Production of educational ouput: Time-series evidence from socioeconomically heterogeneous populations - The case of South Africa, 1927- 1993

This article furthers understanding of the link between inputs and outputs of the educational production process. In contrast to earlier studies, our article employs time-series data in order to investigate the link between educational inputs and outputs and does so over a relatively long time run, from 1910 to 1993. Strong results emerge from our analysis. We find that educational inputs matter in the determination of educational outputs in the case of South Africa. In this we confirm the findings of A. Case and A. Deaton, but our results further strengthen their results.³ In a second important departure from earlier findings, the relationship between inputs and outputs is found to be statistically significant and strong even in schooling systems in which variation in inputs is low-and of the same order as earlier studies that found inputs to be insignificant. A third important innovation of this study is facilitated by the clear separation of the schooling system into racially segregated processes. In estimating separate educational production functions for different racial groupings, we are able to implicity control for at least some of the omitted background variables that are often argued to bias studies at high levels of aggregation.