Gender and technological change: measuring vulnerability to technological displacement in the South African labour market
Date
2022
Authors
Dwolatzky, Leslie
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Across sectors and industries, recent developments in machine learning are having increased applications to processes of economic production. As these systems become more capable of performing complex tasks and become more cost effective, firms are incentivised to incorporate new technologies into their businesses, potentially at the expense of labour. Depending on how easily the occupational tasks that workers perform can be simulated by new technologies, labour-substituting capital has the propensity to become a cause of technological displacement. It is in this context and with reference to the case of the South African labour market that this research project investigates which demographic groups of workers are most vulnerable to replacement by machine learning systems and technological change. Utilising the Brynjolfsson, Mitchell and Rock (2018) ‘suitability for machine learning’ (SML) framework, the project categorises the South African labour force according to the demographic indicators of race, gender and age. Employing relative distribution methods and quantile regression models, the project finds evidence that South Africa’s young, African female population is significantly more vulnerable to replacement by technological systems compared to other demographic groups. This is explained by the types of occupations that South African women perform. Specifically, as the African female population becomes more educated and more sought-after employees in the labour market, they are moving towards more clerical and administrative occupations. While these occupations offer a more stable source of employment and income compared to the elementary occupations they are moving away from, these occupations are theorised to involve tasks which can be more easily performed and automated by new technologies.
Description
A report submitted in partial fulfillment of the requirements for the degree of Master of Arts in the field of e-Science to the Faculty of Humanities, School of Social Sciences University of the Witwatersrand, Johannesburg, 2022