Public Disclosure Authorized JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS* Policy Note No.2 / May 2021 Public Disclosure Authorized Main findings • Women, young workers, and those with low levels of education and limited internet connectivity were dispro- portionately affected by the economic downturn caused by the pandemic-related lockdowns in Latin Amer- ica and the Caribbean. The gender and education differentials persisted throughout the initial six months of the pandemic, and they remain after considering individual, household, and job characteristics. • Most workers employed both before and after the COVID-19 outbreak remained in the same type of job and industry. The Covid-19 crisis also brought historically high levels of absences from work (those not working but with a job to return to), particularly for women and independent workers. Absences preceded job losses, with Public Disclosure Authorized almost one in five absent workers losing their employment after two months. • While countries continue to enhance their vaccination efforts, in the short run, active labor market policies accompanied by financial support and flexible work schedules are likely to remain necessary for the most vulnerable populations. C OVID-19 has upended people’s lives around the world, particularly in terms of employment. To contain the spread of the disease, most Latin American and Caribbean (LAC) governments implemented measures for social distancing, minimized non-essential economic activities, and suspended in-person education services soon after the pandemic began. Despite the expansion of social protection programs and other emergency interventions (e.g., tax reliefs and deferment of social security contributions) for mitigating COVID-19’s socio-economic impacts (De La Flor et Public Disclosure Authorized al., 2021), the region’s economy shrank by almost 7 percent in 2020 (World Bank, 2021a). Around 30 million became unemployed or left the labor force. Even among those who remained employed, working hours declined by 16 percent,1 while labor income in the region dropped by about 10 percent compared to 2019 (ILO, 2020a and 2021). * This note was prepared by Carolina Mejía-Mantilla, Sergio Olivieri, Ana Rivadeneira, Gabriel Lara Ibarra, and Javier Romero, under the guidance of Ximena del Carpio, with the financial support from the Latin American and Caribbean Regional Vice Presidency. This document was edited by Oliver Balch. 1 This is the equivalent of 39 million full-time jobs. (ILO, 2021) 1 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Given the importance of labor income in the region, there are several important questions about the effects of Covid-19 on the labor market. At the outset of the pandemic, 48 percent of Latin American and Caribbean workers2 stopped working and 16 percent lost their job.3 Yet, were job losses similar for all workers? Has the COVID-19 shock exacerbated unfavorable labor market conditions for vulnerable groups over time? What happened to those workers who remained employed throughout the early months of the pandemic? And, what lessons can be drawn from the experience? This note sheds light on these inquiries using household data from the LAC High-Frequency Phone Sur- veys – HFPS (Box 1) – which were collected between May and August of 2020 from 13 countries in the region. Box 1: The LAC High-Frequency Phone Survey The World Bank conducted a multi-round High-Frequency Phone Survey (HFPS) to assess the impact of the coronavirus pandemic on the welfare of Latin American and Caribbean households. Between March and June 2020, the HFPS collected nationally representative information for thirteen countries: Argenti- na, Bolivia, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Paraguay, and Peru. The HFPS provides information on the welfare loss that households and people experienced with respect to food insecurity, changes in employment, income loss, access to health services and education, and coping mechanisms. Additionally, it documents household responses to lock- down and safe-distancing measures imposed by their governments and catalogues the behaviors they un- dertook to mitigate the spread of the disease. These socio-economic impacts are available at the COVID-19 High-Frequency Monitoring Dashboard, which provides 96 harmonized indicators across 50 countries in Latin America and the world. For further information on the LAC HFPS, see World Bank (2021b). 