Kaushik Basu on The Economics of Child Labor via Tyler Cowen. It's subscription only. In addition to what Tyler says,

If controlling child labor is not an end in itself but an instrument for enabling children to grow up into productive and happy individuals, then policies have to be evaluated against this larger yardstick and not just the immediate one of whether it halts child labor. In the poorest regions, society may in fact have to permit children to work a few hours each day. Studies in Peru and Brazil have shown that children's labor is often the only way they can finance schooling for themselves or their siblings and thereby ensure an eventual escape from poverty for their progeny. Such findings raise troubling moral questions, but if policy is to be effective, it must account for the reality of people's lives.

Many commentators have argued--and I agree--that legislative action is not the best way to control child labor, barring some special cases, such as when we have reason to believe that there are multiple equilibria (and so a benign intervention will work). In general, policymakers should work to improve the conditions and earnings of adult laborers so as to diffuse the conditions that foster sending children to work. For instance, during an economic downturn, fluctuations in income may compel parents to withdraw their children from school. Even if the kids return to school later, they find it difficult to catch up and often drop out altogether. Providing parents with access to affordable credit and insurance can help them ride out hard times without resorting to child labor. Small incentives, such as providing children with a midday meal in school or giving parents a subsidy for sending their kids to school, can also sharply reduce child labor, as has been shown in Brazil and Bangladesh.