BBusiness disruptions from social distancing
Mikl´os Koren and Rita Pet˝o ∗ March 2020
Abstract
Social distancing interventions can be effective against epidemics but are potentiallydetrimental for the economy. Businesses that rely heavily on face-to-face communication orclose physical proximity when producing a product or providing a service are particularlyvulnerable. There is, however, no systematic evidence about the role of human interac-tions across different lines of business and about which will be the most limited by socialdistancing. Here we provide theory-based measures of the reliance of U.S. businesses onhuman interaction, detailed by industry and geographic location. We find that 49 millionworkers work in occupations that rely heavily on face-to-face communication or require closephysical proximity to other workers. Our model suggests that when businesses are forcedto reduce worker contacts by half, they need a 12 percent wage subsidy to compensate forthe disruption in communication. Retail, hotels and restaurants, arts and entertainmentand schools are the most affected sectors. Our results can help target fiscal assistance tobusinesses that are most disrupted by social distancing.
Introduction
Social distancing measures are effective non-pharmaceutical interventions against the rapidspread of epidemics [1, 2, 3, 4]. Many countries have implemented or are considering measuressuch as school closures, prohibition of large gatherings and restrictions on non-essential storesand transportation to slow down the spread of the 2019–20 coronavirus pandemic [5, 6, 7, 8].What are the economic effects of such social distancing interventions? Which businesses aremost affected by the restrictions?Past research has analyzed the efficacy of social distancing interventions on reducing thespread of epidemics using the 1918 Spanish Flu in the U.S. [3, 2, 1] and seasonal viral infectionsin France [9]. Our knowledge of economic impacts, however, is limited [10]. For this question,past data may be less relevant, as the importance of face-to-face communication has increasedsteadily in the last 100 years through urbanization [11, 12] and specialization increased inbusiness services as well [13, 14].The starting point of this paper is the observation that many sectors rely heavily on face-to-face communication in the production process [15, 16]. We build a model of communication ∗ Koren: Central European University, KRTK KTI and CEPR. 1051 Budapest, N´ador u. 9., Hungary. E-mail:[email protected]. Pet˝o: KRTK KTI. E-mail: [email protected]. a r X i v : . [ ec on . GN ] M a r o understand how limiting face-to-face interaction increases production costs. Without so-cial distancing, workers specialize in a narrow range of tasks and interact with other workerscompleting other tasks. This division of labor reduces production costs but requires frequentcontact between workers. In the model, the number of contacts per worker is the most frequentin high-population-density areas in businesses where the division of labor is important. Whenface-to-face interaction is limited, these are exactly the businesses that suffer the most.To measure business disruptions from social distancing, we turn to recent data on the taskdescriptions of each occupation [17] and the precise geographic location of non-farm businesses inthe U.S. [18]. We construct three groups of occupations and study their distribution in space.First, some occupations require face-to-face communication several times a week with otherworkers. Examples of these teamwork-intensive occupations include maintenance, personal carerelated occupations and health care professionals. Other occupations require frequent contactwith customers. Counselors, social workers and salespersons are examples of such customer-facing occupations. The third group of workers may need to be in physical proximity of oneanother even if they do not communicate, for example, to operate machinery or access keyresources. Examples of such occupations requiring physical presence are drivers and machineoperators, especially in mining and water transport, where crammed working environments arecommon.The spatial distribution of employment matters because face-to-face interactions are morefrequent in dense cities [15, 16]. So any social distancing intervention imposes disproportionatelymore limitation on urban sectors. We use our theory-motivated measures to estimate whichsectors and which locations will be particularly hurt by social distancing. A model of communication
