key: cord-0913430-tcu5cbgi
authors: Jones, Nicola; Devonald, Megan; Dutton, Rebecca; Baird, Sarah; Yadete, Workneh; Gezahegne, Kiya
title: Disrupted education trajectories: Exploring the effects of Covid‐19 on adolescent learning and priorities for “building back better” education systems in Ethiopia
date: 2021-10-28
journal: Dev Policy Rev
DOI: 10.1111/dpr.12607
sha: c1451fe442d4f62dcec2113746761f3594d5f1c3
doc_id: 913430
cord_uid: tcu5cbgi

MOTIVATION: The Covid‐19 pandemic delivered an unprecedented shock to education systems globally, with school closures affecting 1.6 billion children. Education systems in LMICs are facing significant budget cuts further constraining capacities to adapt to Covid‐19 impacts. The need for evidence to inform policy dialogues about how best to mitigate impacts and support education systems to “build back better” is pressing. PURPOSE: In Ethiopia, schools reopened in October 2020 after a 7‐month pandemic‐related closure. Employing an adapted resilience systems analysis framework, this article focuses on the extent to which Ethiopia’s education system—which has in recent decades seen rapid progress in enrolment rates—has adapted to the impacts of the pandemic on adolescents’ education and learning, and has achieved this equitably. METHODS AND APPROACH: The article draws on mixed‐methods data from Ethiopia collected virtually with a pre‐existing cohort of 3,066 adolescents (1,683 girls & 1,383 boys) during the immediate onset of the pandemic (April–June 2020) and following the reopening of schools (November 2020–February 2021). Adolescent perspectives are complemented by 27 key informant interviews at community and district levels. FINDINGS: Findings highlight that rural adolescents, girls and adolescents with disabilities were less likely to access distance education during school closures due to connectivity challenges and discriminatory norms, and to subsequently re‐enrol. Implementation of adaptive measures, including hygiene guidance, smaller class sizes and catch‐up classes, has been highly uneven, and outreach to support re‐enrolment of socially marginalized adolescents very limited. POLICY IMPLICATIONS: For LMICs like Ethiopia to build back better post‐pandemic and stay on track to achieve Sustainable Development Goal 4’s commitment to inclusive and equitable quality education for all, scaled‐up investments in blended learning approaches, addressing the digital divide, and ensuring targeted outreach and social protection to support re‐enrolment of socially marginalized adolescents is critical.

When the Covid-19 pandemic led to school closures in March and April 2020, around 1.6 billion children globally were estimated to be out of school (around 91% of all students) (Miks & McIlwaine, 2020) . Since then, education systems have experienced the "twin shocks" of long-term school closures and the economic recession that has put additional

This article is protected by copyright. All rights reserved strain on both household and government education budgets (Saavedra, 2020) . Despite the additional resources needed for education systems to adapt to the impacts of Covid-19, a new report found that 65% of low-and lower-middle income countries (LLMICs) are experiencing educational budget cuts due to the pandemic (World Bank Group & UNESCO Global Education Monitoring Report, 2021) .

In light of the pandemic, UNESCO (2020) has estimated that around 24 million learners around the world were at risk of dropping out of school. As expected, the most vulnerable children are most at risk, such as those living in poverty, girls, and children with disabilities (Warren & Wagner, 2020) . There are also significant knowledge gaps in terms of the cumulative impacts of school closures and highly uneven access to distance education on children's and adolescents' educational trajectories (Porter & Pankhurst, 2021) . Addressing these lacunae, however, is of importance for policy and programming interventions in the context of the current pandemic, as well as for disrupted education services in the context of humanitarian shocks more broadly (Hallgarten, 2020) .

To explore these questions, this article focuses on Ethiopia, a low-income country that has seen relatively high investments (4.7% of GDP versus 4.3% sub-Saharan Africa average and a 4.1% LMIC average (World Bank, 2019a) ) and rapid progress in education enrolment over the last two decades (Devonald et al., 2020) . It draws on mixed-methods data from Ethiopia collected virtually with a pre-existing cohort of 3,066 adolescents (1,683 girls & 1,383 boys) in 2020 and 2021 during the immediate onset of the crisis (April-June) and then six months later (November-February) after schools had reopened. The article sheds light on the impacts of the national state of emergency called in response to the pandemic (which closed schools and universities between March and October 2020) on the country's education system. It also examines the extent to which the education system was able to mitigate impacts on adolescents' education and learning, including for the most socially marginalized adolescents in line with Sustainable Development Goal 4's emphasis on inclusive and equitable education for all. To do so, the article employs an adapted framework for resilience systems analysis (OECD, 2014) to understand well the Ethiopian education system is able to respond and adapt to The article is organized as follows: section 2 reviews the literature on past education shocks to understand some of the expected impact pathways of Covid-19, before giving an overview of the current situation in Ethiopia regarding education and Covid-19. Section 3 describes the research methodology used, and section 4 presents the findings drawing on data from the Gender and Adolescence: Global Evidence (GAGE) longitudinal study with adolescents and youth aged 12-20 years, as well as key informants at district and community levels. The article concludes by discussing the policy and programming implications of our findings, reflecting on what lessons can be learnt in terms of promoting resilience to education shocks in LMICs and developing more transformative education systems as part of international efforts to "build back better".

What can we learn from past education shocks?

