The Diffusion of Innovations in Social Networks

This paper determines how different network structures influence the diffusion of innovations. We develop a model of diffusion where: 1. an individual’s decision to adopt a new technology is influenced by his contacts; and 2. contacts can discuss, coordinate, and make adoption decisions together. A measure of connectedness, ‘cohesion’, determines diffusion. A cohesive community is defined as a group in which all members have a high proportion of their contacts within the group. We show a key trade-off: on one hand, a cohesive community can hinder diffusion by blocking the spread of a technology into the group; on the other hand, cohesive communities can be particularly effective at acting collectively to adopt an innovation. We find that for technologies with low externalities (that require few people to adopt before others are willing to adopt), social structures with loose ties, where people are not part of cohesive groups, enable greater diffusion. However, as externalities increase (technologies require more people to adopt before others are willing to adopt), social structures with increasingly cohesive groups enable greater diffusion. Given that societal structure is known to differ systematically along this dimension, our findings point to specialization in technological progress exhibiting these patterns. ∗Bryony Reich, Faculty of Economics, University College London. Email: [email protected]. I would like to thank Alberto Alesina, Antonio Cabrales, Sanjeev Goyal, and Jorgen Weibull for their invaluable guidance and support. I benefited greatly from conversations with and comments of Marco Bassetto, Lars Nesheim and Imran Rasul. I am grateful to Jonathan Newton for numerous interactions at all stages of this project. For helpful comments I would also like to thank Lucie Gadenne, Terri Kneeland, and Sueyhun Kwon, as well as seminar participants at Alicante, Cambridge, INET Contagion Conference, Oxford, PET Luxembourg, and UCL. I gratefully acknowledge financial support from the UK Economic and Social Research Council (grant number ES/K001396/1).