CMPO, University of Bristol

Preliminary points

1. This evidence is about school choice and the 'sorting' or stratification of pupils between schools.

2. There are different ways of deciding which pupils should attend which schools. This might be done on the basis of ability (test results). It might be on the basis of location, or "neighbourhood schooling", whereby each pupil simply attends their nearest school. Or it might be on the basis of choice: each pupil attends the school they choose. When considering the potential impact of choice-based schooling, it is important to compare it to an alternative basis of assignment, and not just consider it in a vacuum. The most obvious comparator today is neighbourhood schooling.

3. The nature of the school assignment policy affects the composition and stratification of neighbourhoods as well as schools. Typically, neighbourhood schooling will lead to more stratified communities as affluent families cluster round the better schools. Choice-based schooling in principle produces more diverse neighbourhoods, as place of residence is divorced from school attended. Any impact on neighbourhoods and on the demand for particular houses will also affect house prices. For example, enhancing the role of choice would remove or diminish the premium from living very close to a good school, reducing the value of such houses.

4. Part of the point of choice-based schooling is that it breaks the link between where you live and which school you attend. In principle, compared to school assignment on test scores or on which house you own, it is therefore more pro-poor. This seems to be a key issue in thinking about social mobility - how to reduce the dependence of the quality of school a pupil attends on the financial circumstances of her family. There are of course issues about how school choice works in practice, which I address below.

Where we are now

5. Our evidence relates to state secondary schools in England. We have not studied primary schools. Our dataset does not contain private schools. About 93% of secondary school students go to state schools.

6. The current situation is one of school choice. The issues are about reforming the system to make choice more even and to make it work better. It is not the case that we are in a system of neighbourhood schooling, and we are considering moving to a choice-based system.

7. Most pupils have "choice" in the sense of available alternatives. 81% of all pupils have 3 schools within 5km of their home, including 99% of pupils in London, 91% in other urban areas, and 42% in rural areas. This is simply how many schools are nearby, not whether they have spare places.

8. School commutes are on average about 1.6km, with 75% travelling less than 3.5km. In London, the average school commute is 1.65km, in other urban areas 1.55km, and in rural areas 2.35km. In London, a quarter of pupils travel more than 3km, in other urban areas 2.8km, and in rural areas 5.7km.

9. Half of all pupils do not go to their nearest school. About a third do not go to one of their nearest three schools. These are striking numbers. It is very far from the case that at the moment everyone goes to their local school. In London, about 27% go to their nearest school, 44% in other urban areas and 59% in rural areas.

10. It is likely that some of this movement may well not be "choice" in the sense of voluntary. That is to say, in a system with relatively fixed school sizes, some pupils may find their local school full and have to go further away. We cannot say from our data what the balance on this is.

Choice and Pupil Sorting

11. The present system is unlikely to realise the pro-poor potential of school choice, because of two factors. First, ability to exercise choice differs between pupils, and second, places in popular schools cannot increase (fast enough), so that some rationing system is used. This is often based on location, so bringing back the importance of owning the right house.

12. We find that the impact of greater availability of choice is to raise sorting. We measure sorting using standard measures of segregation, and we measure choice by the number of schools easily reachable. We focus on post-residential sorting - that is, we compare sorting of pupils in neighbourhoods to the sorting of the same pupils across schools. We show that the greater is choice available, the higher is sorting in schools relative to the sorting in neighbourhoods.

13. We conduct the thought experiment of assigning each pupil to their nearest school, and measure sorting on that basis. Sorting across schools that pupils actually attend is higher in almost all LEAs than it would be under this thought experiment. In some cases, considerably higher. This reinforces the idea that under the current system, the differential availability of choice plus semi-fixed school sizes works to increase sorting. This is just a thought experiment - as argued above, a policy of "each child shall go to their nearest school" is likely to produce even more segregated schools.

14. We analyse the pupils who do not go to their nearest school. They typically travel non-trivial extra distances to school - on average over 2km further. They also typically go to schools that are performing better on the usual league table scores. In fact, 82% of pupils from the poorest third of neighbourhoods go to better (in that sense) schools, and 97% of pupils from the most affluent third of neighbourhoods do. The positive side of this is that most poor pupils are "trading up"; the negative side is the gap in the extent to which this happens between poor and rich.

Issues for a successful school choice policy

15. Clearly different families have different financial capacities to exercise choice. This involves funding school travel and affording high house prices. If a choice policy is successful, then house price gradients around schools may in time become much less pronounced. But different abilities to afford school travel will remain. Thus policy will need to redress this imbalance by providing subsidised school transport for poorer families.

16. The flexible supply of school places is crucial. This is two-sided: popular schools need to be able to offer more places to avoid (much) rationing, and unpopular schools need to be turned around quickly or closed quickly. There are different ways of making more places available in popular schools beyond simply physically expanding that school. If the important factor in a school's success is transferable, then allowing such schools to manage other schools would achieve the same end. The other alternative is simply to create more good schools in areas without them. The policy on Trust status may be designed to enable all these three methods. The key thing for choice to be real is either that there be no rationing of places, or that the rationing not use some factor, such as location, related to income.

17. This relates to the issue of what makes a good school good - is it the leadership and management of a school, the resources available, or the nature of the peer group in the school. To the extent that such things are transferable, popular schools can take over other schools - essentially the popular school becomes bigger quickly. To the extent that such factors are not transferable, expanding or taking over other schools would simply dilute them. The school peer group seems the obvious possibility here. In this case, popular schools would be reluctant to expand. This obviously would mean that a choice policy would not work.

18. The response of parents to choice will also be important. This depends on what parents want from schools. If it is educational quality, then the choice policy may produce that. If it is in fact a peer group for their child that they consider "acceptable", then a choice policy may bring about more mixing of students as the role of address in determining school declines.

19. Trust status may enable popular schools to expand or take over others more easily, helping choice to work better. But the general greater freedom they would have may well exacerbate tendencies to seek out more able pupils. This will work against the idea of choice, and will tend to produce more segregated schools.

Other Issues

20. One argument made in favour of facilitating school choice is that it will increase the competitive pressure felt by schools. This will lead them to 'raise their game' and work harder to raise standards. In principle this argument makes sense - schools have incentives to produce high scores, and parents have some information on schools performance through the school league tables. Evidence from the US in favour of this view is strong. There is little evidence for England on this point and what there is does not offer strong support.

21. Given recent concerns raised by Trevor Phillips about ethnic segregation in schools, this seems to be an unfortunate time to be encouraging the establishment of new faith schools.

Supporting Documents

1. Background Facts for School Choice in England. Simon Burgess, Adam Briggs, Brendon McConnell and Helen Slater. Attached.

2. The Impact of School Choice on Sorting by Ability and Socio-economic Factors in English Secondary Education. Simon Burgess, Brendon McConnell, Carol Propper and Deborah Wilson. Forthcoming as a chapter in an MIT Press book edited by Paul Peterson and Ludger Woessman.

3. Choice: Will more Choice improve Outcomes in Education and Health Care? The Evidence from Economic Research. Simon Burgess, Carol Propper and Deborah Wilson. http://www.bris.ac.uk/Depts/CMPO/choice.pdf