Research & Policy Briefs From the World Bank Malaysia Hub No. 34 May 11, 2020 Which Jobs Are Most Vulnerable to COVID-19? What an Analysis of the European Union Reveals Daniel Garrote Sanchez, Nicolas Gomez Parra, Caglar Ozden, and Bob Rijkers This Research & Policy Brief presents measures of labor market exposure to COVID-19 in the European Union (EU) by identifying jobs in non-essential industries that cannot be performed from home. Jobs most at risk account for 30 percent of all EU employment. These jobs are concentrated in lagging regions; tend to be low paid and less secure; and are disproportionately held by young, poorly educated workers and migrants. In the absence of urgent large-scale remedial action, the COVID-19 crisis is likely to exacerbate preexisting socioeconomic and regional disparities. Introduction Essential Jobs Implementing policies to counter the economic damage inflicted by Government-mandated lockdowns establish which jobs or sectors are the COVID-19 pandemic requires knowing which jobs are most deemed essential to the functioning of the society and the economy. vulnerable. While more than half of all confirmed COVID-19 cases are These administrative decisions are the first set of criteria that this in Europe (as of April 24), not all countries or regions within the same analysis uses to determine the labor market vulnerability of workers. country are affected to the same extent. This is illustrated by map 1 In most countries, for example, doctors, nurses, and other medical which shows the number of confirmed cases per 100,000 habitants professionals delivering critical health care and those in industries for each subregion in Europe. Although the final health outcomes are providing essential goods and services such as food, water, electricity, unknown, they will certainly vary substantially by region. Similarly, not and transportation are allowed—and even encouraged—to continue all workers suffer to the same extent in terms of labor market to go to work. While there are significant overlaps on the “essential” outcomes, even though almost all European economies are in lists of different jurisdictions, there is, inevitably, some degree of lockdowns. Those workers who can work from home and those subjectivity and variation among them. There are even differences employed in essential industries that are kept open can continue to earn a living. In contrast, those who have jobs deemed non-essential across states within the same country, reflecting political, social, and that cannot be performed from home are facing the most significant economic priorities at the local level. In the United States, for job and income losses. example, the federal government has issued a list of 16 sectors that can remain open and continue to operate. This list includes sectors Based on this simple insight, this analysis constructs a new such as food and agriculture, health care and public services, and measure of labor market exposure to COVID-19 and assesses which emergency services. In addition, many states, cities, and counties jobs are most at risk, using data from the most recent 2018 European have declared their own states of emergency, ordering non-essential Labour Force Survey (EU LFS). The objective is to help governments businesses to close. While most local governments follow the federal with limited resources target their support to the regions, sectors, and guidelines, some jurisdictions have produced their own list of occupations that are more severely affected. The analysis essential activities using a detailed classification of industries. demonstrates that, in the absence of urgent large-scale action, the COVID-19 crisis is likely to exacerbate preexisting socioeconomic and The list of “essential sectors” used in this analysis is based on the regional disparities. Young, less-educated workers who are already in decisions of Italy (EU) and the US states of Delaware, Minnesota, and less-secure and low-paying jobs are likely to bear the brunt of the Oklahoma, which have produced lists of which sectors are deemed shock, with lagging regions suffering the worst losses. essential with explicit and highly detailed NAICS (North American Industry Classification System) codes at the 6-digit level (see the reference below table 1). These announcements permit direct Map 1. Spread of COVID-19 in Europe: Confirmed Cases (per 100,000 mapping of essential sectors