Essays on school choice and child development

Student thesis: Doctoral Thesis


This thesis is a collection of three distinct empirical essays on school choice and child development. We employ applied econometric techniques to estimate the causal effect of two aspects of the Scottish educational system on pupils’ behaviour and academic attainment: multi-grade classes and the expansion of free school meals’ provision. In addition, we examine which secondary schools’ characteristics drive households’ residential choices. These three studies contribute to an extensive literature as well as to a series of heated debates. Chapter 1 is titled “School Performance, Non-Cognitive Skills and House Prices.” This chapter is a single authored study of the drivers of secondary school choice in Scotland. In many educational systems, children attend their closest school, with little room for choice. This implies that households decide where to live, at least in part, to secure their children a place at their preferred school. Advocates of school choice motivate its importance in light of improved efficiency, but also better matching effect for single students. Conversely, other experts believe that letting parents decide which school their children should attend may create distortions whereby schools improve under aspects which are preferred by parents, i.e. a more favourable selection, as opposed to improved effectiveness. A wealth of studies shows the complexity of school choice, which lends itself to strategic behaviour. For this reason, studying the drivers of school choice is imperative. Most of the literature to date focuses on school-level average test scores or socio-economic composition as a factor determining school choice. This, however, might reveal very little about the inherent quality of a school. Another strand of the literature focused on measures of value-added (or effectiveness). To date, the evidence points towards parents not caring about objective measures of effective ness, but only about peers’ performance, even if this is mainly driven by selection. One characteristics of these studies, however, is that they focus mostly on measure of value-added which are unknown to parents. Therefore, the question arises whether this lack of an effect is driven by a lack of information, rather than a lack of interest. Furthermore, most of the studies to date focus on a limited number of school performance indicators, thus overlooking multidimensionality of school outputs and whether parents care about these. Finally, despite the extensive evidence on the effects of spillovers in peers’ non-cognitive skills, little is known on whether this dimension of school quality is valued by parents. In this study I examine secondary school choice by estimating house price capitalisation for an array of school characteristics. To overcome the endogeneity of school ‘quality’ I leverage Scotland’s strict residence-based attendance system and employ a boundary regression discontinuity design, whereby I compare prices of property which are located in the same neighbourhood but on opposite sides of catchment area boundaries. This study contributes to the relevant literature in a number of ways. First, it uses a vast array of school outcomes, including post-school destinations, as potential drivers of school choice; second, it attempts to overcome the debate in the literature about the lack of interest in school value-added by using a highly salient and easy-to-understand measure of value-added; third, it uses peers’ non-cognitive skills’ measures as a potential driver of school choice; fourth, it provides nation-wide evidence. I find that peers’ academic performance and socio-economic composition command a house price premium of about 4% for a one-standard-deviation increase. On the other hand, school value-added and non-cognitive skills do not appear to be capitalised into house prices. Chapter 2 is titled “Early-Years Multi-Grade Classes and Pupils Attainment”. This chapter is a co-authored study of the educational gains from sharing the classroom with older, more experienced peers, and focuses on Scotland. There is an extensive literature studying different dimensions of peer effects, e.g. gender, ability, socio-economic background. On the other hand, evidence on age peer effect is relatively scarce. This is despite multi-grade (composite, henceforth) classes, i.e. the practice of grouping pupils from adjacent year groups, being commonly employed in many educational contexts. In this paper we examine whether primary school first-graders experience cognitive gains by sharing the classroom with older peers (second-graders). We match data from the Scottish Pupil Census to academic attainment data obtained from Curriculum for Excellence records, i.e. teacher-assessed attainment in literacy, numeracy, reading, listening and speaking. This empirical exercise carries several methodological challenges. The most relevant is given by the fact that neither are composite classes randomly generated, nor are pupils randomly assigned to them. For instance, selection into composite classes can happen on the account of abilities or to preserve existing networks of friends. To disentangle this endogeneity issue, we make use of an algorithm employed by Scottish Local Authorities, which predicts the creation of composite classes both at the extensive and intensive margin, based on enrolment projections. The algorithm leverages class size caps set by the central government, and it predicts the creation of composite classes any time this enables a saving in the number of classes to be created, and therefore teachers to be appointed. We show how as little as one additional pupil can trigger the creation of at least one composite class across the entire school. Therefore, the creation of composite classes is, in some circumstances, as good as random. We use the prediction from this algorithm as an instrumental variable for the actual composite status. This study contributes to an emerging, small, literature by adding to its external validity. The existing evidence predominantly pertains to rural areas, whereby small cohort sizes dictate the necessity to create fewer classes by pooling different grades. By contrast, composite classes are widespread in Scotland, in rural as well as urban areas. We find that first-graders who share a classroom with second-graders experience educational gains of about 0.2-0.3 of a standard deviation in literacy and numeracy, in line with previous studies. We show that the effects are not driven by gender, socio-economic or urban/rural status. Finally, we do not find evidence of a detrimental effect on older peers (second-graders). Chapter 3 is titled “Does the Provision of Free School Meals Improve School Attendance and Behaviour?” and is also a co-authored work. This chapter is an evaluation of the expansion of free school meals’ provision which took place in Scotland in January 2015. Free school meals (FSM, hereafter) provision is the subject of a long-standing debate. In many contexts, FSM are a means-tested benefits, meaning that their eligibility is contingent to households financial circumstances. Advocates of FSM expansion argue that FSM improve attendance, academic performance, behaviour, diet as well as uptake among previously eligible pupils by reducing stigma. Opponents, on the other hand, point to the high costs associated with such policies, as well as noting the potential detrimental effect on diets. Scientific evidence largely supports the claims of the advocates of such policies. In addition, some studies find that FSM provision leads to improvement in households’ finances. Apart from these studies, most of the evidence is confined to short-term outcomes such as test-scores and overlooks the effects on noncognitive components, which are equally likely to drive future attainment. In this study we estimate the impact of FSM provisions’ expansion on students’ attendance, health related absences, and misbehaviour. We do so by leveraging a change in policy which extended FSM eligibility to all students in the first, second and third grades of primary school (P1 to P3). We estimate a difference-in-differences (DiD) model with variation in treatment intensity. This is determined by the pre-policy share of FSM-takers. Therefore, schools with relatively fewer pupils taking FSM one year before the policy are indeed more exposed to it. The importance of our study can be assessed from a twofold perspective. First, Scotland, in line with UK, reports among the highest rates of child obesity worldwide. Second, outcomes such as attendance and misbehaviour are correlated with the ‘Big Five’ personality traits and therefore non-cognitive skills. Overall, we find precisely estimated null effects of the policy on all dimensions we consider. The absence of effects is not driven by school characteristics which could have plausibly affected the success of the policy, i.e. deprivation, rurality and school resources.
Date of Award15 Feb 2023
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde
SupervisorMarkus Gehrsitz (Supervisor) & Stuart McIntyre (Supervisor)

Cite this