2 In this note, ‘LAC’ or ‘region’ refers to the population of the 13 countries in Latin America and the Caribbean where the High Frequency Phone (HFPS) surveys were conducted. We use the traditional definition of employed population as those who either worked during the week before the survey or did not work but have a job to go back to (they were merely away or absent from work for reasons such as vacation, personal leave, illness, etc.). 3 The difference between the percentage of people who stopped working and those who lost their job is temporary absence: i.e., those workers who did not work the week before the survey but maintained their job attachment. These workers stopped working but did not lose their jobs. We focus on job loss first and shed some light into the group of absent workers in the last part of this note. 2 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Who were most affected by job losses? Women, young workers, and those with low levels of education and internet connectivity were disproportion- ately affected. Two months into the pandemic, around 20 percent of female workers lost the jobs that they had prior to the pandemic, eight percentage points higher than their male counterparts (Figure 1). Similarly, over a quarter of young workers (18 to 24 years old) were left without work, compared to 14 percent of workers aged 25 to 64 years old. Workers with tertiary education and those with internet access at home also experienced job losses at a lower rate. These differentiated impacts remain even after considering the variation in individual, household, and job characteristics. For instance, men were more likely to keep their jobs than women even when controlling for having school-aged children in the household. This suggests that women were more likely to lose their jobs because of in- creased childcare responsibilities, which confirms the results of Cucagna and Romero (2021) using the same dataset. Individuals with tertiary education were almost twice as likely to remain employed than those with primary education, while having internet increased the odds by 83 percent (Annex Table 1).4 Overall, job loss rates declined over time but some of the gaps remained. For instance, a four-point gender gap, a seven-point experience gap, and an eight-point internet access gap were still present in August 2020. Once more, the gender and education gaps observed in August 2020 are robust to the inclusion of controls (Annex Table 2). The former could lead to important setbacks in terms of gender equality in the labor market and with respect to other economic and intrahousehold imbalances (Cucagna and Romero, 2021). Likewise, the observed education gap reflects a pervasive inequality in which individuals with higher education have more stable jobs, work in sectors less affected by containment measures, and are employed in activities more amenable to telework (ECLAC, 2020; ILO 2020c and 2020d). 4 Internet access here could not only measure the ability to work from home – only between 7 and 16 percent of workers in LAC countries have jobs that would allow them to work from home (Delaporte & Werner, 2020) — but it could capture other favorable characteristics that make workers less vulnerable in the first place, such as formal employment. 