The Gendered Impacts of COVID-19 on Labor Markets in Latin America and the Caribbean WORLD BANK Gender Innovation Lab for Latin America and the Caribbean (LACGIL)1 POLICY BRIEF: January, 2021 Key Messages • According to High-Frequency Phone Surveys (HFPS) conducted in 13 countries in LAC, female workers were 44 percent more likely than male workers to lose their jobs at the onset of the COVID-19 crisis. • As the crisis evolved, temporarily unemployed workers started to go back to work. But the di erence in job losses among females and males persisted. • Highly female-intensive sectors—trade, personal services, education, and hospitality—explain 56 percent of all job losses. • Factors associated with resilience to job losses during the crisis di er among males and females. For instance, the presence of school-age children at home is linked with a rise in job losses among females but not among males. • Employment losses as a result of the COVID-19 pandemic may deepen existing gender gaps in the region, thereby creating the need to design policy responses and actions that target women and help ensure an inclusive recovery. The context Across the globe, women often face lower income opportunities percent in 1990 to 53 percent in 2019). This has led to improve- relative to men. The COVID-19 pandemic is establishing new ments in female labor earnings and poverty reduction. But gender dynamics that may exacerbate the existing gender gaps in gaps persist in the region. Men are still far more likely than women economic opportunities in the region. Social distancing measures to participate in the labor force, have formal employment, hold have disrupted social and economic activities, impacting the higher-quality jobs, and work in higher-paying sectors. In addition, income-generating capacity of households. Evidence on previous women—particularly young women—are more likely to be unem- epidemics suggests that this type of shock threatens to roll back ployed (World Bank 2020). gains in women’s economic opportunities (de Paz et al. 2020). Females hold a disproportionate share of occupations requiring face-to-face interactions, such as in retail, personal care, and tourism, making them less likely to work from home and prone to Gender Innovation Lab for Latin America becoming unemployed. Females are more likely than males to be employed in the informal sector and in other vulnerable forms of and the Caribbean (LACGIL) employment. Women also tend to do more unpaid household work than men (about 2.7 hours a day). During the pandemic, The LACGIL supports impact evaluations and inferential unpaid care work has risen because children are out of school, the research to generate evidence on what works in closing elderly are experiencing greater care needs, and health services gender gaps in human capital, economic participation, are overwhelmed (World Bank 2020). social norms, and agency. Additionally, the lab dissemi- nates findings to improve operations and policy making Without well-informed and timely policy responses, the crisis in the design of cost-e ective interventions that tackle could widen gender gaps that persist in the region despite 30 gender inequalities and drive change. years of progress. In the region, for instance, labor force participa- tion among women has risen over the past three decades (from 41 To accomplish this, the LACGIL works in partnership with World Bank units, aid agencies and donors, governments, nongovernmental organizations, private sector firms, and researchers. 1 This note was prepared by Emilia Cucagna and Javier Romero. The team is grateful to Jacobus Joost De Hoop and Ximena del Carpio for very helpful comments. 1 This note explores the impacts of the COVID-19 crisis on labor outcomes among males and females and identifies the dimen- sions that render workers more resilient to job losses. These findings are then used to discuss implications for policymaking. To overcome the scarcity of data generated by the pause in most statistical operations resulting from social-distancing measures, High-Frequency Phone Surveys (HFPS) collected in the region by the Poverty and Equity Global Practice of the World Bank are used. In this way, the note aims to estimate the gendered outcomes in the labor markets associated with the deepest recession since World War II. The data Between May 2020 and August 2020, three waves of phone markets, changes in household incomes, access to services, behav- surveys were conducted in 13 countries in LAC to assess the iors, and knowledge around COVID-19, and demographic and impacts and transmission channels of the COVID-19 crisis in the household characteristics. region.2The first round of surveys was conducted in May 2020 (two months after most countries in LAC2 had declared quaran- The