THE LABOR MARKET IMPACTS OF COVID-19 IN FOUR AFRICAN COUNTRIES (April-October 2020) Evidence from LSMS-Supported High‑Frequency Phone Surveys on COVID-19 Ivette Contreras-Gonzalez, Gbemisola Oseni Siwatu, Amparo Palacios-Lopez, Janneke Pieters, and Michael Weber As part of a global effort led by the World Bank to track the socio-economic impacts of COVID-19, the Living Standards Measurement Study (LSMS) team supports high-frequency phone surveys in Ethiopia, Malawi, Nigeria, and Uganda (among other countries). This brief focuses on the early impacts of COVID-19 on the labor market and their evolution from April to October 2020 using phone surveys in four African countries. SUMMARY OF FINDINGS • The results from high-frequency phone • The level of stringency of lockdown surveys in Ethiopia, Malawi, Nigeria, measures was different in each country. and Uganda, collected from April to Among the four countries, lockdown October 2020, confirm the devastating measures have been the least stringent labor market impacts of the pandemic. in Malawi (according to the Oxford’s These results also shed light on impacts Stringency Index), while Malawi also saw beyond job losses, such as a high share of the lowest share of respondents losing respondents experiencing reductions in their job after the outbreak. income from different sources. • The impact of the pandemic seems to • The first few months of the pandemic extend beyond the labor effects. Between saw the largest impact on labor markets. May and June, approximately 80% of the By May 2020, only four out of ten households in Malawi, Nigeria, and Uganda respondents in Nigeria, six out of ten in reported a decrease in their total income Malawi and Uganda and 8 out of 10 in after the onset of the COVID-19 outbreak. Ethiopia were working. Of all the countries In Ethiopia, 46% of households reported a analyzed, job loss due to the COVID-19 decrease in income. pandemic was highest in Nigeria. • All the countries experienced an increase • In all countries, urban jobs were hit in the share of respondents working in hardest. In urban Nigeria, 56% of people the agricultural sector in the months after with jobs before the pandemic stopped the COVID-19 outbreak. This change of working compared to 40% in rural Nigeria. economic activity could represent a coping This pattern is followed by Uganda, (29% mechanism of the households. vs 11%), Ethiopia (12% vs 6%), and Malawi • The labor market seems to be recovering (8% vs 6%). for household heads and their spouses in • The job impacts are heterogenous in all all countries. However, when considering four countries. Compared to men, women all household members in Nigeria and experienced more job losses in Ethiopia Malawi, the labor market did not appear to (10.1% for female respondents vs. 7.9% for have fully recovered from the shock of the male respondents) and in Uganda (19.1% pandemic in October 2020. vs 14.6%), but less in Nigeria and Malawi. 2 INTRODUCTION Labor markets have been a key This allows for an assessment of the labor transmission mechanism of the economic market impacts over time. Each phone effects of COVID-19. Tracking the survey collected labor information for one development of labor markets is therefore respondent per household (as opposed to paramount to better understand the face-to-face surveys where all household impacts of the pandemic. Documenting members are considered). how the crisis disrupts workers, firms, and labor income can make a valuable While the brief focuses on the evolution and contribution to designing and targeting distribution of the labor market impacts policies that help mitigate the impact of the of COVID-19 in these four countries, it also pandemic and its aftermath. incorporates information on the situation of the respondents before the pandemic. This brief highlights results from the Living It uses data from the Living Standards Standards Measurement Study (LSMS)- Measurement Study - Integrated Surveys on supported High-Frequency Phone Surveys Agriculture (LSMS-ISA) conducted before the (HFPS), conducted from April to October pandemic hit, from which the HFPS samples 2020 in Nigeria, Ethiopia, Malawi, and were drawn. More specifically, it uses data Uganda. Between these months, six rounds from the latest rounds conducted in these of monthly data collection were conducted LSMS-ISA countries: Ethiopia (2018/19), in Nigeria and Ethiopia, five in Malawi, Malawi (2019/20), Nigeria (2018/19), and and four in Uganda.1 The high frequency Uganda (2019/20). of the data collection was designed to closely follow the evolution of the crisis and subsequent policy responses. 