Rigorously estimating the effects of development programs is notoriously difficult. In this paper we present a methodology that borrows from "event studies" commonly used in the finance literature to ascertain the impacts of corporate mergers. In our RETRAFECT methodology a retrospective panel data set is created based on “fundamental” events in the history of surveyed households, events that are discrete, unforgettable, and important to welfare. Based on the relationship between the changes in the estimated probabilities of these events and the timing of the introduction and uptake of a treatment, it is possible to ascertain if the probability of these fundamental events changes significantly after treatment. Our paper presents a series of tests, diagnostics, and corrections to account for potential supply-side and demand-side endogeneity problems in program rollout and adoption, and we caution against causal inference in many types of programs unless specific conditions are satisfied. Our application of the methodology examines housing improvements and purchases of large consumer durables among 1,672 households who gained access to microfinance at different times in Guatemala, India, and Ghana. We carry out estimations using village and country/year-level fixed effects to analyze the timing of these fundamental events relative to the timing of microfinance access and treatment. In our estimations, we find that the probability of a major housing improvement almost doubles from 0.038 to 0.070 in the years subsequent to a household's first microfinance loan, and that the probability of purchasing certain consumer durables increases modestly.
McIntosh, Craig; Villaran, Gonzalo; and Wydick, Bruce, "Microfinance and Home Improvement: Using Retrospective Panel Data to Measure Program Effects on Fundamental Events" (2011). Economics. 18.