This paper proposes a strategic framework to account for fertility choices in polygamous households. A theoretical model specifies the main drivers of fertility in the African context and describes how the fertility of one wife might impact the behavior of her co-wives. It generates predictions to test for strategic interactions. Exploiting original data from a household survey and the Demographic and Health Surveys in Senegal, empirical tests show that children are strategic complements. One wife raises her fertility in response to an increase by the other wife, because children are the best claim to resources controlled by the husband. This result is the first quantitative evidence of a reproductive rivalry between co-wives. It suggests that the sustained high level of fertility in Africa does not merely reflect women's lack of control over births, as is often argued, but also their incentives to have many children. This paper also contributes to the literature on household behavior as one of the few attempts to open the black box of non-nuclear families.

No abstract

International audienceExploiting original data from a Senegalese household survey, we provide evidence that fertility choices are partly driven by women's needs for widowhood insurance. We use a duration model of birth intervals to show that women most at risk in case of widowhood intensify their fertility, shortening birth spacing, until they get a son. Insurance through sons might entail substantial health costs since short birth spacing raises maternal and infant mortality rates

Has massive distribution of insecticide-treated-nets contributed to the reduction in infant mortality in Sub-Saharan Africa over the past 15 years? Using large household surveys collected in 16 countries and exploiting the spatial correlation in distribution campaigns, we estimate the relationship between the diffusion of bednets and the progress in child survival. We find no evidence of a causal link in cities, and no impact either in rural areas with low malaria prevalence. By contrast, in highly malarious rural areas where bednet coverage reached high levels, above 75% of households, infant mortality has been reduced by at least 3 percentage points, which amounts to 25% of the initial mortality. The identified impact is even higher for the children of mothers with no education. It lies at the upper bound found with RCTs, most likely because those were implemented in contexts with lower mortality and/or malaria prevalence.