Abstract
The aim of this empirical study was to calculate private and social rates of return to education and to test the hypothesis that education is a major determinant of earning differentials for blacks in South Africa. Data for 23 278 working men and women were extracted from the 1985 Current Population Survey files. This comprised a sample representative of the black population in the Republic of South Africa excluding the former Transkei, Bophuthatswana, Venda and Ciskei territories. Lifetime earnings profiles were constructed from these data for five educational levels, namely, no schooling up to Standard 1; Standards 2 to 4; Standards 5 to 7; Standards 8 to 9; and Standard 10. Schooling was assumed to account for 60 per cent of the earnings differentials between these profiles. Imputed average household outlays on schooling were taken as the private direct cost of education. These were supplemented by estimates of per pupil spending by the various government departments responsible for black education, to calculate the social costs per year of primary and secondary schooling. Indirect costs in the form of imputed foregone earnings were included from Standard 5 (age 15) onwards. The private internal rates of return to education of males were found to be about 16 per cent at primary level and 24 per cent for secondary schooling. Corresponding social rates of return are about 6 per cent for primary and 15 per cent for secondary education. Estimates for females showed negative rates of return between no schooling and Standards 2 to 4, the private and social rates being estimated at ‐1 per cent and ‐4 per cent respectively; from Standards 2 to 4 to Standards 5 to 7, a private rate of 12 per cent and a social rate of 4 per cent are reported; and for the remaining secondary school phases private rates of about 32 per cent and social rates of about 15 per cent are estimated. The study confirms that education is a major determinant of earnings differentials and education expansion can be regarded as a powerful earnings equaliser.
Notes
Department of Biomathematics and Statistics, University of the North West, Mmabatho. I am indebted to Mr AR Donaldson and Professor A Roux for reviewing my work from time to time and making constructive criticisms and suggestions at different points of its development. Also special thanks go to the University of the North West for financing the research.