Artificial intelligence shock in manufacturing: A threat or an opportunity for South Africa’s Labour Market?

Existing research on artificial intelligence (AI) has focused predominantly on its productivity effects in advanced economies, leaving a critical gap in understanding its labour market consequences in developing nations. This study addresses this gap by examining how AI shocks shape employment dynamics in South Africa’s manufacturing sector—a pivotal industry characterised by structural unemployment and uneven technological readiness. Using quarterly data (2014–2024) and a Structural Vector Autoregression (SVAR) model, we find that a positive AI shock triggers an immediate and statistically significant contraction in employment. This negative effect, consistent with task-based theory, persists over the short to medium term, with variance decomposition showing that AI shocks account for nearly 30 % of forecast-error variance in manufacturing employment over a ten-quarter horizon. These findings demonstrate the distinct risks of automation in developing economies and underscore the urgent need for policies that foster human-AI complementarity, such as reskilling and incentives for labour-augmenting technologies. Beyond its employment effects, the study contributes to open-innovation research by demonstrating how AI diffusion shapes technological learning and adaptive industrial capacity in South Africa’s manufacturing sector. The results provide policy guidance for aligning AI adoption with labour market priorities to support inclusive and innovation-oriented industrial growth.