Network Formation in Large Groups

We conduct an experiment to understand the principles that govern network formation. The design of the experiment builds on a model of linking and efforts taken from Galeotti and Goyal [2010]. In order to reduce cognitive complexity facing human subjects and facilitate learning, we develop a new experimental platform that integrates a network visualization tool using an algorithm of Barnes and Hut [1986] with an interactive tool of asynchronous choices in continuous time. Our experiment provides strong support for macroscopic predictions of the theory: there is specialization in linking and e?orts across all treatments. Moreover, and in line with the theory, the specialization is more pronounced in larger groups. Thus subjects abide by the law of the few. Information on payo?s provided to subjects a?ects their behavior and yields di?erential welfare consequences. In the treatment where subjects see only their own payo?s, in large groups, the most connected individuals compete ?ercely they exert large efforts and have small earnings. By contrast, when a subject sees everyone¡¯s payo?s, in large groups, the most connected individuals engage in less intense competition they exert little e?ort and have large earnings. The e?ects of information are much more muted in small groups.