理解社会安全网与家庭内部食物分配：来自孟加拉国的实验证据

Evidence shows that social protection can improve diets, but little is known about how impacts vary within households, the extent to which the modality of the transfer affects how it is distributed across all household members, whether adding training on the importance of nutrition and diets alters the way transfer resources are allocated within the household, relative to a transfer alone, and if differences in allocations are shaped by differences in livelihood opportunities. We use individual food intake data from two randomized control trials fielded in rural Bangladesh to address these questions. Our results overwhelmingly demonstrate that food gains are distributed equally, regardless of the type of transfers households received (cash, food, or combination), inclusion of nutrition training, regional context, or specific dietary outcome measured. These patterns of findings hold when we consider several extensions: (1) analyzing more aggregated demographic groups; (2) considering alternative measures of diet; (3) analyzing shares rather than levels; (4) considering impacts relative to deprivation at baseline; (5) analyzing impacts on non-food outcomes that can be assigned demographically; (6) re-estimating impacts using alternate samples and alternate estimation models. Where the few significant differences are found, they are often small in magnitude and in favor of children.