We experimentally investigate network formation in an environment where players simultaneously decide on their network connections and contributions to a local public good. Our design is based on the theoretical framework developed by Galeotti and Goyal (2010). The main result in their study is that in every strict equilibrium of the game, there is a limited number of players who contribute to the public good. These players form the core of the network. Other (periphery) players link to the core, and contribute nothing themselves.
In our 2×2 design, we vary the group and distinguish between an environment where there are benefits from an incoming link and one without such benefits. Assuming self-regarding preferences, the theoretical (equilibrium) predictions are the same across the four treatments. Our experimental results show substantial treatment effects, however. In the situation without benefits, equilibrium networks are rarely observed. However, when we introduce these benefits almost all groups converge to an equilibrium network architecture. We attribute this difference across treatments to the relative payoff differences between the core and periphery players. Benefits reverse the relative payoff difference in favor of the core players. Secondly, we find that doubling the group size has no effect on the observed network architecture but does increase public good contributions to levels close to efficiency when there are benefits for an incoming link.