## labor

## Source details

**Canonical URL:** [labor](https://www.imf.org/-/media/files/publications/reo/whd/2020/oct/english/labor.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/reo/whd/2020/oct/english/labor.pdf.md)
- [Structured JSON version](/-/media/files/publications/reo/whd/2020/oct/english/labor.pdf.json)

---

### Introduction and overview of the shock
- The COVID-19 pandemic triggered major economic disruptions and severely affected labor markets in Latin America and the Caribbean (LAC).
- The October 2020 Regional Economic Outlook (REO): Western Hemisphere notes the COVID-19 recession has a much larger decline in consumption in labor-intensive service sectors compared with previous recessions.
- This chapter assesses:
  - the impact on LAC labor markets using labor surveys;
  - how pre-existing labor market features increased vulnerability; and
  - the outlook for employment, income and activity during the recovery.

### Labor market adjustments during the COVID-19 pandemic
- LA5 (Brazil, Chile, Colombia, Mexico and Peru) total employment fell by 30 percent on average between January and May 2020, the largest four-month contraction on record.
- Country-specific observed changes:
  - Brazil experienced the lowest decline in employment over this period, partly due to the government’s emergency employment protection program.
  - Chile, Colombia, and Mexico: employment declined by 20 percent to 30 percent between January and May 2020.
  - Peru (Lima): data for Lima shows a pronounced decline in employment (70 percent).
  - Bolivia: employment fell by 15 percent from February to May.
  - Ecuador: equivalent contraction from December to May/June.
  - Uruguay: employment reduced by 6 percent from February to May.
- Timing and short-term recovery:
  - Employment bottomed-out in May in all countries except Brazil.
  - From May to June, employment increased in some countries as compliance with lockdowns fell and economies gradually reopened.
  - Peru (Lima) experienced the largest month-over-month percentage increase between May and June (118 percent).
  - Chile, Colombia and Mexico: employment grew between 6 percent and 20 percent over the same period.
  - Brazil experienced a further decline in employment in June (one percent).
  - Chile experienced a reversal in employment gains in July.
- Employment levels relative to January 2020:
  - In June 2020 employment was 13 percent to 17 percent lower than January levels in Brazil, Chile, Colombia and Mexico, and 30 percent lower in Peru.
  - Comparator emerging market and advanced economies: on average employment was 1.5 percent below its January level.
- How job losses were absorbed:
  - On average in LA5, for every 100 workers that lost their jobs between January and June, 15 reported being unemployed and 85 reported being out of the labor force.
  - Colombia is the only LA5 country where unemployment played a more prominent role.
  - In many advanced economies, reductions in employment mostly resulted in more unemployed individuals (contrast with LA5).
- Hours worked:
  - The average employed worker in LA5 experienced a significant reduction in weekly hours worked from February to June 2020.
  - The reduction in average weekly hours worked was largest in Brazil and smallest in Chile.
  - A decrease in hours implies total supply of worker-hours fell more than employment.
  - ILO (2020) estimates LAC lost the equivalent of 55 million full-time employees in the second quarter when jointly considering job losses and cuts in hours.
  - Europe and Central Asia and Africa also saw a reduction of 55 million full-time employees in the second quarter, but pre-pandemic employment levels in each of those regions were roughly 30 percent larger than in LAC.

### Key factors affecting employment dynamics during COVID-19
- Channels through which the pandemic affected activity and employment:
  i) Containment policies and social distancing prevented workers in some sectors from participating in productive activities, reducing hours worked.
  ii) Reduced mobility and fear of contagion hindered demand for contact-intensive sectors (hospitality, entertainment, tourism).
  iii) Direct impacts were amplified by intersectoral linkages.
  iv) The initial supply shock could have led to a larger aggregate demand shock (Guerrieri et al., 2020).
- Features that increased LAC’s ex-ante vulnerability:
  - Contact-intensive sectors represent a larger share of GDP and a large share of employment in LAC (magnified by strong intersectoral links, especially in Brazil).
  - Fewer people in LAC work in occupations amenable to remote work:
    - Brazil, Chile, Panama, and Uruguay have the highest share of teleworkable jobs (approximately 25 percent).
    - In other LAC countries less than 20 percent of jobs can be performed remotely.
    - By contrast, emerging markets in Asia and Europe have teleworkable job shares from slightly below 30 percent to more than 50 percent.
  - LAC has lower access to broadband internet, which hampers teleworkability.
- Informality and vulnerability:
  - Informal workers comprise a large share of LAC employment and are more likely to be employed in high contact intensity and low teleworkability jobs.
  - Except for Peru, the share of informal workers employed in contact-intensive occupations is between 5 and 10 percentage points higher than for formal workers.
  - The share of informal workers with high teleworkability jobs is between 20 and 40 percentage points lower than for formal workers.
- Sectoral correlation:
  - There is a negative correlation between contact intensity and teleworkability across sectors.
  - Sectors with large shares of contact-intensive and low-teleworkability occupations include trade, hotels and restaurants, and transport and storage.
  - Sectors with large shares of low-contact intensity and high-teleworkability occupations include finance and insurance, real estate, and information and communications.
  - All sectors remain exposed through input-output linkages to directly affected industries.

