skip to content

The Cambridge-INET Institute

WP Cover

Choi, S., Goyal, S. and Moisan, F.

Connectors and Influencers

WP Number: 1909

Abstract: Classical models of network formation bring out the salience of a `law of the few' property, that is manifest in hub-spoke/star architectures. Existing experiments on network formation find that subjects do not create such networks.

Our paper conducts a network formation experiment on the model of Galeotti and Goyal (2010). The theory predicts that every equilibrium of this game is a `star' network in which the spokes pay for links with a single hub. There are two equilibrium effort configurations: the center makes all the effort (the pure influencer outcome) and the hub makes zero effort (the pure connector outcome). This paper tests these predictions with the help of a new experimental platform with asynchronous activity in continuous time. We vary group size and provision of information of others' payoffs.

Subjects always create networks with specialization in linking. This is consistent with equilibrium prediction. Our second result concerns the interaction of group size and information provision. In a baseline information treatment where subjects only see their own payoffs, they select the pure influencer outcome. By contrast, when we provide information on everyone's payoffs, in large groups, subjects select a pure connector outcome. These behavioural patterns can be accounted for by a decision rule on activity level that combines myopic best response and competition for hub status.

JEL Codes: C92 D83 D85 Z13

Author links: Sanjeev Goyal  

PDF: wp1909.pdf

Open Access Link: 10.17863/CAM.38670

Keynes Fund Project(s):
Experiments on Financial Networks (JHLB)  
Experiments on 'Networked Markets' (JHLR)  

Theme: networks