DETERMINANTS OF PUPILS' EXAM PERFORMANCE IN THE UK AND PAKISTAN

This thesis consists of three chapters exploring the determinants of pupil performance in high-stakes exams in UK and Pakistan. The first empirical chapter looks at the effect of parents’ education on their children’s education adopting Instrumental Variable (IV) methodology. I exploit the 1972 UK RoSLA (raising of school leaving age) as a source of exogenous variation using the Longitudinal Study of Young People in England (LSYPE) dataset linked to the National Pupil Database (NPD). The results show that parental education has a significant and positive impact on their children’s educational outcomes, as measured by performance in GCSE examinations taken at age 16. The subsequent chapter explores key issues relating to teacher quality. Firstly, it investigates the impact of being taught by high quality or low quality teachers on students’ performance in exams, and secondly, which teacher characteristics are associated with student performance. It uses the survey data of 611 pupils from one region of Pakistan studying in Year 9 which is also linked to the administrative data on the students’ exam scores. The unique feature of the data tracks students’ scores across multiple subjects at a single point in time. Teacher fixed effects and pupil fixed effects methods are used. First stage results show that there are significant variations in teacher fixed effects within the schools, suggesting important unobservable differences among teachers. A good teacher, defined as being at the 75th percentile of the teacher fixed effects distribution, is related to an increase in score by 0.15 standard deviations relative to the omitted teacher, while a bad teacher decreases the score by 0.77 standard deviations. Therefore, a pupil having been taught by a good teacher scores 0.92 standard deviations more than the pupil who is taught by a bad teacher (25th percentile teacher) leaving a significant effect on pupil performance. Second stage results suggest that teacher observed characteristics do not explain the variation in teacher quality. The final chapter studies the effect of distance on participation in post-compulsory education in the Pakistan context, and whether socioeconomic characteristics have an effect on achievement in a value-added model taking selection into secondary education into account using a survey dataset on pupils studying in the post-compulsory grade (Year 12) in 2011-2012 from one district of the Punjab province of Pakistan. In this chapter I used two variables as instruments: one is the log of distance to nearest post-compulsory education institution, as a measure of proximity to an educational institution and the second is urban location. The results show that participation and performance in post-compulsory education are two different processes, with participation being driven by availability of post-compulsory institutions within travel distance, while performance once in post-compulsory education is determined by ability.