### Quantification framework for employment and value-added at risk
- Framework adaptation:
  - The chapter adapts the framework in Alfaro et al. (2020) to estimate individual job loss probabilities and aggregate employment and value added at risk.
- Job loss probability depends on:
  (i) a sector-specific demand shock (fear of contagion) and an aggregate demand shock (a Keynesian multiplier),
  (ii) a sector-specific supply shock associated with the state of lockdowns,
  (iii) characteristics of a worker’s occupation (contact intensity, mitigated by teleworkability),
  (iv) impacts on upstream (suppliers) and downstream (buyers) sectors via inter-industry linkages,
  (v) uneven impact of lockdown on firms of different sizes.
- Definitions:
  - Employment at risk: sum of all individuals employed prior to the pandemic, weighted by their job loss probabilities.
  - Value added at risk: derived from employment at risk using sectoral labor intensity.
- Three phases simulated:
  - Lockdown phase: all workers in non-essential sectors (including mining, construction, accommodation, most manufacturing, wholesale and retail trade, transportation, services, arts and entertainment) are subject to lockdowns; demand shocks most negative in transportation and storage, accommodation and food services, wholesale and retail trade, manufacturing, agriculture, arts and entertainment.
  - Selective reopening phase: some sectors (notably manufacturing and construction) reopen; in sectors at least 50 percent open, workers in small firms, informal jobs or self-employed return with no restrictions; risk to medium-sized firms reduces less; demand shock reduced by 25 percent relative to lockdown phase.
  - Advanced reopening phase: supply-side employment risk disappears for sectors reopened in partial opening phase; for remaining sectors supply shock dissipates for self-employed and small firm workers; medium-sized firms face slower reduction in employment at risk due to financial stress and scarring; demand shock assumed 50 percent lower than in lockdown phase.
- Timing correspondence:
  - The three phases broadly correspond to: second quarter of 2020 (lockdown), second half of 2020 (selective reopening), and near future under the baseline (advanced reopening).
  - Country-specific differences exist depending on timing of sectoral openings.
- Limitations:
  - Does not consider employment support measures observed in the region (e.g., Brazil’s emergency employment support program); results may overstate losses, especially in lockdown phase.
  - May overstate speed of recovery in supply capacity in advanced reopening phase; assumes workers at risk can return to same-sector jobs once lockdowns end, whereas permanent sectoral damage and costly reallocation could slow supply recovery.

### Simulation results: employment and value added at risk (LA5 aggregate)
- Employment at risk:
  - Roughly 75 million workers were at risk during the lockdown phase in LA5 countries, or 43 percent of employment at the beginning of 2020.
  - Employment at risk falls to 40 million workers in the partial reopening phase.
  - Employment at risk stays at about 15 million workers in the advanced reopening phase.
- Informality and dynamics:
  - Informal workers face larger employment risk during lockdowns due to occupation mix and weaker linkages to large firms with cash and credit access.
  - Informal workers are the largest contributor to the reduction in employment at risk during reopening phases.
- Persistence drivers in the advanced reopening phase:
  (i) the continuing demand shock affecting sectors directly and indirectly until the pandemic is fully under control,
  (ii) constraints on medium-sized firms rebuilding formal jobs (scarring).
- Aggregate labor metrics:
  - The labor force and employment in LA5 in January 2020 was 203 and 184 million, respectively.

---

### 1. Brazil — key findings
- Employment at Risk by Firm Size:
  - Employment at risk among self-employed individuals and workers in small firms is 3-4 times larger than employment at risk among workers from medium and large firms (Brazil, Colombia, Mexico).
  - In Chile it is 1.4 times larger (mainly self-employed individuals).
  - In Peru the difference is around 6.5 times, due to the high incidence of informality among the self-employed and micro-firms.
- Sectoral impacts:
  - Sectors with low ability to telework and high contact intensity—such as transportation, hospitality (hotels and restaurants), and wholesale and retail trade—face employment at risk of more than 50 percent during the lockdown phase.
  - Sectors not directly affected by lockdowns and/or with higher ability to telework—such as finance, real estate, and education—face more limited risk.
  - In the selective reopening phase reactivation of some sectors reduces employment risk in most industries; some sectors (transport and hospitality) remain high.
  - In the advanced reopening phase employment at risk falls across all sectors.
- Supply vs demand shocks and linkages:
  - The impact of the supply shock outweighs that of the demand shock in the lockdown phase across LA5.
  - Roughly two thirds of employment at risk in LA5 is attributable to lockdowns and their propagation through input-output linkages.
  - The indirect impact through input-output linkages can increase employment at risk by 30 to 40 percent.
- Value added at risk vs employment at risk:
  - Between 25 and 35 percent of value added is vulnerable to lockdowns.
  - Value added at risk falls as economies reopen but even in the advanced reopening phase stands above 5 percent.
  - For all phases, employment at risk is significantly larger than value added at risk.
  - In the LA5 overall:
    - Value added at risk in the early recovery phase is 53 percent of the value added at risk in the lockdown phase.
    - Employment at risk in the early recovery phase is 43 percent compared to the lockdown.
    - In the advanced reopening, value added and employment at risk fall to 24 and 20 percent of the levels from the lockdown period, respectively.
- Comparison with observed losses:
  - Simulations are an upper bound; employment shrank by approximately 20-30 percent in LA5, roughly half to two thirds of the decline suggested by simulations.
  - Simulations do not account for policy responses that likely mitigated declines and may capture workers that experienced reductions in hours.
  - Example: the number of employed individuals in Mexico working 35 hours per week or more fell by approximately 45 percent between May 2019 and May 2020.
- Patterns by worker type:
  - Informal and self-employed workers were the main drivers of employment losses in LA5 in the second quarter.
    - Employment losses among informal and self-employed workers accounted for close to two thirds of all employment losses between the first and second quarters in Brazil, Chile and Peru.
    - In Mexico they accounted for close to 85 percent.
  - In countries with sustained employment improvements since April (Colombia and Mexico), informal and self-employed workers experienced larger employment gains.
  - The recovery may entail lower-paying jobs compared to pre-COVID-19 wages and thus lower productivity; firm closures and layoffs may cause scarring and misallocation.
- Unequal burden:
  - Employment fell more steeply for women than for men; in Brazil, Colombia and Peru the decline in female employment from February to June is approximately 5 percentage points larger than male employment.
  - Workers with only primary education experienced a steeper decline in employment compared to those with secondary education in most countries, except for Mexico.
  - Early estimates show that the pure labor market impact of COVID-19 could lead to a total 23 to 30 million “new poor” in Argentina, Brazil, Colombia, and Mexico combined.

