Privatization of a public good (the management of sewage treatment centers in Dakar, Senegal) leads to an increase in the productivity of downstream sewage dumping companies and a decrease in downstream prices of the services they provide to households. We use the universe of legal dumping of sanitation waste from May 2009 to May 2018 to show that legal dumping increased substantially following privatization–on average an increase of 74%, or an increase of about 1640 trips to treatment centers each month. This is due to increased productivity of all trucks, not just those associated with the company managing the privatized treatment centers. Household-level survey data shows that downstream prices of legal sanitary dumping decreased by 5% following privatization, and DHS data shows that diarrhea rates among children under five decreased in Dakar relative to secondary cities in Senegal following privatization with no similar effect on respiratory illness as a placebo.

We run a randomized controlled trial coupled with lab-in-the-field social network experiments in urban Dakar. Decision spillovers and health externalities play a large role in determining uptake of the sanitation technology, with decision spillovers being largest among households that don’t receive significant subsidies. There is no evidence that the spillovers are explained by social forces in general, nor that they are explained by specific social mechanisms such as learning from others, social pressure, or reciprocity. We do find evidence of a fourth, non-social, mechanism impacting decisions: increasing health benefits. As more neighbors adopt the sanitary technology, it becomes more worthwhile for other households to adopt as well.

Blog and podcast coverage : World Bank Development Impact Blog • NOVAFRICA Development Economics Job Market Talks Podcast

We conduct a randomized experiment with groundnut producers in Senegal to address barriers to quality upgrading and access to international markets. We offer a bundled flexible contract that provides credit, training, and a price premium guarantee to encourage use of a new quality-improving technology. Producers randomly offered the contract are significantly more likely to purchase the technology. In areas where quality is otherwise lower due to agro-climatic conditions, producers are significantly more likely to comply with international standards. Adoption falls two years after our intervention when a price premium guarantee is no longer available, but remains persistently higher among previously-treated producers. We find that producers change their commercialization behavior when the contract is active, opting to sell more to cooperatives and less to other buyers, but this change is small in magnitude and does not persist to the subsequent season.

I study the dynamics of household adoption of a public service in response to short-term subsidies. I exploit spatial variation in exposure to subsidies which induced consumers to use a publicly-provided matching platform for sanitation services in Dakar, Senegal. Using platform administrative data, I show that neighborhoods exposed to short-term subsidies are significantly more likely to use the platform after subsidies end, but this effect declines gradually to zero over time. Following a subsequent city-wide subsidy campaign, increased adoption re-emerges in previously-subsidized neighborhoods. I explore within-neighborhood spillovers as a mechanism and show that a substantial fraction of increased long-run adoption comes from new users.

No single constraint can explain the stagnant agricultural productivity growth in sub-Saharan Africa. Most interventions that relax individual barriers to productivity have delivered disappointing results. We evaluate an at-scale program that targets several productivity constraints with a bundled intervention, using a randomized controlled trial in western Kenya. Program participation increases maize yields by 26%, total maize output by 24%, and profits by 17%. While we cannot directly test whether the program’s success is due to its bundled nature, we find patterns in the data that are consistent with this hypothesis.