income students now have a bigger set of choices than before. This is one of the surprises of research and economics: it is not because something is not looked upon by researchers or does not get any attention, that it is not existing. You can be like an archeologist: you can dig the results up that were unknown until now and they can change the way you see and understand the educational system.
By Thomas Séron
The TSEconomist 