1 Given the social distancing measures implemented by governments to limit the spread of COVID-19, traditional methods of data collection, such as face-to-face (F2F) surveys, have been difficult to implement. Phone surveys are a faster and cheaper alternative to safely eliciting information from individuals, households, firms, and communities; and, therefore, are well-suited for data collection purposes during the COVID-19 pandemic. The downsides are the need to shorten the length of questionnaires, compared to F2F surveys, and for the most part, restrict data collection to only one respondent per household reporting for him or herself. 3 LABOR CONDITIONS BEFORE THE COVID-19 PANDEMIC Prior to the onset of the pandemic, between In all four countries, a large share of the 58% and 70% of the population were working; active population is employed in the the share being highest in Uganda and lowest agricultural sector. Before the pandemic, in Ethiopia. In all four countries, women were Nigeria had the most diverse workforce, less likely to be economically active than men with four out of ten respondents working in (Figure 1), with the gender difference wider in the agricultural sector, and two out of ten in Nigeria and Ethiopia. Ethiopia has the highest the commerce sector. In Ethiopia, seven out share of the economically inactive population of ten respondents worked in agriculture with 31.3%, which is driven by the fact that (Figure 2). four out of ten women do not participate in the labor force (Figure 1). Figure 1. Economically active and inactive population by sex (% of population 15-64 years old) 100 31.3 20.9 40.7 26.4 23.9 28.7 24.0 18.5 29.1 21.5 18.3 24.4 80 60 40 58.3 71.8 46.1 68.5 71.6 65.6 61.3 67.9 55.0 70.5 74.0 67.3 20 0 All Male Female All Male Female All Male Female All Male Female Ethiopia (ESS 2018/2019) Malawi (IHPS 2019) Nigeria (GHS 2018/2019) Uganda (UNPS 2019) Currently working Temporarily absent Unemployed Economically inactive Source: Own calculations based on pre-covid19 surveys. Figure 2. Working respondents by economic sector (% of population 15-64 years old currently working) 100 5.4 11.7 9.0 14.9 14.6 80 5.1 15.6 8.2 4.0 23.8 14.5 60 4.9 4.1 9.5 40 8.0 78.0 63.8 40.7 57.9 20 0 Ethiopia (ESS 2018/2019) Malawi (IHPS 2019) Nigeria (GHS 2018/2019) Uganda (UNPS 2019) Agriculture Manufacturing & Utilities Construction, Transport & Professional Act. Commerce Public Administration Services Source: Own calculations based on pre-covid19 surveys. 4 EARLY IMPACTS OF Malawi’s lockdown measures have THE COVID-19 PANDEMIC – been the least stringent of all these APRIL TO JUNE 2020 countries (Thomas, et al., 2020). This country also experienced the lowest After the COVID-19 outbreak, lockdown share of respondents that lost their measures were implemented to curb job after the outbreak (conditional on contagion and these measures varied by holding a job before COVID-19 outbreak). country. The Oxford’s COVID-19 Government It can be observed that all countries had Response Stringency Index helps to keep 2 implemented lockdown measures by track of different measures implemented by the end of March. However, the levels of each country in a systematic way. The global stringency varied across countries, with stringency index documents government Malawi having the least stringent measures, responses to the pandemic since mid- as the government merely suggested that January 2020. It is a composite measure people stay home but did not implement based on nine response indicators, including a full lockdown, and public transportation school closures, workplace closures and was still running as usual throughout 2020. travel bans, rescaled to a value from 0 to Uganda had the strictest measures, while 100 (100 = strictest)3,3 4 . Nigeria started with stringent measures, but gradually relaxed those over time.4 Figure 3. Oxford’s COVID-19 Government Response Stringency Index (Index, where 100 means that the lockdown measures are the strictest) 100 80 60 40 20 0 Feb 2020 Mar 2020 Apr 2020 May 2020 Jun 2020 Jul 2020 Aug 2020 Sep 2020 Oct 2020 Ethiopia Malawi Nigeria Uganda Source: Our World in Data 2 The COVID-19 Government Response Stringency Index is 4 Thomas, H., Angrist, N., Cameron-Blake, E., Hallas, L., Kira, B., calculated by the Oxford COVID-19 Government Response Majumdar, S., . . . Webster, S. (2020). Variation in Government Tracker (OxCGRT), that collects publicly available information on Responses to COVID-19. Blavatnik School of Government. Retrieved 20 indicators of policies responses that governments have taken from https://www.bsg.ox.ac.uk/research/research-projects/ to respond to the pandemic such as school closures and travel coronavirus-government-response-tracker restrictions. It includes information of 180 countries. See more here: https://www.bsg.ox.ac.uk/research/research-projects/ covid-19-government-response-tracker 3 The nine indicators included are the following: (1) school closings, (2) workplaces closings, (3) cancelling public events, (4) Restrictions on gatherings, (5) closing of public transport, (6) stay at home requirements, (7) restrictions on internal movement, (8) international travel controls, and (9) presence of public information campaigns. More details about the methodology can be found here: https://www.bsg.ox.ac.uk/sites/ default/files/2020-12/BSG-WP-2020-032-v10.pdf 5 The first round of each HFPS shows that a than rural areas. Nigeria, being the most significant share of respondents lost their urbanized country of the four (with 51.2% of jobs for reasons related to COVID-19. its population concentrated in urban areas), In Nigeria, 44.6% of the respondents that was the most affected, 56% of urban workers were working before the outbreak reported employed before the pandemic lost their that they lost their job for reasons related to job compared to 40% of workers in rural the pandemic, while in Uganda, this share areas. Uganda and Ethiopia follow the same was 16.9%, in Ethiopia 8.4%, and in Malawi pattern of job loss (29% and 12% urban vs. 6.6% (See Figure 4). 11% and 6% rural, respectively). Malawi is the least urbanized country (with 17.2% of In all countries, urban areas experienced the population living in urban areas), and it higher job losses related to COVID-19 experienced the lowest share of job loss.5 Figure 4. Working situation after COVID-19 outbreak (% of respondents holding a job before COVID-19 outbreak) 100 5.2 5.1 5.2 2.1 8.4 6.6 16.9 80 44.6 60 40 86.4 88.4 50.2 81.0 20 0 Ethiopia (April/May) Malawi (May/June) Nigeria (April/May) Uganda (June) Working Job loss (related to COVID-19) Job loss (unrelated to COVID-19) Source: Own calculations based on HFPS surveys. Figure 5. Working situation after COVID-19 outbreak, urban and rural (% of respondents working or holding a job before COVID-19 outbreak) 100 6.0 8.1 6.2 11.1 12.5 80 29.1 39.2 56.3 60 40 80.1 89.9 88.4 88.4 39.8 54.9 68.6 87.0 20 0 Urban Rural Urban Rural Urban Rural Urban Rural Ethiopia (April/May) Malawi (May/June) Nigeria (April/May) Uganda (June) Working Job loss (related to COVID-19) Job loss (unrelated to COVID-19) Source: Own calculations based on HFPS surveys. 5 World Bank Indicators (2021). https://data.worldbank.org/ indicator/SP.RUR.TOTL.ZS 6 Compared to men, women experienced in Nigeria, where 16.6% of respondents more job losses in Ethiopia (10.1% who worked in April/May reported having for female respondents vs. 7.9% for a wage job as their main occupation, but male respondents) and Uganda (19.1% 7.4% reported that they were not working vs 14.6%), but less in Nigeria (39.5% vs as usual. In the other countries, a larger 46.0%) and Malawi (6.0% vs 6.9%). It is fraction of workers reported having a wage important to note that these HFPS surveys job in the first round, while the fraction represent one respondent per household, not working as usual was smaller: 3.6% with about 30 percent of respondents being in Ethiopia, 4.7% in Malawi, and 3.6% in female across the countries.6 Uganda (see Figure 7). Among respondents who kept working in In addition to the effect on the labor a wage job, a significant fraction reported market through job loss, there are other not working as