Finding the Influentials that Drive the Diffusion of New Technologies

In this article we put forward a model where aggregate sales are a function of the online search of potential consumers at many locations. We argue that a location may be influential because of its power to drive aggregate sales and this power may be dynamic and evolving in time. Second, the influential locations may produce spillover effects over their neighbors and hence we may observe clusters of influence. We apply Bayesian Variable Selection (BVS) techniques and we use Multivariate Conditional Autoregressive Models (MCAR) to identify influentials locations and their clustering. Our results indicate that the influential locations and their economic value (measured by search elasticities) vary over time. Moreover, we find significant geographical clusters of influential locations and the clusters composition varies during the life-cycle of the consoles. Finally, we find weak evidence that demographics explain the probability of a location to be influential