How chilling are network externalities? The role of network structure

In an influential paper, Goldenberg, Libai, and Muller (2010) use an agent-based model to demonstrate that network externalities have a “chilling” effect on new product diffusion, i.e. they slow down new product adoption since many consumers wait before enough people have adopted. They perform their simulations using theoretical Moore lattices as the underlying social network of consumers. However, it has been demonstrated in other contexts that network structures can significantly affect the dynamics of new product diffusion, and hence it is worth investigating the same considerations for network externalities as well. I use the diffusion model of Goldenberg et al. (2010) to perform simulations on actual social networks to demonstrate that the chilling effect of network externalities is somewhat offset by increasing network size and average degree of the nodes, but accentuated by increased clustering in the network. My simulations also reveal that the diffusion model used by Goldenberg et al. (2010) does not have the chilling effect tautologically “baked” into it; rather network externalities do tend to slow down new product adoption most of the time, but not always.