Arthur Alik-Lagrange soutiendra sa thèse "Trois Essais sur le Workfare dans les Pays en Développement" le jeudi 15 décembre 2016, à 09h30, salle MS 003.
Directeur de Thèse : Pierre Dubois, directeur scientifique TSE, Professeur, Université Toulouse 1 Capitole.
Composition du jury :
- Monsieur Jean-Paul AZAM, Professeur, Université Toulouse 1 Capitole
- Monsieur Clément IMBERT, Professeur, Université de Warwick
- Madame Habiba DJEBBARI, Professeure, Université Aix Marseille
Résumé (en anglais):
In the context of recent forms of anti-poverty public work programs (PWP) in low and middle-income countries, the main rationale for the use of a work requirement at a low-wage rate relies on the implied self-targeting mechanism and the potential stabilization benefits (Ravallion, 1991). PWP screen out high income households who decide to not participate because of high opportunity costs. This revealing mechanism comes at the cost of forgone earnings for low-income households who decide to participate, reducing in turn program’s cost-effectiveness (Besley and Coate, 1992 and 1995) when compared to a benchmark basic income guarantee (BIG).
The rationale for the use of an anti-poverty PWP relies, among others, on two strong assumptions: (i) the existence of heterogeneous opportunity costs increasing along the consumption distribution; and (ii) the potential extra work effort exerted by PWP beneficiaries does not enter into the underlying cost-effectiveness analysis. The first chapter of this thesis revisits the incentives compatibilities analyzed in the seminal work by Besley and Coate (1992), allowing for high unemployment rate among the poor, hence low forgone earnings but higher net extra effort exerted by beneficiaries. Evaluations of workfare programs in poor rural economies have typically ignored two features that policy makers stress: involuntary unemployment and the expected welfare losses from work requirements. We generalize past evaluation theory and methods to incorporate both features, and we show that doing so can switch the policy ranking in favor of BIG over PWP. An empirical case study for India’s massive National Rural Employment Guarantee Scheme (MNREGA) indicates lower impacts on poverty than suggested by past methods, despite a more “poor-poor” incidence. A BIG would dominate net PWP earnings in terms of the impact on poverty for a given budgetary outlay. Another peculiar feature of anti-poverty PWP is that they are usually demand-driven. Households in need for poverty relief apply to low-paid activities depending, among other constraints, on the knowledge they have about their rights under the scheme. Examples include their right to work or the statutory wage rate offered under the scheme.
The second chapter of this thesis analyzes whether knowledge about PWP programs spreads quickly within poor communities or if there are significant frictions. We study the particular case of MNREGA, a decentralized, demand-driven program, where knowledge about beneficiaries’ rights under the scheme might constitute a critical constraint to take-up and transfer to the poor. Using household-level data for the purpose of a randomized control with a pilot information campaign in rural Bihar, India, we combine longitudinal and intra-household observations to estimate the direct knowledge gain from watching an informational movie, while randomized village assignment identifies knowledge sharing between people within villages. Spillover effects are evident, but they are weaker for illiterate and lower caste individuals, especially when also poor; these groups relied more on actually seeing the movie. We reject the null hypothesis of knowledge sharing for the more disadvantaged families, but not otherwise. We contribute to the methodological literature on the identification of spillovers and we show that spillovers can be identified using household panel data. We document sizeable biases in impact estimators, relying on the stable unit treatment value assumption (SUTVA).
The last chapter of this thesis is an impact evaluation of a Colombian workfare program, Empleos en Acción (EA). It was designed to self-select the poor and provide insurance against job losses by informal sector workers, at the possible cost of crowding out private labor effort. We analyze the impact of EA to shed light on (i) whether the program crowds out labor effort by members of the household different from the participant; (ii) whether there are gains in household labor income; (iii) consumption, which is important to assess the role of the program as an insurance mechanism; and finally (iv) whether there are some gains from participating in the program six months after the program has finished. Our results show no evidence of the program crowding out private labor effort by other household members. In addition, we find that the program had large positive transfer benefits, as the program increased individual’s labor income and labor supply in large urban municipalities as well as small rural ones. In small municipalities, the program intention to treat effect increases consumption by 5%, which doubles when focusing only on food consumption. Finally, we do find that the program had a positive effect on individuals’ outcomes as well as on households’ monthly labor income per capita in small rural municipalities six months after the program ended. We shed light on the potential channels explaining this novel result in the literature on PWP evaluation.
