Connectors and Influencers

We consider a setting in which individuals can purchase information at a cost and form costly links to access information purchased by others. The theory predicts that in every equilibrium of this game the network is a `star'. For small groups, there exists a unique purchase configuration -a pure influencer outcome, in which the hub node purchases information while all others free ride. For large groups, there exists, in addition, a pure connector outcome in which the hub purchases no information and the peripheral players purchase information. We test these predictions on a new experimental platform with asynchronous activity in continuous time. We start with a baseline setting where subjects only see their own payoffs. We find that subjects create a star network. In small groups, the hub purchases equilibrium level information, but in large groups the hub purchases excessive information and as a result earns low payoffs. To study the reasons for this excessive investment we propose a treatment in which subjects see everyone's payoffs. We find that in small groups the pure influencer out- come obtains but that in large groups the pure-connector outcome now becomes common, suggesting that information and group size interact in powerful ways to shape networks and payoffs.