UC Berkeley

Developing countries have long relied on foreign aid and international NGOs to provide

supplemental public goods such as communal drinking water points. Exploiting

a geospatial dataset of water projects in Uganda, I consider both eectiveness and

targeting of projects. To evaluate eectiveness as practiced at scale, I use a retrospective

design that links placement of water projects with Demographic and

Health Survey (DHS) clusters, nding that over a ve year span, protected sources

produce statistically signicant improvement of weight-for-age, height-for-age, fever,

and hemoglobin levels, but not of weight-for-height and diarrhea. In exploring poverty

targeting, I link water data with spatial poverty data and argue for a relative

targeting measure that compares dierent entity types to one another in terms of

how progressive their allocation of projects is. Finally, I use a similar procedure to

analyze placement in terms of its correlation with election outcomes.