sample covers an average of 1,000 individuals per country (a tines, in mid-March 2020). The second wave was collected total of 13,152 observations) who were followed during three between June 2020 and July 2020, and the third between July rounds of data collection.3 The sample in each country is national- 2020 and August 2020. The surveys included modules on labor ly representative of individuals ages 18 or more who have access to a phone. The data are derived from probabilistic sampling using random digit dialing.4 The weights have been calibrated to incor- porate population projections of the United Nations Economic Commission for Latin America and the Caribbean (ECLAC). Figure 1a, shows that 63 percent of females reported they had been employed before the pandemic; the share among men was 84 percent. Thus, in the sample, females represented 45 percent of the employed population before the pandemic (figure 1, panel b). FIGURE 1, A. PRE-COVID-19 EMPLOYMENT RATE, BY GENDER 100% 16% 80% 37% 60% 40% 84% 63% 20% 0% Males Females DEFINITIONS Working before Covid-19 Not working before Covid-19 • The analysis is conducted at the regional level by pooling the data on the 13 countries. The countries included in the sample FIGURE 1, B. PRE-COVID-19 SHARE OF THE EMPLOYED are Argentina, Brazil, Bolivia, Colombia, Costa Rica, Chile, POPULATION THAT REPRESENTS EACH GENDER Dominican Republic, El Salvador, Guatemala, Honduras, Mexico, Peru, and Paraguay. Females account for 54 percent of the sample. Consistent with data on the entire region, including the countries outside the sample, men were more likely than women to be employed before the COVID-19 outbreak. • The analysis focuses on individuals who were employed before the pandemic unless otherwise noted. 45% 55% - Job loss or unemployment is defined as not working during the week before the survey. - A temporary job loss has occurred if a respondent reports a job loss but has a job to go back to. - A permanent job loss has occurred if a respondent reports a job loss and has no job to return to. Female Male - Total job losses (the sum of temporary and permanent job losses) have implications for labor income and welfare. Source: Data of HFPS, wave 1. 2 In the note, ‘LAC’ and ‘region’ refer to the population represented by the 13 countries in Latin America and the Caribbean included in the study. 3 The response rates in the second and third rounds were 71.6 percent and 68.8 percent, respectively. 4 The random digit dialing methodology surveys individuals by generating phone numbers at random. For technical details on representativeness, sampling design and weighting of the HFPS, see Flores Cruz (2020), COVID-19 High-Frequency Survey in Latin American Countries. Technical note: Sampling Design and Weighting. 2 What did the study find? The uneven burden of job losses The COVID-19 crisis has disproportionally a ected women, and The depth and breadth of the shock on women are observed this di erence has persisted. Approximately two months after across all countries in the sample. Figure 3 shows that, in all coun- governments had established quarantines in most countries, the tries in the sample, women are more likely than men to lose their first round of data collection took place. This round shows that 56 jobs between round 1 and round 3. The figure focuses on individu- percent of workers had lost their jobs either temporarily or perma- als who were employed before the pandemic and plots, for each nently. This is 44 percent higher than the corresponding rate country, the percentage of women (green) and men (yellow) who among men, 39 percent (figure 2). The employment gap, defined lost their jobs during the crisis either temporarily or permanently. here as the di erence between the female and male unemploy- The gray bar represents the employment gap. The figure is ment rates, stood at 16 percentage points. organized by country and survey round (first and third) and sorted by the level of the employment gap in the first round. Figure 3 also Data from the second and third rounds show that, as individuals suggests that not all countries have been a ected equally. In the started to go back to work, temporary but not permanent job first round at the onset of the crisis, the highest gender gaps were losses declined. However, the gap between men and women found in Honduras and Costa Rica, where women were 25 percent- remained virtually unchanged. By the third round of data collec- age points more likely than men to be unemployed. Bolivia and tion in August 2020, the di erence in total job losses between men Peru exhibited the narrowest gap (10 and 11 percentage points, and women was still 15 percentage points, and the permanent job respectively), but some of the highest overall unemployment