Research
PhD Candidate in Economics
Tinbergen Institute
Vrije Universiteit Amsterdam
Stellenbosch University
Working papers
Making School-Set Assessments Comparable: How School Assessments Hide Learning Inequality
Job market paper
Assessment has been found to raise learning system-wide where its results can be compared across schools, yet at most a quarter of countries test every primary-school leaver comparably. I recover comparable learning outcomes, without a new test, from assessments that are incomparable because their difficulty varies across schools. Because students sit their secondary school’s assessments within months of arriving, I can compare primary schools through students who share a secondary school, and secondary schools through students who share a primary school. Two-way fixed effects chain these comparisons from 1.2 million South African students into a common scale. The gap between rich and poor schools halved on raw marks between 2018 and 2022 but did not narrow on the common scale. Re-ranked on the common scale, about three-quarters of primary schools change learning quintile. The common scale predicts the national secondary leaving examination, sat five years later, better than raw marks do. Easier assessments do not buy a higher-achieving secondary school: at the same raw mark, students whose primary school set easier assessments join lower-achieving classmates. The common scale supplies comparable intake marks, a necessary condition for any value-added measure.
Work in progress
Provision Not Deployment: What a Teacher Post Buys in South African Schools
Draft available on request
South Africa allocates teachers by formula and monitors staffing by the pupil–teacher ratio, not by class size. Rebuilding school timetables from national records for two-thirds of the country’s students, I measure the class each child actually sits in. The pupil–teacher ratio averages 32.8, but the class the average child sits in holds 43.3. The gap is widest in the poorest schools, and a school’s pupil–teacher ratio barely predicts it. To find out why, I exploit the formula that sets each school’s teaching posts from last year’s enrolment, which shifts staffing for reasons unrelated to how the school organises this year’s classes. Using this variation as an instrument, I find that only about half of a post reaches the classroom as smaller classes. The rest is absorbed, largely as free periods. With the teachers they already have, schools could cut the class the average child sits in by up to a fifth. Where the absorbed time is largest, I find no measurable return in learning.
Invisible Borders, Living Institutions: How Apartheid’s Bantustans Still Shape Schooling
Draft available on request
Does a border abolished in 1994 still cause educational disadvantage, and why? I follow each child who changes school across the former KwaZulu border and compare them with themselves, before and after the move, against children on the same path who changed school without crossing. Sorting explains one-half to two-thirds of the raw gap. What remains is causal and runs both ways: crossing into the former homeland raises a child’s chance of repeating a grade by 0.94 percentage points, and crossing out lowers it by 1.19. Despite one national pay scale, schools inside the border pay 9.2 per cent less per teacher, because the more senior teachers leave: a single wage scale laid over unequal places reproduces, through who chooses to teach where, the inequality it was meant to end. Across 21 birth cohorts the gap has not closed.
Book chapters
Bridging the 20-Year Literacy Divide: African Language Reading Progress in South Africa
In Tracking Changes in South African Reading Literacy Achievement, eds. S. van Staden and C. Combrinck, Brill, 2023
Abstract Book (PDF) Publisher
South Africa’s language education policies acknowledge the country’s exceptional linguistic diversity, with all 11 official languages represented. At Grade 4 the language of learning and teaching (LoLT) differs by school. Using PIRLS 2006–2016 data, we describe South Africa’s changes in LoLT, decomposing language proficiency by province, asset ownership quintile, and whether the LoLT matches the learner’s home language. We find three primary results. First, there is a high match between home language and LoLT, which refutes the myth that classrooms are becoming increasingly multilingual, apart from in Gauteng. Second, African languages, especially Nguni languages, are significant contributors to the improvement seen in PIRLS between 2006 and 2016. Third, there is still a significant gap in performance between English and Afrikaans compared to all other South African languages. At the 2010–2016 rate of progress, it would take another 20 years before the fraction of Nguni language speakers that meet the Low International Benchmark would be the same as Afrikaans or English speakers in 2016.
