I'll inject a passage from my development economics textbook to this discussion, especially since it's quite relevant. Naturally, because economists view the world in terms of trade-offs, it'll probably make everyone mad. Horseshoe theory at its finest. I typed this really fast, my apologies for any errors. But first, a shorter explanation of why child labour exists in relatively broad and general terms (caveating, as is especially relevant in the GA, the non-existence of slavery). It doesn't exist because people are evil and assign labour to children. It doesn't exist because their parents are evil. It exists because households attempting to meet budget constraints in extremely low productivity labour market environments have to leverage their only advantage to survive: their labour. Returning to the passage.
Working conditions are often horrendous; the ILO resorts that some of its surveys show that more than half of child labourers toil for none or more hours [*392]per day. The worst forms of child labour endanger health or well-being, involving hazards, sexual exploitation, trafficking, and debt bondage. In a 2011 publication, the ILO reported that every year, about 22,000 children die as a result of work-related accidents. Clearly, child labour is not an isolated problem but a widespread one...
Nevertheless, it is not obvious that an immediate ban on all forms of child labour is always in the best interests of the child. Without work, a child may become severely malnourished; with work, school fees as well as basic nutrition and health care may be available. But there is a set of circumstances under which both the child labourer and the whole may be unambiguously better off with a ban on child labour: multiple equilibria. Kaushik Basu ahas provided such an analysis and we shall first consider his simple model which shows how this problem may arise.
To model child labour, we make two important assumptions: First, a household with a sufficiently high income would not send its children to work. As one might hope, there is strong evidence that this is true, at least most of the time. Second, child and adult labour are substitutes. In fact, children are not as productive as adults, and adults can do any work that children can do. This assertion is not an assumption; it is a finding of many studies of the productivity of child labourers in many countries. It is is important to emphasise this, because one rationalisation for child labour often heard is that children have special productive abilities, such as small fingers, that make them important for the production of rugs and other products. However, there is no support for this view. In essentially every task that has been studied, including carpet weaving, adult labourers are significantly more productive.
For Kaushik Basu's model, refer Kaushik Basu,
Child labour: cause, consequence, and cure, with remarks on international labour standards, 37 J Economic Literature 1101 (1999). At a more broad level, the generally accepted means to solve the issue – which is that all families would be better off if children did not work, since it would depress the labour supply amongst especially low-productivity workers, raising their wages – is to create a conditional cash transfer programme which would give people money and in-kind transfers (e.g. school meals and healthcare) for sending their children to school. This is a response to the fact that for many families, the opportunity cost of sending their children to school, with the direct costs, is too large for them to bear. Lowering that cost is therefore imperative for long-term human capital development policy.
There is some debate over whether unconditional cash transfers to basically all poor people in these areas is as or more effective than conditional cash transfers. Empirical evidence from randomised controlled trials shows that UCT is more effective at long term improvements which reduce the likelihood of teenage pregnancies and marriages. The cost of those UCTs as well is not large relative to those in more conditional cash transfer programmes, with something on the order of 4 dollars per month to parents and 1 dollar per to school-age children. See Sarah Baird et al.,
Cash or condition? Evidence form a cash transfer experiment, 126 Quarterly J Economics 1709 (2011).