## 1. Tele-workability by Occupation ## Source details **Canonical URL:** [1. Tele-workability by Occupation](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020088-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020088-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020088-print-pdf.pdf.json) --- ### I. Introduction - Social distancing policies for COVID-19 affect a large share of workers worldwide, with millions unemployed and many jobs at risk. - Research goals: - Identify which types of jobs are most at risk given requirements for physical presence or high personal proximity. - Examine how “tele-workability” depends on worker characteristics: age, educational attainment, gender, employment status, and earnings. - Compare feasibility to work remotely across advanced and emerging economies to inform social protection and labor market policies. - Approach overview: - Construct a new index of “tele-workability” for 35 advanced and emerging market economies using a task-based approach. - Data sources: occupation-level tele-workability classification from Dingel and Neiman (2020) for the US; individual-level data from OECD’s PIAAC. - Estimation uses an Expectation Maximization (EM) algorithm to map 6-digit SOC occupation-level measures to individual-level PIAAC observations, producing worker-level tele-workability scores that range between 0 and 1. ### II. Data and Methodology - Data sources and features: - Dingel and Neiman (2020): O*NET-based occupational feasibility of working from home, indexed 0 or 1 at 6-digit SOC codes; based on “work context” and “generalized work activities.” - OECD PIAAC: nationally representative individual-level data for 35 countries; includes demographics, occupation (2-digit ISCO), sector, and detailed task frequency items (five-response frequency scale). - Crosswalk and mapping: - PIAAC 2-digit ISCO occupations map to multiple 6-digit SOC measures, allowing individual workers to be associated with multiple tele-workability probabilities. - Estimation approach: - Iterative EM algorithm using a weighted Generalized Linear Model (GLM) with equal initial weights across duplicates. - Individual-level regression specification: - t_ij = sum_{n=1}^N β_n X_in + ε_ij - i denotes individuals; j denotes duplicates when multiple probabilities map to one individual; t_ij is tele-workability score; X_in contains N individual, job, and task characteristics; β_n are parameters. Tele-workability probability lies between 0 and 1. - Regressors include gender, education, income deciles, immigration status, age; workplace skills (use of computer, Internet, email, spreadsheets, communication software, presentation skills); work characteristics (physical/manual work, autonomy over schedule and task performance, setting schedules of others, training others). - Weight update formula (Ibrahim, 1990): - w_ij = f(t̂ − t_ij | x_in, β_n) / sum_{n=1}^N f(t̂ − t_ij | x_in, β_n) - f(.) is standard normal density; t̂ is predicted value. - Once parameters converge, β_n applied to all PIAAC countries to estimate individual tele-workability across the full sample. ### III. Main Findings #### Tele-workability Index and Occupations - Occupation-level patterns: - Lowest tele-workability: elementary occupations (e.g., janitors, construction laborers, street vendors), then plant and machinery operators, craft and related trades workers (e.g., mechanics, garments workers). - Highest tele-workability: professionals, managers, officials and legislators. - Cross-country variation: - Turkey exhibits lower tele-workability scores across occupations. - Countries with significantly lower average tele-workability: Turkey, Chile, Mexico, Ecuador, and Peru. - Within advanced economies, Greece and Italy among the lowest; Nordic countries and Singapore among the highest. - Sector-level distribution: - Lowest tele-workability concentrated in: accommodation and food services; transportation; wholesale and retail trade; health and social services; manufacturing. - Highest tele-workability concentrated in: information and communication; finance and insurance; professional services. - Negative association between level of economic development and feasibility to work remotely within sectors. - Countries with higher index scores even in less telework-able sectors: Finland, Singapore, Lithuania. #### Who Holds These Jobs? - Gender: - Men, on average, are less likely than women to be engaged in tasks that can be performed at home (Japan an exception with no statistically significant difference). - Women’s employment