Although there are many unknowns in the Covid-19 education recovery, evidence from past health crises (that also resulted in school closures) gives an indication of some of the potential challenges. One key example is the Ebola epidemic in West Africa from 2014-16, which closed schools for more than 33 weeks (United Nations Development Group (UNDG), 2015) . During this time, many countries de-prioritized education; girls' This article is protected by copyright. All rights reserved enrolment, which was already lower than that of boys before the Ebola crisis, did not return to pre-crisis levels (Malala Fund, 2020) . In one study in Sierra Leone, school closures resulted in a 16% decrease in school enrolment (Bandiera et al., 2018) . During the return to school in Liberia, the lack of Ebola protocols in some schools led to students enrolling in private schools or dropping out altogether (Santos & Novelli, 2017) . School closures have also heightened other risks, including teenage pregnancies (Dickson et al., 2018) , familial and sexual violence (Plan International, 2015; United Nations Development Programme (UNDP), 2015) and negative impacts on children's mental health (Hallgarten, 2020b; Baird et al., 2020; Banati et al., 2020) . Disrupted education can have prolonged effects on children's learning outcomes. Analysis of the impact of a 14-week school closure in Pakistan after the 2005 earthquakes found that four years on, children in affected areas were 1.5 years behind in their schooling. Andrabi et al. (2020) suggest this was due to a lack of adaptation to students' reduced learning levels when schools reopened.

However, research also suggests that both educational and life skills support can be beneficial in reducing some of these impacts. In Sierra Leone, a programme that provided life skills, sexual and reproductive health, and vocational learning during the Ebola crisis found that in villages participating in the programme, school enrolment fell by half the rate of that of control villages (Bandiera et al., 2018) . Additionally, an education programme in Sierra Leone--which provided radio programmes and listening groups on education, health, and life skills for children affected by the Ebola crisis--reported a 35% increase in enrolment in participating schools, an increase in learning outcomes, as well as an increase in self-confidence among students who participated in the programme (Dyson & Amara, 2017) .

What is emerging research saying about the impact of the Covid-19 shock on education in LMICs?

The short-term impacts of the pandemic on adolescents' access to learning have been farreaching: nearly 95% of the world's students were out of school in the first quarter of 2020, and by November that year, schools in 30 countries--educating a third of the world's children, mostly in LMICs--were still closed (or had reopened but subsequently closed again) (United Nations (UN), 2020; United Nations Children's Fund (UNICEF), 2020). With 500 million children unable to participate in remote learning because they lack access to digital devices, online connectivity or family/teacher support, there are growing concerns that tens of millions of young people will drop out of school permanently UN, 2020; UNICEF, 2020; Clark et al., 2020) . This gap in educational opportunities is likely to result in longer-term welfare and human capital losses, particularly for younger adolescents (Fuchs-Schündeln et al., 2020) and for girls, on account of entrenched inequalities and digital divides (Ellanki et al., 2021; Mathias et al., 2020) .

Although research on the impact of school closures on educational achievement is thin since the closures are ongoing, emerging evidence suggests that gaps in attainment are likely to be larger among children and adolescents in LMICs due to challenges in delivering online learning alternatives, such as low-quality streaming, limited interaction with instructors (stemming from among others connectivity challenges, limited training for

This article is protected by copyright. All rights reserved teachers on remote education, and limited motivation due to poor remuneration) and poorly adapted materials (Hadullo et al., 2018) . Recent research also suggests that already more disadvantaged students in low-income settings are faring worse during the closures (Asanov et al., 2021; Maldonado & De Witte, 2020) , especially adolescents with special learning needs (Akmal et al., 2020; Amin et al., 2020; Bellerose et al., 2020; UN, 2020) . Adolescents with disabilities face greater risks due to a lack of individualised learning in most remote learning platforms and limited capacities of caregivers to provide appropriate support at home (McClain-Nhlapo, 2020) . Moreover, gender differences are also apparent; girls in LMICs may fare worse than boys because they are often expected to spend time at home doing domestic work rather than studying, and are at greater risk of child marriage if they are out of school. However, boys also face specific risks. Involvement in child labour is higher among boys than girls across all ages and the number of children in child labour has risen by an estimated 8.9 million children during the pandemic (ILO &UNICEF, 2021) Research has also highlighted a return to traditional gender norms during the pandemic, with implications for learning and school re-enrolment. Young Lives data from Ethiopia shows that 70% of girls spent more time on household work compared to only 36% of young men (Porter & Pankhurst, 2021) . School engagement is also linked with psychosocial outcomes; an increase in school engagement is correlated with increased hope and social competence, which in turn increase well-being (Demirci, 2020) . By contrast, Banati et al. (2020) suggest that during school closures, adolescents in Côte d'Ivoire, Ethiopia, and Lebanon have all reported heightened rates of poor mental health linked to separation from school peers and uncertainty over their educational futures. Similarly, Baird et al. (2020a) underscore that in Bangladesh and Jordan increased social isolation during the pandemic has exacerbated adolescent girls' psychosocial problems, particularly older girls due to the intersecting impacts of both cultural and pandemic-related mobility restriction

The reopening of schools in Ethiopia following the Covid-19 national emergency

In Ethiopia, schools closed from 16 March 2020 and started to reopen seven months later, in October 2020. The World Bank outlined the Covid-19 policy response in three overlapping phases: (1) coping (involving school closures and remote learning); (2) managing continuity (involving a gradual return to school with a period of uncertainty); and (3) improving and accelerating (building stronger education systems) (World Bank, 2020a). During the "coping" phase, Ethiopia's Ministry of Education advised primary school children to follow distanced education via radio, with secondary school children advised to take up television (TV) lessons (Tibebu Tiruneh, n.d.) . However, access to remote learning can present significant challenges for children and adolescents in LMICs, and the Covid-19 pandemic has highlighted the widening digital divide 1 . In Ethiopia, while 42% of urban households have access to radio, only 29% of rural households have access. Similarly, 61% of urban households have access to TV, compared to only 11% of rural households (Central Statistical Agency of Ethiopia (CSA) & ICF, 2017).