---

### 4. Teleworkability by Gender — findings and implications
- Teleworkability and contact-intensity definitions:
  - Teleworkability is based on Dingel and Neiman (2020); high contact intensity occupations are defined as in Leibovici, Santacreu, and Famiglietti (2020).
- Distributional impacts and policy cushioning:
  - Large employment protection and social assistance programs in some countries partly offset, and in some cases erased, the distributive impact of the shock in the short run.
  - Lustig et al. (2020) simulate that policy support programs in Argentina and Brazil erased most of the increase in poverty due to COVID-19, while in Colombia it mitigated it.
- Short-term outlook determinants:
  - Duration of economic constraints and extent of lockdown measures.
  - International spillovers via value chains and infection cases in other countries.
  - Frictions in search and matching processes may slow job creation even after demand rebounds; labor informality may imply quick initial recovery via informal jobs but slower subsequent formal job creation.
- Long-term risks and scarring:
  - Conventional hysteresis and structural reallocation risks apply.
  - COVID-19-specific risks: prolonged lockdowns increase probability job losses become permanent; recoveries may spur automation replacing jobs; LAC’s high share of routine-intensive jobs implies large potential job losses, particularly for women.
  - Women have borne most additional household and childcare duties; closure of in-person schooling increases childcare duties and may reduce female labor force participation persistently.
- Policy implications and structural priorities (four priorities):
  - Favor creation of formal jobs.
  - Strengthen workers’ safety nets.
  - Level the playing field for female workers.
  - Promote a “green recovery.”
- Specific policy directions:
  - Fine-tune policies to each country’s reopening phase to promote formality while fostering short-term growth.
  - Strengthen and extend effective coverage of unemployment insurance (UI) schemes.
  - Modernize regulatory frameworks to extend social protection coverage to workers in gig economy / digital platforms.
  - Remove distortions in high-skill abstract task-intensive and service-oriented sectors to increase incentives for women to acquire human capital and enter the labor force.
  - Use the recession to incentivize low-carbon activities and energy efficiency, including through carbon taxes; carbon tax may lower relative distortion between formal and informal sectors.
- Employment protection programs — Latin America examples:
  - Common measures: wage subsidies and loans to support employment retention (Argentina, Brazil, Chile, Colombia, Mexico, Peru); expansions of unemployment insurance in some countries (Argentina, Chile, Colombia).
  - Brazil’s emergency wage subsidy program:
    - Budget allocation of 0.7 percent of GDP.
    - Allows subsidized reduction in working hours or suspension of work contracts for up to 6 months (suspension fully subsidized for small firms).
    - As of early July, over 9.1 million workers have benefited from the scheme.
    - Beneficiary breakdown: roughly half through a reduction in hours and half from a complete suspension.
    - Overall coverage: around 10 percent of all jobs in Brazil and more than 25 percent of all formal private sector jobs have benefited from the program.
- Unemployment insurance schemes in LA5 before and during COVID-19:
  - Prior to the pandemic, only Brazil, Chile, and Colombia had traditional UI programs among LA5.
  - Typical features: only private sector employees eligible; minimum contribution period about one year; triggered by involuntary job losses.
  - Replacement rates: Brazil and Chile average replacement rates comparable to US and Canada (around 50 percent); somewhat below France and Germany (around 60 percent).
  - Duration:
    - Brazil: maximum duration of payments of only 5 months.
    - Chile: duration of 1 year.
    - Colombia: duration of 6 months.
  - Effective coverage (fraction of unemployed workers who receive UI):
    - Brazil: 4 percent.
    - Colombia: 5 percent.
    - US: 26 percent (for comparison).
    - Canada and France: 40 percent (for comparison).
    - Chile: 46 percent.
  - Implications: Informality likely behind low effective coverage; low UI eligibility implies limited incentives for active job search.
  - Pandemic-era international examples:
    - US: CARES Act expanded UI, including Pandemic Unemployment Insurance (PUA) and USD 600 per week Pandemic Unemployment Compensation Payment (PUCP) until end-July.
    - Canada: Canada Emergency Response Benefit (CERB), weekly payments of CAD 500 for 28 weeks; Canada Recovery Benefit (CRB) with CAD 400 weekly for 6 months for those ineligible.
    - Europe: expanded job retention schemes (Kurzarbeit, Activité Partielle).

