Search

Despite limited uptake, index insurance is often seen as one of the most remarkable innovations of the past decades to help smallholder farmers manage risks. This paper uses a Bayesian hierarchical model to aggregate evidence from existing experiments and assess the external validity of their results. The findings show positive but highly heterogeneous responses to index insurance across experiments. Interventions expanding access to index insurance typically boost productive investments by 0.06–0.11 standard deviation on average. However, treatment effects display substantial heterogeneity and there is no evidence that this heterogeneity can be meaningfully explained by basic household characteristics. The existing evidence base thus offers limited insights to predict the impact of index insurance in new settings. The paper concludes that governments and development agencies should remain cautious before investing in the widespread expansion of index insurance.

Afin de lutter contre l’abandon scolaire, le gouvernement marocain a mis en place un programme de cash-transfer conditionnel à partir de 2008 : le programme Tayssir. L’objectif initial du programme semble être atteint mais qu’en est-il de la qualité de l’éducation reçue par les élèves qui restent grâce au transfert ? C’est ce que nous nous proposons de regarder en comparant les résultats aux examens de fin de primaire et de fin de collège des élèves dans les communes bénéficiaires et non bénéficiaires. Comme le programme Tayssir ciblait jusqu’en septembre 2018 les communes dont le taux de pauvreté est supérieur à 30%, nous utilisons une méthode d’échantillonnage qui repose sur l’idée que les communes qui sont proches du seuil de ciblage sont suffisamment similaires pour servir de groupe de comparaison. Nous trouvons qu’à la fin du primaire, le programme Tayssir n’a pas d’effet sur les performances des filles mais que les garçons ont des résultats en moyenne moins bons dans les communes bénéficiaires du programme que dans les communes non bénéficiaires. La raison est qu’en réduisant l’abandon scolaire, le programme Tayssir a permis aussi à des élèves en difficulté de rester scolarisés. Le défi du gouvernement est alors désormais de faire en sorte que l’ensemble des élèves marocains acquièrent des compétences en essayant d’adapter la pédagogie pour ne laisser aucun enfant sur le bas-côté., Revue Réflexions Économiques, No 1 (2020): L'éducation en questions

In Tunisia, while social protection and labor programs are in place, severe challenges including inefficiency, fragmentation, and inequity limit the country’s ability to respond to increasing social needs. Gender issues are also one of the critical areas since young women are experiencing even more severe challenges getting into the tight labor market than young men. Unemployment in the MENA region has been a challenge for some time, markedly during the Arab Spring, resulting in the need to create over 50 million jobs in the region in the next decade, to ensure socio-political stability. Unemployment rates are highest in rural and low-income areas. It is in this context that a pilot project of Community Works andLocal Participation (CWLP) was initiated in rural Jendouba in 2015. It was financed by the Japan Social Development Fund (JSDF) through the World Bank and implemented by the Tunisia Republic’s Ministry of Vocational Training and Employment (MVTE). A rigorous randomized control trial (RCT) was embedded in the second phase of the CWLP roll-out (starting in late 2015 and early 2016) and carried out by the World Bank’sDIME Department in partnership with MVET’s ONEQ. The study’s main objective was to capture the effects of CWLP’s cash for work activities. The results of this study, based on a detailed survey of over 4,000 participants and non-participants 6-12 months after completion of project activities, suggested that in general, the CWLP has had positive impacts on the economic well-being of beneficiaries and to a small extent on social and psychological well-being. However, these results also raised concerns that these positive effects may not persist in the long-run, particularly for women who still face huge constraints participating in the tight labor market, which has yet to fully recover to pre-Jasmine revolution levels.