allowed to stay open onto economic data habitants) that prevent the measures in this study from suffering from interpretative coding error. This study classifies a sector as essential if No data Less it was listed as being such in each of the lists issued by Italy, Delaware, Minnesota, and Oklahoma. (The qualitative pattern of results obtained is very robust to using alternative measures, such as only considering a sector as essential if it appears on at least two of the lists.) As such, the measure used here includes only those sectors that appear on all lists and are thus unanimously deemed essential. This More measure most closely follows the list of essential sectors issued by Italy, one of the worst affected countries, which has put in place one of the most stringent shutdowns observed to date. Appendix table A1 presents the correlation between the different classifications in different jurisdictions. A related complication is that some countries have issued lists of sectors that are allowed to stay open provided they follow social distancing practices (such as deliveries only, or in-person attendance Source: Observatoire Coronavirus–Le Grand Continent (accessed April 24, 2020). only for emergencies). This analysis uses only the regulations on Affiliation: Daniel Garrote Sanchez (Social Protection and Labor, dgarrotesanchez@worldbank.org) , Nicolas Gomez Parra (Development Research Group, ngomezparra@worldbank.org), Caglar Ozden (Development Research Group, cozden@worldbank.org), Bob Rijkers (Development Research Group, brijkers@worldbank.org). Acknowledgement: We are grateful to Ana Fernandes, Chisako Fukuda, Aart Kraay, Norman Loayza, Harry Moroz, Nina Rahman, Achim Schmillen, and Joana Silva for useful comments and discussions. We acknowledge the financial support from the Multi-Donor Trust Fund on International Trade and Knowledge for Change Program. Objective and disclaimer: Research & Policy Briefs synthesize existing research and data to shed light on a useful and interesting question for policy debate. Research & Policy Briefs carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions are entirely those of the authors. They do not necessarily represent the views of the World Bank Group, its Executive Directors, or the governments they represent. Which Jobs Are Most Vulnerable to COVID-19? What an Analysis of the European Union Reveals Table 1. Share of 6-Digit Industries that are Considered Essential, Italy Map 2. Percent of Jobs Considered Essential, European Union, Norway, and the U.S. States of Delaware, Minnesota, and Oklahoma and Switzerland, 2018 Share State/country (percent of total number of 6-digit NAICS) Restrictions % European Union 65 - 72 Italy 48.0 507 60 - 65 55 - 60 United States 50 - 55 Delaware 81.4 860 45 - 50 40 - 45 Minnesota 63.2 668 No data Oklahoma 78.1 826 Source: For Italy, del Rio-Chanona et al. (2020); for the United States, Delaware Division of Public Health; Minnesota Executive Department; Oklahoma Department of Commerce; and U.S. Cybersecurity and Infrastructure Security Agency. Note: There are 1057 6-digit North American Industry Classification System (NAICS) codes in total. strictly essential sectors. Thus, applying this study’s measure of essential sectors to other areas will inevitably entail extrapolative error. Given the high correlation across the different lists and the fact that the analysis conservatively includes only those sectors that appear on all lists, this study’s measure provides a decent first-order approximation of what sectors are likely to be included in lists of essential sectors issued by other jurisdictions. The next challenge relates to the classification of industries in the EU LFS dataset. In order to create a harmonized EU-wide dataset, EU LFS provides the sector of each worker only at a relatively aggregated level (the 1-digit Nomenclature of Economic Activities (NACE) Source: Authors’ calculations based on data from Eurostat, European Union category). Therefore, this study uses the classification by 6-digit NAICS Labor Force Survey (2018a). codes and maps it to 1-digit NACE industries. The weights for each Note: Data are for 2018. 