Although academic research on the COVID-19 pandemic began with a slow start, policymakers were grappling with challenges about the economic effects of the national containment strategy. Since real-time macroeconomic indicators will take some time to come in, we published a policy brief on April 6thbased on research we began in March detailing the costs of the shutdown. We found that the average county will experience a 5% decline in their annual gross domestic product (GDP) growth rate for each month of an economic shutdown, which comes to a little over $1 trillion per month. However, counties vary significantly based on their industry composition and their exposure to the pandemic.
Our methodological approach builds heavily on the occupational index of digital skill intensity from Gallipoli and Makridis (2018). We use this index to compute the fraction of workers within an industry that is considered digitally intensive and subsequently apply it to real GDP data across the county and industry pairs from the Bureau of Economic Analysis. Figure 1 plots these digital-intensity shares. That allows us to gauge the exposure of each county to industries that vary in their digital intensity, reflecting the reality that counties with higher concentrations of industries that rely less on digital skills will have a tougher time adapting to the containment strategy and the temporary move to online goods and services.
Figure 1: Shares of Digital Workers, by Industry, from Gallipoli and Makridis (2018)
Because of significant cross-county differences in the exposure to industries that are more, versus less, digitally-intensive, we find that there is substantial heterogeneity in the economic effects of the shutdown across space. Figure 2 plots the distribution of growth rate declines across counties for both a one month and two-month economic shutdown, given that we are now going into the second month of the national quarantine. While the average county experiences a 10% decline in their growth rate under two months of a shutdown, some counties in the 5thpercentile, for example, experience a 15% decline.
Figure 2: Distribution of Counterfactual County Real GDP Growth (1 and 2 Month Containment)
We also explored the cross-sectional relationship between simulated county GDP declines and various county characteristics. Figure 3 presents these correlations through a series of bin-scatterplots, which is a non-parametric way of representing observations based on their similarities along a specified dimension (e.g., simulated GDP growth). First, counties with higher shares of digital workers are less affected. While this is partially a mechanical function of the model, it also is consistent with the reality that working from home is a smaller disruption in counties with higher industry concentrations of digital workers. Second, counties with lower median household income will be more adversely affected, which also reflects the fact that lower-income counties have fewer digital jobs to cushion against the shock to the physical mobility of people and goods (Gallipoli and Makridis, 2018).
Figure 3: Correlations Between County Characteristics and Simulated 1-month GDP Decline
Admittedly, our approach has at least two limitations. For starters, we abstract from non-linearities and the role of uncertainty. Moreover, we abstract from inter-sectoral complementarities. We, therefore, interpret our estimates as a lower bound on the economic effects of the quarantine, complementing what is now becoming a larger body of literature on the economic ramifications of the pandemic and the forced shutdown. Some of these studies look retrospectively at the 1918 Influenza as a historical case-study, finding that it led to an 8% decline in consumption (Barro et al., 2020) and an 18% decline in manufacturing output. (Correia et al., 2020). Other studies have found similar declines in current consumption (Baker et al., 2020) and explain how supply-side shocks can adversely affect aggregate demand (Guerrieri et al., 2020; Eichenbaum et al., 2020).
Our paper, however, provides valuable information down to the county-level that will prove very useful for guiding the re-opening of the economy across locations. In particular, our results provide estimates of the potential benefits of re-opening at a county-level. These can be paired with data on the number of infections and population density to estimate probabilistic models about the risk of contagion that could arise from a staggered relaxation of social distancing measures.
We hope that the availability of new crowd-sourced data will allow for more robust testing and modeling of the economic effects and the policy surrounding the optimal re-opening of the economy. While our estimates suggest that the containment strategy has had significant costs, we are optimistic that these measures will not last forever: our country is reaching an inflection point, and we will soon be on the path towards a revival of mind and spirit that may pave the way for generations to come.
Barro, Robert J., Ursua, Jose F., and Weng, Joanna, “The coronavirus and the Great Influenza Pandemic: Lessons from the “Spanish flu” for the coronavirus’ potential mortality and economic activity,” NBER working paper (2020).
Correia, Sergio, Luck, Stephan, and Verner, Emil, “Pandemics Depress the Economy, Public Health Interventions Do Not: Evidence from the 1918 Flu”, Working paper (2020).
Eichenbaum, Martin S., Rebelo, Sergio, and Trabandt, Mathias, “The macroeconomics of epidemics,” NBER working paper (2020).
Gallipoli, Giovanni and Makridis, Christos, “Structural Transformation and the Rise of Information Technology,” Journal of Monetary Economics 97 (2018), pp. 91-110.
Guerrieri, Veronica, Lorenzoni, Guido, Straub, Ludwig, and Werning, Ivan, “Macroeconomic implications of COVID-19: Can negative supply shocks cause demand shortages,” Working paper (2020).