---

### Box 3 — Evidence from Mexico’s labor survey (main results)
- Data and methodology:
  - Data source: Mexico’s ETOE labor monthly survey (April through June), following workers interviewed in the regular March survey.
  - Outcomes: experiencing a full employment spell (employed in March, April, May, and June); experiencing a reduction in hours in June relative to March; experiencing a reduction in hourly labor income relative to March.
  - Controls: age, gender, educational attainment, type of occupation (teleworkable or contact-intensive), sector, and firm size.
  - Observations: 8,946.
  - R-squared reported for columns (1)–(6): 0.10, 0.11, 0.01, 0.01, 0.02, 0.024.
- Selected econometric coefficients and interpretations:
  - Male:
    - Employed March through June: 0.111***, 0.124***.
    - Reduced hours in June compared to March: -0.0442***, -0.0509***.
    - Reduction in income since March: -0.0505***, -0.0534***.
  - Tertiary education:
    - Employed March through June: 0.140***, 0.0972***.
    - Reduced hours in June compared to March: -0.0770***, -0.0545***.
    - Reduction in income since March: -0.0461***, -0.0357***.
  - Age groups (relative to baseline):
    - 25–40 years: Employed: 0.173***, 0.163***; Reduced hours: -0.0552***, -0.0501***; Reduction in income: -0.0261*, -0.0239.
    - 40–55 years: Employed: 0.179***, 0.167***; Reduced hours: -0.0599***, -0.0538***; Reduction in income: -0.0215, -0.0190.
    - Over 55 years: Employed: -0.00753, -0.0221; Reduced hours: -0.0334*, -0.0262; Reduction in income: 0.0979***, 0.101***.
- Role of firm size:
  - Employed by a mid-sized firm in March:
    - Employed March through June: 0.174***, 0.171***.
    - Reduced hours in June compared to March: 0.0005600, 0.000888.
    - Reduction in income since March: -0.0556***, -0.0563***.
  - Employed by a large firm in March:
    - Employed March through June: 0.213***, 0.215***.
    - Reduced hours in June compared to March: -0.0114, -0.0165.
    - Reduction in income since March: -0.0950***, -0.101***.
  - Interpretation: larger firm size increases the chance of a full employment spell and reduces the chance of reductions in hours; larger firms show larger negative effects on hourly wages in some specifications.
- Role of occupation:
  - Contact-intensive job in March:
    - Employed March through June: -0.0187*.
    - Reduced hours in June compared to March: 0.0322***, 0.0322***.
    - Interpretation: contact-intensive occupations are associated with lower full-employment spells and higher probability of hours and wage cuts.
  - Teleworkable job in March:
    - Employed March through June: 0.126***.
    - Reduced hours in June compared to March: -0.0671***.
    - Reduction in income since March: -0.0322**.
    - Interpretation: teleworkable occupations are associated with higher likelihood of full-employment spells and lower likelihood of reductions in hours and wages.
- Gender and education disparities:
  - Women: less likely to experience employment losses but more likely to suffer reductions in hours and income.
  - Primary and secondary educated workers: similar pattern to women—less likely to lose employment outright but more likely to see reductions in hours and income.
  - Tertiary education: higher probability of full employment spell and lower likelihood of reductions in hours and income.
- Additional notes:
  - Constant terms (columns (1)–(6)): 0.263***, 0.247***, 0.683***, 0.702***, 0.714***, 0.731***.
  - Observations are consistently 8,946 across regressions.
  - Box prepared by Samuel Pienknagura. Source: IMF staff calculations.

---

### Annex Figure 3.3 — Colombia: Labor productivity and employment at risk (lockdown phase)
- Sectors displayed:
  - Agriculture; Natural resources; Manufacturing; Construction; Trade; Transportation; Hotels and restaurants; Information and communication; Finance; Real estate; Administrative, professional, and other services; Public sector; Education; Healthcare; Household workers.
- Axes and tick labels preserved exactly:
  - Horizontal axis: Log   of  value  added   per  worker (billions  of pesos)
    - Tick labels: 0.0 0.5 1.0 1.5 2.0 2.5
  - Vertical axis: Employment  at  risk  (millions)
    - Tick labels: 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5
- Visual emphasis: figure maps each sector by its log of value added per worker (billions of pesos) and employment at risk (millions) during the lockdown phase.
- Sources: National authorities; and IMF staff calculations.

*International Monetary Fund | October 2020 — “Introduction” (labor - Introduction)*

### Introduction

### Introduction

### Overview of the shock and scope
- The COVID-19 pandemic triggered major economic disruptions and severely affected labor markets in Latin America and the Caribbean (LAC).
- The October 2020 Regional Economic Outlook (REO): Western Hemisphere notes the COVID-19 recession has a much larger decline in consumption in labor-intensive service sectors compared with previous recessions.
- This chapter assesses: the impact on LAC labor markets using labor surveys; how pre-existing labor market features increased vulnerability; and the outlook for employment, income and activity during the recovery.

### Labor market adjustments during the COVID-19 pandemic
- LA5 (Brazil, Chile, Colombia, Mexico and Peru) total employment fell by 30 percent on average between January and May 2020, the largest four-month contraction on record.
- Country-specific observed changes:
  - Brazil experienced the lowest decline in employment over this period, partly due to the government’s emergency employment protection program.
  - Chile, Colombia, and Mexico: employment declined by 20 percent to 30 percent between January and May 2020.
  - Peru (Lima): data for Lima shows a pronounced decline in employment (70 percent).
  - Bolivia: employment fell by 15 percent from February to May.
  - Ecuador: equivalent contraction from December to May/June.
  - Uruguay: employment reduced by 6 percent from February to May.
- Timing and short-term recovery:
  - Employment bottomed-out in May in all countries except Brazil.
  - From May to June, employment increased in some countries as compliance with lockdowns fell and economies gradually reopened.
  - Peru (Lima) experienced the largest month-over-month percentage increase between May and June (118 percent).
  - Chile, Colombia and Mexico: employment grew between 6 percent and 20 percent over the same period.
  - Brazil experienced a further decline in employment in June (one percent).
  - Chile experienced a reversal in employment gains in July.
- Employment levels relative to January 2020:
  - In June 2020 employment was 13 percent to 17 percent lower than January levels in Brazil, Chile, Colombia and Mexico, and 30 percent lower in Peru.
  - Comparator emerging market and advanced economies: on average employment was 1.5 percent below its January level.
- How job losses were absorbed:
  - On average in LA5, for every 100 workers that lost their jobs between January and June, 15 reported being unemployed and 85 reported being out of the labor force.
  - Colombia is the only LA5 country where unemployment played a more prominent role.
  - In many advanced economies, reductions in employment mostly resulted in more unemployed individuals (contrast with LA5).
- Hours worked:
  - The average employed worker in LA5 experienced a significant reduction in weekly hours worked from February to June 2020.
  - The reduction in average weekly hours worked was largest in Brazil and smallest in Chile.
  - A decrease in hours implies total supply of worker-hours fell more than employment.
  - ILO (2020) estimates LAC lost the equivalent of 55 million full-time employees in the second quarter when jointly considering job losses and cuts in hours.
  - Europe and Central Asia and Africa also saw a reduction of 55 million full-time employees in the second quarter, but pre-pandemic employment levels in each of those regions were roughly 30 percent larger than in LAC.

