AGENT-BASED SIMULATION OF INNOVATION DIFFUSION WITHIN SOCIAL NETWORKS.

Abstract

An agent-based model is developed to study the diffusion process of an innovation. Agents make decision at each time step to adopt or not based on economic utility maximization objective. To model real-life social networks, a small-world network is employed because of its clustering behaviour amongst agents. Diffusion curves are then plotted and compared against by varying several variables, such as randomness in the network, extent of heterogeneity amongst consumers, and strength of local network effects. Results are consistent with literature. This paper then extends to compare several promotional strategies to determine the optimal strategy that can penetrate the market successfully within the shortest time. The results suggest that firms should increase the number of seeds to target to, subject to their budget constraints. The optimal seeds identification is found to be based on betweenness centrality ranking, which would allow the seeds to have the greatest influence over the network.