Network Security

I build a model in which helpful tips and harmful viruses spread through a large social network, modeled as a random graph. The network is generated by the equilibrium actions of individual agents who form links to other agents in the hope of receiving tips. These links also expose agents to viruses, a risk they can mitigate by investing in security. Equilibrium choices at the individual level aggregate to produce tip and virus “outbreaks”; I characterize how frequent and severe these outbreaks will be. There are both positive and negative network externalities — as a denser network spreads both tips and viruses more widely —as well as a positive externality from investments in security, and their interplay yields rich and complex comparative statics. For example, while a policy intervention to subsidize security will generally improve welfare, under broad conditions it will also make virus outbreaks more common and more severe. Finally, I determine when the network will be resilient, in the sense that connectivity does not collapse if the ratio of viruses to tips grows large.