Linking an Unjournal.org evaluation (package here) of one of the papers mentioned in this article.
Akram et al. (2014) subsidized transport in rural villages in Bangladesh, which increased the proportion of households sending a migrant to a city by 30 percentage points.
For example, an RCT by Akram et al. (2014)[21] provided transport subsidies to treated villages in rural Bangladesh, which increased the proportion of households sending a migrant to a city by 30 percentage points (which persisted in follow-up years). While the individuals who moved saw benefits, so did those that remained, as agricultural wages in the treated villages increased by roughly 4-6 percent as more laborers left for the city[21]. By boosting benefits for non-movers, internal migration increases its impacts even further and bolsters the cost-effectiveness calculations.
Evidence Action shut down the charity No Lean Season (the precise intervention in this paper) partly because of “Mixed evidence of impact” (see GiveWell’s report here). But the decision was also partly due to particular problems (allegations of fraud and mismanagement by the implementing partner. Thus, we suspected there may still be a strong case that this could be a cost-effective intervention. The present paper reported substantial positive effects, and was heavily cited. Thus we believed it deserved more careful evaluation, and if there were substantial flaws, these should be publicly shared.
[We only provided one evaluation]… Although we normally seek two or more strong evaluations, we struggled to find evaluators who could provide informed constructive criticism and appraisal. We might have persisted further, however…
After further consultation with an applied researcher in this area, we were advised that the GiveWell’s decision was more about issues having to do with the failure of the program to scale up, rather than its earlier performance as considered in the paper evaluated here. Follow-on work thus seems more relevant for further evaluation, such as “Delegation Risk and Implementation at Scale: Evidence from a Migration Loan Program in Bangladesh” (Mitchell et al, 2023)
Footnote: Mobarak also wrote a nontechnical comment about the difficulties of scaling up interventions (largely based on his experience here).
Linking an Unjournal.org evaluation (package here) of one of the papers mentioned in this article.