International audience

This thesis provides three empirical essays on the design and evaluation of social safety nets. Chapter 1 adds to the literature on the performances of targeting methods in general and Proxy Means Testing in particular. Using a unique survey experiment conducted in Tanzania, it investigates whether and to what degree Proxy Means Testing targeting performances are biased when household consumption data are subject to non-random errors. The results indicate that Proxy Means Testing performances are quite vulnerable to non-random errors when the objective is to target absolutely poor households, but remain largely unaffected when the objective is to target a fixed share of the population. Chapter 2 studies the impact on migration of a cash-for-work program in Comoros that randomly offered poor households cash transfers in exchange for their participation in public works projects. Using first-hand data, this chapter shows that the program increased migration to Mayotte – the neighboring and richer French Island. Between 2016 and 2018, treated households received up to USD320 in cash and, as a result, were three percentage points more likely to have a household member migrating to Mayotte (a statistically significant 38 percent increase relative to the control group). This result appears to be driven by the alleviation of liquidity and risk constraints to migration. Chapter 3 explores the productive effects of cash-for-work programs in the context of the Productive Safety Net Project in Ethiopia. With more than 8 million beneficiaries, the Productive Safety Net Project is among the largest safety net programs in Africa. It is also often considered as Africa’s largest climate change adaptation program due to its focus on activities such as land improvements and soil and water conservation measures. This chapter relies on satellite and geo-referenced data to evaluate the effects of these activities and overcome the lack of household data. Difference-in-differences estimates covering whole Ethiopia over the 2000-2013 period show no evidence to support that public works had measurable impacts on agricultural productivity and resilience to climate shocks., Cette thèse fournit trois essais sur le design et l'évaluation des filets sociaux de sécurité. Le Chapitre 1 contribue à la littérature sur les performances des méthodes de ciblage en général et sur le Proxy Means Testing en particulier. En utilisant une enquête expérimentale en Tanzanie, ce chapitre cherche à mesurer si les performances de ciblage du Proxy Means Testing sont biaisées lorsque les données de consommation sont sujettes à des erreurs de mesure non-aléatoires. Les résultats indiquent que les performances du Proxy Means Testing sont assez vulnérables aux erreurs de mesure non-aléatoires quand l'objectif est de cibler les ménages pauvres dans l'absolu, mais qu'elles restent en grande partie non affectées quand l'objectif est de cibler une part fixe de la population. Le Chapitre 2 étudie l'impact sur la migration d'un programme argent-contre-travail aux Comores. Ce programme a alloué de manière aléatoire à des ménages pauvres des transferts monétaires en échange de leur participation dans des travaux publics. En utilisant des données de première main, ce chapitre montre que le programme a augmenté la migration vers Mayotte -- l'île Française voisine et plus riche. Entre 2016 et 2018, les ménages traités ont reçu jusqu'à 320USD et, par conséquent, étaient trois points de pourcentage plus susceptibles d'avoir un membre du ménage qui migre à Mayotte (une hausse statistiquement significative de 38 pourcent comparé au groupe de contrôle). Ce résultat semble être expliqué par la réduction des contraintes de liquidité et de risque à la migration. Le Chapitre 3 explore les effets productifs des programmes argent-contre-travail dans le contexte du Productive Safety Net Project en Ethiopie. Avec plus de 8 millions de bénéficiaires, le Productive Safety Net Project est parmi les plus grands programmes de filets sociaux d'Afrique. Il est aussi souvent considéré comme le programme d'adaptation au changement climatique le plus large d'Afrique avec ses activités concentrées sur l'amélioration des terres et des mesures de conservation des sols et des eaux. Des estimations en différence-en-différence couvrant toute l'Ethiopie sur la période 2000-2013 ne montrent aucune évidence pour supporter que les travaux publics aient eu des impacts mesurables sur la productivité agricole et sur la résilience aux chocs climatiques.

International audience

Proxy Means Testing (PMT) is a popular method to target the poor in developing countries. PMT usually relies on survey-based consumption data and assumes random measurement errors – an assumption that has been challenged by recent literature. Using a survey experiment conducted in Tanzania, this paper brings causal evidence on the impact of non-random errors on PMT performances. Results show that non-random errors bias the coefficients from PMT models, resulting in a 5 to 27 per cent reduction in PMT predictive performances. Moreover, non-random errors induce a 10 to 34 per cent increase in the incidence of targeting errors when poverty is defined in absolute terms. More reassuringly, impacts on the ranking of households are smaller and essentially non-significant. Taken together, these results indicate that PMT performances are quite vulnerable to non-random errors when the objective is to target absolutely poor households, but remain largely unaffected when the objective is to target a fixed share of the population.

Proxy Means Testing (PMT) is a popular method to target the poor in developing countries. PMT usually relies on survey-based consumption data and assumes random measurement errors – an assumption that has been challenged by recent literature. Using a survey experiment conducted in Tanzania, this paper brings causal evidence on the impact of non-random errors on PMT performances. Results show that non-random errors bias the coefficients from PMT models, resulting in a 5 to 27 per cent reduction in PMT predictive performances. Moreover, non-random errors induce a 10 to 34 per cent increase in the incidence of targeting errors when poverty is defined in absolute terms. More reassuringly, impacts on the ranking of households are smaller and essentially non-significant. Taken together, these results indicate that PMT performances are quite vulnerable to non-random errors when the objective is to target absolutely poor households, but remain largely unaffected when the objective is to target a fixed share of the population.