Mathematical Stunting in South Africa: An Analysis of Grade 5 Mathematics Outcomes in TIMSS 2015 and 2019
In Early Grade Mathematics in South Africa, eds. H. Venkat and N. Roberts, Oxford University Press, 2022
Abstract Book (PDF) Publisher
TIMSS data show stagnation in Grade 5 mathematics outcomes in South Africa from 2015 to 2019. We analyse specific TIMSS items relating to number and fractions and show that even before the learning losses induced by Covid-19, most Grade 5 learners were effectively at Grade 2 or 3 level. Two in three Grade 5 learners (61%) could not do basic multiplication, and three in four (75%) were unable to answer a Grade 3 subtraction problem. The stagnation is robust to sampling and item-scaling interrogations, and the percentage of Grade 5 children in classes of more than 50 rose from 16% in 2015 to 34% in 2019. As there have been no other identifiable and significant changes over this period, we argue that increases in class size are likely to have contributed to the stagnation, exacerbating the endemic deficits in teachers’ content knowledge and pedagogical content knowledge.
Policy papers
Education, Inclusive Growth, and Fiscal Consolidation
SA-TIED Working Paper 247, UNU-WIDER, 2025
This report explores strategies for improving education outcomes in South Africa, with a view to strengthening economic growth and employment over the long run. It argues that the key to more rapid educational progress, both in the quality of schooling broadly considered and in expanding higher education outcomes, is foundational literacy and numeracy. A special focus is proposed on ‘marginal learners’: those who qualify for admission to higher education but whose likelihood of successfully completing degrees must be improved before university entry. Opportunities should be expanded in high-functioning schools, and stronger links are needed between post-school education and training and workplace requirements. The report examines the importance of education for growth and employment, the current state of education in South Africa, and education policy reforms that take into account South Africa’s current fiscal constraints.
Inclusive Development in South Africa: Elements of a Structural Reform Strategy
SA-TIED Working Paper 273, UNU-WIDER, 2025
Rationalising Micro, Multigrade and Under-Subscribed Schools in South Africa: A Pathway to Fiscal Efficiency and Educational Equity
Teacher Demographic Dividend Research Note 19, 2024
Summary Note (PDF)
South Africa’s fiscal challenges and the imperative to improve educational outcomes are renewing interest in reducing the number of small, multigrade, and under-subscribed schools. At present, 23% of all schools are classified as multigrade, despite serving only 5% of the nation’s learners, yet they command a disproportionate share of operational, maintenance, and leadership costs. Spatial analyses suggest that approximately 2,400 small, multigrade schools may lie within reasonable walking distance of neighbouring schools. Rationalisation could free up financial and human resources, especially principals’ posts and fixed-cost budgets, for reinvestment in teaching and learning. When accompanied by community engagement, reliable learner transport or boarding, and support for receiving schools, it can advance both fiscal efficiency and educational quality.
Education and Growth: Some Ideas on How to Improve Results
Bureau for Economic Research, 2025
Teacher Knowledge and Teacher Age: What Are the Levels of In-Service Teacher Knowledge?
Teacher Demographic Dividend Research Note 10, 2022
Master’s thesis
Invisible Borders – Persisting Scars: How the Apartheid Homelands Define Contemporary South Africa
MSc thesis, Wageningen University & Research, 2021
Abstract Thesis (PDF)
I use a spatial regression discontinuity design to examine the long-run welfare reductions caused by the apartheid homelands (1948–1994), the only areas in apartheid South Africa where Black African people could reside and own land. The homelands have caused a persisting reduction in educational attainment (decreasing the school completion rate by 2.17%) and in education inputs (increasing students per teacher by 7.66%), while doubling the number of schools per square kilometre. They have also doubled long-run population density, and erosive agricultural practices have reduced topsoil quality. A historical analysis points to the limited size of the homelands, denaturalisation, the migrant labour system, apartheid rural policy, Bantu education and property dispossession as likely causes. The welfare of South Africans today is significantly reduced by living just 5 km on the wrong side of a now non-existent homeland border.