Blissful Ignorance : the Value of Information in a Pandemic

This paper studies how partial information regarding the true number of infected affects optimal mitigation and testing policies during the COVID-19 pandemic. We start by documenting two motivating observations which highlight the value of information: First, an overreaction in mitigation at the onset of the pandemic compared to its later stages; Second, a tendency for what we call "blissful ignorance," where less testing is associated with fewer mitigation measures in place. We show that these can be justified through the lens of optimal policies under partial information. Specifically, we develop an epidemiological model where the true number of infected can be partially inferred from two signals: hospitalization and testing. An egalitarian planner can decide on the degree of mitigation and testing, which affect infection rates and signal noises about the infected. Using the calibrated model, our main results show that the planner is willing to give up 17% of output for testing to eliminate the uncertainty, provided the full enforcement of mitigation. Absent testing, an overreaction of up to 35% in mitigation can partially replace this information role of testing. Finally, when mitigation is not enforceable, the planner optimally remains blissfully ignorant by reducing the number of tests