usual (i.e., temporarily not factors that may affect households’ working or working fewer hours). This income such as reduced business activity impact of the pandemic was also stronger or a decrease in working hours. Figure 6. Working situation last week, by gender (% of respondents working or holding a job before COVID-19 outbreak) 100 7.9 6.9 6.0 14.6 10.1 19.1 80 39.5 46.0 60 40 87.4 83.3 88.8 87.6 48.4 56.8 83.9 78.3 20 0 Male Female Male Female Male Female Male Female Ethiopia (April/May) Malawi (May/June) Nigeria (April/May) Uganda (June) Working Job loss (related to COVID-19) Job loss (unrelated to COVID-19) Source: Own calculations based on HFPS surveys. Figure 7. Wage workers (as % of respondents working) 30 20 22.3 19.8 9.1 15.1 10 7.4 3.6 4.7 3.6 0 Ethiopia (April/May) Malawi (May/June) Nigeria (April/May) Uganda (June) Not working as usual Working as usual Source: Own calculations based on HFPS surveys. 6 More specifically: 25% of the respondents in Nigeria are female, while this number is 28% in Ethiopia, 40% in Malawi, and 49% in Uganda 7 During the first round, the respondents the working population in Ethiopia was were asked about changes on income from concentrated in agriculture, a sector that different sources, compared to the income was affected relatively less compared to the that they received in the previous year, to commerce and services sectors. The share of avoid a seasonality effect. households that experienced a reduction in farming income was also greater in Nigeria, During the first round, approximately Malawi, and Uganda, compared to Ethiopia. 80% of the households in Malawi, Nigeria, The same trend is reported for the share and Uganda reported a decrease in their of households with a reduction in wage total income following the COVID-19 income. However, the share of households outbreak. In Ethiopia, 46% of households that experienced a decrease in non-farming reported income loss (Figure 8). This could family businesses surpassed 80% in all the reflect that before the pandemic 78% of countries analyzed. Figure 8. Changes in income after outbreak Share of households with a loss Household farming, livestock, or fishing income of income after outbreak (as % of HHs with this source of income in the last 12 months) 100 100 80 80 41.4 72.6 72.6 60.1 60 60 40 40 46.1 81.8 80.0 76.4 20 20 56.8 22.3 18.2 34.5 0 0 1.9 5.0 9.3 5.4 Ethiopia Malawi Nigeria Uganda Ethiopia Malawi Nigeria Uganda (April/May) (May/June) (April/May) (June) (April/May) (May/June) (April/May) (June) Any income decreased after outbreak Increased No change Decreased Non-farm family business income Wage employment income (as % of HHs with this source of income in the last 12 months) (as % of HHs with this source of income in the last 12 months) 100 100 80 80 85.1 83.5 84.6 90.2 34.4 58.4 58.4 57.9 60 60 40 40 20 20 65.0 37.0 37.0 37.4 11.2 9.4 13.6 6.9 0 0 Ethiopia Malawi Nigeria Uganda Ethiopia Malawi Nigeria Uganda (April/May) (May/June) (April/May) (June) (April/May) (May/June) (April/May) (June) Increased No change Decreased Increased No change Decreased Source: Own calculations based on HFPS surveys. 8 THE EVOLUTION OF THE respondents working was 71% in June and COVID-19 PANDEMIC – rose to 85% in October, while in Uganda APRIL TO OCTOBER 2020 this indicator went from 70% in June to 89% in October (See Figure 9).7 However, it is Following the early impacts of the important to keep in mind that changes in pandemic, the working situation of the the working situation may be caused by respondents has gradually recovered. By seasonal fluctuations unrelated to COVID-19. May 2020, only four out of ten respondents in Nigeria were working, and a similar The duration of joblessness between share lost their job right around that time. April and October 2020 also varies By October 2020, the labor market had by country, reflecting some level of recovered, with 87% of all respondents instability in the labor markets. In Nigeria, back in the active labor force. In Ethiopia, only 28.2% of respondents in the panel 64% of respondents were working in May, remained working during all rounds, while and 87% in October. In Malawi, the share of 36.8% stopped working during one round Figure 9. Distribution of respondents (% of respondents by country and