Child labour is not uniquely a contemporary phenomenon and, for economic historians, sociologists and anthropologists, it is not a new subject. For economists, it is. A lively interest in the economics of child labour is less than a decade old but has been growing exponentially, in line with public interest in the subject. Discussion of child labour has been stimulated by contemporary debates on globalisation and international labour standards. For instance, the suggestion that the US ban imports from developing countries that are produced with the help of child labour has motivated analyses of the rationale for and the likely consequences of legal actions of this sort. Economists have produced models focused on trade and labour markets at one level and on decision making within the household that supplies child labour on the other. They have investigated the predictions of these models using both cross-country (macro) data and household and individual level (micro) data. This paper draws largely on the recent economics literature although it reflects insights and empirical findings from across disciplines.
One reason, it may be argued, that child labour has become such a fertile research topic in economics is that it has arisen at a time when the literature on endogenous growth1 has matured and there is fairly widespread agreement in the field that the cornerstone of pro-poor growth strategies is investment in human capital. This investment occurs largely in children, in building their health and educational capital. Thus the inherent value of education apart, it is recognised that it has a functional role in encouraging productivity and growth. Even in societies where nobody is very poor in absolute terms, there is a case for subsidising education that rests on positive externalities associated with it.2 In developing countries, this case is strengthened by the fact that many poor households are credit-constrained and simply unable to afford, by their own means, to send their children to school. Their children therefore grow up to be relatively poor and this poverty is potentially perpetuated across the generations. With large sections of society caught in poverty traps of this nature, overall economic growth and human development is constrained. International organizations of varied persuasions (such as the ILO, UNICEF and the World Bank) have adopted the goal of reducing child labour and currently have fairly large programmes directed at this goal. While research is now fairly active, this is an area in which policy has run ahead of research. There is now an urgent need to identify which of a number of intuitively plausible policies is likely to be most effective. This, as we shall see, depends on understanding the nature and causes of child labour.
In setting out an analytical structure for thinking about why children work, it is useful to go down to the level where parents are faced with a choice between sending a child to work or to school. This choice can be structured in terms of constraints, incentives and agency. The common presumption is that it is the constraints of poverty that drive children into work. If this is so, the appropriate policy will attempt to relax this constraint by, for example, offering cash transfers to very poor households. However, if schools are costly, unavailable or of poor quality then, even when the household is not very poor, it may be rational to send the child to work. This is especially so for work, like farm work, which can produce experience gains that are valuable if the child grows up to inherit the farm. This is the case where incentives dominate: the rewards to work exceed the rewards to schooling. In this case, policies that make schooling more attractive are likely to reduce child labour. A third interesting possibility relates to parent agency. Parents typically decide whether the child works and their interests may not always coincide exactly with the child’s interests. If parent altruism were limited then a case could be made for legislative action involving a ban on child labour or compulsory schooling laws. Table 1 summarises these alternative hypotheses, showing how they produce dramatically different policy implications. Region-specific research is therefore needed to establish whether policy should primarily address poverty, inadequate schooling or the freedom that parents have to decide what their children do.
The empirical literature is now quite vast, including studies from a number of countries in each continent. A striking feature of this literature is the sheer variety of results that it has produced. It is unclear whether this reflects the diversity of country experiences of child labour or whether, instead, it reflects weaknesses in methodology.3 This paper will attempt to cull from the corpus of results whatever patterns are discernible and likely to be policy-relevant.
The rest of the paper is organised as follows. Section 2 presents an overview of the incidence and nature of child labour. Section 3 briefly contrasts child labour in Asia and Africa. A review of theoretical models of child labour is presented in Section 4, where the discussion is aimed at bringing out the policy relevance of the different models that have been proposed. The evidence is surveyed in Section 5, which relies primarily upon recent microeconometric research. Section 6 presents an in-depth analysis of the data for Ghana and Pakistan in order to illustrate contrasts between an African and an Asian country. In Section 7 we bring theory and evidence together and, using the Pakistan data, illustrate the potential for research to inform policy. Section 8 concludes with a broader discussion of what we have learnt so far that is relevant to policy making in this area.