When workers communicate with others, they can divide labor more effectively. Productioninvolves sequentially completing tasks indexed by z ∈ [0 , Z < Z γ /γ , where γ > γ , the more specialized each worker will bein a narrower set of tasks. Without loss of generality, we normalize the wage rate of workers toone so that all costs are expressed relative to worker wages.Once the range of tasks Z is completed, the worker passes the unfinished product on to an-other worker. This has a cost of τ , which can capture the cost of communicating and interactingacross workers. The determinants of communication cost will be parametrized later. After allthe tasks are completed, another step of communication with cost τ is needed to deliver theproduct to the customer. This cost leads to the Marshallian externality that firms want to beclose to their customers and customers want to be close to their suppliers [21, 22].The firm will optimally decide how to share tasks between workers. The key trade-off iseconomizing on the cost of communication while exploiting the division of labor [20]. Let n denote the number of workers involved in the production process. Because workers are2ymmetric, each works on Z = 1 /n range of tasks before passing the work to the next worker.Production involves n − Z = 1 /n ), vertical movement represents interaction ( τ ). Wenote three potential interpretations of our model. First, when workers work in teams, theycan efficiently divide labor among themselves (panel A). The benefit of a larger team is betterspecialization. Second, communication may involve the customer (panel B). The benefit ofmore frequent interaction with the customer is a product or service that is better suited to theirneeds. Third, workers may need access to a key physical resource, such as a machine, vehicleor an oil well (panel C). In this case, even if they do not communicate, they may be subject tosocial distancing measures. The key assumption behind all three interpretations is that frequentinteraction increases productivity, whether happening between workers, between workers andcustomers, or between workers and machines.W1W2W3... 1 /n /n /n /n /n /n /n /n /n /n /n Patterns of interaction in the workplace.
Horizontal movement representsproduction, vertical movement represents interaction. (A) Each worker W works on a range1 /n of tasks, passing work n − c ( τ ) = min n nτ + 1 γ n − γ , (1)where total communication costs are nτ and production costs are nZ γ /γ with Z = 1 /n .Given the strict convexity of this cost function, and ignoring integer problems, the first-ordercondition is necessary and sufficient for the optimum, n ∗ ( τ ) = τ − / (1+ γ ) . (2)The number of worker contacts is decreasing in the cost of communication, expressed relativeto worker wage. When the division of labor is important, γ is high, and the number of contactsdoes not depend very strongly on communication costs.The total cost of producing one good can be calculated by substituting in (2) into (1), c ( τ ) = τ χ /χ, (3)where χ = γ/ (1 + γ ) ∈ (0 ,
1) measures the importance of division of labor. This unit costfunction is the same as if workers and communication were substitutable in the production3unction in a Cobb-Douglas fashion. Indeed, χ captures the share of costs associated withcommunication and can be calibrated accordingly. The cost of communication
Businesses may differ in χ , how important communication is in their production process. An-other source of heterogeneity is the cost of communication between workers, which we specifyfurther as follows.When workers meet face to face, communication costs depend inversely on the populationdensity in the neighborhood of the firm, τ = d − ε with ε >
0. This captures the Marshallianexternality of knowledge spillovers [21], which happen more easily in densely populated areas[15, 23, 24, 16]. Contacts will be more frequent in dense areas, n ∗ ( d ) = d ε (1 − χ ) . (4)The unit cost of production is c ( d ) = d − εχ /χ. (5)Firms in dense areas face lower unit costs [25], but this agglomeration benefit may be offsetby higher wages and land rents in places with high population density [26, 27, 28]. In spatialequilibrium (not modeled here), firms with high communication needs will choose to locate inhigh-density areas [16]. Social distancing measures