### Key factors affecting employment dynamics during COVID-19
- Channels through which the pandemic affected activity and employment:
  i) Containment policies and social distancing prevented workers in some sectors from participating in productive activities, reducing hours worked.
  ii) Reduced mobility and fear of contagion hindered demand for contact-intensive sectors (hospitality, entertainment, tourism).
  iii) Direct impacts were amplified by intersectoral linkages.
  iv) The initial supply shock could have led to a larger aggregate demand shock (Guerrieri et al., 2020).
- Features that increased LAC’s ex-ante vulnerability:
  - Contact-intensive sectors represent a larger share of GDP and a large share of employment in LAC (magnified by strong intersectoral links, especially in Brazil).
  - Fewer people in LAC work in occupations amenable to remote work:
    - Brazil, Chile, Panama, and Uruguay have the highest share of teleworkable jobs (approximately 25 percent).
    - In other LAC countries less than 20 percent of jobs can be performed remotely.
    - By contrast, emerging markets in Asia and Europe have teleworkable job shares from slightly below 30 percent to more than 50 percent.
  - LAC has lower access to broadband internet, which hampers teleworkability.
- Informality and vulnerability:
  - Informal workers comprise a large share of LAC employment and are more likely to be employed in high contact intensity and low teleworkability jobs.
  - Except for Peru, the share of informal workers employed in contact-intensive occupations is between 5 and 10 percentage points higher than for formal workers.
  - The share of informal workers with high teleworkability jobs is between 20 and 40 percentage points lower than for formal workers.
- Sectoral correlation:
  - There is a negative correlation between contact intensity and teleworkability across sectors.
  - Sectors with large shares of contact-intensive and low-teleworkability occupations include trade, hotels and restaurants, and transport and storage.
  - Sectors with large shares of low-contact intensity and high-teleworkability occupations include finance and insurance, real estate, and information and communications.
  - All sectors remain exposed through input-output linkages to directly affected industries.

### Quantification framework for employment and value-added at risk
- The chapter adapts the framework in Alfaro et al. (2020) to estimate individual job loss probabilities and aggregate employment and value added at risk.
- Job loss probability depends on:
  (i) a sector-specific demand shock (fear of contagion) and an aggregate demand shock (a Keynesian multiplier),
  (ii) a sector-specific supply shock associated with the state of lockdowns,
  (iii) characteristics of a worker’s occupation (contact intensity, mitigated by teleworkability),
  (iv) impacts on upstream (suppliers) and downstream (buyers) sectors via inter-industry linkages,
  (v) uneven impact of lockdown on firms of different sizes.
- Definitions:
  - Employment at risk: sum of all individuals employed prior to the pandemic, weighted by their job loss probabilities.
  - Value added at risk: derived from employment at risk using sectoral labor intensity.
- Three phases simulated:
  - Lockdown phase: all workers in non-essential sectors (including mining, construction, accommodation, most manufacturing, wholesale and retail trade, transportation, services, arts and entertainment) are subject to lockdowns; demand shocks most negative in transportation and storage, accommodation and food services, wholesale and retail trade, manufacturing, agriculture, arts and entertainment.
  - Selective reopening phase: some sectors (notably manufacturing and construction) reopen; in sectors at least 50 percent open, workers in small firms, informal jobs or self-employed return with no restrictions; risk to medium-sized firms reduces less; demand shock reduced by 25 percent relative to lockdown phase.
  - Advanced reopening phase: supply-side employment risk disappears for sectors reopened in partial opening phase; for remaining sectors supply shock dissipates for self-employed and small firm workers; medium-sized firms face slower reduction in employment at risk due to financial stress and scarring; demand shock assumed 50 percent lower than in lockdown phase.
- Timing correspondence:
  - The three phases broadly correspond to: second quarter of 2020 (lockdown), second half of 2020 (selective reopening), and near future under the baseline (advanced reopening).
  - Country-specific differences exist depending on timing of sectoral openings.
- Limitations of the framework noted in the chapter:
  - Does not consider employment support measures observed in the region (e.g., Brazil’s emergency employment support program); results may overstate losses, especially in lockdown phase.
  - May overstate speed of recovery in supply capacity in advanced reopening phase; assumes workers at risk can return to same-sector jobs once lockdowns end, whereas permanent sectoral damage and costly reallocation could slow supply recovery.

### Simulation results: employment and value added at risk
- LA5 aggregate results:
  - Roughly 75 million workers were at risk during the lockdown phase in LA5 countries, or 43 percent of employment at the beginning of 2020.
  - Employment at risk falls to 40 million workers in the partial reopening phase.
  - Employment at risk stays at about 15 million workers in the advanced reopening phase.
- Informality and dynamics:
  - Informal workers face larger employment risk during lockdowns due to occupation mix and weaker linkages to large firms with cash and credit access.
  - Informal workers are the largest contributor to the reduction in employment at risk during reopening phases.
- Persistence in the advanced reopening phase is related to:
  (i) the continuing demand shock affecting sectors directly and indirectly until the pandemic is fully under control,
  (ii) constraints on medium-sized firms rebuilding formal jobs (scarring).
- Aggregate labor metrics:
  - The labor force and employment in LA5 in January 2020 was 203 and 184 million, respectively.