This article is protected by copyright. All rights reserved Recent research has highlighted some of the challenges in access to remote learning in Ethiopia. Findings from a Young Lives phone survey show that by late 2020, 40% of 19-year-olds still in education had not engaged in any form of learning (including online learning) since the start of school closures (Porter & Pankhurst, 2021) . Similarly, GAGE research found that although in urban areas 86% of students reported doing something to continue schooling, this was mainly studying on their own initiative, and only 8% reported tuning into Ministry of Education TV or radio programmes (Baird et al., 2020b) . Schools and teachers felt unsupported at this time. Young Lives found that although most primary schools (89%) were offering some remote support during school closures, only a small percentage of head teachers felt confident in the school's ability to support teachers, and 41% had not received any guidance about how to support students during the closures (Outhred et al., 2020) . Adolescents with disabilities face significant challenges due to school closures and a lack of specialised remote services (Tibebu Tiruneh et al., 2021) . This has affected learning outcomes; a study of 7,500 children with disabilities in grades 1 to 4 in Ethiopia found that their average numeracy achievements after school reopened was significantly lower compared to children of the same grade levels before the pandemic (2018-2019) (Ibid.).

Ethiopia is currently in the second of the three phases, "managing continuity", and is seeing a gradual return to school. During the reopening of schools, 88% of children have re-enrolled at primary level, 75% at pre-primary and 66% at secondary (Bizuwork & Sewunet, 2020) . A self-reported adolescent survey in Ethiopia by the Center for Global Development (CGD) and Malala Fund found that 2% of children said they would be unlikely to return to school, with girls slightly more likely to say they would not return (Akmal et al., 2020) . Fear of Covid-19 was cited as the main reason why they would not return (mainly cited by girls), followed by work commitments to support their family (mainly cited by boys) (Ibid). School principals and teachers have highlighted a number of challenges during the return, such as limited infrastructure to implement Covid-19-related safety measures, risk of school dropout (with girls and rural students perceived as most likely to drop out), decreased student performance, and the importance of including catchup lessons (Yorke et al., 2020) . In addition to the "twin shock" to education systems of the Covid-19 pandemic and its impacts on household livelihoods, the ongoing political unrest represents a third shock likely to considerably affect access to education in some parts of the country (UNICEF, 2021).

To prevent transmission of Covid-19 during school closures in Ethiopia, the Ministry of Education developed an Environmental and Social Management Framework (ESMF) to mitigate the potential risks involved in its Covid-19 Education Emergency Response project, which has three components: (1) maintaining learning trajectories and safety during school closures; (2) supporting readiness for school reopening and mitigating learning loss; and (3) system-level resilience and project coordination (Ministry of Education, 2020b). The ESMF outlines a number of measures to mitigate the impacts of Covid-19 such as introducing social distancing and mask protocols in schools, ensuring widespread access to hand hygiene facilities, and deploying gender-based campaigns, including raising awareness of the importance of continuing with schooling for prevention of early marriage and teenage pregnancy (ibid.). However, it is unclear how these measures can be actioned in schools without significant investment in school infrastructure. For

This article is protected by copyright. All rights reserved example, one study reports that 74% of schools in urban areas and 87% of schools in rural areas do not have handwashing facilities (Kim & Rose, 2020) .

Before Covid-19 there was already increasing consensus as to a learning crisis in Ethiopia, manifested in slow learning progress, automatic promotion policies, socioeconomic disadvantage and lack of home support. This has led to a situation where only a very small percentage of 11-year-olds achieve basic literacy and numeracy skills, and by age 15, most rural students are three years below the expected learning outcome in maths (Oketch et al., 2020; Presler-Marshall et al., 2021) . Although access to education in Ethiopia has improved significantly in recent years, there is substantial regional variation, with lower enrolment rates in rural regions such as Afar, where only 20% of children are enrolled in upper-primary school (Devonald et al., 2020; Ministry of Education, 2020a) . Against this backdrop of a broader educational crisis--which is mirrored in many LMICs (World Bank, 2019b)--there are growing voices calling for education systems to use the return to schools as an opportunity to "build back better" by strengthening systems and infrastructure and providing more equitable access (World Bank, 2020a) . This includes building on remote learning platforms (with a focus on removing the digital divide) to reach the most excluded children and adolescents (Saavedra, 2020) .

This article draws on an adapted framework for resilience systems analysis (Figure 1) , which outlines three different levels of capacities that are instrumental in producing a resilient system: (1) absorptive capacity--the ability of a system to prepare for, mitigate or prevent the impacts of a shock through intentional protective action to ensure stability; (2) adaptive capacity--the ability to make changes to moderate any potential damage to the system and adjust to the changing situation; and (3) transformative capacity--the ability to make fundamental changes to transform the system so that it is more equitable and fit for purpose (OECD, 2014) . Resilience is defined here using the OECD definition: "the ability of households, communities and nations to absorb and recover from shocks, whilst positively adapting and transforming their structures and means for living in the face of long-term stresses, change and uncertainty" (Mitchell, 2013, p. 4) . In this paper, we primarily focus on the adaptive capability due to our emphasis on the challenges with returning to school post-lockdown. We have split this capacity into the two phases outlined by the World Bank; 1) coping (involving school closures and remote learning) and (2) managing continuity (involving a gradual return to school with a period of uncertainty).

In the Ethiopian context, the risk landscape consists of a triple shock--school closures and economic recession due to the Covid-19 pandemic, alongside pre-existing political insecurity and ethnic conflict--each of which can result in disruption to school systems. In terms of adaptive capacity, we consider various coping mechanisms to prevent the initial impacts of the crisis in the form of distance/remote education and outreach by school teachers. Further adaptive mechanisms are introduced once schools are reopened, including the introduction of catch-up lessons, social distancing and hygiene measures, and re-enrolment outreach. Finally, transformative capacity involves more fundamental changes to education services to deliver more equitable access and enhanced learning

This article is protected by copyright. All rights reserved outcomes through gender and social norm change, and improvements to school quality, governance and accountability. We reflect on this third dimension in the discussion section.