We study the effect of a social distancing intervention that puts an upper limit N on thenumber of face-to-face contacts. Firms can mitigate the disruption from this measure by movingcommunication online, but this is more costly per contact than face-to-face communication.The optimal number of contacts without social distancing is given by Eq (4). Firms with n ∗ > N are limited by social distancing. Without moving communication online, their unit costwill increase to c (cid:48) = N d − ε + N − γ /γ , which is greater than the optimal cost, c (cid:48) ( d ) c ( d ) = χ Nn ∗ ( d ) + (1 − χ ) (cid:18) Nn ∗ ( d ) (cid:19) − γ > . (6)The first term of the weighted average is less than one, representing a reduction in communica-tion costs once the number of contacts is limited. The second term is greater than one due tothe fact that every worker has to complete a wider range of tasks than before, and they lose thebenefit of specialization. Because n ∗ is the cost-minimizing communication choice of the firm,the second term dominates, and production costs increase with social distancing.If the firm chooses to use telecommunication, the cost per contact will be T > d ε (otherwisethe firm would have used telecommunication before). The proportional increase in productioncosts in this case is given by c (cid:48)(cid:48) ( d ) c ( d ) = T χ d εχ > . (7)4n both cases, the cost increase is highest for communication-intensive firms (large γ and χ )and those operating in a high-density area (high d and hence high n ∗ ).Fig 2 displays the ratio of production costs under social distancing to the optimal productioncosts as a function of density. Firms in low-density areas are unaffected by social distancingsince they do not have many contacts anyway. Those in intermediate-density areas would sufferless of a cost increase by switching to telecommunication. Firms in the highest-density areaswill stick to face-to-face communication, which is still the more efficient form of communicationdespite restrictions. However, they will suffer the greatest cost increase. c (cid:48) /c density1 social distancingtelecommunicationFigure 2: Both social distancing and telecommunication hurt firms in dense areasmore.
See Eq (6) and Eq (7) for the relative production costs under the two interventions.
Data and methodology
To estimate the potential disruptions from social distancing, we need a measure of the im-portance of worker interaction (corresponding to χ in the model) and its cost (captured bypopulation density d ).Let ξ o denote an indicator equal to one if occupation o is interaction-intensive and zero oth-erwise. For industry i , χ i = (cid:80) o s io ξ o measures the fraction of workers in affected occupations,with s io denoting the employment share of occupation o in industry i .We use the Occupational Information Network (O*NET) [17] to measure the characteristicsof a given occupation, similarly to previous studies [29, 30, 31]. The O*NET dataset containsdetailed standardized descriptions on almost 1,000 occupations along eight dimensions. Wefocus on job characteristics that are related to recent social distancing measures, while priorwork focused mainly on measuring offshorability of the given tasks [29, 30].Social distancing interventions limit the interaction between people and regulate physicalproximity between individuals. We thus focus on three related job characteristics based onwork context and work activity described in O*NET. The first two indicators capture howcommunication-intensive the job is. Communication can be of two types: internal communica-tion with co-workers ( teamwork ) or external communication directly with customers ( customer-facing ). The third indicator takes into consideration the possibility that workers may need tobe in physical proximity of one another even if they do not communicate. We create an in-dex that shows how important physical presence is to perform a given job. Table 1 details5he specific O*NET indexes that contribute to each of our three measures. As social distanc-ing measures only limit personal communication, for communication indexes, we require thatthe necessary face-to-face communication happens at least several times a week. In teamwork,face-to-face meetings can often be substituted by more structured communication, for whichworking from home is not as disruptive. To allow for this possibility, we only classify occupa-tions as teamwork-intensive where both emails and letters and memos are less frequent forms ofcommunication than face-to-face meetings. This excludes most managers and certain businessservices. Similarly, for physical presence, we require at least a certain degree of proximity toother workers which corresponds to working in a shared office.Table 1: Definition of social distancing indexes.Index Tasks Context
