IMPACT OF COVID-19 ON LEARNING Evidence from Six Sub-Saharan African Countries Hai-Anh H. Dang, Gbemisola Oseni, Alberto Zezza, and Kseniya Abanokova This document has been prepared by Hai-Anh Dang (Senior economist; corresponding author), Gbemisola Oseni (Senior economist), Alberto Zezza (Program Manager), and Kseniya Abanokova (Junior research fellow). The first three authors are with Development Data Group, World Bank, the last author is with Higher School of Economics, National Research University, Russia. We would like to thank Raka Banerjee for excellent editorial help. And we would like to thank the UK Foreign, Commonwealth & Development Office for funding assistance. BACKGROUND This brief offers new analysis on the impacts of the COVID-19 pandemic on learning The COVID-19 pandemic has wreaked outcomes for six sub-Saharan African havoc upon global learning, with many countries. We analyze detailed household- countries facing severe school disruptions level data from several rounds of panel and closures. An emerging literature based phone surveys collected by the World Bank on household survey data points to the in Burkina Faso, Ethiopia, Malawi, Mali, pandemic as having exacerbated inequalities Nigeria, and Uganda. in education and learning in countries from Italy to Denmark, the United Kingdom, and These surveys were first implemented the United States. between late April and early June 2020, after school closures due to the pandemic. For example, Bacher-Hicks et al. (2021) 1 In each survey round, the surveyed find that high income areas in the United households were asked a set of core States with greater internet access and questions on topics such as knowledge of better equipped schools had substantially COVID and mitigation measures, access to larger increases in search intensity for online educational activities during school closures, learning resources than poorer areas. Jæger dynamics of employment, household and Blaabæ (2020)2 observe that in Denmark, income and livelihood, income loss and families with higher levels of socioeconomic coping strategies, and received assistance. status took out more digital children’s books, both before and during the COVID-19 To further enrich the analysis, we match lockdown, than poorer families. During these phone surveys with useful variables COVID-19, the baseline gradient increased: in the pre-COVID-19 Living Standard families with high socioeconomic status Measurement Study – Integrated Surveys consistently took out more digital children’s on Agriculture (LSMS-ISA) panel household books than their low socioeconomic status surveys, such as household consumption counterparts in each of the three phases of quintiles and the education levels of the COVID-19 lockdown. household heads.3 1 Bacher-Hicks, A., Goodman, J. and Mulhern, C. (2021). Inequality 3 For more information on the LSMS-ISA surveys, see in household adaptation to schooling shocks: COVID-induced http://www.worldbank.org/lsms. online learning engagement in real time. Journal of Public Economics, 193: 104345. 2 Jæger, M. M. and E. H. Blaabæk (2020). Inequality in learning opportunities during COVID-19: Evidence from library takeout. Research in Social Stratification and Mobility, 100524. 2 In particular, we investigate the following also show the absolute differences between research questions: this rate and the enrollment rate – the • What are the impacts of COVID-19 on share of households with children attending school enrollment and attendance? school – before schools were closed due to • Does the pandemic result in learning COVID-19 (red dots). inequalities in terms of access to different learning activities and contact with teachers? In other words, these differences represent • Do these impacts vary for different types the reduction in learning activities that can of households, such as rich versus poor be directly attributed to school closures households, and less educated parents caused by the pandemic. versus more educated parents? • To what extent do community Figure 1 ranks the frequency of the learning characteristics such as area of residence activities in a decreasing order and shows mitigate or exacerbate these impacts? that Uganda and Nigeria have the highest rates at approximately 60 percent each, EDUCATIONAL OUTCOMES followed by Mali (36 percent), Burkina Faso BEFORE AND AFTER COVID-19 (32 percent), Ethiopia (28 percent), and Malawi (20 percent). This shows that the We show in Figure 1 the participation rate negative impacts of the pandemic have of the share of households with children been substantial, with the differences engaged in any learning activity after schools before and after the onset of the were closed due to COVID-19 (blue bars) for pandemic ranging from more than the six countries. To compare with a similar 30 percentage points (Uganda) to almost educational outcome before COVID-19, we 80 percentage points (Malawi). Figure 1. Any learning activities after COVID-19 vs. Difference with school enrollment before COVID-19, by country 60 80 70 50 60 AFTER COVID-19 (%) 40 50 Gap (pp) 30 40 30 20 20 10 10 0 0 Uganda Nigeria Mali Burkina Faso Ethiopia Malawi After COVID-19 Gap 3 Figure 2 looks more closely at the specific On the other hand, Mali makes the most use learning activities and suggests that the of teacher assignments but hardly relies on most popular types of educational mobile learning apps and educational TV activities are teacher assignments or radio programs. If we consider all these and educational radio and television specific activities as equally important, we programs. However, countries exhibit can obtain the average ranking for the five significant heterogeneity in terms of these activities as a whole for each country.4 specific learning activities. Using this average ranking, Nigeria comes first (1.8), followed by Ethiopia In particular, when compared to the other (2.6), Burkina Faso (3), Mali and Uganda countries, Nigeria leads in using mobile (both have 4.4), and finally Malawi (4.8). learning apps and tutoring sessions, while Burkina