3 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Figure 1. Evolution of Job Loss Rates by Individual Characteristics JOB LOSS BY SEX JOB LOSS BY AGE Female Male Age [18-24] Age [25-54] Age [55-64] 20.3% 26% 18.1% 12.0% 11.0% 14.5% 14.6% 12.3% 8.8% 8.6% 13.8% 8.1% 6.8% 8.6% 7.7% MAY - JUN JUN - JUL JUL - AUG MAY - JUN JUN - JUL JUL - AUG JOB LOSS BY EDUCATION LEVEL JOB LOSS BY INTERNET ACCESS Primary or less Secondary Post-sec. or more No internet With internet 22.4% 27.0% 18.6% 12.1% 11.9% 17.0% 15.7% 10.5% 14.0% 7.1% 5.0% 8.7% 7.6% MAY - JUN JUN - JUL JUL - AUG MAY - JUN JUN - JUL JUL - AUG Note: Working-age population (18 to 64 years old). Wave 1 was collected in May-June, wave 2 in June-July, and wave 3 in July-August 2020. For wave 1, job loss is calculated by comparing the present to the pre-pandemic situation of workers. For waves 2 and 3, we compare to waves 1 and 2, respectively. Source: LAC High-Frequency Phone Surveys (HFPS) 2020. 4 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Job losses are conditioned by employment characteristics such as formality, employment status, and econom- ic sector, as these convey different economic risks. In general, independent workers (i.e., self-employed and those with a family business), who are more likely to work informally, exhibited higher job-loss rates than wage employees (Figure 2). This is consistent with the finding that in Peru, formality (defined as having a written contract or a tax iden- tification number) was more important for job retention than the ability to work from home (Cueva et al., 2021). Sim- ilarly, 20 percent of those working in construction lost their jobs, while the rate reached 25 percent for those working in food preparation, accommodation, and tourism. Independent workers in these subsectors had a lower probability of keeping their employment, even after controlling for other relevant variables using econometric analyses (Annex Tables 1 and 2). Figure 2. Evolution of Job Loss Rates by Characteristics of Previous Work, 2020 JOB LOSS BY STATUS IN EMPLOYMENT Employee Self-employed/Family business 17.1% 15.1% 11.3% 10.2% 7.8% 7.1% M A Y-J U N J U N -J U L J U L -A U G JOB LOSS BY INDUSTRY May-Jun Jun-Jul Jul-Aug 25.3% 19.9% 19.7% 16.6% 16.5% 15.4% 15.0% 14.2% 13.2% 12.3% 10.9% 11.0% 10.5% 11.1% 10.2% 10.1% 9.6% 8.9% 8.8% 8.7% 7.6% 6.9% 6.9% 5.2% 4.4% 3.7% 2.5% Agriculture & mining Manufacturing Trade Construction Transportation Accomodation, food Education & health Public Other services & tourism administration Note: Working-age population (18 to 64 years old). Wave 1 was collected in May-June, wave 2 in June-July, and wave 3 in July-August 2020. For wave 1, job loss is calculated by comparing the present to the pre-pandemic situation of workers. For waves 2 and 3, we compare to waves 1 and 2, respectively. Source: LAC High-Frequency Phone Surveys (HFPS) 2020. 5 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 What happened to those who remained employed? Most workers employed both before and after the COVID-19 outbreak remained in the same type of job and industry. In the region, an average of 5 percent of the employed population changed their pre-pandemic job (with the precise figure varying from 3 to 9 percent in the individual countries).5 However, transitions between types of employment were relatively small (Table 1), contrary to what was widely hypothesized. By August 2020, around 8 percent of wage workers before the pandemic had started a business on their own, while nearly 5 percent made the reverse transition. This resulted in a net increase of 3 percentage points in self-employment. Yet, 87.5 percent of workers observed in August 2020 retained the status they had before the pandemic. Similarly, transitions between economic sectors were relatively small between March and May 2020. By July 2020, however, this transition rate had increased to 15 percent. Notably, 6 percent of the employed population left the services sector for the construction and manufacturing sector. Table 1. Transitions between status in employment Employment May-Jun 2020 Employment Jul-Aug 2020 Employment Self-employed Wage Employment Self-employed Wage Total Total Pre-pandemic / Fam. Business Employee Pre-pandemic / Fam. Business Employee Self-employed Self-employed 35.1 0.4 35.4 31.3 4.8 36.2 /Fam. Business /Fam. Business Wage Employee 1.9 62.6 64.6 Wage Employee 7.7 56.1 63.8 Total 37.0 63.0 100.0 Total 39.0 61.0 100.0 Source: LAC High-Frequency Phone Surveys (HFPS) 2020. 