Teamwork Work With Work Group or Team Face-to-faceProvide Consultation and Advice to Others discussionsCoordinating the Work and Activities of Others several times a weekGuiding Directing and Motivating Subordinates & more often thanDeveloping and Building Teams emails, letters, memosCustomer Deal With External Customers Face-to-facePerforming for or Working Directly with the Public discussionsAssisting and Caring for Others severalProvide Consultation and Advice to Others times a weekEstablishing and Maintaining Interpersonal RelationshipsPresence Handling and Moving Objects Density ofOperating Vehicles, Mechanized Devices or Equipment co-workersRepairing and Maintaining Electronic Equipment like sharedRepairing and Maintaining Mechanical Equipment office or moreInspecting Equipment, Structures, or MaterialEach social distancing index (column 1) is created as an arithmetic average of the componentindexes (column 2). To be classified an affected occupation, the average has to exceed 62.5and the work context index has to exceed the threshold in column 3.We aggregate the measures to 6-digit occupation codes (Standard Occupational Classifica-tion; 2010-SOC). We have information on the relevance of teamwork, customer contact andphysical presence for 809 occupations in SOC 2010 codes.Teamwork and customer contacts are highly correlated (Fig 3), but they are conceptuallydifferent. While all medical occupations require teamwork and customer contact, supervisors ingeneral are working in teams but do not often communicate directly with customers. Machineoperators and production workers in general are at the bottom of both of the distributions. Asmanagers can substitute personal communication with emails, they are not considered in generalas teamwork-intensive occupations according to our definition. Given the high correlation be-tween the two types of communication, we often refer to communication-intensive occupations6hat are either teamwork-intensive or customer-facing.
Healthcare social workersElementary school teachersDentistsPsychiatristsSurgeonsAthletic trainersFirefightersChefs and head cooksCouriersPostal service mail carriers Textile machine operatorsNuclear power reactor operatorsChemical equipment operators C u s t o m e r c on t a c t
20 40 60 80Teamwork
Figure 3:
Teamwork and customer contact are highly correlated.
Each circle representsan occupation. Teamwork and customer contact indexes are constructed as explained in maintext.As a next step, we calculate for each sector the share of workers whose job requires a highlevel of teamwork, customer contact and physical presence. We use the same sectoral break-down as the Current Employment Statistics (CES) [32]. As all the indexes are an absolute valuerunning from 0 to 100, we use 62.5 as a cutoff to define a job to be teamwork-intensive, cus-tomers contact-intensive or job that require physical presence from the worker. The occupationstructure of the industries are retrieved from the official industry-occupation matrix [33], weuse the employment statistics by occupation-industry for February 2020.Based on the share of relevant occupations in industry employment, the most teamwork-intensive sectors are, for example “Hospitals,” “Accommodation” and “Motion picture andsound recording industries.” In contrast, teamwork is not important in sectors like “Forestryand logging” and “Fishing, hunting and trapping.” Customer contact is relevant in sectors like“Hospitals” and “Retail”, while it is not relevant is sectors line “Truck transportation,” and“Forestry and logging.” Physical presence is relevant in sectors like “Truck transportation,”“Repair and maintenance,” mining in general, while it is not relevant in finance and informationtechnology sectors.