The diffusion of mobile money: Evidence from a lab experiment in the field

The success story of M-PESA in Kenya raised hopes that massive mobile money adoption would promote financial inclusion of the poor. This is particularly true in sub-Saharan Africa where the poor had very limited access to formal financial services before the introduction of mobile money technology. In Mozambique, recent research has shown considerable adoption of mobile money in rural areas within one year of its introduction. However, it remains to be understood how information and adoption diffusion occurs in the case of this innovation.

The current project proposes to understand how information about newly provided mobile money services flows within social networks in rural villages of Mozambique and how social influence affects the adoption of mobile-money services such as electronic money transfers and electronic savings. Particular attention will be given to how many peers in someone’s social network need to receive information and/or adopt the new service for adoption to follow.

The project will target 1000 participants of 51 treatment areas of a field experiment on the introduction of a mobile money service in the provinces of Maputo, Gaza, and Inhambane. Although adoption was considerable, with 64 percent of the treated individuals undertaking at least one transaction in the six months following the introduction of the service, we do not know much about how information and adoption diffused within each of the treated locations.

For this research project, we will assign anonymous social links in a randomised manner between the participants in the previous study. By varying the structure of the social network, we will be able to assess the role of network density, clustering and centrality in the diffusion of information and adoption. Having set up the social network, we will ‘seed’ chosen individuals with information about distinct mobile money services. We will then observe how this information diffuses to others, and how recipients of the information and their network neighbours adopt the new service. Furthermore, we will test for the presence of positive externalities in information and for the possibility of strategic complements or substitutes in adoption. Finally, we will also test for social thresholds in adoption behaviour (which should be affected by clustering and network density) and investigate whether nodes that have been randomly assigned a more central position are more influential.

Our research is being developed in partnership with the Central Bank of Mozambique, who are particularly aware of the enormous potential the adoption of mobile money in rural areas may represent in terms of the country financial inclusion. First, as the number of electronic transactions increases, the Central Bank’s control of monetary aggregates will improve, thereby sharpening the monetary authority’s policymaking capacity. Second, as new users adopt and diffuse mobile money services, they will become better acquainted with the provision and use of financial services in general. This in turn is expected to promote the progressive adoption of traditional financial services by mobile money users and others in their social networks.