month) 100 80 60 40 64 85 86 87 88 87 71 71 73 79 85 42 71 82 86 85 87 70 86 89 89 20 0 May June July August September October May June July August May June July August September October September October May June July August September October Ethiopia Malawi Nigeria Uganda Working Job loss Not working Note: Months that are left blank are months when no data was collected. The share of respondents classified in the category “job loss” refers to the ones that were working in the previous round but stopped working during the round presented in the figure. In the case of the first round, the working situation is compared to mid-March 2020. Source: Own calculations based on HFPS surveys. 7 The figures in this section includes only the respondents that participated on HFPS in all the rounds from May to October. It includes 1,451 respondents in Nigeria for 6 rounds, 2,219 respondents in Ethiopia for 6 rounds, 1,108 respondents in Malawi for 5 rounds, and 1,669 respondents in Uganda for 4 rounds. 9 and 23.9% during two or more rounds (see As of October 2020, the share of Figure 10). In Malawi slightly less than half respondents with a job reached its pre- of the respondents was working during all pandemic level in all countries. Figure 11 rounds (46.9%), 31.9% stopped working in shows the working situation of respondents one round, and only 5.5% stopped working at two different points in time, during the in two or more rounds. In Ethiopia most pre-COVID-19 survey and in October 2020. It of the respondents were working in all should be highlighted that the pre-COVID-19 rounds (56.4%), 18.0% stopped working survey was collected in different months for one round, and 18.6% were in and out than the HFPS, and thus this indicator may of the labor force. In Uganda, 60.5% of also capture seasonal fluctuations unrelated respondents were working during all the to COVID-19 . rounds and only 2.1% stopped working for two or more rounds. Figure 10. Respondents by labor status in HFPS (% of respondents) 100 12.7 8.5 10.1 18.6 2.1 80 5.5 2.2 23.9 24.3 18.0 31.9 60 36.8 40 20 56.4 46.9 28.2 60.5 0 Ethiopia Malawi Nigeria Uganda Working in all rounds Stopped working in one round Stopped working in two or more rounds In and out of the labor force Not worked in any round Source: Own calculations based on HFPS surveys. Figure 11. Working respondents (% of respondents by country in pre-covid19 surveys and HFPS in October 2020) 100 80 60 61 87 83 85 85 87 85 89 40 20 0 ESS HFPS IHPS HFPS GHS HFPS UNPS HFPS (2018/2019) (October 2020) (2019) (October 2020) (2018/2019) (October 2020) (2019) (October 2020) Ethiopia Malawi Nigeria Uganda Note: This figure includes only the respondents that participated on HFPS in all the rounds from May to October. Source: Own calculations based on pre-covid19 and HFPS surveys. 10 DATA ON ALL WORKING-AGE In Nigeria and Malawi, when considering ADULTS – SPECIAL CASE OF all household members, the labor market NIGERIA AND MALAWI does not appear to have fully recovered from the shock of the pandemic. Figures Data for up to 6 household members (15-64 12 and 13 present the differences in the years old) was collected in September 2020 working situation of the respondents for Nigeria, and in October 2020 for Malawi, versus the other HH members for Nigeria in addition to the data collected for the and Malawi, respectively. The share of main respondents. Collecting data for more respondents with a job is higher than the household members allows for a broader share of other household members working analysis when looking at differences in the in Nigeria as well as in Malawi. Furthermore, labor market characteristics, such as gender when we compare the working situation of and education. Since the main survey respondents and household members in the respondents tend to be household heads HFPS to the working situation reported in the (predominantly male and older) and thus pre-COVID-19 survey, we observe that the differ in important ways from the overall share of individuals working is lower in the population of working-age adults, they may HFPS. This suggests that, in both countries, not fully reflect the employment situation the labor market conditions are worse than for the working-age population. what is inferred from the October HFPS