Education and HIV incidence among young women: causation or selection?*

Several studies report that schooling protects against HIV infection in Sub-Saharan Africa. This study examines the effect of secondary school attendance on the probability of HIV incidence among young women aged 15-24, using panel data from rural KwaZulu-Natal in South Africa. Three approaches are used to distinguish causation from selection: instrumentation to identify the causal effect, a fixed effects model to control for constant unobserved factors and assessments of the bias from selection on unobserved variables. Although there is a strong negative association between secondary school attendance and HIV incidence, we are not able to find support for a causal effect. Thus, there is no evidence that interventions that increase secondary school attendance in KwaZulu-Natal would mechanically reduce HIV risk for young women. Our focus on school attendance, in contrast to studies that analyze school attainment, might explain the negative finding.