“Economic Activity and COVID-19 Transmission: Evidence from an Estimated Economic-Epidemiological Model”

Research
1

Hi everyone - @Doug_Hanley @Daniel_Kolliner_University_of_Maryland Mark Bognanni and I released a preliminary version of our new working paper: “Economic Activity and COVID-19 Transmission: Evidence from an Estimated Economic-Epidemiological Model”

Abstract: We develop and estimate a spatial model of the joint evolution of economic variables and the spread of COVID-19 across U.S. counties. Agents in the model optimally trade-off engaging in market consumption with the risk of contracting the disease. To motivate the model, we use three novel county-level data sets to document key empirical relationships between non-pharmaceutical interventions (NPIs), health, mobility, and employment outcomes at a daily frequency. We investigate the relative importance of NPIs, such as stay-at-home orders, and endogenous social distancing. Finally, we use our estimated model to address key issues in battling the pandemic: How will the spread of the disease and economic activity evolve if current NPIs are relaxed? What are optimal implementable NPI policies, taking into account the trade-off between the spread of the virus and economic activity?

Any comments are most appreciated!

http://perseus.iies.su.se/~kmitm/covid.pdf

2

Congratulations on the finished research! Curious, would you be interested in hosting a short seminar on your work for Consortium members? I’ll DM you the details.

3

Hi @Kate_Blumberg_SafeGraph - yes, we’d definitely be interested. Please DM me with the info

4

Very clever modeling! Well done.

5

Hi @Kurt_Mitman_IIES_Stockholm, thanks for sharing your paper! I particularly like how there was a focus on the ability to inform policy. I have two questions:

  1. All else equal, could optimal NPIs vary by the type of county? For example, a county with a dense metropolitan area vs a sparsely populated rural county? Or maybe this is naturally accounted for in the model.
  2. Are there long-term economic consequences of individual endogenous responses in the scenarios the virus “slow burns”?
    Thanks!
6

Hi @Ryan_Kruse_MN_State, thanks for your questions.

  1. We estimate the “beta” from the SIR model at the county level, basically to allow for the transmission rate per infect person to be higher in some areas than others (we check ex-post that these are pretty correlated with population density. so e.g., the NYC counties show up at the top). So we allow the optimal NPI policies to depend on the beta. You’re right that high-beta places (where transmission rates are higher) have tighter optimal policies than low-beta areas.
  2. Right now we don’t have long-run economic consequences (besides just the accumulated loss in output over the 10 years until it finally ends). That’s something that we’d like to try to capture in a future version of the paper
7

Please tell us the answers @Kurt_Mitman_IIES_Stockholm! TLDR: How will the spread of the disease and economic activity evolve if current NPIs are relaxed? What are optimal implementable NPI policies, taking into account the trade-off between the spread of the virus and economic activity?

8

@Kurt_Mitman_IIES_Stockholm Great paper with sobering implications. So basically for “kill zone” to occur, economic activity should be reduced to below 20%? Is this time constrained to rapid eradication (say within this year) or not? If not, then maybe you can lax this and see how long this will protract? It just seems rather impossible to me that 20% is implementable, also corroborated by your own facts of reduced economic activity and output (still much above 20%).

9

Hi @Ruowei_Yang_UM_Baltimore - I agree that 20% doesn’t seem feasible. What we’re working on now is looking at different lockdown paths assuming some maximal tolerance for lockdowns. Say, you aren’t willing to implement something that reduces activity below 40% (or in general X%), what’s the best thing that you can do. Will report back!

10

@Kurt_Mitman_IIES_Stockholm it seems like at this point, states and other entities are pushing forward regardless of the existence of testing capacity or other considerations. So it would be extremely useful to estimate for a given level of lockdown (which can be calibrated empirically), how many excess deaths vs. economic impacts there would be relative to the optimal policy.