Faso leads in using educational television or radio programs. Figure 2. Share of households with children participating in specific learning activities during post-COVID-19 school closures Panel A: Assignments by teacher Panel B: Mobile learning apps 40 40 Percent (%) Percent (%) 30 30 20 20 10 10 0 0 Mali Ethiopia Nigeria Burkina Uganda Malawi Nigeria Ethiopia Malawi Uganda Burkina Mali Faso Faso Panel C: Educational TV programs Panel D: Educational radio programs 40 40 Percent (%) Percent (%) 30 30 20 20 10 10 0 0 Burkina Nigeria Uganda Ethiopia Malawi Mali Burkina Nigeria Ethiopia Uganda Malawi Mali Faso Faso Panel E: Sessions with tutor 40 Percent (%) 30 20 10 0 Nigeria Ethiopia Mali Burkina Malawi Uganda Faso 4 For example, Figure 1 indicates that for the educational activity “Assignments by teacher” (panel A), Mali has a rank of 1, Ethiopia has a rank of 2, Nigeria has a rank of 3, and so on. The rankings for the other activities are similarly obtained, and the overall ranking for each country is obtained by averaging the rankings for the five activities. 4 We plot the number of contacts with To what extent are these differences driven teachers since schools were closed for by gaps in area of residence, education, or an average household in Figure 3, where wealth that existed before the pandemic? types of contacts with teachers include: no Put differently, did the pandemic create new contacts, SMS, online applications, email, types of learning inequalities or did it simply mail, telephone, WhatsApp, Facebook, and deepen existing inequalities? other methods. All countries are similar in terms of showing a low number of We examine these questions by considering contacts with teachers after the onset of the pre-COVID19 urban-rural, education, and the pandemic. wealth divides. In Figure 4 above, we plot the participation rate in any learning activity after Figure 3 shows that Nigeria leads with the pandemic (blue bars) and the absolute around 1.2 contacts with teachers, while differences between this rate and the school the corresponding figure for the remaining enrollment rate before the pandemic (red countries hovers around one. dots). Put differently, Figure 4 is Figure 1 disaggregated by urban-rural residence areas. Figure 3. Average number of contacts with teacher after COVID-19 1.5 Number of contacts 1 0.5 0 Nigeria Burkina Faso Mali Uganda Malawi Figure 4. Any learning activities after COVID-19 vs. difference with school enrollment before COVID-19, by residence areas Panel A: Rural Panel B: Urban 70 80 70 80 60 70 60 70 After COVID-19 (%) After COVID-19 (%) 50 60 50 60 Gap (pp) Gap (pp) 40 50 40 50 30 40 30 40 20 30 20 30 10 20 10 20 0 10 0 10 Uganda Nigeria Mali Burkina Ethiopia Malawi Nigeria Uganda Ethiopia Mali Burkina Malawi Faso Faso After COVID-19 Gap After COVID-19 Gap 5 While Figure 4 shows a similar country have the highest participation rate out of ranking as Figure 1, two key differences all urban households, but rural Ugandan stand out. First, urban households have households have the highest participation consistently higher participation rates rate out of all rural households. While Malawi post-COVID-19 in all six countries. The has the lowest rates across all countries gap between urban and rural households for both urban and rural households, the ranges from 11 percentage points (Mali) to remaining five countries change rankings 27 percentage points (Ethiopia). depending on whether urban or rural households are considered. Second, for the same country, urban households can perform better or worse We further plot the share of learning than rural households when compared with activities by pre-COVID-19 household similar households in different countries. consumption quintiles and household head For example, urban Nigerian households education level in Figures 5 and 6. Figure 5. Any learning activities after the pandemic, by pre-COVID-19 consumption quintile 80 60 Percent (%) 40 20 0 Nigeria Uganda Mali Burkina Faso Ethiopia Malawi Q1: Poorest Q2: Poorer Q3: Middle Q4: Richer Q5: Richest Figure 6. Any learning activities after the pandemic, by pre-COVID-19 household head education level 80 60 Percent (%) 40 20 0 Nigeria Uganda Mali Burkina Faso Ethiopia Malawi No formal education Primary Incomplete secondary Complete secondary Post-secondary 6 The overall patterns for the six countries overall participation rate compared to the indicate that richer households or latter two countries. These differences raise households with more educated heads particular concerns about the high level of have more learning activities for their inequality in learning activities for Malawi. children. The gaps between the richest and the poorest consumption quintiles range NEXT STEPS FOR POLICY from 13 percentage points (Malawi) to 31 percentage points (Uganda). As an early study analyzing the impacts of COVID-19 on learning activities in a multi- The gaps between households where heads country, sub-Saharan African setting, our have post-secondary education relative results suggest that policy interventions to those with no formal education range are needed to both counteract the negative from 17 percentage points (Nigeria) to effects of the COVID-19 pandemic on 48 percentage points (Malawi). Notably, the education for all students as well as to help participation rate in learning activities for them compensate for lost instruction time. the most educated households in Malawi is Support should be focused on households nearly 60 percent, almost six times higher that are poor, have less education, or than the corresponding figure for the least reside in rural areas. These vulnerable educated households. In fact, while the most groups have been hardest hit and are most educated households in Malawi have higher in need of protection against deepening participation rates than those of Burkina levels of learning inequalities caused by the Faso and Ethiopia, Malawi has a lower COVID-19 pandemic. 7 IMPACT OF COVID-19 ON LEARNING EVIDENCE FROM SIX SUB-SAHARAN AFRICAN COUNTRIES MAY 2021 www.worldbank.org/lsms