*Italic: International Monetary Fund | October 2020 — “Introduction” (labor - Introduction)*

### 1.  Brazil

### 1. Brazil

### Employment at Risk by Firm Size
- Employment at risk among self-employed individuals and workers in small firms is 3-4 times larger than employment at risk among workers from medium and large firms (Brazil, Colombia, Mexico).
- In Chile it is 1.4 times larger (mainly self-employed individuals).
- In Peru the difference is around 6.5 times, due to the high incidence of informality among the self-employed and micro-firms.
- Note: LA5 = Brazil, Chile, Colombia, Mexico, Peru.

### Sectoral Impacts during Lockdown and Reopening Phases
- Sectors with low ability to telework and high contact intensity—such as transportation, hospitality (hotels and restaurants), and wholesale and retail trade—face employment at risk of more than 50 percent during the lockdown phase.
- Sectors not directly affected by lockdowns and/or with higher ability to telework—such as finance, real estate, and education—face more limited risk.
- In the selective reopening phase (example shown for Colombia), reactivation of some sectors (manufacturing, construction, and part of trade) reduces employment risk in most industries.
- Employment at risk rises further or stays high in a few sectors (such as transport and hospitality) during selective reopening because the direct impact of the lockdown continues.
- In the advanced reopening phase employment at risk falls across all sectors.

### Supply vs Demand Shocks and Input-Output Linkages
- The impact of the supply shock outweighs that of the demand shock in the lockdown phase across LA5.
- Roughly two thirds of employment at risk in LA5 is attributable to lockdowns and their propagation through input-output linkages.
- The preponderance of the supply shock fades as economies enter the selective reopening phase.
- The indirect impact of supply and demand shocks through input-output linkages can increase employment at risk by 30 to 40 percent (the difference between the purple and grey bars in the simulation).

### Value Added at Risk vs Employment at Risk
- Between 25 and 35 percent of value added is vulnerable to lockdowns.
- Value added at risk falls as economies reopen but even in the advanced reopening phase stands above 5 percent.
- For all lockdown/reopening phases, employment at risk is significantly larger than value added at risk—reflecting higher labor intensity in sectors more vulnerable to lockdowns.
- In the LA5 overall:
  - Value added at risk in the early recovery phase is 53 percent of the value added at risk in the lockdown phase.
  - Employment at risk in the early recovery phase is 43 percent compared to the lockdown.
  - In the advanced reopening, value added and employment at risk fall to 24 and 20 percent of the levels from the lockdown period, respectively.

### Comparison of Model Predictions with Second Quarter Employment Data
- Simulation results show larger employment at risk during the lockdown phase compared to employment losses registered in LA5 when lockdowns were most stringent; simulations are an upper bound of actual employment losses.
- Employment shrank by approximately 20-30 percent in LA5, roughly half to two thirds of the decline suggested by the employment at risk simulations.
- Simulations do not account for policy responses that likely mitigated the decline in employment, and may capture workers that experienced reductions in hours.
- Example: The number of employed individuals in Mexico working 35 hours per week or more fell by approximately 45 percent between May 2019 and May 2020—a decline comparable to the share of employment at risk in the second quarter.

### Patterns by Worker Type and Recovery Implications
- Informal and self-employed workers were the main drivers of employment losses in LA5 in the second quarter.
  - Employment losses among informal and self-employed workers accounted for close to two thirds of all employment losses between the first and second quarters in Brazil, Chile and Peru.
  - In Mexico they accounted for close to 85 percent.
- In countries with sustained employment improvements since April (Colombia and Mexico), informal and self-employed workers experienced larger employment gains—suggesting the recovery may follow patterns similar to simulations.
- The recovery may entail lower-paying jobs compared to pre-COVID-19 wages and thus lower productivity; firm closures and layoffs may cause scarring and misallocation affecting the pace of recovery.

### Unequal Burden: Gender, Education, and Informality
- Employment fell more steeply for women than for men; in Brazil, Colombia and Peru the decline in female employment from February to June is approximately 5 percentage points larger than male employment.
- Workers with only primary education experienced a steeper decline in employment compared to those with secondary education in most countries, except for Mexico.
- The share of female workers employed in occupations with high contact intensity is larger than the share of male workers; female workers also have more teleworkable occupations in part.
- The ability to work remotely is an advantage favoring workers with tertiary education; contact-intensive jobs are more likely among workers with secondary education.
- The link between job losses, educational attainment, and informality highlights the regressive nature of the shock and is expected to exacerbate inequality and increase poverty.
- Early estimates show that the pure labor market impact of COVID-19 could lead to a total 23 to 30 million “new poor” in Argentina, Brazil, Colombia, and Mexico combined.

*Source: IMF staff calculations; LA5 = Brazil, Chile, Colombia, Mexico, Peru.*

### 4.  Teleworkability by Gender

### 4.  Teleworkability by Gender

### Teleworkability, gender, and distributive impacts
- Teleworkability is based on Dingel and Neiman (2020); high contact intensity occupations are defined as in Leibovici, Santacreu, and Famiglietti (2020).
- Large employment protection and social assistance programs in some countries partly offset, and in some cases erased, the distributive impact of the shock, at least in the short run.
- Lustig et al. (2020) simulate the combined effect on poverty and inequality of the COVID-19 shock and assistance programs:
  - Policy support programs in Argentina and Brazil erased most of the increase in poverty due to COVID-19, while in Colombia it mitigated it.
  - Social indicators may further deteriorate as costly assistance programs are unwound and if the effects of the shock persist longer than expected.

