The Structure of Information Networks

We develop a model of information acquisition in a network where agents pay for all the information they acquire including those through indirect links. The cost of information depends on the value of information itself and the distance it traverses in the network. We find that when the costs of information increase with distance, the complete network is the only Nash network. When costs of information decrease with distance capturing delay, all equilibrium information networks are minimally connected, though not all trees are Nash. We analyze the popular star and chain networks and identify strict Nash networks. We show that there is almost no divergence between efficient and equilibrium information networks. We explore the implications of a spatial model and information decay and discuss the relationship with experimental evidence.