“Hospitals” score high on all three measures because communication in health care teamsand with patients is important, and doctors and nurses work in close physical proximity toothers. We nonetheless remove this sector from the analysis because of its inevitable direct rolein combating the epidemic which is not captured well in a simple model of communication.To measure the heterogeneity in the cost of communication, we measure population densityin the neighborhood of businesses. The assumption in the model is that communication costs arelower in dense areas. This is consistent with the fact that communication-intensive sectors such7s business services and advertising, together with central administrative offices of productionfirms tend to be concentrated in high-density areas [34].The location of sectors comes from the County Business Patterns (CBP) data for 2017 [18].For a finer spatial resolution, we use the data tabulated by ZIP-Code Tabulation Areas. TheCBP lists the number of establishments of a certain size for each ZIP-code and NAICS industrycode. Because establishment sizes are given in bins (e.g., 1–4 employees), we take the midpointof each bin as our estimated employment (e.g., 2.5 employees). In small industries and ZIPcodes, the Census omits some size categories to protect the confidentiality of businesses. Weimpute employment in these plants from the national size distribution of plants in the sameNAICS industry. Our estimated industry-level employment is a very good approximation toofficial employment statistics [32]. The correlation between our estimates based on CBP andthe employment reported in CES is 0.98.To understand the heterogeneity across regions, we average the share of workers in teamwork-intensive, customer-facing and physical presence occupations across industries active in each ZIPcode. For each of the three occupation groups, let ξ og denote the indicator whether occupation o belongs to group g . The industry share of workers in the occupation group is given by (cid:80) o s io ξ og . We compute the regional share of the occupation group as an employment-weightedaverage across industries, (cid:80) i l ir (cid:80) o s io ξ og / (cid:80) i l ir , where l ir is the estimated employment ofindustry i in region (ZIP code) r . Because we only have industry but not occupation data bylocation, we have assumed that the occupation distribution within the sector, s io , does not varyacross locations. The fact that the CBP tabulates employment by establishments rather thanfirms makes this a good approximation. For example, a sporting good producer may also havean administrative office and a retail store in different locations, but these will be classified intheir respective NAICS sectors rather than in sporting good manufacturing.We use population density to measure how locations differ in the costs of communication.We also experimented with using employment densities instead of population densities. Resultswere very similar, as the two measures are highly correlated with the exception of very highemployment-density urban centers where population is more sparse [35]. Counterfactual calculations
To gauge the magnitude of the effect of social distancing, we calibrate the parameters of themodel and compute the effect of a policy that limits the number of worker contacts. At the sametime, we let the government introduce a proportional wage subsidy λ to help offset the costsfrom lower interaction. With this subsidy, the cost of labor will be (1 − λ ). We ask what level of λ would compensate businesses for the communication disruption caused by social distancing.Using the cost change in Eq 6, we can express λ ir = 1 − − χ i − χ i N/n ∗ ir (cid:18) Nn ∗ ir (cid:19) γ i > . (8)The compensating wage subsidy increases in the importance of communication χ i and the opti-mal number of contacts n ∗ ir , and decreases in the number of allowed contacts N . The subscripts8ote that communication share is industry specific and the optimal number of contacts is bothindustry and region specific.We calibrate the upper limit on personal contacts N such that the overall number of con-tacts in the economy, when averaged across ZIP codes and industries, is reduced by half, (cid:80) i,r l ir min { N, n ∗ ir } = 0 . (cid:80) i,r l ir n ∗ ir . Due to the inherent nonlinearity of the model, otherinterventions will have different effects.To calibrate the importance of communication χ i , note that it is the cost share of commu-nication, and can be correspondingly calibrated to the employment share of communication-intensive occupations in industry i . Here we take all occupations that are either teamworkintensive or customer facing.Population density in region r can be measured directly in the data, as explained above.The only remaining parameter to calibrate is the elasticity of the communication externality ε . We rely on previous estimates of agglomeration effects that capture the elasticity of totalfactor productivity with respect to population density [25]. In our model, the elasticity of unitcosts (which can be construed as the inverse of productivity) with respect to density is − εχ (Eq (5)). We calibrate ε = 0 .