5 Using data from HFPS in the COVID-19 Monitoring Global Dashboard, De La Flor et al. (2021) consider job changes to be between 4 and 14 percent in LAC countries, as a proportion of only those who actually worked (excluding those absent from work). In this note, absentees are included as part of the employed population. 6 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 At the outset of the pandemic, one in four workers did not work during the reference week despite having a job to which they expected to return. Individuals absent from work are traditionally considered as part of the employed population. However, the COVID-19 pandemic increased the number of absences from work as well as the level of uncertainty as to whether and when people could return to their jobs.6 While the proportion of absent workers de- creased from 25 percent in May 2020 to 17 percent in August 2020, over two-thirds of those absent in May continued to be absent three months later (Figure 4). Women and independent workers reported higher absence rates, which underscores their vulnerable position in the labor market. Similarly, a large proportion of those working in the service and tourism sectors also reported being absent: trade (43 percent); tourism, accommodation, and food preparation (57 percent); and transportation (42 percent). Figure 4. Composition of the Employment to Population Ratio, 2020 72.6% 72.7% 70.0% 20.8% 17.3% 25.0% 51.8% 55.4% 45.0% May-Jun Jun-Jul Jul-Aug At Work Absent from Work Employment rate Source: LAC High-Frequency Phone Surveys (HFPS) 2020. 6 The International Labor Organization addressed this issue in their Technical Note, “Monitoring labour markets amid lockdowns to contain the COVID-19 virus: Essential labour force survey content and treatment of special groups”. Available at: https://rtc-cea.cepal.org/sites/default/files/rtc_connected/files/ILO-techni- cal%20note-april-2020_0.pdf 7 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Figure 5. Transition of absent workers in May/June 2020 PEOPLE ABSENT FROM WORK Throughout the pandemic, job losses were pre- MAY-JUN 2020 ceded by absences from work. Only half of absent Did not have workers went back to work between May and August a job in Jul/Aug 2020 (Figure 5). Of those who did not return to work, meanwhile, almost 4 out of 10 did not do so because 18.4% they had lost their job. This evidence underlines the At work uncertainty associated with this status (i.e., absence 50.1% in Jul/Aug from work). In circumstances of high and pervasive 31.6% economic uncertainty, absent workers should there- Absent from work in Jul/Aug fore be considered as part of the target population for income-protection policies and even for training and job-placement interventions. Source: LAC High-Frequency Phone Surveys (HFPS), 2020. Lessons learned High levels of uncertainty characterized the labor markets of the region during Covid-19. Millions of jobs were lost and absence rates were unprecedentedly high. Governments made significant efforts to buffer the impacts of the crisis by activating social protection policies and by targeting workers not usually covered by social insurance schemes, such as informal workers (De La Flor et al., 2020; IADB, 2020). In the short term, financial support and flexible work sched- ules will likely continue to be necessary for the most vulnerable. Over the medium run, it is important that countries complement social assistance from formal wage employment with other alternative sources (World Bank, 2019; World Bank 2021c). Equally important is the promotion of programs to reskill or upskill the labor force so as to reintegrate workers (World Bank, 2019; Beylis et al., 2020). Female workers in the region were amongst the most negatively impacted, reversing the achievements of re- cent decades. Our results show that women’s employment was disproportionally affected. This was partly because of an increase in household and caregiving responsibilities and partly due to a decline in the activities in which they were predominantly employed (especially informal activities in the services sector).7 The longer it takes women to re-enter the labor market, the larger the consequences on their future working conditions and in overall productivity. 