### Short-term outlook for employment recovery
- Three key determinants of the speed of recovery:
  - Duration of economic constraints associated with the pandemic and extent of lockdown measures; sectors unable to operate at full capacity will limit labor demand.
  - International spillovers via value chains and infection cases in other countries can delay recovery, important for countries dependent on exports and tourism.
  - Frictions in search and matching processes may slow job creation even after labor demand rebounds (Buckman et al., 2020; Kandoussi and Longot, 2020). Labor informality in LAC may imply:
    - Quick initial recovery in employment via informal jobs.
    - Slower subsequent growth as formal jobs are created with longer lags.
- Limiting losses in formal employment during the contraction is crucial to prevent a slow-paced recovery.

### Long-term risks and potential scarring
- Conventional recession channels applicable to COVID-19: hysteresis in business cycle dynamics, structural reallocation, falling investment, loss of firm-worker matches, destruction of on-the-job experience (tenure premia), and long unemployment spells (Cerra et al., 2020; Portes, 2020).
- COVID-19-specific considerations:
  - Longer lockdown/pandemic duration increases probability that job losses become permanent; prompt recoveries elsewhere often reflect recalls of workers (Jones et al., 2020).
  - Recoveries historically spur permanent replacement of jobs by automation (Cores et al., 2020; Jaimovic and Siu, 2020). LAC risks:
    - High share of routine-intensive jobs implies large potential loss of jobs, particularly for women (Beylis et al., 2020; Brussevich et al., 2019).
    - Automation adoption in emerging markets has been slower due to low wages, informality, and scarcity of advanced technical skills; some hardest-hit industries (hospitality, trade, construction) are not easily automatable.
    - High macroeconomic uncertainty reduces incentives to invest in expensive machinery.
  - Female labor force participation risk:
    - Women have borne the bulk of additional house- and family-related activities during the pandemic (Andrew et al., 2020; Del Boca et al., 2020; Kalenkoski et al., 2020).
    - Closure of in-person schooling increases childcare duties, potentially reducing female labor force participation persistently.
    - A fall in female participation is particularly relevant for LAC where female entry into the labor force has been a key driver of employment growth (Busso and Romero Fonseca, 2015).

### Policy implications and structural priorities
- Four important structural aspects for recovery stimulus policies:
  - Favor creation of formal jobs.
  - Strengthen workers’ safety nets.
  - Level the playing field for female workers.
  - Promote a “green recovery.”
- Specific policy directions:
  - Fine-tune policies to each country’s reopening phase to promote formality while fostering short-term growth.
  - Strengthen and extend effective coverage of unemployment insurance (UI) schemes to provide buffers as the economy shifts.
  - Modernize regulatory frameworks to extend social protection coverage to workers in gig economy / digital platforms:
    - These platforms pose challenges for traditional UI but increase visibility of economic activity and labor earnings.
  - Remove distortions in high-skill abstract task-intensive and service-oriented sectors to increase incentives for women to acquire human capital and enter the labor force (Petrongolo and Ngai, 2017; Bhalotra and Fernández, 2018).
  - Use the recession as an opportunity to incentivize low-carbon activities and energy efficiency, including through carbon taxes (October 2020 WEO, Chapter 3).
    - Carbon tax may lower relative distortion between formal and informal sectors, potentially expanding the formal sector (Bento et al., 2018).
    - Consider policies to support job transitions and targeted cash transfers financed perhaps by carbon tax revenues.

### Employment protection programs (Box 1) — Latin America examples
- Common measures: wage subsidies and loans to support employment retention (Argentina, Brazil, Chile, Colombia, Mexico, Peru); expansions of unemployment insurance in some countries (Argentina, Chile, Colombia).
- Brazil’s emergency wage subsidy program:
  - Budget allocation of 0.7 percent of GDP.
  - Allows subsidized reduction in working hours or suspension of work contracts for up to 6 months (suspension fully subsidized for small firms).
  - As of early July, over 9.1 million workers have benefited from the scheme.
  - Beneficiary breakdown: roughly half through a reduction in hours and half from a complete suspension.
  - Overall coverage: around 10 percent of all jobs in Brazil and more than 25 percent of all formal private sector jobs have benefited from the program.

### Unemployment insurance schemes in LA5 countries before and during COVID-19 (Box 2)
- Prior to the pandemic, only Brazil, Chile, and Colombia had traditional UI programs among LA5.
- General features:
  - Similar structure and terms to US and Canada but slightly less generous than some European countries like France and Germany.
  - Typically only private sector employees eligible; minimum contribution period about one year; triggered by involuntary job losses.
  - Brazil and Chile average replacement rates comparable to US and Canada (around 50 percent) and somewhat below France and Germany (around 60 percent).
- Duration and coverage specifics:
  - Brazil: maximum duration of payments of only 5 months.
  - Chile: duration of 1 year.
  - Colombia: duration of 6 months.
- Effective coverage (fraction of unemployed workers who receive UI):
  - Brazil: 4 percent.
  - Colombia: 5 percent.
  - US: 26 percent (for comparison).
  - Canada and France: 40 percent (for comparison).
  - Chile: 46 percent.
- Implications of low coverage:
  - Informality likely behind low effective coverage; informal employees and self-employed typically do not contribute to UI.
  - Low UI eligibility implies limited incentives for active job search; many displaced workers may be classified as out of the labor force during the pandemic, reflecting limited practical difference between being unemployed or out of the labor force without access to UI.
- Pandemic-era expansions and international comparisons:
  - US expanded UI via CARES Act: extended weeks, Pandemic Unemployment Insurance (PUA), and USD 600 per week Pandemic Unemployment Compensation Payment (PUCP) until end-July.
  - Canada temporarily replaced usual EI with Canada Emergency Response Benefit (CERB): weekly payments of CAD 500 for 28 weeks; introduced Canada Recovery Benefit (CRB) with CAD 400 weekly for 6 months for those ineligible.
  - Europe expanded pre-existing job retention schemes (Kurzarbeit, Activité Partielle) to preserve worker-firm matches and provide partial compensation for reduced hours.