02 so that, across all ZIP codes and industries, a regression of logmodel-implied productivity ( εχ i ln d r ) on log population density ln d r yields an elasticity of 0.04[25, page 60].Given these parameter values, we compute the compensating wage subsidy for each industryin each ZIP code using Eq 8. We report employment-weighted averages of this across sectorsand across locations. Results
Table 2 displays the top five and the bottom five industries by 2-digit NAICS industries as sortedby the percentage of workers in communication-intensive occupations, excluding hospitals andclinics. Across industries, retail trade, accommodation and food services, arts, entertainment,and recreation, other services and educational services have the highest share of communication-intensive workers, exceeding 35 percent. Transportation, production, construction and agri-cultural industries are less reliant on face-to-face communication. This heterogeneity acrossindustries is important to understand the effect of social distancing measures.Fig 4 plots the share of workers in the three affected occupation groups across ZIP codesby population density. Customer-facing occupations are overrepresented in dense areas. In thehighest population density ZIP codes, 43 percent of workers are employed in customer-facingoccupations. Teamwork-intensive occupations are broadly distributed in space.In the calibrated model, a social distancing policy that puts a cap on interactions per workersuch that the total number of interactions drops by half nationwide is compensated a 12.2percent wage subsidy. The distribution of the compensating wage subsidy is, however, unequalover space and across industries. New York City, with a population density about 20 times theaverage U.S. city, would require a 13.3 percent wage subsidy. By contrast, the compensatingwage subsidy in agriculture, transportation and manufacturing would be less than 6 percent9able 2:
Retail, professional services, finance and restaurants are the most commu-nication intensive. CommunicationIndustry Teamw. Custom. Overall Presence
Retail trade 13 67 68 5Accommodation & food services 8 50 51 1Arts, Entertainment, and Recreation 12 40 42 2Other Services (except Public Admin.) 12 38 41 12Educational Services 15 35 37 1...Wholesale Trade 8 16 20 12Transportation and Warehousing 8 10 16 32Manufacturing 7 6 11 10Construction 15 5 18 28Agri., forestry, fishing & hunting 4 4 7 23“Teamw.” and “Custom.” show the percentage of workers in teamwork-intensive andcustomer-facing occupations, respectively. “Overall” shows the percentage of workers incommunication-intensive occupations that are either teamwork-intensive or customer-facing.It is less than the sum of the two indexes because some occupations rely on both types ofcommunication. “Presence” shows the percentage of workers whose jobs require physicalpresence in close proximity to others.(Table 3).
Discussion and conclusions
The main cost of social distancing in our model is insufficient division of labor. This mechanismis motivated by [19] and captures the same trade-off as [20]. Our contribution is specifying thecost function in such a way that can be easily mapped to the data.More broadly, our argument is that frequent interaction increases productivity irrespectivewhether it is happening between workers, between workers and customers, or between workersand machines. In the main part of the empirical analysis, we focused only on the first twotypes of interactions, while we were silent about the third. But social distancing measures alsoaffect sectors where workers need to be in physical proximity of one another even if they do notcommunicate, for example, to operate machinery or access key resources. This is relevant insectors like “Mining, Quarrying, and Oil and Gas Extraction” and “Transportation” while it isnot relevant in sectors like “Finance and Insurance” and “Professional, Scientific, and TechnicalServices.” As can be seen from Fig 4, occupations requiring physical presence have the highestshare in less dense areas where production and mining plants are located. (Farms are notincluded in the CBP.) The share of these occupations in the most dense areas is only 3 percent.10 S ha r e o f w o r k e r s ( pe r c en t )
10 100 1000 10000 100000Population density (person/km2, log scale)Teamwork Customer contactPhysical presence
Figure 4:
Urban areas employ more workers in customer-facing occupations, lessin occupations requiring physical presence.