7 ILO (2020c), Cucagna and Romero (2021) and Khamis et al. (2021) reach similar conclusions. 8 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Their return can be accelerated by active labor market policies in the short run and by medium-term policies aimed at increasing access to care services for children and the elderly. Similarly, maintaining and improving teleworking (ideally, with flexible work schedules for both men and women) could play an important role in balancing professional and family responsibilities (ILO, 2020d). After the initial impact of the pandemic on young workers, the slow recovery represents a challenge. At the same time, we observe that job losses were disproportionately high for under-25s. Because of the pronounced eco- nomic downturn and an expected slow recovery, there are fewer entry-level vacancies and fewer hires in general. Additionally, firms are more likely to re-hire or hire more experienced workers (ECLAC and ILO, 2020b). As with women, policies must actively support their placement and improve their employability to avoid affecting their employment and income trajectories in the future. Subsidies to reduce the cost of hiring young workers during the first stages of economic recovery have proven to be effective (ECLAC and ILO, 2020b). The pandemic demonstrated the urgency to expand access to digital technologies and revealed the impor- tance of digital skills. Our results show that connectivity and the possibility of teleworking protected some of the jobs in the region. Moreover, digital services and ecommerce have surged during the pandemic. However, more than half of the region’s population cannot take advantage of the digital economy. Countries must promote digital skills among the workforce. Similarly, they must expand connectivity and increase affordable access to the internet by investing in infrastructure for digital transformation. 9 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 References Beylis, G., Fattal-Jaef, R., Sinha, R., Morris, M. and A. R. Sebastian. 2020. Going Viral: COVID-19 and the Accelerated Transformation of Jobs in Latin America and the Caribbean. World Bank Latin American and Caribbean Studies. Washington, DC: World Bank. Available at: https://openknowledge. worldbank.org/handle/10986/34413. License: CC BY 3.0 IGO Cucagna, E. and J. Romero. 2021. The Gendered Impacts of COVID-19 on Labor Markets in Latin America and the Ca- ribbean. World Bank, Washington, DC. Available at: https://openknowledge.worldbank.org /handle/10986/35191. License: CC BY 3.0 IGO. Cueva, R., Del Carpio, X. and H. Winkler. 2021. The Impacts of COVID-19 on Informal Labor Markets: Evidence from Peru. Mimeo. De La Flor, L., Mujica, I., Fontenez, M., Newhouse, D., Rodriguez Alas, C., Sabharwal, G. and M. Weber. 2021. Taking Stock of COVID-19 Labor Policy Responses in Developing Countries. Jobs Watch COVID-19. World Bank, Washington, DC. Available at: https://openknowledge.worldbank.org/ handle/10986/35331. License: CC BY 3.0 IGO. Delaporte, I. and W. Peña. 2020. Working From Home Under COVID-19: Who Is Affected? Evidence From Latin American and Caribbean Countries. GLO Discussion Paper, No. 528. Global Labor Organization (GLO), Essen. Available at: http://hdl.handle.net/10419/216106 Economic Commission for Latin America and the Caribbean (ECLAC) and International Labour Organization (ILO). 