Replication package — Experimental impact study of the Café Lumière micro-grid project in rural Madagascar

This repository contains the data and Stata code required to reproduce the quantitative results of the impact evaluation of the Café Lumière project, a solar-powered micro-grid and energy-platform initiative implemented by the NGO Electriciens sans Frontière (ESF) and financed by the Agence Française de Développement (AFD) in six villages of the Vakinankaratra and Itasy regions of Madagascar between 2018 and 2021. The evaluation follows a randomized controlled trial (RCT) design at the village level, with twelve comparable localities identified through a common pre-feasibility study and treatment randomly assigned to half of them. Outcomes are measured using two waves of household and individual surveys collected in 2017–2018 and in 2023, complemented by locality-level data. The package is organised into two folders. The Data folder contains three Stata datasets: a panel of adults (Panel_adultes.dta), a household-level panel (Panel_menages_v2.dta), and an unbalanced panel of children (UnbalancePanel_enfants.dta). The Scripts folder contains five .do files intended to be run sequentially: household-level analysis (1_Reg_Households.do), individual-level analysis (2_Reg_Individuals.do), child-level analysis (3_Reg_Children.do), robustness checks (4_Robustness.do), and Wild Bootstrap inference for cluster-robust p-values (5_Reg_WildBootStrap.do). Regression output is written to a RegTables subfolder. This package is released to support transparency and replication of the published findings on access to basic electricity services, secondary income-generating activities, household wealth, and social outcomes (health, education, public safety) attributable to the Café Lumière intervention. For questions, contact Vincent Nossek ([e-mail]).