The impact of early grade repetition on test scores: Evidence from a regression discontinuity design in South Africa

This paper provides the first causal evidence on the effects of early grade repetition in South Africa. Using a large administrative dataset covering six provinces from 2017–2023, I implement a fuzzy regression discontinuity (RD) design exploiting published promotion thresholds to estimate the impact of repeating Grade 1 or Grade 4 on subsequent test scores. Grade 1 repetition raises achievement in Home Language, Mathematics, and English First Additional Language by over one standard deviation in the following grade, with effects diminishing but remaining sizeable until at least three grades after the repetition. Grade 4 repetition yields smaller initial gains, which fade less over time. The initial effects of Grade 1 repetition are larger than those reported in the most comparable RD studies, reflecting rapid cognitive development in early primary years, while the Grade 4 effects are large but consistent with international evidence. The findings indicate that repetition can be an effective remedial tool even in settings with limited structured support for repeaters, especially in the context of literacy deficits amongst English learners. The paper contributes to the literature on grade repetition, skill formation, and education policy in middleincome countries, providing evidence relevant for the design of promotion rules and remedial strategies.