6-digit NAICS category within each 1-digit NACE industry are calculated using the detailed US employment data from the 2019 Once the share of the essential workers in each of the NACE 1-digit Occupational Employment Statistics (OES) issued by the US Bureau of sectors is determined, the share of essential workers in each of the Labor Statistics. The concordance between these two classifications is statistical regions (Nomenclature of Territorial Units for Statistics quite similar if the weights based on the individual labor force surveys [NUTS2] regions) within the European Union (and Norway and of other EU countries are used. The results are shown in table 2. All Switzerland) can be calculated. These shares are presented in map 2. the workers in several of the 1-digit categories are considered essential. These categories include utilities (such as electricity, water, More than half of all jobs in the EU (58 percent) are in sectors and sewerage), as well as public administration, health considered essential. The share of employment in essential industries administration, and transportation. varies significantly across geographic regions and tends to increase Table 2. Share of Labor Force Working in Essential Sectors, Italy and the U.S. States of Delaware, Minnesota, and Oklahoma NACE-1 Employment in digit essential sectors Sector category (% of workers in the sector) A 88.9 Agriculture, Forestry and Fishing B 26.3 Mining and Quarrying C 24.2 Manufacturing D 100.0 Electricity, Gas, Steam and Air Conditioning Supply E 100.0 Water Supply; Sewerage, Waste Management and Remediation Activities F 55.7 Construction G 20.0 Wholesale and Retail Trade; Repair of Motor Vehicles and Motorcycles H 98.7 Transportation and Storage I 0.0 Accommodation and Food Service Activities J 69.4 Information and Communication K 99.8 Financial and Insurance Activities L 0.0 Real Estate Activities M 80.5 Professional, Scientific and Technical Activities N 11.8 Administrative and Support Service Activities O 100.0 Public Administration and Defense; Compulsory Social Security P 100.0 Education Q 100.0 Human Health and Social Work Activities R 0.0 Arts, Entertainment and Recreation S 11.3 Other Service Activities T -- Activities of Households as Employers and for Own Use U -- Activities of Extraterritorial Organizations and Bodies Source: Authors' calculations based on del Rio-Chanona et al. (2020) for Italy, and Delaware Division of Public Health; Minnesota Executive Department; and Oklahoma Department of Commerce; U.S. Bureau of Labor Statistics; and U.S. Cybersecurity and Infrastructure Security Agency for the United States. Note: There are no concordances with 6-digit North American Industry Classification System (NAICS) for the last two Nomenclature of Economic Activities (NACE) sectors (T and U). 2 Research & Policy Brief No.34 with income. (Darker blue means the share of essential workers Figure 1. Home-Based Work versus Face-to-Face Interactions for among all workers is higher in that region.) The share is higher in Various Sectors Belgium, France, Luxembourg, the Netherlands, and the Scandinavian countries, and much lower in Southern European countries such as a. Share of jobs that are amenable to home-based Italy and Spain. Furthermore, there is substantial variation within work versus face-to-face jobs countries. (NACE 1-digit) Home-Based Work and Face-to-Face Jobs 80 ICT The second measure of exposure to vulnerability is based on the Finance Education Share of telework jobs (%) nature of the jobs themselves, as opposed to government mandates. 60 Int. Orgs. Professionals Two different criteria are used in the literature. The first is the feasibility of home-based work. Dingel and Neiman (2020) use Art / recreation Real Estate Government information from characteristics of more than 900 occupations based 40 Electricity/gas on two surveys from the US Department of Labor, Employment and Trade Other Training Administration’s Occupational Information Network Transport Admin Health & social 20 (O*NET). When answers to those surveys reveal that an occupation Manufacture Mining Water/waste mgmt requires daily “work outdoors” or that “operating vehicles, Construction Agriculture Hospitality HH activities mechanized devices, or equipment is very important to that 0 occupation’s performance,” they determine that the occupation 20 30 40 50 60 70 cannot be performed entirely from home. This study translates Intensity of face-to-face interactions (0-100) those occupations, based on the Standard Occupational Classification (SOC) system used in the United States, to the occupation classifications system used in Europe (the International Standard