School choice, competition and ethnic segregation in Lancashire : evidence from structural models of two-sided matching

• Main Author: Weldon, Matthew Philip
• Other Authors: Titman, Andrew ; Sherlock, Christopher
• Published: Lancaster University 2017
• Subjects: 370.15
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.718654

This thesis uses administrative data from the National Pupil Database (NPD) to examine the phenomenon of ethnic sorting into English secondary schools. The thesis investigates the contribution of parental decision-making to observed patterns of school segregation, and also argues that sorting influences the efficacy of competition-based policies for ed- ucational improvement. Focusing upon three towns with ethnically-mixed populations in the North West of Eng- land – Blackburn, Oldham and Preston – the thesis investigates the hypothesis that segregation can be explained by the aggregate effect of individual choices. It is proposed that parents’ preferences for the ethnic characteristics of school peer groups, and socio- economic differences in parental regard for school test score performance, are important factors in contributing to uneven sorting in schools. The impact of sorting on market structure is also examined, both in relation to the amount of choice that parents en- joy in segregated school markets, and the amount of competition that schools face from neighbouring schools. In order to study school choice in the three markets, the thesis develops a random utility model for parental preferences, incorporating proximity, test scores, and preferences for ethnic peer groups. However, data on individuals’ choice of school are lacking in England, and data on pupil allocations from the NPD do not relate straightforwardly to preferences, since popular schools have limited capacity. To deal with this limitation of the data, the thesis adopts a structural game-theoretic model of two-sided matchings, and develops an innovative method to estimate the parameters of the random utility model. In developing the method the thesis considers the inferential properties – notably identification and large-sample bias – of two-sided empirical models, and provides simulation evidence to support the use of such models for estimation of preferences in large many-to-one markets. The thesis contributes to the, hitherto primarily US-based, evidence on ethnic self- segregation in schools. It concludes that there is robust evidence that preferences for ethnic peer groups contribute to sorting in the markets studied. The thesis also presents evidence that different socio-economic groups place different weight upon academic per- formance when choosing a school. In relation to the effects of sorting on competition, the thesis shows that peer preferences and sorting reduce the potential for schools to compete on quality. Finally, the thesis concludes that preference-induced sorting not only hinders the establishment of self-improving school markets, but also inhibits the inter-group contact necessary for socially-cohesive, diverse communities to flourish.