2020. Employment trends in an unprecedented crisis: policy challenges. Employment Situation in Latin America and the Caribbean, No. 23 (LC/TS.2020/128). Santiago. Available at: https://www.ilo.org/santiago/publicaciones/ coyuntura-laboral-am%C3%A9rica-latina-caribe/WCMS_760452/lang--en/index.htm Inter-American Development Bank. 2020. Social Policies in Response to Coronavirus #2, Labor markets of Latin Amer- ica and the Caribbean in the face of the impact of COVID-19. Available at: https://publications.iadb.org/publica- tions/english/document/Labor-markets-of-Latin-America-and-the-Caribbean-in-The-Face-of-The-Impact-of- COVID-19.pdf International Labour Organization (ILO). 2021. ILO Monitor: COVID-19 and the world of work. Seventh edition. 25 Jan- uary 2021. Available at: https://www.ilo.org/global/topics/coronavirus/impacts-and-responses/WCMS_767028/ lang--en/index.htm -------------------------------. 2020 (a). ILO Monitor: COVID-19 and the world of work. Sixth edition. 23 September 2020. Available at: https://www.ilo.org/global/topics/coronavirus/impacts-and-responses/WCMS_755910/lang- -en/index.htm 10 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 -------------------------------. 2020 (b). Monitoring labour markets amid lockdowns to contain the COVID-19 virus: Essential labour force survey content and treatment of special groups. Technical Note, April 2020. Available at: https://rtc-cea.cepal.org/sites/default/files/ rtc_connected/files/ILO-technical%20note-april-2020_0.pdf -------------------------------. 2020 (c). Labour Overview for Latin America and the Caribbean [Executive Summary]. ILO Regional Office for Latin America and the Caribbean. Available at: https://www.ilo.org/lima/publicaciones/WCMS_764633/lang--en/index.htm -------------------------------. 2020 (d). The COVID-19 response: Getting gender equality right for a better future for women at work. ILO Policy Brief, May 2020. Available at: https://www.ilo.org/global/topics/coronavirus/WCMS_744685/lang--en/index.htm Khamis, M., Prinz, D., Newhouse, D., Palacios-Lopez, A., Pape, U. and M. Weber. 2021. The Early Labor Market Impacts of COVID-19 in Devel- oping Countries: Evidence from High-Frequency Phone Surveys. Jobs Working Paper; No. 58. World Bank, Washington, DC. Available at: https://openknowledge.worldbank.org/handle/10986/35044. License: CC BY 3.0 IGO World Bank. 2021 (a). Global Economic Prospects. January 2021. Washington, DC: World Bank. doi: 10.1596/978-1-4648-1612-3. License: Creative Commons Attribution CC BY 3.0 IGO. -------------------------------. 2021 (b). LAC COVID-19 High-Frequency Phone Surveys 2020, Technical note. Poverty and Equity Global Practice, LAC. Mimeo. -------------------------------. 2021 (c). Lives or Livelihoods? The Costs of Staying Healthy. Poverty and Equity Global Practice, LAC. Mimeo. -------------------------------. 2019. World Development Report 2019: The Changing Nature of Work. Washington, DC.: World Bank. doi: 10.1596/978-1-4648-1328-3. License: Creative Commons Attribution CC BY 3.0 IGO 11 www.worldbank.org /BancoMundial @BancoMundialLAC JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Appendix Table 1: Logistic regression odds ratios for keeping employment between pre-pandemic and May/June 2020 Pr(keep_job) (1) (2) (3) (4) (5) (6) Male 1.4272 * 1.3946 * 1.4272 * 1.4140 * 1.3933 * 1.4278 * (0.2031) (0.2006) (0.2289) (0.2312) (0.2283) (0.2371) Child 0.6854 ** 0.6903 ** 0.6455 *** 0.6452 *** 0.6343 *** 0.6712 ** (0.0803) (0.0828) (0.0835) (0.0855) (0.0849) (0.0914) Male#Child 1.5193 * 1.5378 * 1.6441 * 1.6638 * 1.6982 ** 1.6804 * (0.2720) (0.2789) (0.3244) (0.3333) (0.3414) (0.3423) Age 1.1484 *** 1.1438 *** 1.1653 *** 1.1621 *** 1.1629 *** 1.1628 *** (0.0263) (0.0269) (0.0298) (0.0307) (0.0307) (0.0306) Age2 0.9985 *** 0.9986 *** 0.9983 *** 0.9984 *** 0.9983 *** 0.9983 *** (0.0003) (0.0003) (0.0003) (0.0003) (0.0003) (0.0003) Edu_second 1.3919 ** 1.3426 * 1.3085 1.2643 1.2603 1.2375 (0.1781) (0.1777) (0.1888) (0.1885) (0.1864) (0.1898) Edu_third 2.3590 *** 2.2508 *** 1.9751 *** 1.8863 *** 1.8667 *** 2.0905 *** (0.3263) (0.3234) (0.3109) (0.3057) (0.3011) (0.3492) Internet 