Locally weighted polynomial regression ofaverage share of teamwork-intensive occupations, customer-facing occupations and occupationsrequiring physical presence across sectors within the ZIP code (bandwidth = 0 . The five most affected sectors require more than 20 percent wage subsidy.Industry Wage subsidy Employment
Retail Trade 22.1 15,659Accommodation and Food Services 17.7 14,379Arts, Entertainment, and Recreation 15.1 2,494Other Services (except Public Admin.) 14.5 5,939Educational Services 13.8 3,838...Wholesale Trade 7.7 5,936Construction 6.8 7,646Transportation and Warehousing 5.9 5,523Manufacturing 4.5 12,861Agriculture, Forestry, Fishing and Hunting 2.6 55
Average 12.2 116,496 “Wage subsidy” displays the percentage decrease in labor costs necessary to compensatebusinesses when worker contacts are reduced by half. “Employment” is the February 2020employment of the sector in thousands [32]. The last row shows the employment-weightedaverage wage subsidy. Table excludes hospitals, clinics, and government establishments whichare not present in CBP. 11o a greater or lesser extent, all sectors will be affected by social distancing. Some sectorsare hit by the intervention due to restricted face-to-face communication, others are hit due torestricting physical proximity of people. Some sectors are less affected across all dimensions.Examples include “Fishing, hunting and trapping,” “Printing and related support activities,”and manufacturing in general.We see four avenues for further research. The first concerns the demand side of the economy,which we have mostly neglected by focusing on the production function. The employmentresponse to a production disruption greatly depends on the elasticity of demand. We hypothesizethat sectors like schooling and health care have inelastic demand and will continue to employmany workers despite significant disruptions. However, personal services, small grocery storesand restaurants may face more elastic demand and respond to large production cost increasesby laying off a significant fraction of their work force.The second direction concerns the interaction between sectors and regions. Whenever pro-ductivity in any business drops, this shock can propagate to its buyers and suppliers. Theaggregate consequences of the epidemic will hence be modulated by input-output linkages be-tween sectors, regions and countries [36, 37, 38].The third and forth directions concern the long-run response of businesses as they try tobecome more resilient to such shocks in the future. Whether the share of telecommunicationremains large in the long run depends crucially on how easily it substitutes for face-to-faceinteraction. In our model, the two are perfect substitutes with the only distinction that theefficiency of face-to-face meetings improves with population density, whereas telecommunicationdoes not depend on agglomeration [23, 24, 16]. Previous work has found face-to-face commu-nication to be more effective in high-intensity communication which is particularly helpful toovercome incentive problems in joint production [39, 40]. Data on internet flows suggests thattelecommunication is not a good substitute for face-to-face meetings [41]. None of these papersdiscuss disruptions from social distancing measures. Further study of the modes of communi-cation in the O*NET occupation survey can shed light on whether telecommunication can actas a low-cost substitute for face-to-face meetings.Fourth, businesses may change their location in response to perceived threats and disrup-tions. As we discussed, epidemics have a disproportionate effect on cities. So it is conceivablethat in a post-pandemic spatial equilibrium (not modeled here, but see [16]), the agglomerationpremium falls and firms find it less attractive to locate in cities. A poignant point of comparisonis the increased threat of terrorism in major cities following devastating attacks on New York,Washington, London, Paris, Madrid, Moscow and Mumbai. The general conclusion about terrorthreat is that cities have remained resilient and a robust attractor of businesses [42, 43]. Wespeculate that epidemics and social distancing can be more detrimental to cities than terrorthreats, because they tear apart the very fabric of urban life. However, we have limited data tomake further predictions.
Supporting information Available at https://github.com/ceumicrodata/social-distancing/blob/master/data/derived/industry-index.csv . S2 Full table of ZIP-codes. Social distancing exposure by location.
Available at https://github.com/ceumicrodata/social-distancing/blob/master/data/derived/location-index.csv . S3 Data repository. Replication code and data.
Replication code and data are availableat https://github.com/ceumicrodata/social-distancing . Acknowledgments
We thank G´abor B´ek´es, P´eter Harasztosi, P´eter Kar´adi, Bal´azs Lengyel, D´avid Kriszti´an Nagy,and Andrea Weber for comments.