We use this framework to explore how well the education system in Ethiopia is able to respond and adapt to the Covid-19-related shocks, and the impacts on adolescents' education and learning. 

This article draws on mixed-methods data with 3,066 adolescents in Ethiopia as part of GAGE--a longitudinal research programme that aims to understand 'what works' to empower and enhance the capabilities of 20,000 adolescent boys and girls across six LMICs. In this article, we focus on data collected through phone surveys and qualitative in-depth interviews exploring adolescents' experiences during the pandemic.

Data was collected at two points during the pandemic with a subset of the 7,500 adolescents participating in the GAGE study in Ethiopia who were initially recruited in late

This article is protected by copyright. All rights reserved 2017/early 2018 (see Jones et al., 2018 ) (see also Figure 2 ). A first round of qualitative phone interviews were undertaken from April to June 2020 with 86 adolescent girls and 84 boys from four urban sites (Debre Tabor and Ebenat in Amhara region, Batu in Oromia region and Dire Dawa City Administration) and in six rural communities across South Gondar, Amhara region (2), East Hararghe, Oromia region (2) and Zone 5, Afar region (1) in order to capture diversity in the short-term educational response to Covid-19 and related education and learning outcomes (see Table 1 ). In line with the "leave no one behind" Agenda for Sustainable Development, the sample includes a subset of the most socially marginalized adolescents such as those with physical and visual disabilities, and married adolescents. These findings were triangulated with 27 key informant interviews with community leaders, teachers and school principals, and local government officials.

A second round of data collection starting in November 2020 followed the reopening of schools across the country. Quantitative data was collected with 3,066 adolescents from December 2020 to February 2021 via phone in Debre Tabor, Batu, and Dire Dawa as well as in the two rural areas (South Gondar and East Hararghe) (see Table  2 ). In November and December 2020, the same subset of adolescents interviewed during the first round of qualitative data collection were re-interviewed by phone, and 27 key informant interviews were also undertaken to understand changes in the education system response.

The GAGE survey includes adolescents in two age cohorts; at baseline, this involved younger adolescents aged 10-12 years and older adolescents aged 15-17. At the time of the Covid-19 survey, most adolescents were aged 13-20. The phone survey included questions about adolescents' experiences during the pandemic, with specific questions on education and learning during school closures (for more information on the survey, see Baird et al., 2021) . In the qualitative data collection, where possible, researchers carried out phone interviews in their local language with participants whom they had previously had contact

This article is protected by copyright. All rights reserved with through GAGE participation. For more information on the tools used during the phone interviews, see Małachowska et al., 2020. 

The quantitative analysis aims to describe the experience of adolescents who returned to school in Ethiopia after schools reopened following Covid-19-related closures, and those who attended school before school closures but did not return when schools reopened. For both groups, multivariate regression analysis was used to determine significant

This article is protected by copyright. All rights reserved differences between those living in rural and urban locations, between age cohorts, and between girls and boys. Additionally, we determined significant demographic differences between adolescents who returned to school and those who did not, using multivariate linear regression. Regression analysis controlled for the current age of the adolescent. All demographic comparisons between groups discussed in the text are significant at 95% confidence or higher. Sample weights were used to make results representative of the study areas. Statistical analyses were conducted using Stata15.1.

Qualitative interviews were transcribed and translated into English and coded with the qualitative software MAXQDA using an iterative thematic approach. Debriefing sessions were held to discuss initial findings, and these fed into the development of a thematic codebook to ensure that local specificities were included.

Ethical approval was obtained through the ODI Ethics Review and ethics committees within the Ethiopian regional health bureaus across all sites for both the qualitative and survey research. The GAGE research design and tools were approved by the George Washington University Committee on Human Research's Institutional Review Board (071721) and the Addis Ababa University College of Health Sciences Institutional Review Board (113/17/Ext). Written consent for those over 18 and assent (with parental consent) for those under 18 had already been secured during previous rounds of data collection; however, the Covid-19 phone interviews were preceded by conversations with participants and their caregivers in which consent and assent was secured once more. For adolescents without mobile phones we worked closely with community facilitators to undertake outreach about the phone interviews and to provide access to a phone in a Covid-19secure way (i.e. by providing disposal gloves, masks and hand sanitiser to participants). Due to the challenges of maintaining privacy while undertaking phone interviews, sensitive topics were only discussed if the respondent was comfortable doing so and researchers were certain that the respondent had full privacy. The research team also offered flexible times including evenings or calls made over several days so that the interviews could be fitted in around paid or domestic and care work responsibilities.

We now discuss our findings on the adaptive capacities of the Ethiopian education system to withstand the Covid-19 shock. We also reflect on their implications for policy and programming, alongside the potential for more transformative capacities within the Ethiopian education system in the Discussion section.

During the coping stage, intentional protective actions are put in place to deal with specific shocks. In the case of Covid-19, schools across Ethiopia closed in March 2020 as part of

This article is protected by copyright. All rights reserved measures under the rubric of the National State of Emergency aimed at controlling the spread of the virus.