rates loss rate a ected one woman in five. in the region. FIGURE 2.: FEMALE WORKERS ARE MORE LIKELY THAN MALE WORKERS TO FACE TEMPORARY AND PERMANENT JOB LOSSES 60% 56% 50% 49% 42% 40% 39% 35% Females 34% Males Percent 30% 27% 28% 26% 23% 20% 21% 21% 21% 19% 16% 13% 13% 12% 10% 0% R1 R2 R3 R1 R2 R3 R1 R2 R3 TOTAL JOB LOSSES TEMPORARY JOB LOSSES PERMANENT JOB LOSSES Source: Data of HFPS, waves 1-3. FIGURE 3. IN ALL COUNTRIES, FEMALES ARE MORE LIKELY THAN MALES TO FACE JOB LOSSES 100% 30 80% 25 Job losses (percent) (pecentage points) Employment gap 20 60% 15 40% 10 20% 5 0% 0 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 R1 R3 HONDURAS COSTA DOMINICAN GUATEMALA PARAGUAY COLOMBIA MEXICO EL CHILE ECUADOR ARGENTINA PERU BOLIVIA RICA REPUBLIC SALVADOR Gender gap (right) Male Female Source: Data of HFPS, waves 1-3. 3 What is driving employment losses? Women tend to work in sectors that rely more heavily on ing 60 percent of females before the crisis. This pattern suggests face-to-face interactions and are thus more vulnerable to social that gender gaps in labor outcomes are being exacerbated as a distancing measures. Indeed, the most female-intensive sectors result of the COVID-19 crisis. Job losses among females not only explain most of the observed job losses (figure 4). Although the widen economic gender gaps but also might exacerbate other impacts vary by sector, 56 percent of the job losses were concen- intrahousehold imbalances by reducing women’s empowerment, trated in trade, personal services, education, and hotels and lessening women’s intrahousehold bargaining power, and aggra- restaurants, according to the first wave of data collection. These vating intimate partner violence (Manser and Brown 1980; Perova are four of the five most highly female-intensive sectors, employ- and Reynolds 2017). FIGURE 4. THE MOST FEMALE-INTENSIVE SECTORS EXPLAIN A LARGE SHARE OF THE TOTAL JOB LOSSES TRADE 20% PERSONAL SERVICES Share of total job losses (wave 1) 15% 10% EDUCATION HOTELS AND RESTAURANTS MANUFACTURE CONSTRUCTION 5% TRANSPORTATION OTHER PUBLIC ADM HEALTH AGRICULTURE PROFESSIONAL ACT ELECTRICITY, WATER TURISM FINANCIAL SERVICES MINING 0% 0% 10% 20% 30% 40% 50% 60% 70% Share of employment that is female (before COVID-19) Source: Data of HFPS, wave 1. Note: Job loss is defined as not working during the week before the survey. Pre-pandemic wage employment, Internet access, and education which are associated with types of work that are less likely to be are found to increase the probability of remaining employed, but conducted remotely. some e ects are di erent among men and women (figure 5). Using a multivariate regression approach, the study explores the Additionally, the analysis finds that high educational attainment factors that correlate with the probability of remaining employed is associated with greater resilience to job loss.6 This result may through the pandemic (that is, across the three survey waves).5 be related to the type of work carried out by well educated work- Relative to self-employment, pre-pandemic wage employment is ers. Such workers may be involved in performing relatively more associated with a greater likelihood of remaining employed after nonroutine cognitive employment tasks. These results may also be the COVID-19 crisis, and this is the largest di erence found in the consistent with the higher productivity of more highly educated model. It is likely that pre-pandemic wage employment operates workers and the evidence that unemployment decreases as the through the increased job security that formal jobs o er and the years of education rise. ability of firms to adapt to the crisis relative to self-employees, who are more likely to own informal firms. Women are more likely than men to lose jobs because of increased childcare household needs given that social norms The study finds that internet access, which facilitates the possi- encourage women to become the primary family caregivers.7 bility of working remotely, is an important determinant of Although the presence of school-age children in the household is employment during the pandemic. The results show that access to not a factor associated with the probability of remaining employed the Internet raises the probability of remaining employed. Howev- two months after the onset of the COVID-19 crisis (May 2020) er, the e ect is not statistically significant among women. This may (figure 6), caregiving became a more relevant factor associated be explained by the fact that, before the COVID-19 crisis, women with job losses as the pandemic persisted.8 worked mostly in sectors that require face-to-face interactions, FIGURE 5. FACTORS ASSOCIATED WITH REMAINING EMPLOYED DURING THE CRISIS 0.30 0.20 0.10 0.00 Female Male -0.10 Source: Data of HFPS, waves 1–3. CHILD IN SCHOOL AGE HIGH EDUCATION ACCESS TO INTERNET WAGE EMPLOYEE Note: The bars represent the coe cient LEVEL of selected variables in a multivariate regression. The diagonal stripes indicate a lack of statistical significance (p > .1). 