References [1] Bootsma MCJ, Ferguson NM. The effect of public health measures on the 1918 influenzapandemic in US cities. Proc Natl Acad Sci. 2007;.[2] Markel H, Lipman HB, Navarro JA, Sloan A, Michalsen JR, Stern AM, et al. Nonpharma-ceutical interventions implemented by US cities during the 1918-1919 influenza pandemic.JAMA. 2007;298(6):644–654.[3] Hatchett RJ, Mecher CE, Lipsitch M. Public health interventions and epidemic intensityduring the 1918 influenza pandemic. Proc Natl Acad Sci. 2007;104(18):7582–7587.[4] Wilder-Smith A, Freedman DO. Isolation, quarantine, social distancing and communitycontainment: pivotal role for old-style public health measures in the novel coronavirus(2019-nCoV) outbreak. J Travel Med. 2020;27(2).[5] Anderson RM, Heesterbeek H, Klinkenberg D, Hollingsworth TD. How will country-basedmitigation measures influence the course of the COVID-19 epidemic? Lancet. 2020;.[6] Cohen J, Kupferschmidt K. Mass testing, school closings, lockdowns: Countries pick tacticsin ‘war’ against coronavirus. Science. 2020;.[7] Thompson SA, Serkez Y. How Has Your State Reacted to Social Distancing? The NewYork Times. 2020;.[8] Mayor de Blasio Issues New Guidance to New Yorkers;2020. .139] Adda J. Economic Activity and the Spread of Viral Diseases: Evidence from High Fre-quency Data. Q J Econ. 2016;131(2):891–941.[10] Wren-Lewis S. The economic effects of a pandemic. In: Baldwin R, Weder di Mauro B,editors. Economics in the Time of COVID-19. London, UK: CEPR Press; 2020. p. 109–112.[11] Henderson JV. Cities and Development. J Reg Sci. 2010;50(1):515–540.[12] Henderson V. Urbanization in Developing Countries. World Bank Res Obs. 2002;17(1):89–112.[13] Herrendorf B, Rogerson R, Valentinyi ´A. Growth and Structural Transformation. In:Aghion P, Durlauf SN, editors. Handbook of Economic Growth. vol. 2. Elsevier; 2014. p.855–941.[14] Duarte M, Restuccia D. Relative Prices and Sectoral Productivity. J Eur Econ Assoc.2019;.[15] Charlot S, Duranton G. Communication externalities in cities. J Urban Econ.2004;56(3):581–613.[16] Tian L. Division of Labor and Productivity Advantage of Cities: Theory and Evidencefrom Brazil; 2019.[17] National Center for O*NET Development. O*NET OnLine; 2020.[18] U S Bureau of the Census. County Business Patterns; 2017.[19] Smith A. An Inquiry into the Nature and Causes of the Wealth of Nations (Book I).London: W. Strahan; T. Cadell; 1778.[20] Becker GS, Murphy KM. The Division of Labor, Coordination Costs, and Knowledge. QJ Econ. 1992;107(4):1137–1160.[21] Marshall A. Principles of Economics (eighth edition). London: MacMillan; 1920.[22] Krugman P. Geography and Trade. Cambridge: MIT Press; 1991.[23] Rossi-Hansberg E, Wright MLJ. Urban Structure and Growth. Rev Econ Stud.2007;74(2):597–624.[24] Ioannides YM, Overman HG, Rossi-Hansberg E, Schmidheiny K, Moav O. The Ef-fect of Information and Communication Technologies on Urban Structure. Econ Policy.2008;23(54):201–242.[25] Ciccone A, Hall R. Productivity and the Density of Economic Activity. Am Econ Rev.1996;86(1):54–70.[26] Eberts RW, Gronberg TJ. Wage gradients, rent gradients, and the price elasticity ofdemand for housing: An empirical investigation. J Urban Econ. 1982;12(2):168–176.1427] Madden JF. Urban wage gradients: Empirical evidence. J Urban Econ. 1985;18(3):291–301.[28] Combes PP, Duranton G, Gobillon L. The Costs of Agglomeration: House and Land Pricesin French Cities. Rev Econ Stud. 2019;86(4):1556–1589.[29] Firpo S, Fortin NM, Lemieux T. Occupational Tasks and Changes in the Wage Structure.IZA; 2011. 5542.[30] Autor DH, Dorn D. The Growth of Low-Skill Service Jobs and the Polarization of the USLabor Market. Am Econ Rev. 2013;103(5):1553–1597.[31] Jin B, McGill A. Who is most at risk in the coronavirus crisis: 24 million ofthe lowest-income workers; 2020.