In place of in-person formal education, the Ministry of Education introduced remote learning through TV and radio, and private schools also used social media platforms such as Telegram. The survey findings show that among students who were previously enrolled in formal schools (n=2678, 87%), 78% tried to continue learning during school closures. The percentage of students who reported engaging in any form of learning from home varied by subgroup: 84% of those living in urban areas reported continuing their learning at home using any method compared to just 73% of adolescents in rural areas. Younger adolescents were more engaged with learning, with 74% of younger adolescents in rural areas reporting engaging in any learning activities during school closures compared to 64% of older adolescents, and 92% of younger and 80% of older adolescents in urban areas. Furthermore, adolescents who returned after schools reopened were significantly more likely to report engaging in learning activities during school closures than those who had not yet returned to school at the time of the survey (79% vs. 65%)

In rural areas, adolescents in the qualitative research emphasized that radio and self-study were the most common approaches. For example, as a 14-year-old boy from East Hararghe explained, "Our teacher informed us to attend education via radio. I learnt it via radio." Others, particularly older adolescents, described continuing their education through self-study using textbooks. However, in some urban (mainly private) schools, students explained that they were given more hands-on support from teachers through provision of recorded lessons, reading materials and assignments. As a 14-year-old boy from Batu explained: "[The teachers] would record the video and our parents would go and fetch them. It was only relevant courses that they recorded, such as mathematics and science. They gave us more than 30 videos. Then, assignments would be given out of those courses covered by the video and we would do those assignments at home."

The quantitative findings highlight the challenges involved in accessing distance education provisioning. The most commonly reported method of home study by students was using their own textbooks (69%), with few adolescents reporting that their primary method of learning was tuning in to Ministry of Education TV or radio programming (3%) or working on assignments provided by their school (2%). Older adolescents in Dire Dawa-the largest urban centre in our Ethiopian sample--reported substantially greater access to online learning platforms (21%) compared to older adolescents in other urban locations (9.5%). This is most likely due to better overall digital connectivity in the city, but also better levels of support from teachers. While students faced numerous challenges in continuing to learn while school was closed, in rural areas 19% reported the lack of clear guidance as their biggest challenge compared to 7% in urban areas. Our qualitative interviews with adolescents and key informants highlighted that there was often no interaction with teachers during school closures, especially in rural areas, as many teachers had moved back to their place of origin during the pandemic (because they could not work

This article is protected by copyright. All rights reserved and were better able to receive and provide support to family members in their home communities). As a result, some rural students were unaware of the distanced education available. In urban areas, the absence of an established system for student-teacher interaction (besides TV learning) led to poor support from teachers.

Our qualitative findings further unpack the many challenges involved in students accessing remote learning. The overwhelming majority of students had very limited access to the internet, electricity, and technological devices such as TV and radio. Key informants highlighted a large digital divide between rural and urban areas, which is likely to further increase inequalities between students. As one key informant explained, "It exacerbates the previous education inequalities so much. For example, students that learn in Addis Ababa and rural regions are incomparable. All of them are going to sit the same national examination. Those from the big town have access to technology; students from rural areas do not have access to radio. How do you compare students from those areas?" This was especially apparent in pastoralist Afar region, where most adolescents lacked access to remote learning and distanced education. An 18-year-old boy explained, "No. I didn't participate in distance education or any other type of learning after the school closed due to the pandemic. None of my classmates are participating in distance education." Generally low levels of parental literacy, especially in rural and pastoralist communities, further limited adolescents' access to support with their studies at home.

Mobility restrictions for girls, largely due to fears over safety and gender norms that place a premium on girls' honour, , in urban areas can also impact access to remote learning. Adolescents in our qualitative sample highlighted challenges in obtaining remote learning materials or accessing Wi-Fi due to these restrictions. A 19-year-old girl commented: "I think it is easier for boys [to attend distanced learning] unless they are negligent, because they can easily go out and get Wi-Fi access, they can even have other options." Flexibility in time use can also impact adolescents' ability to engage with distance education. Gender norms, which define household work as the responsibility of women and girls, can reduce the time available for remote learning. A 15-year-old girl recounted her experience: "I had chores to do so I didn't have the time [to do distanced education]."

For urban adolescents who did access distance learning programming, the quality of the education provided was also a concern. Many described difficulties in following the online learning and limited support from teachers during school closures. A 15-year-old girl from Debre Tabor explained, "I probably watched the programme for a week or two. It wasn't as clear as the face-to-face lessons at the school and it runs very fast." Other issues concerned language, as some remote learning platforms only offered classes in English (which can exclude students with limited English language skills), and inconsistent TV programming schedules. The level of support given also seemed to differ by school, with large differences perceived between private and public schools.

Inclusive services were also a major challenge. In the Government of Ethiopia's Education Sector COVID 19-Preparedness and Response Plan there is no mention of children with disabilities. Adolescents with disabilities underscored that remote learning services had not been adapted to meet their needs, even among those attending inclusive education centres. As a result, many adolescents with visual and hearing disabilities were

This article is protected by copyright. All rights reserved unable to engage with distanced education programming. A 15-year-old boy noted: "I have a hearing disability and that is why I couldn't hear and I never follow the programme."

During the managing continuity stage, there is a gradual return to school with some uncertainty. After a seven-month break in in-person education, in October 2020 the Ethiopian government began to reopen schools. As part of this process, it introduced a range of measures to tackle the negative impacts of Covid-19-related school closures and to ensure the continuation of education during the "new normal" of the pandemic.

To help reduce the spread of Covid-19, the Ministry of Education introduced protective measures, including social distancing (through double-shift systems, building more classrooms and/or using temporary schools), the distribution of face masks and other personal protective equipment (PPE), and providing soap and water tanks. Our survey data show that the overwhelming majority of adolescents who returned to school reported wearing a mask most of the time when outside the house (90.9%) and teachers wearing masks at school (82%). However, our qualitative findings indicated that in practice, maskwearing was more uneven and becoming less common. Whereas in some schools, protection measures were enforced (with students being refused entry without masks), in other areas (particularly rural areas) students were no longer wearing masks at all in school. A 19-year-old boy from South Gondar explained, "The community start saying corona [virus] does not exist anymore and I think the students also start thinking that and also no one was questioning them when students came to school without masks. Because of that, they stop wearing masks." Furthermore, although some adolescents (mainly older students who were taking national exams) highlighted that they had been given masks, many were not, and had to buy their own masks, adding to their education expenses. A 16-year-old girl from South Gondar explained, "We are forced to wear a new face mask every day. It is expensive. It is a must to buy. They are telling us to use one mask for one day." Some respondents also highlighted a lack of handwashing facilities in schools or very limited provisioning, with our survey findings indicating that only 56% of students have access to soap and water for handwashing while at school. Teachers, mainly in rural areas, explained their frustration with the lack of resources, training and support they have been given to implement Covid-19 protective measures. A school principal from East Hararghe said "We did not get support so far. The woreda [district] give us a mask for those students [who sit national exams], no sanitizer, and no masks for the remaining students and no water sources. An NGO [non-governmental organisation] provides us with a water container, but we do not have water in it…. Also, there is no training given to teachers to help implement new measures related to 

