High school dropout in South Africa: evidence from individual-level panel data

This paper examines the determinants of high school dropout, using longitudinal data from South Africa. Employing a random-effects probit model, we account for possible selection bias in the decision to dropout from high school by estimating both observable and unobservable factors that determine high school dropout. We allow for both parametric and non-parametric distributions of individual-specific unobserved heterogeneity in our estimations. Empirical results show that ignoring unobserved individual differences understates the likelihood of dropping out of high school and the probability of completing high school education. Results further identify little or no mother's education, being black or coloured, poor academic performance, low innate ability, teacher's absenteeism, and living in an urban area as observable factors that drive selection into high school dropout.