often concentrated in the public sector, care and education; men more in construction, transportation, manufacturing. - Caveat: access to digital tools matters; women may be at higher risk where a digital gender divide exists. - Age and education: - Older workers (aged 60 and above) are, on average, slightly more likely to hold higher tele-workability jobs compared to younger workers (under 30), though results vary by country. - Workers without a college degree are significantly less likely to work from home across most countries. - Having a college degree greatly improves likelihood of working remotely across all age groups; older workers with lower education still have higher tele-workability reflecting lifecycle occupational progression. - Job characteristics: - Part-time and temporary workers are less likely to work remotely; part-time workers in Singapore and Korea particularly lower tele-workability. - Workers in SMEs (less than 250 workers) are less likely to be in telework-amenable jobs compared to larger enterprises. - Immigration status: - Foreign-born individuals, on average, are more likely to be in occupations less amenable to teleworking, more so in Europe than the US. Exceptions: Peru and Mexico where foreign-born have higher tele-workability on average. - Earnings distribution: - Strong correlation between hourly earnings and tele-workability: workers in the bottom two deciles are significantly less likely to work remotely than those in the top two deciles. - Workers in bottom earnings quintiles concentrated in low-teleworkability occupations and sectors; these workers often have limited financial buffers. #### Distributional Metrics and Inequality - Cross-country disparity metric: - Compute ratio = (Mean tele-workability in top hourly earnings decile) / (Mean tele-workability in bottom hourly earnings decile). - Countries with wider disparities include: Hungary, Slovenia, Netherlands, and the US. - Example contrast: average tele-workability lower in Turkey than Singapore, but earnings disparity between top and bottom deciles is significantly higher in Singapore. #### Estimates of Employment at Risk - Threshold calibration: - Use US BLS sector-level employment changes between February 2020 and April 2020 (adjusted for 2017–2019 average) to set sector-specific tele-workability thresholds associated with highest risk of job loss. - Apply thresholds to full country sample and scale with ILO 2019 employment by country and sector. - Adjust for country-specific stringency of lockdown measures using Oxford University’s Coronavirus Government Response Tracker for April 2020, normalized to US value. - Key quantitative estimates: - Over 97.3 million workers are at high risk of layoffs and furlough across the 35-country sample. - This corresponds to more than 15 percent of the workforce in the sample. - US accounts for over 21 million of these jobs, or about 13 percent of employment (aligned with April 2020 BLS Employment Situation report). - Worst-hit sectors across 35 countries: - Restaurants and accommodation: close to 17.6 million workers at risk. - Wholesale and retail: over 13.9 million workers at risk. ### IV. Conclusions and Policy Implications - Vulnerability profile: - Workers most vulnerable to stringent social distancing policies: young, less educated, part-time or temporary employment, earnings toward the bottom of the distribution, employed in SMEs, foreign-born in many countries. - Hardest-hit occupations and sectors coalesce these vulnerable characteristics (e.g., accommodation and food services, retail). - Cross-country and informal sector concerns: - In emerging market economies (e.g., Turkey, Peru, Mexico) fewer than half the households have access to a computer at home; informal working arrangements and high shares of self-employed and casual work limit remote work feasibility and access to formal insurance and health care. - Informal workers face higher risk of falling into poverty and greater obstacles to recovery; social protection responses must reach these workers and their families quickly and effectively. - Long-term risks and policy needs: - The impact on employment and sectoral job distributions depends on severity and duration of containment measures and economic contractions. - Past crises suggest long-lasting negative effects on future earnings and job security, particularly for low-income and precariously employed workers. - Changed consumer preferences (e.g., greater e-commerce reliance) may permanently reduce demand for some services (retail, tourism, dining