Classification of Occupations, ISCO-08) at the 3-digit level of b. Share of essential jobs versus jobs that are amendable granularity. to home-based work A second and related measure is based on the extent of (NACE 1-digit) face-to-face interactions in various occupations (Avdiu and Nayyar 100 Health Water/waste mgmt& social Electricity/gas Transport Government Finance Education 2020; Blinder 2006). Blinder originally constructed his measure to Agriculture determine whether an occupation could be moved offshore. He argues 80 Share of essential jobs (%) Professionals that those occupations that require face-to-face interaction with the consumer (such as retail, health care, or education services) or require 60 ICT inputs specific to a location (such as construction or agriculture) are Construction not offshorable. This study modifies these categorizations to focus only 40 on those occupations that require face-to-face interaction. Using the same rationale, Avdiu and Nayyar (2020) create an index of Mining face-to-face interaction that varies from 0 to 1. 20 Manufacture Trade Admin Other Different factors determine the amount of face-to-face interaction 0 required in different jobs. Tasks that involve (1) establishing and 0 Hospitality 20 40 Real Art / recreation Estate 60 80 maintaining personal relationships; (2) assisting and caring for others; Share of telework jobs (%) (3) performing for or working directly with the public; and/or (4) selling to or influencing others typically require more extensive personal interaction. Consequently, these jobs are more susceptible to COVID-19–induced labor market disruptions, social distancing, and c. Share of essential jobs versus face-to-face jobs other similar behavioral changes. The feasibility of home-based work is correlated with the extent of face-to-face interaction required, as is (NACE 1-digit) shown in figure 1. Information, communication, and technology (ICT) 100 Electricity/gas Water/waste mgmt Government Finance Transport Health & social Education and professional and scientific jobs can more easily be provided from Agriculture home and require little face-to-face interaction. On the other hand, 80 Share of essential jobs (%) Professionals hospitality, food services, and health and social services are not amenable to home-based work and require extensive face-to-face 60 ICT interactions. However, in certain industries, the two measures Construction diverge. For example, the majority of manufacturing jobs require 40 physical presence in the place of work but do not demand extensive face-to-face interaction between workers or workers and consumers. Mining Conversely, education services are amenable to home-based work, 20 Manufacture Trade but they still require significant face-to-face interactions. Whether Admin Other they can be performed remotely depends on technology, training of 0 Art / recreation Real Estate Hospitality the service providers as well as willingness of students. 20 30 40 50 60 70 Intensity of face-to-face interactions (0-100) The relationships between how essential a sector is and the extent of home-based work, as well as face-to-face interactions, are presented in panels b and c of figure 1, respectively. Since governments (not Source: Authors’ calculations based on data from Eurostat, European Union markets) determine whether a sector is essential by decree, sectors Labor Force Survey (2018a). tend to be at the extremes—they are either essential or not. In Note: Data are for 2018. All data are at the Nomenclature of Economic Activities contrast, variation in the extent of home-based work or face-to-face (NACE) 1-digit level. HH = household; ICT = information and communications technology. interactions across sectors is less extreme. As can be seen in different 3 Which Jobs Are Most Vulnerable to COVID-19? What an Analysis of the European Union Reveals Map 3. Home-Based Work and Face-to-face Interactions, European Union, Norway, and Switzerland, 2018 a. Percent of jobs that are amenable to home-based telework b. Intensity of face-to-face jobs (index 0-100) % (Index 0-100) 50 - 60 50 - 55 45 - 50 48 - 50 40 - 45 46 - 48 35 - 40 45 - 46 30 - 35 44 - 45 25 - 30 43 - 44 20 - 25 42 - 43 10 - 20 40 - 42 No data 35 - 40 No data Source: For panel a, authors’ calculations based on data from Eurostat, European Union Labor Force Survey (2018a) and Dingel and Neiman (2020) methodology. For panel b, authors’ calculations based on data from Eurostat, European Union Labor Force Survey (2018a) and Avdiu and Nayyar (2020) methodology. Note: Data are for 2018. panels of figure 1, the measure of which sectors are essential is only In the EU, 35 percent of all jobs can be done at home (this share is weakly correlated with home-based work and face-to-face very similar to Dingel and Neiman’s (2020) finding that 37 percent of interaction. This in turn implies that merely relying on whether a US jobs can be performed at home). Jobs in the ICT, finance, and sector can be performed from home and/or requires face-to-face education sectors are highly amenable to working from home. On the interaction will provide only a very partial picture of what jobs are at other hand, jobs in agriculture and hospitality (hotels, restaurants, risk because of COVID-19. bars) are less amenable to home-based work. The feasibility of Map 4. Jobs Most at Risk, European Union, Norway, and Switzerland a. Percent of jobs non-essential and not amenable to telework b. Percent of jobs non-essential and with extensive face-to-face interactions % % 40 - 50 30 - 50 35 - 40 25 - 30 33 - 35 23 - 25 31 - 33 21 - 23 29 - 31 19 - 21 27 - 29 17 - 19 25 - 27 15 - 17 20 - 25 13 - 15 10 - 20 10 - 13 No data No data Source: Authors’ calculations based on data from Eurostat, European Union Labor Force Survey (2018a). Note: Data are for 2018. 4 Research & Policy Brief No.34 home-based work increases with income. Richer and typically Figure 2. Job Exposure versus Regional Development Northern European countries such as Denmark, the Netherlands, Norway, Sweden, and Switzerland are characterized by a greater a. Share of jobs that are both non-essential and not prevalence of work that can be done from home, whereas the poorer amenable to home-based work versus Southern European countries and the new member states typically regional GDP per capita have relatively fewer jobs that can be done from home, as shown in % of jobs non-essential & non amenable to telework 50 map 3, panel a. GR62 GR42 ES70 ES53 The prevalence of jobs requiring little face-to-face interaction is RO42 not necessarily correlated with income. In fact, it is highest in Central 40 HU22 PT15 ES52 ITI3 CZ08 European countries such as Czech Republic, Hungary, and the Slovak BG34 SK02 CZ05 HU21 ES22 BG42 BG32 RO12 CZ07 ITH3 PT11 CZ04 ITF1 ITI1 ITH5 Republic, due to a higher share of manufacturing jobs, as seen in map GR43 CZ03 ES23ITH4 BG33 HR03 ITI2 ES41 ITC1 ES21ITC4 RO31 SK03RO11 ITF4 PL22 ES12 ES24 ES51 ITH1 GR64 ES11 FRC2 PL41 PL43 ES61 3, panel b. BG31 ITF3PT16SI03 ITF5 CZ06 ES42 ITF2 HU31SK04 PL52 HU12 PL61 IE04 PT30 ITC2 AT30 PL82 ITG2 BG41 PL51 ES13 HU33 CY00 DEG0 ITH2 HU32 ES62CZ02 ITC3 AT20 DE10 HU23 PL62 RO22 PL71 PL21 ITF6 HR04 EE00 IE05 GR54 GR51 PL92LT02 PT18 BE25 GR41PL72 GR52PL42 ES43 PL63 FRF3 DEE0 FRG0 DEC0 GR61 DE80 DED0 FRF1 DE90 BE22 DEB0 DK03 DE20 RO32 ITG1 LV00 DK05 30 SI04 DEA0 The next step is to combine the sectors deemed essential on the GR63 GR65 GR53ES64 ES63 FRD1 PT20 DK02 GR30FI1C ITI4 FI19 DE50 FRI3FRH0 FRF2 PT17 FRK2 DK04AT10 FRK1 UKC0 PL84 RO41 UKL0 UKF0 UKE0UKG0 DE40UKM0 DEF0 BE23 DE70 FRC1 SE21 BE21 HU11 BE10 governments’ lists with these two criteria, which are based on the PL81 UKN0 FI1D UKK0 FRD2 SE31 NL00 FRY1 FRM0 BE34 UKD0 FRI1 SE32LT01 ES30 SE23 IE06 RO21 FRJ1BE33 FRB0 FRE1 FRL0 DE60 BE32 FRI2 DE30 SK01CZ01 economic nature of different jobs. Combining these measures helps FRE2 BE35 FRJ2 UKJ0 FRY2 UKH0SE12 SE22 SE33 BE24 DK01 identify which jobs are most at risk because of the pandemic and FI20 FI1B FRY3 BE31PL91 UKI0 SE11FR10 related remedial measures such as social distancing, decline in travel, 20 FRY4 LU00 and mandatory lockdowns. Panel a in map 4 presents the share of jobs 9 9.5 10 10.5 11 11.5 that cannot be performed from home in non-essential industries in Log GDP per capita (purchasing power parity) each statistical (NUTS2) region in the European Union. On average, such jobs account for 30 percent of all employment in the EU. The ratio is higher in Southern and Eastern Europe. It is between one-third b. Share of jobs that are both non-essential and not to half of all jobs in large parts of Southern Europe (Greece, Italy, amenable to home-based work versus regional poverty Portugal and Spain) and Eastern Europe (Czech Republic, Hungary, Romania and Slovak Republic). In contrast, the share of vulnerable % of jobs