Modeling adaptive forward-looking behavior in epidemics on networks

Incorporating decision-making dynamics during an outbreak poses a challenge for epidemiology, faced by several modeling approaches siloed by different disciplines. We propose an epieconomic model where high-frequency choices of individuals respond to the infection dynamics over heterogeneous networks. Maintaining a rational forward-looking component to individual choices, agents follow a behavioral rule-of-thumb in the face of limited perceived forecasting precision in a highly uncertain epidemic environment. We describe the resulting equilibrium behavior of the epidemic by analytical expressions depending on the epidemic conditions. We study existence and welfare of equilibrium, identifying a fundamental negative externality. We also sign analytically the effects of the behavioral rule-of-thumb at different phases of the epidemic and characterize some comparative statics. Through numerical simulations, we contrast different information structures: global awareness - where individuals only know the prevalence of the disease in the population - with local awareness, where individuals know the prevalence in their neighborhood. We show that agents' behavioral response through forward-looking choice can flatten the epidemic curve, but local awareness, by triggering highly heterogeneous behavioral responses, more effectively curbs the disease compared to global awareness.