data obtained from the respondents only (as summarized above in Figure 11). Figure 12. Respondents from HFPS and HH members Figure 13. Respondents from HFPS and HH members in Nigeria in Malawi (% of respondents or/and HH members as applicable) (% of respondents or/and HH members as applicable) 100 100 12.0 7.8 5.3 20.3 14.8 27.2 5.4 30.2 9.4 80 7.1 80 42.0 8.0 4.3 5.7 60 60 5.9 40 40 76.8 84.7 64.8 71.3 67.8 82.9 49.6 61.6 20 20 0 0 Both Only Other Both Both Only Other Both respondents + respondents HH members respondents + respondents + respondents HH members respondents + HH members HH members HH members HH members GHS Panel IHPS Wave 4PP September 2020 2019 October 2020 2018/2019 Currently Temporarily Unemployed Economically Currently Temporarily Unemployed Economically working absent inactive working absent inactive Source: Own calculations based on HFPS and pre-covid19 surveys Source: Own calculations based on HFPS and pre-covid19 surveys 11 The data on all working-age adults In Malawi, females are a long way from captured in Nigeria and Malawi also shed returning to levels reported before the further light on gender inequalities. In COVID-19 outbreak. As of October 2020, Nigeria, as of September 2020, 77.6% of the male respondents are 4.4 percentage male individuals (main respondents plus points less likely to work (66.5-70.9), which other HH members) were working, while in is similar to the gap for males in Nigeria. the pre-COVID-19 survey, this indicator was However, while in 2019, 64.8% of the 82.3%. Consequently, there is still a gap of female individuals were working, this went 4.7 percentage points with the pre-COVID-19 down to 55.6% in October 2020. Therefore, level. In the case of female individuals, this the female gap reaches the value of 9.2 gap is 6.3 percentage points (65.4%-71.7%). percentage points. Figure 14. Respondents from HFPS and HH members Figure 15. Respondents from HFPS and HH members in Nigeria by gender in Malawi by gender (% of respondents and HH members) (% of respondents or/and HH members) 100 100 9.4 14.4 11.2 18.2 24.4 30.1 26.1 34.3 6.9 6.1 80 1.4 5.1 80 11.8 0.7 8.1 2.2 4.1 2.3 0.4 8.3 4.6 5.1 3.6 60 60 6.4 82.3 71.7 77.6 65.4 70.9 64.8 66.5 55.6 40 40 20 20 0 0 Male Female Male Female Male Female Male Female GHS Panel Wave 4 PP HFPS R5 IHPS HFPS R5 2018/2019 September 2020 2019 October 2020 Currently Temporary Unemployed Economically Currently Temporary Unemployed Economically working absent inactive working absent inactive Note: This figure is restricted to those that are included in both sur- Note: This figure is restricted to those that are included in both survey veys (GHS-Panel Wave 4 PP 2018/2019 and Nigeria HFPS R5) (IHPS 2019 and Malawi HFPS R5) Source: Own calculations based on HFPS and pre-covid19 surveys Source: Own calculations based on HFPS and pre-covid19 surveys 12 CONCLUSIONS • As the pandemic is still ongoing, policies • In addition to the COVID-19 impact through need to focus on the vulnerable groups job loss, there are other factors that may that have been hit hardest. This brief affect households’ income such as reduced shows that working-age women are business activity or a decrease in working recovering slower in terms of their hours. Consequently, public policies focused economic participation in labor markets, on the recovery should also consider these and that the urban sector has been hit other factors beyond job losses. the hardest. • Data for all household members, available • There is substantial variation in job losses for Nigeria and Malawi, suggest the across the four countries. In Nigeria, labor market has not yet fully recovered. 44.6% of the respondents that were Furthermore, the crisis is entrenching working before the outbreak reported preexisting labor market gender inequality. that they lost their job for reasons related A similar analysis for the other countries is to the pandemic, while in Uganda, this needed to disentangle the differentiated share was 16.9%, in Ethiopia 8.4%, and in impacts that may be hidden when Malawi 6.6%. considering just one respondent. 13 THE LABOR MARKET IMPACTS OF COVID-19 IN FOUR AFRICAN COUNTRIES (APRIL-OCTOBER 2020) EVIDENCE FROM LSMS-SUPPORTED HIGH‑FREQUENCY PHONE SURVEYS ON COVID-19 JUNE 2021 WWW.WORLDBANK.ORG/LSMS