out, personal services), implying persistent labor market shifts. - Policy responses should account for demographic and distributional considerations both during the crisis and in its aftermath to cushion short-term shocks and support reallocation. ### Annex 1. Occupational Classifications (ISCO taxonomy) - Professional - Science and engineering professionals - Health professionals - Teaching professionals - Business and administration professionals - Information and communications technology professionals - Legal, social and cultural professionals - Technicians & assoc. prof. - Science and engineering associate professionals - Health associate professionals - Business and administration associate professionals - Legal, social, cultural and related associate professionals - Information and communications technicians - Legislators, senior officials, & managers - Chief executives, senior officials and legislators - Administrative and commercial managers - Production and specialized services managers - Hospitality, retail and other services managers - Clerks - General and keyboard clerks - Customer services clerks - Numerical and material recording clerks - Other clerical support workers - Crafts & trade - Building and related trades workers, excluding electricians - Metal, machinery and related trades workers - Handicraft and printing workers - Electrical and electronic trades workers - Food processing, wood working, garment and other craft and related trades workers - Service, shop, & market - Personal service workers - Sales workers - Personal care workers - Protective services workers - Plant/machine operators - Stationary plant and machine operators - Assemblers - Drivers and mobile plant operators - Elementary - Cleaners and helpers - Agricultural, forestry and fishery laborers - Laborers in mining, construction, manufacturing and transport - Food preparation assistants - Street and related sales and service workers - Refuse workers and other elementary workers ### Annex 2. Country estimates of tele-workability differences by worker characteristics (highlights) - Full sample (All Countries) regression coefficients: - Male: -0.069*** - No College Degree: -0.115*** - Below 30 years of age: -0.038*** - Above 60 years of age: 0.009*** - Bottom two income deciles: -0.116*** - Top two income deciles: 0.181*** - Born abroad: -0.067*** - Firms of less than 250 workers: -0.078*** - Part-time: -0.057*** - Selected country notes (coefficients reproduced exactly as reported): - Japan: Male: 0.084***; No College Degree: 0.038***; Above 60 years of age: -0.059***; Born abroad: 0.088***; Part-time: -0.032*** - Mexico: Male: -0.027***; No College Degree: -0.079***; Below 30 years of age: 0.033***; Above 60 years of age: -0.049***; Bottom two income deciles: 0.060***; Born abroad: 0.203***; Firms <250: 0.043***; Part-time: 0.033*** - Turkey: Male: -0.017***; No College Degree: -0.035***; Below 30 years of age: 0.038***; Bottom two income deciles: 0.055***; Top two income deciles: -0.046***; Firms <250: 0.051*** - United States: Male: -0.025***; Below 30 years of age: -0.048***; Above 60 years of age: 0.071***; Bottom two income deciles: -0.044***; Top two income deciles: 0.025***; Firms <250: -0.014***; Part-time: -0.029*** - Note: Full sample of 35 countries contains 151,177 observations. Blank cells in the original table indicate lack of variable availability for some countries. ### Annex 3. Gender differences across sectors and occupations (distributional notes) - Occupations ordered from lowest tele-workability to highest: Elementary, Plant/Machinery, Crafts/trade, Service/Shop/Market, Technicians, Clerks, Legislators/Managers, Professionals. - Sectors ordered from lowest tele-workability to highest: Accommodation/Food, Construction, Transportation, Admin./Support, Wholesale/Retail, Manufacturing, Health/Social Serv., Public Admin./Defence, Education, Professional, Finance/Insurance, Information/Comm. - Source: PIAAC survey and staff calculations. ### Annex 4. Employment at risk estimates by sector (35-country sample) - Employment at risk (in thousands) | Share of employment at risk - Accommodation/Food Services: 17,666 | 0.47 - Wholesale/Retail Trade: 13,902 | 0.15 - Manufacturing: 12,258 | 0.14 - Other Services: 11,209 | 0.31 - Health/Social Services: 9,784 | 0.14 - Transportation: 7,642 | 0.15 - Education: 7,340 | 0.15 - Construction: 6,693 | 0.15 - Administrative/Support Activities: 5,587 | 0.09 - Public Administration/Defense: 2,674 | 0.08 - Finance/Insurance: 1,414 | 0.07 - Agriculture: 936 | 0.03 - Utilities: 118 | 0.02 - Mining: 108 | 0.04 - Note: Estimates are based on the sample of 35 countries available in the PIAAC survey. *Source: wpiea2020088-print-pdf (PIAAC survey; Dingel and Neiman (2020); BLS (2020); ILO (2020); Oxford University Coronavirus Government Response Tracker (April 2020); staff calculations).