This article is protected by copyright. All rights reserved

The government committed to constructing additional classrooms to allow for greater physical distancing among students. However, although the Ministry of Education set a maximum of 25 students per classroom and one student per desk, this was not feasible in many schools due to the limited resources available. Our qualitative findings also indicated that regional governments put varying emphasis on this protective measure, with more schools in Afar and Oromia having additional classrooms constructed compared to Amhara. A key informant from South Gondar (Amhara) explained that "When students started education, only 30 students were learning in a single classroom but after a while the number increased to 60 in order to correct the problem the schools were having with classrooms and teachers." These findings are supported by the quantitative data, with 34% of students in Oromia reporting new classroom construction compared to 15% in Amhara. We found no evidence, however, that additional classrooms had been added for students with disabilities (who in grades 1-4 participate in 'special needs education' classes before being integrated into mainstream classes from grade 5) or that specific attention had been paid to the construction of disability-accessible classes.

The education system changes have had some positive impacts on adolescents. Among those who had returned to school, 76% reported smaller class sizes. For some, smaller class sizes reduced opportunities for being distracted by other students. As a 17year-old girl from Debre Tabor explained, "We used to sit three to a desk but now it is two. This is good for learning because when we are many, we were distracted by playing, so this helps to focus on our education and there is less disturbance." Adolescents in Afar preferred attending education in the afternoon shift as it gave them time to travel to school (many have to travel long distances on foot), highlighting some of the potential benefits of a more flexible approach to school start times. However, in some areas (such as South Gondar), later start times made adolescent girls vulnerable to the risk of sexual violence when returning from school later, and as a result some girls had to leave class early, further undermining learning.

To help support students' transition back to school some schools provided catch-up sessions, though this appeared to vary significantly by region and by school. In South Gondar, 74% of students reported that their schools offered tutorials compared to just 24% of schools in East Hararghe. In South Gondar and some schools in Afar, students described receiving 45 days of catch-up lessons. However, in other areas, catch-up sessions were much shorter or non-existent, with automatic promotion to the next grade. A 14year-old girl from East Hararghe described the situation as follows: "This year we started grade 7 though we did not cover lessons from grade 6… They did not teach us the remaining topics from last year, we started directly from grade 7." Some students also highlighted a lack of time to fully grasp the catch-up lessons, with teachers going through the material very quickly. Among those who were offered catch-up lessons, 97% of students reported participating.

This article is protected by copyright. All rights reserved However, qualitative findings show that most respondents felt education quality had declined since schools reopened due to teachers rushing through lessons and missing certain subjects, as a 14-year-old boy from Batu explained: "Last year we were learning without rushing. This year it is so fast." Several adolescents noted that their ability to learn was being affected by a lack of available teachers to cover the increased number of classes (teacher absenteeism was a frequently mentioned problem across communities) and pressure to catch up on the education that had been missed.

Adolescents with disabilities faced significant challenges in keeping up with their lessons and teachers did not seem to be accommodating their needs. They also lacked social support from their peers due to social distancing measures. As an 18-year-old boy with a visual disability from South Gondar explained, "The teachers are teaching really fast to cover grade 7 topics. They don't have the time to give a tutor for grade 6 topics. But when they teach us quickly, it is difficult to understand, especially for me because I am blind. The students were ignoring me a lot. They just say hi but they completely stopped showing me the classrooms, bringing the chairs, and also they don't even ask me how I was doing… which was different from previous years, and that was making me feel sad. They were trying to prevent themselves from getting corona [virus] ." Adolescents with disabilities can feel left behind in their education and unable to get the appropriate support they require, which in turn can have spillover impacts on their psychosocial well-being.

During the return to school it is critical to provide outreach to disadvantaged students to re-enrol, to prevent them dropping out altogether. However, evidence of this happening was limited, and there was a lack of specific outreach to adolescents with disabilities and girls at risk of child marriage. We even found that some adolescents in rural areas were unaware schools had reopened. As an 18-year-old boy from Afar explained, "I didn't hear about the school reopening till I came here this morning. No one told me about the school reopening because I live in a remote area." In some areas, students in government schools received school materials to encourage re-enrolment, but this was rare.

Across our research sites, adolescents and key informants highlighted that a significant number of students had dropped out of school since the onset of Covid-19. Our survey results found that only three-quarters (75.9%) of adolescents who were in school before the onset of the pandemic had returned, with a significant gap in favour of boys returning--82.1% compared to 70.8% of girls. Of note is that 80.6% of those who did not return had reported (prior to schools reopening) that they wanted to return to school, but 23.2% had already anticipated not being able to return to school. This was in turn borne out in our findings about re-enrolment whereby, controlling for age, living in urban areas is associated with a 9% greater chance of returning to school compared to in rural areas.

Adolescents who did return to school were 1.5 years younger on average--14.9 years compared to 16.6 years--suggesting that older adolescents were under more pressure to drop out permanently, likely due to paid work, migration for work, care work or

This article is protected by copyright. All rights reserved marriage pressures as evidenced in our qualitative findings, which revealed that during school closures, marriage and work pressures were significant. While adolescents with disabilities were less likely to be in school before the pandemic, among those who were attending school before, the proportion of adolescents with disabilities who returned to school was the same as those without disabilities.

