Investing in agriculture when it is worth it. Evidence from rural Uganda

One of the reasons for the persistent low agricultural productivity in Sub-Saharan Africa is the lack of adoption of pro table agricultural technologies. Yet, what is pro table in a controlled experimental setting may not be pro table in a real-world setting. Estimating the returns to a single input is, in fact, challenging as farmers may respond to adoption by re-optimizing the use of other inputs. This paper explores farmers behavioral response to a positive random shock on future productivity by disentangling inputs returns from farmers' response. Using a unique household panel dataset collected in rural Uganda, I proxy a future productivity shock with the birth of a female calf against that of a male calf. Calves have no technical returns, but female calves will become cows producing milk, providing a stable source of income, while bulls and oxen are of little use in this context. The main OLS and di erence-in-di erences results show the existence of a crowd-in e ect. Farmers react to the birth of a female calf by increasing inputs' expenditures. They invest more on their cattle's health, increase hired labor and are more willing to pay for cattle-related investments but not for other activities. This increase in investments leads to an increase in milk production and revenues that lasts over time. Further results show that economies of scale associated with the number of female animals seem to explain this behavioral response.