non-essential & non amenable to telework 50 jobs is significantly lower in Scandinavia, France, Germany, and the GR62 GR42 United Kingdom. ES53 ES70 RO42 Next, the study combines the non-essential criteria with the 40 HU22 CZ08 ITI3 PT15 ES52 requirement for extensive face-to-face interactions. Jobs are RO12 HU21 BG34 BG42 BG32 SK02 CZ04 CZ05 CZ07 ITH3 ITF1 PT11 ES22 considered to be extensive face-to-face jobs if they require more CZ03 ITI1 ITH4 ITH5 ES23 GR43 BG33 ITC4 ITC1 ITI2 ES41ES21 HR03 RO11 RO31 SK03 ITH1 ES51 ES12 PL22 ITF4ES24 GR64 FRC2 PL41 ES11 PL43 face-to-face interactions than the average occupation. The results are BG31 CZ06 SI03 ITF3 ITF5PT16 ES42 ES61 HU12 SK04 AT30 IE04 HU31 ITF2 ITC2 PL52 PL61 BG41 ITG2 PT30PL82 PL51 ES13 CZ02 HU33 AT20 DEG0 CY00 DE10 ITH2 ES62 HU32 presented in panel b of map 4. While the share of the vulnerable jobs RO22 EE00 HU23 IE05 ITC3 PL21HR04 PL62 ITF6 PL71 LT02 BE25 GR51 GR54 DEE0 DEC0 PL42PT18 PL92 PL72 GR52 FRF3 GR41FRG0 PL63 ES43 RO32 GR61 DE20DED0 DK03 BE22 DE80 DE90 DEB0 FRF1 DK05 ITG1 30 LV00 DEA0 GR53 SI04 is lower in this case, the overall pattern is qualitatively similar. The RO41 UKC0 UKM0 GR30 UKF0 UKE0 DK02 GR65 AT10 DK04 UKL0 BE23DE40 DEF0 GR63 CH05 DE70 ITI4 FI1C DE50 FRI3 FRF2 FRK2 FRK1 FRD1 ES63 FI19 FRH0 PT17 PL84 PT20 ES64 HU11 UKG0 BE21 BE10 SE21 FRC1 FRD2 main exception is that Central and Eastern European countries are LT01 UKK0 UKN0 UKD0 RO21 CZ01 NO04IE06 CH06 BE34 CH02 BE33 CH03 NO03 BE32 SE32 DE60 FRB0 SE31 FRM0 FRI1 SE23 FRL0 CH07 FRJ1 FI1D NL00 ES30 FRY1 FRE1 FRI2 PL81 DE30 now less exposed because a larger share of their jobs are in the SK01 NO05 CH01 FRE2 UKH0 NO06 DK01 SE12 BE24 UKJ0 BE35 NO02 FRJ2 SE33 SE22 FRY2 NO07 manufacturing sector. These jobs are not easily amenable to FI1B FI20 CH04 PL91 FRY3 BE31 UKI0 NO01 SE11 home-based telework arrangements, but they do not require FR10 20 LU00 FRY4 extensive face-to-face interactions, either. In other words, factories 0 10 20 30 40 % of temporary jobs can weather social distancing requirements more easily (assuming they are considered essential) when compared to many services. This can be a saving grace for Eastern European countries in this crisis. c. Share of jobs that are both non-essential and not amenable to home-based work versus Regional Income Levels and Labor Market Vulnerability share of temporary jobs % of jobs non-essential & non amenable to telework 50 The preceding maps point to a disturbing pattern: European regions that are already economically disadvantaged are also likely to be GR42 GR62 tormented by the greatest labor market pain inflicted by COVID-19. RO42 The share of jobs that are susceptible to losses due to COVID-19 is PT15 HU22 strongly negatively correlated with regional GDP per capita. Panel a of ITI3 40 SK02 BG34 HU21 ITH3 RO12 BG32 BG42 PT11 figure 2 plots every NUTS2 region based on the share of jobs that are BG33 ITH4 ITC4 ITH5 ITC1 ITI1 HR03ITI2 RO11 ITF1 GR43 both non-essential and not amenable to home-based work versus ITH1 ITC2 SK03 FRC2 SK04 PL22 IE04 RO31 PT16ITF2 GR64 PL43 PL41 ITF5 HU12BG31 ITF3 PL61 ITF4 PL52 HU31 BG41 PL51 PT30 PL82 (log) GDP per capita. A 10 percent increase in regional GDP per capita ITG2 ITH2 DE10 DEG0 CY00 HU33 ITC3 RO22 HU32 EE00 IE05 GR54 PT18 LT02 HR04 BE25 GR51 DEE0PL21 PL71 PL62 ITF6HU23 DEC0 FRF3 GR41 DE90 PL92 PL42 GR52 FRG0 PL72 is associated with a 0.5 percentage point reduction in jobs at risk. 30 PT17 RO32 DE20 PL63DEB0 GR30DK02 DEA0 PT20 DK03 DK05 LV00 FRF1 BE22 DE50 ITI4 DK04 DED0 GR63 CH05 FRH0 GR65 DE80 FRI3 FRF2 GR61 GR53 FRD1 ITG1 UKC0 FRK2 FRK1 PL84 UKL0 BE23 DE40 UKM0UKF0 UKG0 DEF0 DE70 UKE0 RO41 HU11 BE10 BE21 NL00 FRC1 UKK0 UKN0 FRD2 CH02 PL81 IE06 UKD0 FRM0 CH06 This association between the share of vulnerable jobs and income SK01 LT01 UKH0 DE60 CH03 DE30 CH01 FRE2 FRB0 DK01 BE34 FRI1 FRL0 BE32BE33 FRE1CH07 FRI2 RO21 FRJ1 FRY1 levels also holds within countries. Panel b of figure 2 plots the share of PL91 UKJ0 BE24 CH04 FRJ2 BE35 FRY2 jobs at risk against regional poverty, proxied as the share of workers in BE31 FRY3 UKI0 FR10 20 FRY4 the bottom three deciles of the national earnings distribution. Clearly, LU00 10 20 30 40 50 poorer regions have more jobs at risk. This mean the COVID-19 crisis % of lowest three income deciles will likely exacerbate preexisting regional disparities. Source: For panel a, authors' calculations based on data from Eurostat, European