Child marriage was a common driver for school dropout, largely due to family pressures and deep-rooted norms that favour marriage over education for girls, as a school principal explained, "It has a big impact in our school. Sixteen girls got married from this school. Girls that are under the [legal] marital age [18 years in Ethiopia] got married. The community did not understand the value of education. They were considering that school may never open again, and so many girls got married." Our quantitative survey found that 9% of unmarried female adolescents identified pressure to marry as a concern for other females in their community, with half of those noting that this pressure had increased since March 2020. Our qualitative findings (especially in rural areas) indicate that adolescent girls tended to identify two or three classmates (on average) who had not returned to school. A 12-year-old girl in rural East Hararghe noted that most of the female students in her year group had dropped out, many due to marriage: "We were learning in grade 6 [before ]… There were 50 girls and 30 boys learning in my class… but now only 10 [girl] students are coming, many girls dropped out of school." In other cases, marriage pressures resulting in dropouts seemed to stem as much from broader social norms that value marriage over education for girls, as from direct family pressure. For example, an 18-year-old girl from Batu noted that "Many students returned back to school but many also got married… All of the girls who got married are in my age range… Instead of sitting idle without school, they decided to get married… I asked one of the girls what is the advantage of getting married and she argued that it is better to establish a household to get more freedom." In addition, girls face further pressure to take on household care work, and this can hinder their return to school. A 14-year-old girl explained: "I dropped out of school because of corona, then my parents told me to help with childcare. I am out of school for a long time. I want to go back to school."

Covid-19-related economic constraints placed additional pressures on households, and we found that some adolescents were unable to return to school because of deepening household poverty and inability to cover education-related costs. While our survey findings did not reveal differences in terms of mean household asset levels or participation in the PSNP public works scheme, our qualitative findings underscored that loss of household income as well as individual employment precluded some young people from being able to afford school supplies, and thus they were delaying re-enrolment. As an 18-year-old boy from East Hararghe explained, "I did not start school yet. I do not have money to buy school materials. Previously I had work but now that work has ended. So I do not have money."

Others noted that they were under pressure to continue paid work they had taken on during the pandemic to support their households. Our qualitative findings highlight that a number of adolescent girls had been compelled to migrate to nearby towns or cities to take on paid domestic work when schools closed. As a 14-year-old girl in Debre Tabor, whose family had arranged for her to take on paid domestic work so that she could support them during the economic downturn, explained: "I started working in other people's houses after the school closed in February [2020] Several respondents reported that a downturn in local employment opportunities had been compounded by ongoing challenges related to violence and political unrest, and that together these factors had hindered their return to school. When adolescents were asked whether violence or unrest had been a factor in their not returning to school, almost one-fifth agreed (19.7%). There were no significant differences between boys and girls or between urban and rural adolescents who reported violence as a barrier to re-enrolling, emphasizing the widespread nature of political unrest and insecurity across the study communities.

These findings on political unrest were also echoed in the qualitative data. For example, a 15-year-old girl living in a rural community in East Hararghe noted how, "We have been challenged by corona [virus] and the security problems for a long time… School closed for a long time because of corona, students forgot what they learned without going to school. … We were not going to places of work. Because of that we lack money and it exacerbates our poverty… I want to attend school and reach a good position. But I do not have materials to go to school. I do not have money. I am staying at home…."

A key component into a resilient system in Ethiopia has been to scale up national social protection programmes aimed not only at supporting vulnerable households' basic economic and food security needs, but also at helping cover education-related expenses to facilitate participation in distance education. In Ethiopia, the primary social protection response to Covid-19 has been through the flagship Productive Safety Net Programme (PSNP), which reaches 8 million rural households (Araya et al., 2020) , and the Urban PSNP, which reaches nearly 500,000 households (see Table 3 ). Recent analysis by the World Bank and the International Food Policy Research Institute (IFPRI) found that the PSNP offset food insecurity and reduced health and education expenditure during the pandemic (Abay et al., 2020) . However, other analysts highlighted that existing programmes have limited ability to respond to people who are newly vulnerable, especially in urban contexts (see Baird et al., 2020b) . Moreover, adolescents and youth are especially likely to be excluded from social assistance provisioning because targeting is at the household level, and adolescent or youth-headed households are typically not recognised and/or the PSNP rosters are not regularly updated to enable vulnerable adolescents to enrol and receive support in a timely way (Jones et al., forthcoming) .

Although the PSNP is one of the largest social protection programmes in sub-Saharan Africa, adolescents in our research sites underscored the dearth of social protection support available since Covid-19-related lockdowns were introduced. For example, an 18year-old internally displaced girl in East Hararghe noted, "Previously we were getting food aid biweekly. But now it has been two months since we got food aid… [Other internally displaced households] borrow from others and repay their debt when they get aid in future… When they get aid in the form of grain, they sell it and pay back their debt and buy what they need, like soap, with the rest of the money." Some adolescents with disabilities had been receiving school stipends before the pandemic but noted that these were suspended once schools closed. A 19-year-old boy with a visual disability from Debre Tabor explained how "Previously, they were giving us 350 birr [8 USD] per month. But it ceased after September [2020] . They have not told us the reason either." Due to a lack of regular social protection support, many young people with disabilities emphasized that they were compelled to depend on relatives and neighbours for charitable support and that this dependency was very stressful.

