What are the long-run economic costs of COVID-19? While the virus will eventually pass, an event of this magnitude could leave lasting effects. A shift in confidence and fear could prevent firms and consumers from rebounding to their old investment and spending habits. A recent paper (Kozlowski, Veldkamp, and Venkateswaran, 2020a) formalizes this discussion and quantifies these effects. The authors use a standard economic and epidemiological framework, with one novel channel: a “scarring effect.” Scarring is a persistent change in beliefs about the probability of an extreme, negative shock to the economy.  

Perception can be everything: People observe new events and use these experiences to inform their expectations. For example, if you haven’t seen many pandemics, you think pandemics are rare. However, when you see a pandemic, you come to believe that pandemics are not as rare as you previously thought. As an analogy, imagine you see the outcome of many rolls of a die. You don’t know how many sides the die has or if the sides are weighted. However, you see numbers one through six come up many times, about equally, so you think this is a standard six-sided die. One time, though, a seven comes up. As a result, you revise your belief and now think the die has at least seven sides. Even if you don’t see a seven again, the knowledge that the number seven can be rolled stays with you. It would probably affect your willingness to bet on the outcome of a die roll, too. Seeing the COVID-19 pandemic was like seeing a seven come up on what you thought was a six-­sided die.

Consciously or not, we all use past events to inform our beliefs, like econometricians do. Rare events are those for which we have little data. In turn, the scarcity of data makes new rare events particularly informative, so rare events trigger larger belief revisions. Furthermore, because it will take many more observations of non-rare events to convince someone that a rare event really is unlikely, these changes in expectations are particularly persistent.1

The work by Kozlowski, Veldkamp, and Venkateswaran (2020a) embeds a belief-updating tool in a macroeconomic model with epidemics. Belief updating can generate large and persistent economic losses well after an epidemic is over because agents think that epidemics are more likely after seeing one. The figure shows predictions about the response to the COVID-19 crisis. GDP drops 9 percent in 2020 and recovers gradually but does not go back to its previous trajectory. It persistently remains 4 percent below the pre-COVID steady state. The discounted value of the lost output over the future is almost 10 times the 2020 drop, and belief revisions account for the bulk of the losses.