Regions most susceptible to labor market pain are the ones in Union Labor Force Survey (2018a) and National account (2018b). For panel b and which jobs already tend to be more precarious and less protected. c, authors’ calculations based on data from Eurostat, European Union Labor Panel c of figure 2 documents a positive association between the Force Survey (2018a). Note: Data are for 2018. share of jobs at risk and the share of temporary workers, who can be 5 Which Jobs Are Most Vulnerable to COVID-19? What an Analysis of the European Union Reveals Figure 3. Vulnerability by Socioeconomic Status a. Vulnerability of jobs by income decile b. Vulnerability of jobs by age group 80 80 Share of employment (%) Share of employment (%) 60 60 40 40 20 0 20 1 2 3 4 5 6 7 8 9 10 15-19 20-24 25-29 30-39 40-49 50-64 Income decile (1 lowest - 10 highest) Age group (years) No telework (NT) Non-essential (NE) No telework (NT) Non-essential (NE) Face-to-face (FtF) NT and NE Face-to-face (FtF) NT and NE FtF and NE FtF and NE c. Vulnerability of jobs by education level d. Vulnerability of jobs by place of birth 80 80 Share of employment (%) Share of employment (%) 60 60 40 40 20 20 0 < Upper secondary Upper secondary Tertiary EU native-born EU28 migrants Non-EU28 migrants Education level No telework (NT) Non-essential (NE) No telework (NT) Non-essential (NE) Face-to-face (FtF) NT and NE Face-to-face (FtF) NT and NE FtF and NE FtF and NE Source: Authors’ calculations based on data from Eurostat, European Union Labor Force Survey (2018a). Note: Data are for 2018. more easily fired than workers with permanent contracts. levels, as is shown in panel b of figure 3. Unlike the health risks of Vulnerability to COVID-19 has induced further vulnerability in COVID-19, which are concentrated among the elderly and increase employment. steeply with age, the economic risks are concentrated among the young, and decline with age, as shown in panel c of figure 3. Migrants, Income Distribution and Labor Market Vulnerability especially those from non-EU countries, are also more likely to be employed in risky occupations that are most exposed to This section investigates which workers are most at risk. The EU LFS COVID-19–induced job losses, as shown in panel d of figure 3. reports the income decile of each wage earner in his/her respective country, so poverty comparisons can be performed. Appendix table Conclusion and Policy Recommendations A2 reports labor market vulnerability by European country. In high income European countries, COVID-19–induced labor market Figure 3, panel a shows that workers with the lowest pay suffer the pain is disproportionately borne by young and poorly educated highest vulnerability. The share of workers who cannot work from workers. These workers are already employed in low-paying jobs, live home, are working in non-essential sectors, and/or are working in jobs in regions that are already lagging and are subject to a greater requiring extensive face-to-face interaction all sharply decline with prevalence of temporary employment contracts. The COVID-19 crisis income. Workers in the bottom decile are more than twice as likely to is bound to exacerbate inequality, both within and across countries, be at risk than those in the top income bracket because 42 percent of unless dramatic remedial action is undertaken immediately. all workers in the bottom earning decile are employed in jobs in non-essential industries that cannot be performed at home, whereas While the insights and patterns presented in this analysis are such jobs account for only 16 percent of employment among workers based on the data from the European Union (EU) countries (plus in the top income decile. Perhaps not surprisingly, the probability of Switzerland and Norway), the patterns are likely to be similar in other being employed in a job that cannot be done from home and is high-income countries such as the United States, Canada, Australia, 6 non-essential is significantly larger for workers with low education Japan or Singapore. In contrast, labor market shocks are likely to be Research & Policy Brief No.34 different, possibly more severe in lower-income countries. There are Appendix. Supporting Data various reasons for this conjecture. We observe lower penetration of high-speed internet services and other technologies needed to Table A1. Correlation between Classifications of Essential Sectors perform