Education and labour market outcomes

Abstract

Unemployment is widespread in Africa, particularly in South Africa, where high levels contribute to poverty and inequality, correlating with poor educational outcomes. This thesis explores the impact of education on labour market outcomes in South Africa using data from the South African General Household Survey (GHS). It is comprised of three stand-alone papers each focussing on a specific topic, which corresponds to chapters 2, 3 and 4.

Chapter 2 estimates the effects of education on the probability of employment. I use the compulsory schooling reform of 1997 to identify the effect of education on employment outcomes. The mandatory schooling policy required all children aged 7 and older to be in school up until they reached the age of 15 or Grade 9. Using the GHS data for the period 2014-2018 and the two-stage least squares technique, the results indicate that compulsory schooling increased average educational attainment by between 0.04 and 0.18 years. However, the effect differs by gender: an additional year of compulsory schooling raised education by 0.04 to 0.20 years for males and by 0.09 to 0.17 years for females.

On average, an additional year of education increases the probability of employment by 7.3 to 28.4 percentage points, depending on the survey year. When disaggregated by gender, education is associated with a 3.0 to 18.0 percentage point increase in employment probability for women and a 3.0 to 13.9 percentage point increase for men. The results further reveal substantial variation in the effect of education on employment across race and gender. Black African and White males employment gains from education are larger than for females while among individuals of mixed race, there is no clear gender pattern. Theses results emphasize the essential contribution of education to improving employment opportunities in South Africa. In addition, they underscore the importance of targeted education policies that addresses gender and racial disparities in employment.

In Chapter 3 I explore whether the 1997 Compulsory Schooling Law helps explain gender-based differences in wage distributions and whether labour market selection effects influence this relationship. The analysis employs conditional quantile regression and a selection-adjusted approach that accounts for non-random selection into employment. While compulsory schooling laws aim to retain children in school, based on the assumption that increased schooling enhances employment prospects and earnings, South Africa’s education system under performs relative to many other countries, which may weaken the intended impact of compulsory schooling reforms on labour market outcomes.

Using the GHS data for 2018 the study reveals that returns to education are generally higher for women than for men, that the compulsory schooling law raised these returns, and that returns vary across both the wage distribution and levels of education — suggesting that schooling contributes to wage inequality. Additionally, adjusting for non-random selection into employment lowers the estimated returns for women but does not change the relative pattern of returns before and after the compulsory schooling policy was implemented. Expanding access to higher education for previously disadvantaged groups and enhancing the quality of skills training through targeted policies could contribute to narrowing labour market earnings disparities.

Chapter 4 examines how education impacts labour market earnings among Black South Africans. According to the country’s compulsory schooling law, children must start Grade R or Grade 1 in the year they turn six or seven, provided their birthday falls on or before June 30 of the admission year. I take advantage of the natural variation in educational attainment created by this school entry cut-off to estimate the returns to education using a regression discontinuity design, drawing on data from the GHS for the years 2014 to 2018. This study adds to the sparse literature on the effects of school entry policies in Sub-Saharan Africa.

The results reveal signs of birth date manipulation around January 1st, possibly to benefit from school admission rules. To minimize bias from this potential manipulation, I exclude individuals born on January 1st from the analysis sample. This approach helps reduce distortions in the estimated treatment effects. The evidence suggests that starting school at a younger age is linked to fewer years of schooling completed and lower labour market earnings. However, among those who began school early, education has a statistically significant positive effect on earnings. These results highlight key policy concerns about the optimal school starting age and emphasize the importance of providing additional support to younger entrants who might face academic disadvantages.