Abstract
We test the impact of a reciprocal adult labor program, Ajuda Mútua (AM), on child labor and schooling. AM was introduced into the province of Nampula in Mozambique, an area where farm production relies on child labor, potentially due to labor and financial market failures. Using difference in differences, we estimate that AM reduces child labor by eight percentage points. We argue that AM reduces child labor by providing low-cost adult labor and potentially increasing farm productivity. We benchmark the AM results against the impact of Village Saving and Loan Associations (VSLA) and AM and VSLA in combination (VAM). Neither VSLA nor VAM reduce child labor. If credit is used in a way that increases labor demand beyond what can be accommodated by AM labor, child labor may increase. We conclude that addressing labor market failures may be more successful at reducing child labor than addressing financial market failures. Results on schooling are mixed.

۱- Introduction
In 2020, around one in children worldwide and one in children in sub-Saharan Africa were involved in child labor. While child labor rates have steadily declined between 2012 and 2020 worldwide, child labor rates in sub-Saharan Africa went up over the same period. In sub-Saharan Africa in 2020, 81.5 percent of child labor was in agriculture, with children often working unpaid on the family farm (ILO and UNICEF, ILO & UNICEF, 2021).

Labor market failures play an important role in the pervasiveness of farm child labor (Basu et al., 2010, Benjamin, 1992, Bhalotra and Heady, 2008, Bhalotra and Heady, 2003, Bharadwaj, 2015, Dumas, 2007, Dumas, 2013, Dumas, 2020). Labor market failures occur when households are not able or willing to exchange labor through the labor market (De Janvry, Fafchamps, & and Sadoulet, 1991). Farm households may not be able to exchange labor if the agricultural cycles they experience are synchronized: labor demand and supply peak and trough at the same time. In this case households may experience excess labor demand in the peak season and use child labor as a buffer. Pressures to use child labor as a buffer will be greater if cash-based labor markets are absent or external (non–household) labor cannot be found. Even if households do not experience synchronised agricultural cycles and external labor is available, households may not be willing to use external labor, still preferring child labor. This can happen when there is no cash to pay for external labor or external labor is too costly, for example due to lower search and monitoring costs (Bharadwaj, 2015, Skoufias, 1995).1 Proposed solutions that encourage external labor usage include long-term personalized contracts and sharecropping (Braverman and Stiglitz, 1982, Eswaran and Kotwal, 1985, Fehr and Gächter, 2000). However, empirical evidence on the impact of these programs on farm labor, including child labor, is scarce.