5 The analysis relies on a linear probability model estimated by ordinary least squares. The dependent variable is a dichotomous variable that indicates whether the individual remained employed across the three rounds of data collection. The analysis is restricted to individuals employed before the pandemic. The following explanatory variables are included: fixed e ects for country, industry, age in years, household composition (number of male household members, number of female household members, number of household members ages 65 or more), number of bedrooms (which, together with household composition, can be interpreted as a proxy for wealth), and type of job (self-employed, wage employee, family business, and other), and dummy variables for urban residence, high educational attainment (measured as completion of at least secondary education), access to the Internet, and whether a school-age child is in the household (defined as children ages 5–18). The model is estimated separately for males and females. 6 High educational attainment is defined as completion of at least secondary education. 7 Regression results are not statistically significant among men. 8 This is consistent with only 3 percent of women (and 1 percent of males) who report that childcare is the main reason explaining their job losses at the time of the first survey wave. 4 FIGURE 6. IMPACT OF A SCHOOL-AGE CHILD IN THE HOUSEHOLD, BY SURVEY ROUND 0.10 0.06 0.02 Female Male -0.02 Source: HFPS, waves 1–3. Note: The two bars on the left represent the coe cient of selected variables in a multivariate regression on the likelihood -0.06 of retaining a job during the first survey round. The two bars on the right represent the coe cient of selected variables in a multivariate regression on the likelihood of retaining a job over the -0.10 three waves of data collection. The diagonal stripes indicate a lack of ROUND 1 ROUNDS 1 TO 3 statistical significance (p > .1) First signs of recovery? Most sectors do not show signs of a recovery among female workers as of August 2020, and those that do mostly involve lower-quality jobs. Only 42 percent of individuals who had returned to work by August 2020 and who had been employed pre-COVID-19 are working in the same sectors in which they had worked before the pandemic. Essential sources of employment among women, such as trade, personal services, and education, are still at considerably low levels of operation. Also, the distribu- tion of women workers across types of jobs has changed. Before COVID-19, 61 percent of female workers were employed as paid employees, and 33 percent were self-employed (figure 7, panel a). By August 2020, 53 percent were paid employees, and 38 percent were self-employed (figure 7, panel b). FIGURE 7, A. TYPE OF EMPLOYMENT BEFORE COVID-19 FIGURE 7, B. TYPE OF EMPLOYMENT AFTER COVID-19 6% 9% 33% 53% 38% 61% Self-employed Employee Other Self-employed Employee Other Source: HFPS, waves 1–3. 5 Policy recommendations • The COVID-19 pandemic has the potential to widen existing • Ensuring the availability of disaggregated and representative gender gaps in economic opportunities. This note shows that data may help in the design of more specific and well targeted gender gaps in LAC are being worsened as a consequence of policies. Country-specific policies may be needed to maximize the pandemic. Understanding the gender implications of the impact of policy actions. HFPS data can be used at the coun- COVID-19 is key to informing the design of e ective policy try level to identify specific gender gaps and tailor policy responses. E ective responses and actions should focus on responses. The transmission channels identified in this note are creating the conditions and the incentives for women to work. focused on the economic conditions of women. However, the Given women´s greater involvement with the care of children COVID-19 crisis may also be influencing women’s agency and and the elderly, as well with household tasks, policies may incor- human endowments, such as through health care and education. porate elements aimed at restoring household dynamics and This possibility drives the need for gender disaggregated and incentives that encourage women to work. The findings of this representative data that can assist in measuring the impact of note suggest that policy should encompass recovery plans with the crisis on such dimensions. These dimensions include school a gender angle. dropouts and educational attainment, early childhood develop- ment among boys and girls, access