The network at work: Diffusion of banana cultivation in Tanzania

This paper investigates the role of networks for diffusion of improved banana cultivation introduced by an agricultural project in Tanzania. In the existing literature on networks and technology adoption, network effects are interpreted as learning. I show that a farmer's network can affect the adoption of a new crop not only through social learning, but also by providing necessary inputs for adoption. I set up a simple model for adoption and derive similar model implications for the provision of either inputs or information through the network. Empirically, I find that a farmer is 37 percentage points more likely to adopt banana cultivation if there is at least one project participant growing bananas in the farmer's network. I use three falsication tests to support causal interpretation of the network effect on adoption. Provision of inputs (banana seedlings) through networks is found to play an important role for the network effects found.