1.8839 *** 1.9005 *** 1.7711 *** 1.8045 *** 1.8300 *** 1.7021 *** (0.2023) (0.2071) (0.2177) (0.2266) (0.2310) (0.2247) Wage_employee 1.0708 0.8419 3.8093 ** 3.9721 ** (0.0960) (0.0973) (1.7734) (1.9401) Agriculture & mining 0.7741 0.7366 3.2501 * 4.2411 ** (0.1846) (0.1862) (1.7050) (2.3428) Manufacturing 0.8577 0.8586 1.7811 2.1351 (0.2073) (0.2117) (1.1350) (1.4879) Construction 0.4926 * 0.4653 ** 2.1885 2.3011 (0.1422) (0.1374) (1.3125) (1.4664) Trade, transp. & comm. 0.8151 0.7823 2.9061 * 3.3690 * (0.1423) (0.1461) (1.3546) (1.6553) Public admin. & utilities 0.9840 1.0087 1.6683 1.7576 (0.2690) (0.2823) (1.2489) (1.2498) Accom., food, other services 0.6221 ** 0.6153 ** 3.2063 * 3.3241 * (0.1001) (0.1043) (1.5118) (1.6395) Wage#Agric 0.1874 ** 0.1466 ** (0.1101) (0.0891) Wage#Manuf 0.4652 0.3491 (0.3200) (0.2602) Wage#Const 0.1821 * 0.1782 * (0.1238) (0.1266) Wage#Trade 0.2502 ** 0.2185 ** (0.1252) (0.1144) Wage#PublicAdmin 0.5646 0.4920 (0.4541) (0.3786) Wage#ServicesOther 0.1573 *** 0.1547 *** (0.0786) (0.0804) _cons 0.1091 *** 0.1148 *** 0.1381 *** 0.1752 ** 0.0434 *** 0.0178 ***   (0.0509)   (0.0552)   (0.0754)   (0.1008)   (0.0304)   (0.0131)   Country-level fixed effects No No No No No Yes N 9615 9400 8303 8088 8088 8088 Chi 2 221.9727 210.2265 188.1927 175.5991 191.1116 296.9252 p 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Pseudo R2 0.0595   0.0577   0.0642   0.0625   0.0674   0.0963   Exponentiated coefficients; Standard errors in parentheses. * p<0.05 ** p<0.01 *** p<0.001 12 JOBS INTERRUPTED: THE EFFECTS OF COVID-19 IN THE LAC LABOR MARKETS Policy Note No.2 / May 2021 Appendix Table 2: Logistic regression odds ratios for keeping employment between June/July and July/August 2020 Pr(keep_job) (1) (2) (3) (4) (5) (6) Male 1.7261 * 1.7189 * 1.9583 * 1.9632 * 1.9810 * 1.9686 * (0.4640) (0.4736) (0.5716) (0.5864) (0.5976) (0.5900) Child 0.7421 0.7518 0.7605 0.7666 0.7523 0.7773 (0.1594) (0.1659) (0.1795) (0.1859) (0.1835) (0.1904) Male#Child 1.0268 0.9758 1.0294 0.9657 0.9620 0.9553 (0.3526) (0.3404) (0.3830) (0.3650) (0.3656) (0.3531) Age 1.1006 * 1.0821 1.0990 1.0768 1.0732 1.0702 (0.0481) (0.0477) (0.0548) (0.0541) (0.0536) (0.0535) Age2 0.9990 0.9992 0.9990 0.9992 0.9993 0.9993 (0.0005) (0.0005) (0.0006) (0.0006) (0.0006) (0.0006) Edu_second 1.1115 1.1337 1.1598 1.2001 1.2403 1.2329 (0.2309) (0.2357) (0.2618) (0.2715) (0.2807) (0.2844) Edu_third 2.3980 *** 2.5016 *** 2.0362 * 2.1128 ** 2.1077 ** 2.1197 ** (0.5928) (0.6272) (0.5698) (0.5875) (0.5884) (0.6043) Internet 1.7615 ** 1.6978 * 1.3357 1.2615 1.2582 1.1617 (0.3846) (0.3762) (0.3285) (0.3244) (0.3271) (0.3174) Wage_employee 1.2633 1.1980 5.7126 * 6.4426 * (0.2103) (0.2224) (4.2658) (4.8480) Agriculture & mining 0.4342 0.3233 ** 0.7518 0.9163 (0.1941) (0.1288) (0.5723) (0.7018) Manufacturing 0.4091 0.3082 * 2.8583 2.6170 (0.2040) (0.1475) (2.8341) (2.6003) Construction 0.3800 0.2505 ** 4.7729 5.2078 (0.2152) (0.1332) (4.5563) (5.0730) Trade, transport. & comm. 0.4318 * 0.3154 *** 1.2247 1.3737 (0.1676) (0.1075) (0.8853) (1.0042) Public admin. & utilities 0.4315 0.2879 * 0.6541 0.7919 (0.2301) (0.1472) (0.7806) (0.9126) Accom., food, other services 0.5756 0.4032 ** 1.8389 1.9534 (0.2156) (0.1316) (1.3371) (1.4472) Wage#Agric 0.7114 0.5502 (0.6330) (0.4858) Wage#Manuf 0.0717 * 0.0727 * (0.0790) (0.0802) Wage#Const 0.0347 ** 0.0325 ** (0.0384) (0.0361) Wage#Trade 0.1987 * 0.1842 * (0.1631) (0.1512) Wage#PublicAdmin 0.3647 0.2853 (0.4773) (0.3641) Wage#ServicesOther 0.1618 * 0.1599 * (0.1309) (0.1314) _cons 0.5525 *** 0.7024 1.3858 2.6968 0.7449 0.5126 (0.4845)   (0.6366) (1.5498) (2.8934) (0.8728) (0.6167) Country-level fixed effects No No No No No Yes N 5432 5339 4391 4302 4302 4302 Chi 2 47.4631 47.6021 59.1239 70.7262 97.5483 143.8856 p 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Pseudo R 2 0.0501   0.0516   0.0511   0.0563   0.0678   0.0878   Exponentiated coefficients; Standard errors in parentheses. * p<0.05 ** p<0.01 *** p<0.001 13