* ### 1. Tele-workability by Occupation ________________________________________________________ 8 ### 1. Tele-workability by Occupation ### Scope - Contains 8 main items enumerated 1 through 8. - Contains 4 annexes: Annex 1 through Annex 4. ### Purpose (inferred from structure) - Focuses on tele-workability indicators broken down by occupation, sector, GDP per capita (PPP), worker characteristics, age group and education level, hourly earnings decile, and earnings-decile distributions. - Includes cross-country comparisons and employment-at-risk estimates. - Provides occupational classifications and country-level estimates, and analyzes gender differences across sectors and occupations. ### Document structure (high level) - Main items: 1 through 8 covering tele-workability metrics and distributions. - Annexes: 1. Occupational Classifications; 2. Country Estimates of Tele-workability Differences by Worker Characteristics; 3. Gender Differences Across Sectors and Occupations; 4. Employment at Risk Estimates. *Source: wpiea2020088-print-pdf - 1. Tele-workability by Occupation.* ### References __________________________________________________________________________________ 23 ### wpiea2020088-print-pdf - References __________________________________________________________________________________ 23 ### I. Introduction - Social distancing policies for COVID-19 affect a large share of workers worldwide, with millions unemployed and many jobs at risk. - Key research goals: - Identify which types of jobs are most at risk given requirements for physical presence or high personal proximity. - Examine how “tele-workability” depends on worker characteristics: age, educational attainment, gender, employment status, and earnings. - Compare feasibility to work remotely across advanced and emerging economies to inform social protection and labor market policies. - Methodological overview: - Construct a new index of “tele-workability” for 35 advanced and emerging market economies using a task-based approach. - Data sources: occupation-level tele-workability classification from Dingel and Neiman (2020) for the US; individual-level data from OECD’s PIAAC. - Estimation uses an Expectation Maximization (EM) algorithm to map 6-digit SOC occupation-level measures to individual-level PIAAC observations, producing worker-level tele-workability scores that range between 0 and 1. ### II. Data and Methodology - Data sources and features: - Dingel and Neiman (2020): O*NET-based occupational feasibility of working from home, indexed 0 or 1 at 6-digit SOC codes; based on “work context” and “generalized work activities.” - OECD PIAAC: nationally representative individual-level data for 35 countries; includes demographics, occupation (2-digit ISCO), sector, and detailed task frequency items (five-response frequency scale). - Crosswalk and mapping: - PIAAC occupations (2-digit ISCO) map to multiple 6-digit SOC measures, allowing individual workers to be associated with multiple tele-workability probabilities. - Estimation approach: - Iterative EM algorithm using a weighted Generalized Linear Model (GLM) with equal initial weights across duplicates. - Individual-level regression specification: - t_ij = sum_{n=1}^N β_n X_in + ε_ij - i denotes individuals; j denotes duplicates when multiple probabilities map to one individual; t_ij is tele-workability score; X_in contains N individual, job, and task characteristics; β_n are parameters. Tele-workability probability lies between 0 and 1. - Regressors include gender, education, income deciles, immigration status, age; workplace skills (use of computer, Internet, email, spreadsheets, communication software, presentation skills); work characteristics (physical/manual work, autonomy over schedule and task performance, setting schedules of others, training others). - Weight update formula (Ibrahim, 1990): - w_ij = f(t̂ − t_ij | x_in, β_n) / sum_{n=1}^N f(t̂ − t_ij | x_in, β_n) - f(.) is standard normal density; t̂ is predicted value. - Once parameters converge, β_n applied to all PIAAC countries to estimate individual tele-workability across the full sample. ### III. Main Findings H3: Tele-workability Index and Occupations - Occupation-level patterns: - Lowest tele-workability: elementary occupations (e.g., janitors, construction laborers, street vendors), then plant and machinery operators, craft and related trades workers (e.g., mechanics, garments workers). - Highest tele-workability: professionals, managers, officials and legislators. - Cross-country variation: - Turkey exhibits lower tele-workability scores across occupations. - Countries with significantly lower average tele-workability: Turkey, Chile, Mexico, Ecuador, and Peru. - Within advanced economies, Greece and Italy among the lowest; Nordic countries and Singapore among the highest. - Sector-level distribution: - Lowest tele-workability concentrated in: accommodation and food services; transportation; wholesale and retail trade; health and social services; manufacturing. - Highest tele-workability concentrated in: information and communication; finance and insurance; professional services. - Negative association between level of economic development and feasibility to work remotely within sectors. - Countries with higher index scores even in less telework-able sectors: Finland, Singapore, Lithuania. H3: Who Holds These Jobs? - Gender: - Men, on average, are less likely than women to be engaged in tasks that can be performed at home (Japan an exception with no statistically significant difference). - Women’s employment often concentrated in the public sector, care and education; men more in construction, transportation, manufacturing. - Caveat: access to digital tools matters; women may be at higher risk where a digital gender divide exists. - Age and education: - Older workers (aged 60 and above) are, on average, slightly more likely to hold higher tele-workability jobs compared to younger workers (under 30), though results vary by country. - Workers without a college degree are significantly less likely to work from home across most countries. - Having a college degree greatly improves likelihood of working remotely across all age groups; older workers with lower education still have higher tele-workability reflecting lifecycle occupational progression. - Job characteristics: - Part-time and temporary workers are less likely to work remotely; part-time workers in Singapore and Korea particularly lower tele-workability. - Workers in SMEs (less than 250 workers) are less likely to be in telework-amenable jobs compared to larger enterprises. - Immigration status: - Foreign-born individuals, on average, are more likely to be in occupations less amenable to teleworking, more so in Europe than the US. Exceptions: Peru and Mexico where foreign-born have higher tele-workability on average. - Earnings distribution: - Strong correlation between hourly earnings and tele-workability: workers in the bottom two deciles are significantly less likely to work remotely than those in the top two deciles. - Workers in bottom earnings quintiles concentrated in low-teleworkability occupations and sectors; these workers often have limited financial buffers. H3: Distributional Metrics and Inequality - Cross-country disparity metric: - Compute ratio = (Mean tele-workability in top hourly earnings decile) / (Mean tele-workability in bottom hourly earnings decile). - Countries with wider disparities include: Hungary, Slovenia, Netherlands, and the US. - Example contrast: average tele-workability lower in Turkey than Singapore, but earnings disparity between top and bottom deciles is significantly higher in Singapore. H3: Estimates of Employment at Risk - Threshold calibration: - Use US BLS sector-level employment changes between February 2020 and April 2020 (adjusted for 2017–2019 average) to set sector-specific tele-workability thresholds associated with highest risk of job loss. - Apply thresholds to full country sample and scale with ILO 2019 employment by country and sector. - Adjust for country-specific stringency of lockdown measures using Oxford University’s Coronavirus Government Response Tracker for April 2020, normalized to US value. - Key quantitative estimates: - Over 97.3 million workers are at high risk of layoffs and furlough across the 35-country sample. - This corresponds to more than 15 percent of the workforce in the sample. - US accounts for over 21 million of these jobs, or about 13 percent of employment (aligned with April 2020 BLS Employment Situation report). - Worst-hit sectors across 35 countries: - Restaurants and accommodation: close to 17.6 million workers at risk. - Wholesale and retail: over 13.9 million workers at risk. ### IV. Conclusions and Policy Implications - Summary of vulnerability profile: - Workers most vulnerable to stringent social distancing policies: young, less educated, part-time or temporary employment, earnings toward the bottom of the distribution, employed in SMEs, foreign-born in many countries. - Hardest-hit occupations and sectors coalesce these vulnerable characteristics (e.g., accommodation and food services, retail). - Cross-country and informal sector concerns: - In emerging market economies (e.g., Turkey, Peru, Mexico) fewer than half the households have access to a computer at home; informal working arrangements and high shares of self-employed and casual work limit remote work feasibility and access to formal insurance and health care. - Informal workers face higher risk of falling into poverty and greater obstacles to recovery; social protection responses must reach these workers and their families quickly and effectively. - Long-term risks and policy needs: - The impact on employment and sectoral job distributions depends on severity and duration of containment measures and economic contractions. - Past crises suggest long-lasting negative effects on future earnings and job security, particularly for low-income and precariously employed workers. - Changed consumer preferences (e.g., greater e-commerce reliance) may permanently reduce demand for some services (retail, tourism, dining out, personal services), implying persistent labor market shifts. - Policy responses should account for demographic and distributional considerations both during the crisis and in its aftermath to cushion short-term shocks and support reallocation. *Source: PIAAC survey; Dingel and Neiman (2020); BLS (2020); ILO (2020); Oxford University Coronavirus Government Response Tracker (April 2020); staff calculations (as presented in wpiea2020088-print-pdf).* ### Annex 1. Occupational Classifications ### Annex 1. Occupational Classifications ### Occupational taxonomy (ISCO) - Professional - Science and engineering professionals - Health professionals - Teaching professionals - Business and administration professionals - Information and communications technology professionals - Legal, social and cultural professionals - Technicians & assoc. prof. - Science and engineering associate professionals - Health associate professionals - Business and administration associate professionals - Legal, social, cultural and related associate professionals - Information and communications technicians - Legislators, senior officials, & managers - Chief executives, senior officials and legislators - Administrative and commercial managers - Production and specialized services managers - Hospitality, retail and other services managers - Clerks - General and keyboard clerks - Customer services clerks - Numerical and material recording clerks - Other clerical support workers - Crafts & trade - Building and related trades workers, excluding electricians - Metal, machinery and related trades workers - Handicraft and printing workers - Electrical and electronic trades workers - Food processing, wood working, garment and other craft and related trades workers - Service, shop, & market - Personal service workers - Sales workers - Personal care workers - Protective services workers - Plant/machine operators - Stationary plant and machine operators - Assemblers - Drivers and mobile plant operators - Elementary - Cleaners and helpers - Agricultural, forestry and fishery laborers - Laborers in mining, construction, manufacturing and transport - Food preparation assistants - Street and related sales and service workers - Refuse workers and other elementary workers ### Annex 2: Country estimates of tele-workability differences by worker characteristics Note: Table contains estimates of country differences from regressions where tele-workability index is regressed on one of the worker characteristics, with an interaction term for the country capturing country differences from the full sample mean. Each worker characteristic variable takes on a value of 1, indicating whether an individual belongs to a given group, or 0. Blank cells are not estimated for a given country due to lack of variable availability in the data. Full sample of 35 countries contains 151,177 observations. - Header (worker characteristics): - Male | No College Degree | Below 30 years of age | Above 60 years of age | Bottom two income deciles | Top two income deciles | Born abroad | Firms of less than 250 workers | Part-time - All Countries: -0.069*** -0.115*** -0.038*** 0.009*** -0.116*** 0.181*** -0.067*** -0.078*** -0.057*** - Austria: 0.022* 0.060*** -0.009 