Programme evaluation is an important tool to inform policymakers (governments, aid donors and the development community at large) about the efficient allocation of resources and for the improvement of existing policies. This has put the programme evaluation at core of recent research and debate in development economics. With that respect several studies have attempted to estimate the impact of nutrition programmes and child health in developing countries. As part of these programmes, school feeding and deworming are considered as important driving forces to increase enrollment and attendance and improve student performance. This research seeks to assess the impact of school feeding and deworming (as a package of programmes or multiple treatments) on school performance (test scores), retention (drop-out rates) and repetition rates while elaborating on the determinants of school performance. To this end, we use recent experimental data from Senegal. It was collected as part of an experimental programme (Policy Impact Evaluation Research Initiative) conducted by the Consortium for Social and Economic Research (CRES) and the Senegalese Ministry of Education.

Packages of programmes are usually implemented by governments or NGOs because of their efficiency and also because single programmes are costly compared to a package. However, researchers usually analyse these programmes separately to have the effect of the package because of the unavailability of appropriate tools enabling to estimate the package effect and disentangle the effect of programmes within a unified setting. We propose to fill this gap by developing a multiple treatments framework which can offer the estimation of a wide and rich range of treatment effects. These effects range from package effect to exclusive ones (disentangled effects) to several relative effects, while allowing to tackle issues related to complementarity or substitutability of programmes.

The contributions of this research will depart from the previous literature in several aspects. From a methodological point of view, i) we aim at extending the standard (one treatment) structural selection model of evaluation to the case of multiple treatments for cross-section, ii) develop the same methodology for panel data allowing for unobserved heterogeneity and dynamics. iii) In terms of policy analysis, this research will bring new insights to the empirical literature on the evaluation of development programmes in Africa in particular those relating to nutritional and health programmes. It will also inform policymakers (government and NGOs) about the effectiveness of programmes implemented in recent years and guide them in choosing among several options depending on the objectives.