“The Effects of Shocks on Social Networks: A Comparative Analysis of Experimental and Simulated Behavior in Small Networks.”
Status: Under review
Abstract: The COVID-19 pandemic imposed severe restrictions on social contact, drastically increasing the cost of face-to-face interaction. At the same time, social distancing made alternative forms of communication (e.g., online teaching and meetings) more common. This demonstrates how shocks can affect the utility people assign to their social ties. Past research suggests that, despite the cost of forming ties, people can benefit directly from the ties they choose to form, but also indirectly through triadic closure or spillover ties across layers in a multiplex network. We know less, however, about the manner in which people react to shocks and how social networks reorganize during and following shocks. We study the effect of shocks on social networks via a human experiment and agent-based model. In our design, human subjects (and simulated agents) can form ties with each other to increase their utility. However, the costs associated with social ties limit the number and type of ties that agents can form. After several rounds, some or all subjects are shocked: the cost of ties (e.g., the risk of infection) dramatically increases or decreases. We show that shocks have three unexpected effects. First, shocks spread: non-shocked subjects act more like shocked agents when other subjects are shocked. Second, a shocked subject’s behavior changes more dramatically when more subjects are shocked. Third, experimental subjects’ utilities significantly outperform myopically rational agents when tie-costs are low. Yet human subjects continue to cooperate excessively when tie-costs are high, yielding negative utility. We discuss the implications of these results for real-world networks, in the context of shocks such as the COVID-19 pandemic.
This paper is a result of work supported by the SPINS research group at UC Davis.