Abstract

This paper studies the issue of child labour through an integrated economic and philosophical lens. We combine six waves of the Kenya Demographic and Health Survey (1993–2022) to construct a birth-cohort pseudo-panel for long-run structural analysis, with the Kenya Continuous Household Survey (2020–2022) and county-level sugarcane production data for short-run behavioural analysis. We model the dynamic relationship between poverty, educational deprivation, and child labour and examine how sugarcane-sector shocks interact with this system. The long-run results imply persistence across the three outcomes, while the dominant eigenstructure does not fully support a simple self-reinforcing cycle, instead revealing a more complex pattern of co-movement. In the short-run analysis, a shift-share instrumental-variable strategy does not provide sufficient evidence to identify a causal effect of sugarcane exposure: restricting the analysis to the 15 counties with genuine sugarcane variation produces weak instruments and unbounded Anderson–Rubin confidence sets.
Philosophically, the paper argues that child labour cannot be adequately justified as a voluntary choice. Structural deprivation and adaptive preferences undermine the conditions for meaningful agency, thereby violating both rights-based and capability-based normative standards.