work at home. A larger share of the workers is employed in Jurisdiction Consolidated Oklahoma Minnesota Delaware lower-paying, informal (face-to-face) jobs. Even manufacturing jobs, which are considered not to require face-to-face interaction in Europe, Oklahoma, USA 0.2060* might be performed in more crowded factories. As a result, mobility restrictions like the ones imposed by many non-OECD countries, will Minnesota, USA 0.6457* 0.1423* have more severe impacts on low skilled and poorer workers. Delaware, USA 0.1389* 0.4053* 0.2594* These factors need to be considered when policy responses are Italy 0.8813* 0.1870* 0.4735 0.1288* designed in both high-income and developing countries. Fiscal Source: Author's calculations based on del Rio-Chanona et al. (2020) for Italy; stimulus packages can specifically target regions, sectors or and Delaware Division of Public Health; Minnesota Executive Department; and occupations that specifically suffer from these labor market Oklahoma Department of Commerce U.S. Bureau of Labor Statistics; del Rio-Chanona (2020); U.S. Cybersecurity and Infrastructure Security Agency for disturbances. Similarly, targeted cash transfer programs and social the United States. safety nets (in response to COVID-19 induced shocks) need to be Note: Only “strict” measures of essential sectors are used for this table (all the designed with job vulnerability considerations in mind. Finally, in the jurisdictions deem these sectors to be essential). Correlations are calculated for the binary indicators. long run, job vulnerability needs to be an explicit consideration of * p < 0.05. education and labor market policies. Table A2. Labor Market Vulnerability by European Country Percent of total employment Not home-based Face-to-face Not home-based Non-essential Face-to-face Country and and jobs jobs jobs non-essential jobs non-essential jobs Austria 66 44 43 32 19 Belgium 63 39 44 27 17 Bulgaria 72 47 43 37 21 Croatia 69 44 43 33 20 Cyprus 66 44 47 33 23 Czech Rep. 70 47 38 35 17 Denmark 62 40 46 28 18 Estonia 64 46 40 32 18 Finland 62 40 43 27 17 France 63 38 43 26 16 Germany 63 43 42 30 18 Greece 69 42 47 32 22 Hungary 71 44 40 34 17 Ireland 64 43 46 30 21 Italy 69 47 44 35 21 Latvia 67 42 44 30 19 Lithuania 66 43 42 30 18 Luxemburg 51 29 44 19 14 Netherlands 60 40 46 28 19 Norway 62 37 46 24 16 Poland 68 43 40 32 18 Portugal 69 45 44 34 20 Romania 77 42 39 33 17 Slovak Rep. 74 44 40 35 17 Slovenia 65 45 41 33 18 Spain 70 47 47 35 23 Sweden 60 37 44 24 16 Switzerland 60 40 43 26 17 United Kingdom 59 40 46 26 18 Source: Authors’ calculations based on data from Eurostat, European Union Labor Force Survey (2018a). Note: Data are for 2018. References Avdiu, B., and G. Nayyar. 2020. “When Face to Face Interactions Become an Occupational Hazard: Eurostat. 2018a. “European Union Labor Force Survey” https://ec.europa.eu/eurostat/web/lfs/overview Jobs in the Time of Covid-19” Future Development, March 30. ---------. 2018b.“National accounts.” https://ec.europa.eu/eurostat/web/national-accounts Blinder, A.. 2006. “Offshoring: The Next Industrial Revolution?” Foreign Affairs 85: 85 (2, March- Le Grand Continent. 2020. "Observatoire Coronavirus" . April 24. April), pp. 113-28. https://legrandcontinent.eu/fr/observatoire-coronavirus/ Delaware Division of Public Health. 2020. “List of Essential and Non-essential Business” Minnesota Executive Department. 2020. “Executive Order 20-33 Critical Sector Descriptions.” https://coronavirus.delaware.gov/resources-for-businesses/ https://bit.ly/2KuwLD1. Oklahoma Department of Commerce. 2020. “Oklahoma Essential Industries List.” del Rio-Chanona, R. M., P. Mealy, A. Pichler, F. Lafond, and D. F. Farmer. 2020. “Supply and Demand https://www.okcommerce.gov/wp-content/uploads/Oklahoma-Essential-Industries-List.pdf Shocks in the COVID-19 Pandemic: An Industry and Occupation Perspective.” COVID Economics United States, Bureau of Labor Statistics. 2020. “May 2019. Occupational Employment Statistics Vetted and Real-Time Papers, Issue 6, Centre for Economic Policy Research (CEPR) Press. (OES).” https://bit.ly/3aBChyD. Dingel, J., and B. Neiman. 2020. “Which Jobs Can Be Performed from Home?” Covid Economics United States, Cybersecurity and Infrastructure Security Agency, 2020. “Guidance on the Jobs in Vetted and Real-Time Papers, Issue 13, Centre for Economic Policy Research (CEPR) Press. the Time of Covid-19” https://bit.ly/2VXxvHj 7