Child Work and Schooling: The Role of Household Asset Profiles and Poverty in Rural Ethiopia

Abstract

Child labour is commonly associated with poverty. This is consistent with the expectation that the supply of child labour will fall as incomes increase. However, the empirical evidence for this link is weak. We thus seek to extend the theoretical and empirical framework to better address demand determinants for child labour. We argue that this demand is household-specific given that in Ethiopia, as in most other developing countries, child labour is overwhelmingly performed for the child's own household in the absence of a smoothly functioning child labour market. A simple agricultural household model with a missing labour market shows that household asset portfolios and household composition are the principal determinants of child labour demand. Multinomial logit, mixed logit and simultaneous equation models are used to analyze child time use decisions in the context of rural Ethiopia while addressing issues of income endogeneity and the failure of the independence of irrelevant alternations hypothesis. Our results suggest that the demand for child labour plays a major role in child time-use decisions and that this demand varies substantially between households according to their asset profiles and household composition. In addition, by adequately addressing the demand side, we actually find support for a poverty--child labour link. These results imply that in pursuing asset accumulation-based poverty alleviation policies, attention should be paid to the possibility that this will encourage households to withdraw their children from school in order to take advantage of the increased returns. Copyright 2007 The author 2007. Published by Oxford University Press on behalf of the Centre for the Study of African Economies. All rights reserved. For permissions, please email: journals.permissions@oxfordjournals.org, Oxford University Press.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 105 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.