Open Data Portal

Citation Information

Type Working Paper - SALDRU Working Paper No. 251
Title Returns to English skills in the South African labour market
Publication (Day/Month/Year) 2019
While there are 11 official languages in South Africa, English remains the dominant language in the country’s economic and political sphere. Internationally, there is a large body of research providing evidence that proficiency in the country’s dominant language is associated with high labour market returns in terms of earnings and employment. However, the literature in this area of research is considerably lacking in South Africa. This paper aims to extend this literature by analysing the effect of English proficiency on wages and employment probabilities amongst a nationally representative sample of working-aged males in the country. The analysis makes use of Waves 1-5 of the National Income Dynamics Study (NIDS) survey, which is the first survey in South Africa to collect data on English proficiency. Several econometric techniques are employed in an attempt to deal with measurement error and endogeneity which can lead to biased estimates of the proficiency effect in the wage and employment models, including the use of fitted values, instrumental variables, matching on household and spouse characteristics, district fixed effects, and the Heckman selection technique. It is found that, in the absence of these methods, both the wage and employment estimates are biased downwards – in other words, the bias from measurement error dominates the bias due to endogeneity overall. The results provide strong evidence of positive effects of English language proficiency on both employment probabilities and wages. Specifically, a one unit increase in the fitted value of proficiency is associated with a 23-25 percentage point increase in the probability of employment, and a wage premium of 33 percent.

Related studies

Kahn, Amy, Nicola Branson, and Murray Leibbrandt. "Returns to English skills in the South African labour market." SALDRU Working Paper No. 251 (2019).
Copyright DataFirst, University of Cape Town