to health care services, time • Implementing safety net programs that target the most a ect- spent doing household chores, potential changes in social norms ed women during the crisis: female-headed households, and attitudes, and the greater risk of violence against women informal and domestic workers who do not benefit from social and girls. protection coverage, and unemployed females; could help households mitigate the negative shock and continue to invest in children (Vandeninden et al., 2019; Rutkowski, 2020). More- over, cash transfer programs may also support self-employed women in restarting their businesses, considering that most female-intensive sectors are also more prone to be a ected negatively by social distancing measures (de Paz et al, 2020). Furthermore, social programs could include training and coach- ing components, incentives for formalization, business plan STAY CONNECTED competitions, and access to finance (Rutkowski and Bousquet 2019). Since cash transfers can be implemented with limit Visit the LACGIL website for more information. person-to-person interactions and at a low cost, they can be E-mail: lacgenderlab@worldbank.org e ective measures to implement in a COVID-19 context. ACKNOWLEDGMENTS • In the medium term, policies could aim to raise the resilience of This work has been funded by the Umbrella Facility for Gender self-employed and less well educated workers, especially Equality (UFGE), which is a multidonor trust fund administered women. This might include providing liquidity and other financial by the World Bank to advance gender equality and women’s support through lines of credit or financial services for wom- empowerment through experimentation and knowledge creation en-owned firms. These initiatives may be also combined by aimed at helping governments and the private sector focus training and mentoring provision, incentives for formalization, policies and programs on scalable solutions with sustainable business plan competitions (Rutkowski and Bousquet, 2019). outcomes. The UFGE is supported with generous contributions Additionally, policies might also facilitate access to formal from Australia, Canada, Denmark, Germany, Iceland, Latvia, the caregiving support and elderly care after economic activities Netherlands, Norway, Spain, Sweden, Switzerland, the United outside the household resume and promote measures to recog- Kingdom, the United States, and the Bill and Melinda Gates Foun- nize, reduce, and redistribute the burden of unpaid work within dation. households (Vaeza, 2020). The findings, interpretations, and conclusions expressed in this brief are entirely those of the authors. They do not necessarily reflect the views of the World Bank, its a liated organizations, the Executive Directors of the World Bank, or the governments they represent. This material should not be repro- duced or distributed without the UMBRELLA FACILITY FOR GENDER EQUALITY World Bank's prior consent. 1818 H. St NW Washington, DC 20433 6 REFERENCES de Paz, Carmen, Miriam Müller, Ana María Muñoz-Boudet, and Isis Gaddis. 2020. “Gender Dimensions of the COVID-19 Pandemic.” Policy Note (April 16), World Bank, Washington, DC. Manser, Marilyn, and Murray Brown. 1980. “Marriage and Household Decision-Making: A Bargaining Analysis.” International Economic Review 21 (1): 31–44. Perova, Elizaveta, and Sarah Anne Reynolds. 2017. “Women’s Police Stations and Intimate Partner Violence: Evidence from Brazil.” Social Science and Medicine 174 (February): 188–96. Reynolds, Sarah Anne, Elizaveta Perova, and Ian M. Schmutte. 2020. The Gender Wage Gap and Violence Against Women in Brazil. Work in progress. World Bank, Washington, DC. Rutkowski, Michal. 2020. “How Social Protection Can Help Countries Cope with COVID-19.” World Bank Blogs: Voices, April 15. https://blogs.world- bank.org/voices/how-social-protection-can-help-countries-cope-covid-19. Rutkowski Michal, and Franck Bousquet. 2019. “Social Protection: Protect- ing the Poor and Vulnerable during Crises.” World Bank Blogs: Develop- ment for Peace, December 12. https://blogs.worldbank.org/dev4peace/so- cial-protection-protecting-poor-and-vulnerable-during-crises. Vaeza, María Noel. 2020. “Gender and COVID-19 in Latin America and the Caribbean: Integrating Gender into the Response Framework.” Regional O ce for the Americas and the Caribbean, UN Women, Clayton, Panama City, Panama. Vandeninden, Frieda, Rebekka Grun, and Amina Semlali. 2019. The Way Forward for Social Safety Nets in Burkina Faso. Washington, DC: World Bank. World Bank. 2020. “Closing Gender Gaps in Latin America and the Caribbean.” Report 128525, World Bank, Washington, DC. Copley, A., Decker, A., Delavelle, F., Goldstein, M., O'Sullivan, M., & Papineni, S. (2020). COVID-19 Pandemic Through a Gender Lens. 7