0.023 -0.003 0.012 -0.005 0.032** 0.041*** - Belgium: 0.005 0.094*** 0.018 0.060 -0.002 -0.012 -0.025 0.031* 0.035** - Canada: -0.018*** 0.006 -0.043*** -0.004 -0.026*** -0.002 0.066*** -0.010 - Chile: -0.011 0.050*** 0.033*** -0.043** 0.020* -0.010 0.017 -0.014 0.032** - Cyprus: -0.043 0.035 0.043 -0.068 0.038 -0.049 0.055 0.047 0.019 - Czech Republic: -0.002 0.026* 0.034** -0.029 0.021 -0.012 0.017 0.054*** 0.030 - Germany: 0.025*** 0.063*** -0.017*** -0.010 0.002 0.036*** -0.049*** -0.015*** 0.024*** - Denmark: 0.003 0.035** -0.065*** 0.040 -0.028* 0.012 0.013 -0.029* 0.009 - Ecuador: -0.054*** -0.070*** 0.024*** -0.049 0.025* -0.038*** -0.010 -0.022 - Spain: 0.000 -0.092*** 0.011 0.008 0.006 0.022*** -0.036*** 0.015* 0.005 - Estonia: -0.036 0.024 0.026 -0.008 0.006 -0.057* -0.002 0.031 0.048 - Finland: -0.005 0.058*** -0.045*** 0.018 -0.010 -0.012 -0.008 0.001 0.003 - France: -0.017*** 0.027*** 0.029*** 0.010 0.018*** 0.021*** -0.019*** 0.021*** 0.022*** - United Kingdom: 0.001 -0.001 -0.006 -0.008 -0.038*** 0.045*** 0.046*** -0.004 -0.003 - Greece: -0.008 0.013 0.013 0.026 0.050*** -0.027 -0.018 0.010 -0.010 - Hungary: -0.038*** -0.023** 0.034** -0.035* -0.013 0.046*** 0.080** 0.001 -0.002 - Ireland: -0.006 -0.010 -0.010 -0.053 -0.011 0.000 0.025 -0.027 -0.035* - Israel: -0.028** -0.011 -0.049*** 0.012 -0.024 0.009 0.023 -0.030* -0.013 - Italy: -0.028*** -0.089*** 0.015** 0.015 0.016** -0.014** -0.058*** 0.011 0.037*** - Japan: 0.084*** 0.038*** -0.004 -0.059*** 0.037*** -0.036*** 0.088*** -0.028*** -0.032*** - Kazakhstan: -0.057*** 0.023** 0.079*** -0.059** 0.053*** -0.071*** -0.003 0.049*** 0.024 - Korea: 0.050*** 0.030*** 0.088*** -0.118*** -0.004 -0.061*** -0.078*** -0.011 -0.050*** - Lithuania: -0.051** 0.024 0.048** -0.010 0.024 0.007 0.015 0.038 0.050 - Mexico: -0.027*** -0.079*** 0.033*** -0.049*** 0.060*** -0.031*** 0.203*** 0.043*** 0.033*** - Netherlands: 0.022*** -0.057*** -0.045*** 0.000 -0.058*** 0.003 0.012 -0.003 0.049*** - Norway: 0.024* -0.016 -0.071*** 0.024 -0.032* -0.012 0.008 -0.006 -0.023 - New Zealand: -0.016 -0.023 -0.037** 0.012 -0.031 0.028 0.083*** -0.041* 0.006 - Peru: 0.007 0.040*** 0.057*** 0.001 0.160*** -0.002 0.033*** - Poland: -0.053*** 0.011 0.038*** -0.017 0.003 0.004 0.145** 0.062*** 0.043*** - Singapore: 0.004 -0.033** 0.049*** -0.084*** -0.065*** -0.011 0.096*** 0.003 -0.074*** - Slovak Republic: -0.027* -0.065*** 0.052*** -0.003 -0.004 -0.018 0.065 0.070*** 0.011 - Slovenia: -0.030 0.074*** 0.007 0.060 -0.034 0.037 -0.071* 0.102*** 0.032 - Sweden: 0.001 0.044*** -0.045*** 0.017 0.003 -0.015 0.010 0.005 -0.008 - Turkey: -0.017*** -0.035*** 0.038*** -0.001 0.055*** -0.046*** 0.000 0.051*** 0.016 - United States: -0.025*** 0.002 -0.048*** 0.071*** -0.044*** 0.025*** -0.004 -0.014*** -0.029*** ### Annex 3: Gender differences across sectors and occupations (distributional notes) - Source: PIAAC survey and staff calculations. - Occupations are ordered by level of tele-workability from lowest to highest; darker colors indicate higher concentration of workers in a given occupation. - Occupations listed (from lowest tele-workability to highest): Elementary, Plant/Machinery, Crafts/trade, Service/Shop/Market, Technicians, Clerks, Legislators/Managers, Professionals. - Sectors are ordered by level of tele-workability from lowest to highest; darker colors indicate higher concentration of workers in a given sector. - Sectors listed (from lowest tele-workability to highest): Accommodation/Food, Construction, Transportation, Admin./Support, Wholesale/Retail, Manufacturing, Health/Social Serv., Public Admin./Defence, Education, Professional, Finance/Insurance, Information/Comm. ### Annex 4: Employment at risk estimates by sector - Employment at risk (in thousands) | Share of employment at risk - Accommodation/Food Services: 17,666 | 0.47 - Wholesale/Retail Trade: 13,902 | 0.15 - Manufacturing: 12,258 | 0.14 - Other Services: 11,209 | 0.31 - Health/Social Services: 9,784 | 0.14 - Transportation: 7,642 | 0.15 - Education: 7,340 | 0.15 - Construction: 6,693 | 0.15 - Administrative/Support Activities: 5,587 | 0.09 - Public Administration/Defense: 2,674 | 0.08 - Finance/Insurance: 1,414 | 0.07 - Agriculture: 936 | 0.03 - Utilities: 118 | 0.02 - Mining: 108 | 0.04 - Source and data notes: - Source: PIAAC Survey; BLS; ILO; and staff calculations. - Note: Estimates are based on the sample of 35 countries available in the PIAAC survey. *Annex 1. Occupational Classifications — wpiea2020088-print-pdf* --- _Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020088-print-pdf.pdf_