Previous studies of peer-to-peer technology diffusion have primarily focused on the decision of potential adopters. Equally relevant for observed diffusion in many contexts is the willingness of incumbent adopters to actively share technology. Morgan Hardy and Jamie McCasland (2016) report the results of a field experiment that
considers both parties involved in the diffusion process. Specifically, the authors develop a new weaving technique and randomly seed both training and a limited number of one-time technique-specific contracts in a real network of garment making firm owners in Ghana. They find that firms that need the technology to complete the contract learn it from firms that received training; however, firms selected to receive only training and no contract offer are much less likely to share the technology than those selected to receive both. They document spillover effects in both
learning and teaching from baseline technology sharing contacts, but also find that networks display dynamic properties: a large number of firm owners generate new technology sharing contacts in response to the experiment. Teaching patterns to both pre-existing connections and new contacts follow the overall pattern, in which firms selected to receive both training and experimental demand are most likely to share the technology. The authors interpret these findings as evidence that experimental competition disincentivized diffusion, which is supported with additional evidence exploiting random order size and order timing. Finally, they develop a model
which conceptualizes observed diffusion patterns as endogenously resulting from both static dyad-specic and dynamic technology-specic costs and benefits to both the potential learner and the potential teacher.