Fig. 1

(Upper panel) Mechanisms studied in the paper. Our main interest lies in estimating (1), the causal effect of schistosomiasis on agriculture. To achieve identification we use a set of variables (instruments) that influence disease intensity without directly affecting agriculture: we use the densities of the different snail species that act as intermediate hosts for the Schistosoma. (2). Poverty is shown to have a reinforcing effect on the burden the disease exerts on agricultural production, as well as being its consequence (3). Water resources development (4) is shown to boost agriculture and development, but also to increase the adverse effects of schistosomiasis via both the increase of snail habitat and human-water contact. (Lower panel) Data overview. a Villages included in the agricultural surveys, the capital Ouagadougou (white point) and level 1 (regions, black lines) and level 2 (provinces, white lines) administrative subdivisions. b River network (blue lines, width proportional to upstream area) and water resources infrastructure in the country. c Estimated schistosomiasis prevalence up to 2010. d Estimated schistosomiasis prevalence for 2011–2017