Global Poverty Research Group

Human capital acquisition and school quality in South Africa

The Research question

Human capital is both a dimension of well-being, incorporated for example in the UNDP human development index, and a potentially important determinant of well-being, reflected in the view that investment in human capital underpins the ability of poor countries to grow. How human capital is supplied and how it is used is very dependent on the institutions which characterise a society. As part of our research we proposed to study the links between human capital formation, institutional structure and well-being. One specific example of institutional structure is the issue of school quality. In almost all countries, higher quality schools coexist with lower quality ones. What are the implications of these differences for human capital acquisition by the poor?

     As has already been noted in research reported on above, the characteristics of blacks and whites explain only a relatively small part of the very large race gap in unemployment and earnings in South Africa. Building on this work, further analysis by Geeta Kingdon and John Knight, has explored the particular role of school quality in explaining the black-white gaps in unemployment and earnings in South Africa. The objective was to examine to what extent the part of the wage (and unemployment) gap between blacks and whites in South Africa that is not explained by black-white differences in observed characteristics (such as education and location), can be explained by the differing quality of schooling faced by black and white workers when they were students.

Measurement of schooling quality has been contentious in the literature. While early studies of the educational production function concluded that school inputs were not good indicators of schooling quality as they were not well related to student learning, carefully conducted recent studies show that class-size is an important determinant of student achievement levels.  Following the recent well-accepted work of Anne Case and Angus Deaton on school quality (which shows that class size is an important determinant of schooling enrolment, grade attainment and cognitive skill levels in South Africa), pupil-teacher ratios were used an indicators of schooling quality. District level data on race-specific pupil-teacher ratios in 1991 was matched with 1999 household survey data on individuals aged 20-30 years old in 1999 since individuals in this age range in 1999, would have been of school-going age in 1991, the year to which the school quality data pertain.

     The data suggests that poor quality schooling is responsible for a very large part of the black disadvantage in employment vis a vis white persons in South Africa. Inclusion of the school quality measure in the unemployment probit (and in the earnings function) greatly increases the ‘explained’ proportion of the black- white gap in unemployment (earnings). Thus, the data shows that the non-explained (potentially discriminatory) parts of the race gap in unemployment and earnings are much smaller than might have been concluded without access to schooling quality information. The results support the conclusion that the labour market disadvantage of blacks relative to whites in contemporary South Africa is due, in large measure, to prior discrimination in the schooling system which consigned blacks to poorer quality schooling.

     While the results greatly reduce the potentially discriminatory part of the race gap in labour market outcomes, the picture is not a happy one. Firstly, a good part of the race gap in unemployment incidence and earnings remains even after controlling for black-white differences in schooling quality. It is possible that this residual gap is due to a number of productivity-related factors which remain unobserved in our data but which employers may at least partly observe at the time of recruitment, such as motivation, ambition, and drive. However, if we accept the premise that it is implausible to ascribe wholesale differences between races in average levels of motivation, drive, and ambition, then the residuals here may well reflect employer discrimination.  Secondly, much of even the so-called ‘explained’ component is due to racial differences in characteristics that are the product of prior discrimination in the schooling system and in the wider political system.

     Whether the large black-white gap in unemployment is due to employer discrimination against blacks or due to other factors is another policy relevant question. If much of the race gap in unemployment is due to labour market discrimination by employers, then policy makers may wish to tighten employment equity regulations. But if much of the black-white unemployment rate gap is due to pre-labour market discrimination in the schooling system, then the policy implications are fundamentally different: they would consist of improving the quality of education for black children. The fact that inclusion of school quality greatly lowers the portion of the black-white unemployment gap that is not explained by black-white differences in observed characteristics suggests that the considerable differences in school quality faced by black and white children are responsible for a good part of the blacks’ higher unemployment rates.  It suggests that a powerful way of improving employment equity by race in the future would be to equalise the quality of education for black and white children.

Recent papers

Fassler, M., G. Kingdon, and J. Knight, ‘Transitions from unemployment to employment in South Africa’, mimeo, Department of Economics, University of Oxford, May 2002.

Kingdon, G. and J. Knight, ‘Quality of schooling and the race gap in labour market outcomes in South Africa’, mimeo, Department of Economics, University of Oxford, October 2002.

Reseachers to contact for this project

Geeta Kingdon & John Knight