School Types. Evidence on the Relative Effectiveness of Different School Arrangements

Organised by

• Department of Economics

We estimate the relative effectiveness of several secondary school types operating in the English school market, where differentiation, accountability and parental choice have been prevalent for over three decades. We overcome the self-selection of pupils into different school types by leveraging the quasi-random pupil-level variation in the probability of being assigned to a given school type that is generated by the Gale-Shapley centralized mechanism that assigns pupils to schools and governs oversubscriptions. To break ties, the various schools use a mixture of entry test scores, home-school distance and lottery numbers. We efficiently combine these sources of variation using the recent advancements in "Research Design Meets Market Design" – RDMD – methods developed in Abdulkadiroglu et al., 2022. Our evidence is based on population-level administrative data for the 2010 cohort of applicants to secondary schools in Birmingham (about 15,000 pupils). We find that selective grammar schools provide substantial academic returns – despite being attended by the most talented students. Stark changes in peer composition for attendees hint at ability peer effects as the most likely mechanism generating this result. Autonomous academy schools do not generate positive returns – irrespective of whether they are 'conversions' or 'take-overs'. Finally, we find non-positive and insignificant effects also for faith school attendance. This is the first attempt at estimating the effects of different school types within the same education market (Birmingham) and using a single identification strategy (RDMD), based on the assignment mechanism.

Co-authors: Filip Gonshorek (ZEW), Thilo Klein (ZEW) and Olmo Silva (LSE)