*Source: Regional Economic Outlook: Western Hemisphere, October 2020 — chapter excerpt "4. Teleworkability by Gender."*

### Box 3. The Role of Teleworkability, Contact-Intensity and Firm Size in Determining Workers’

### Box 3. The Role of Teleworkability, Contact-Intensity and Firm Size in Determining Workers’ Employment Risk: Evidence from Mexico’s Labor Survey

### Data and methodology
- Data source: Mexico’s ETOE labor monthly survey (April through June), following workers interviewed in the regular March survey.
- Outcomes analyzed:
  - Experiencing a full employment spell (employed in March, April, May, and June).
  - Experiencing a reduction in hours in June relative to March.
  - Experiencing a reduction in hourly labor income relative to March.
- Controls included: age, gender, educational attainment, type of occupation (teleworkable or contact-intensive), sector of employment, and firm size.
- Sample size and fit:
  - Observations: 8,946
  - R-squared: 0.10, 0.11, 0.01, 0.01, 0.02, 0.024 (for columns (1) through (6), respectively)
- Statistical notation: Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

### Main econometric results (selected coefficients from Box Table 3.1)
- Male
  - Employed March through June: 0.111***, 0.124***
  - Reduced hours in June compared to March: -0.0442***, -0.0509***
  - Reduction in income since March: -0.0505***, -0.0534***
- Tertiary education
  - Employed March through June: 0.140***, 0.0972***
  - Reduced hours in June compared to March: -0.0770***, -0.0545***
  - Reduction in income since March: -0.0461***, -0.0357***
- Age groups (relative to baseline)
  - 25–40 years: 0.173***, 0.163***; Reduced hours: -0.0552***, -0.0501***; Reduction in income: -0.0261*, -0.0239
  - 40–55 years: 0.179***, 0.167***; Reduced hours: -0.0599***, -0.0538***; Reduction in income: -0.0215, -0.0190
  - Over 55 years: -0.00753, -0.0221; Reduced hours: -0.0334*, -0.0262; Reduction in income: 0.0979***, 0.101***

### Role of firm size
- Employed by a mid-sized firm in March
  - Employed March through June: 0.174***, 0.171***
  - Reduced hours in June compared to March: 0.0005600, 0.000888
  - Reduction in income since March: -0.0556***, -0.0563***
- Employed by a large firm in March
  - Employed March through June: 0.213***, 0.215***
  - Reduced hours in June compared to March: -0.0114, -0.0165
  - Reduction in income since March: -0.0950***, -0.101***

- Interpretation from regressions:
  - Larger firm size increases the chance of a worker experiencing a full employment spell.
  - Larger firm size reduces the chance of experiencing reductions in hours.
  - Effects on hourly wages are not statistically significant in some specifications (noted in narrative).

### Role of occupation: teleworkability and contact-intensity
- Contact-intensive job in March
  - Employed March through June: -0.0187*
  - Reduced hours in June compared to March: 0.0322***, 0.0322***
  - Narrative interpretation: Workers in contact-intensive occupations are less likely to experience a full employment spell and more likely to suffer hours and wage cuts.
- Teleworkable job in March
  - Employed March through June: 0.126***
  - Reduced hours in June compared to March: -0.0671***
  - Reduction in income since March: -0.0322**
  - Narrative interpretation: Workers in teleworkable occupations are more likely to experience a full employment spell and less likely to experience reductions in hours and wages.

### Gender and education disparities
- Women (narrative summary from regressions)
  - All else equal, women are less likely to experience employment losses but more likely to suffer reductions in hours and income.
- Educational attainment
  - Workers with primary and secondary educational attainment face similar patterns as women: less likely to lose employment outright but more likely to see reductions in hours and income.
  - Tertiary education is associated with higher probability of full employment spell and lower likelihood of reductions in hours and income (see coefficients above).

### Additional table details and notes
- Constant terms (columns (1)–(6)): 0.263***, 0.247***, 0.683***, 0.702***, 0.714***, 0.731***
- Observations are consistently 8,946 across regressions.
- The analysis controls for sector of employment and other individual characteristics as recorded in the March survey.

*This box was prepared by Samuel Pienknagura. Source: IMF staff calculations.*

### Annex Figure 3.3. Colombia: Labor Productivity and Employment at Risk

### Annex Figure 3.3. Colombia: Labor Productivity and Employment at Risk

### During the Lockdown Phase by Sector
- Sectors shown:
  - Agriculture
  - Natural resources
  - Manufacturing
  - Construction
  - Trade
  - Transportation
  - Hotels and restaurants
  - Information and communication
  - Finance
  - Real estate
  - Administrative, professional, and other services
  - Public sector
  - Education
  - Healthcare
  - Household workers

- Chart axes and labels (preserved exactly as shown):
  - Horizontal axis: Log   of  value  added   per  worker (billions  of pesos)
    - Tick labels: 0.0 0.5 1.0 1.5 2.0 2.5
  - Vertical axis: Employment  at  risk  (millions)
    - Tick labels: 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5

- Visual emphasis:
  - The figure maps each listed sector by its log of value added per worker (billions of pesos) on the horizontal axis and employment at risk (millions) on the vertical axis during the lockdown phase.

### Key metadata and provenance
- Sources: National authorities; and IMF staff calculations.

*International Monetary Fund | Regional Economic Outlook: Western Hemisphere | October 2020*

---


_Source: https://www.imf.org/-/media/files/publications/reo/whd/2020/oct/english/labor.pdf_