Schools vary in quality, the position of local schools in league tables are readily available, and (in England) there has long been an element of parental choice. The policy rationale for allowing school choice is that: increasing competition between schools will incentivise increased school quality; it allows better matches given unobservable variation in preferences, and it may increase equality of opportunity when school quality varies, especially by geographical location.
How school places are allocated is a question with potentially important consequences. In England, high-performing schools tend to be oversubscribed. This work studies how parental preferences for (secondary) schools and their admission criteria interact to determine the mix of pupils within schools – together with the mechanism design used. The literature on mechanism design suggests that Roth’s DA (deferred acceptance) mechanism dominates all others. The theory also suggests that missing out on one’s preferred school will be worse under alternatives to DA. We estimate the determinants of preferences, in particular the trade-off between quality and proximity, and the consequences for the mix of pupils in schools using data on preference lists for all parents in England matched to the National Pupil Database that records the attainments and backgrounds of all pupils in English schools. Under the DA method (now universal in England) theory says that preferences should be unaffected by constraints – parents do not have an incentive to strategize in making choices (ie take account of the probabilities of admission). This property resolves the identification problem and has been relied on in all empirical work. However, the theory assumes that parents list ALL schools in rank order. In practice, lists are limited (to as few as 3 in some areas). We use several identification strategies that do not rely on the theory, and reject this assumption empirically. DA also implies that, conditional on preferences, admission only depends on the criteria used for admission – in fact we show that faith schools and academy schools (the English equivalent of charter schools in the US), which receive some additional information directly from parents, do manage to cherry-pick high ability pupils. We also find widespread irrationality in parental preferences and misleading information provided via at least some of the websites that are employed to run the DA mechanism.
This raises the second question: does it actually matter which school one attends and does choice (and the mechanism design) matter. The existing literature is ambiguous when it comes to educational outcomes. Here we use a cohort study (of about 10% of pupils in 5% of schools) who were admitted several years before DA become compulsory for all schools – around half of the areas used DA while the rest used a first preference (FP) mechanism. This data is also linked to NP that allows us to geolocate both pupil and schools. We exploit the richness of the data and a matching method that controls for location to construct counterfactual first best schools. We look at both short term educational outcomes and the long-term consequences of missing out on a place in a preferred school. We find that DA works well and there are modest adverse effects of missing out which are confined to places with short lists, while missing out when FP is used has large negative outcomes.
Our results highlight the fact that schools matter. Missing out on a preferred school due to small differences, such as distance from a school or the use of strategic vs true preferences, can have a long run effects on pupil life chances.
The evidence points to the potential for making DA work better and the web-based nature of the implementation of DA in England suggests that parents could be nudged into making “better” choices.
Bio: I have wide research interests across applied microeconomics: education economics, labour economics, and risk and gambling. I have held a variety of advisory positions in Whitehall, and I founded and ran a consulting company that specialized in the design of lottery games from 1992 to 2007.
I have been an academic economist for over 40 years: currently Professor of Economics at Lancaster. From 1998 to 2009 I was Professor of Economics at Warwick. From 1987 to 2008 I was Professor of Economics at Keele. From 1976 to 1987 I was a Lecturer at Manchester University. I have had the honour of being been a Research Fellow at the Institute for Fiscal Studies in London for most of my career. I am a Founding Fellow at the IZA in Bonn where I am currently the Research Director in Education Economics, and I am a Fellow of the European Economics Association. I have been visiting professor at Toulouse U, Paris II, Sydney U, UNSW, Melbourne, Wollongong, Arhus U, SFI in Copenhagen, Princeton U, and at the Department of Work and Pensions.
I have supervised 25 PhD students, many of which can now be found in prestigious positions around the world. I have four grandchildren, three children, and one wife.