Our findings highlight that while the Ethiopian education system introduced multiple adaptive measures to facilitate continuity of learning for children and adolescents following the onset of the Covid-19 pandemic, there were significant inequities. Rural adolescents, adolescent girls, and adolescents with disabilities were not only less likely to have accessed distance education during school closures due to connectivity challenges, lack of adaptation for disability-related challenges and discriminatory gender norms, but are also less likely to have re-enrolled following the return to school. Adaptive measures undertaken to promote continuity of learning following the reopening of schools while mitigating the ongoing risk of Covid-19 infection, including hygiene measures, smaller class sizes and catch-up classes, have been implemented in a highly uneven manner within and

This article is protected by copyright. All rights reserved across regions. Our findings indicate that catch-up class provisioning in rural areas has been especially inadequate, given that most young people in rural communities were unable to access distance education and were out of school for over six months. Of most concern from an equity perspective, our data underscores that there has been a dearth of outreach to support the re-enrolment of the most socially marginalized adolescents, especially older girls at risk of child marriage and adolescents with disabilities.

The implications of our findings for education systems in LMICs like Ethiopia to build back better after the Covid-19 pandemic and develop the system resilience needed to achieve Sustainable Development Goal 4's aim of providing inclusive and equitable quality education for all, point to the urgency of more consistent implementation of Covid-19 hygiene and physical distancing measures as well as scaled-up investments in blended learning approaches (World Bank, 2020b) . To mitigate the impacts of potential future school closures, no-tech and low-tech approaches to distance and classroom-based education should be prioritized in recognition of adolescents' limited access to connectivity and devices. These in turn need to be buttressed by active teacher support (either virtual or physically distanced) to address the limited learning support that many adolescents received during school closures highlighted by our findings (see also Guglielmi et al., 2021, including for a discussion on the costing implications of such measures). It is imperative to pay particular attention to tackling the digital divide that exacerbates the educational disadvantages faced by rural adolescents, girls, and adolescents with disabilities in particular-in terms of physical and social access as well as capacities to use technology (see also Jones et al., 2021) . A Universal Design for Learning approach which entails making learning content available in various formats and on a variety of platforms to cater for a wider range of learning needs (Mcclain-Nhlapo et al., 2020) , such as including sign language interpreters for TV remote learning or using innovative approaches such as using solar powered MegaVoice devices to provide students with visual disabilities access to reading materials as pioneered by SENTIgray an organisation in Ethiopia. In addition, it is also important to invest in additional resources-such as constructing additional classrooms and recruiting more teachers-that will allow a more consistent approach to protective measures in schools. Investments in targeted outreach efforts with families and communities to highlight the longer-term risks of child work and child marriage and to support the re-enrolment of the most socially marginalized following school closures is also essential.

Finally, to bring about transformative effects for adolescent education and ensure education systems are better able to adapt to future shocks, such efforts must be accompanied by interventions within schools and communities to address discriminatory norms that undermine the educational futures of adolescents with disabilities and older adolescent girls in particular. Our findings highlight that they underpin the compounded challenges these cohorts face in crisis contexts like the Covid-19 pandemic. Investments in social protection for education that help to tackle poverty-related barriers to education (including cash transfers and school feeding), and that can be rapidly scaled in crisis contexts are also key. Given the high levels of poor learning outcomes in pre-pandemic Ethiopia (Presler-Marshall et al., 2021) , building strong school systems that deliver improved school quality-through improvements in school governance mechanisms and teacher capacities (especially with regard to remote learning)-is essential in ensuring

COVID-19 and food security in Ethiopia: Do social protection programs protect? Policy Research Working Paper 9475

Who is going back to school? A four-country rapid survey in Ethiopia

Bangladesh: COVID-19 knowledge, attitudes, practices & needs -Responses from three rounds of data collection among adolescent girls in districts with high rates of child marriage. COVID-19 Research & Evaluations brief

Human capital accumulation and disasters: Evidence from the Pakistan earthquake of 2005. RISE Working Paper, Series 20/039. Blavatnik School of Government at the University of Oxford

Ethiopia's Productive Safety Net Programme and adolescent wellbeing: evidence from the GAGE study

Remotelearning, time-use, and mental health of Ecuadorian high-school students during the COVID-19 quarantine

Social isolation and disrupted privacy: Impacts of COVID-19 on adolescent girls in humanitarian contexts

Adolescents' experiences of covid-19 and the public health response in urban Ethiopia. Gender and Adolescence: Global Evidence

Intersecting vulnerabilities: The impacts of COVID-19 on the psycho-emotional lives of young people in low-and middle-income countries

The economic lives of young women in the time of Ebola: Lessons from an empowerment program

Pre-pandemic influences on Kenyan girls' transitions to adulthood during COVID-19

Schools reopening restores normalcy to children amid lingering COVID-19 risks

Ethiopia demographic and health survey

A future for the world's children? A WHO-UNICEF-Lancet Commission

School engagement and well-being in adolescents: Mediating roles of hope and social competence

Leave no adolescent behind: Addressing inequalities in educational attainment in rural Ethiopia

Recovering from the Ebola virus disease: Rapid assessment of pregnant adolescent girls in Sierra Leone

Final evaluation report: Increasing access, retention and performance in primary education

Listening to Young Lives at work in India: Third call. Covid-19 phone survey headlines report

Young lives, interrupted: Short-term effects of the COVID-19 pandemic on adolescents in low-and middle-income countries

The long-term distributional and welfare effects of Covid-19 school closures. Working paper

Reimagining Girls' Education: Solutions to Keeping Girls Learning during Emergencies. UNICEF and ODI

Factors affecting asynchronous e-learning quality in developing countries university settings

evaluations-education-response-ebola International Labour Office (ILO) and United Nations Children's Fund (UNICEF). (2021)

Listening to young people's voices under covid-19. Exploring adolescents' experiences and priorities in Ethiopia under covid-19. Policy brief. Gender and Adolescence: Global Evidence

Intersecting barriers to adolescents' educational access during COVID-19: exploring the role of gender, disability and poverty

Rethinking social protection for vulnerable youth in Bangladesh, Ethiopia and Lebanon in the context of COVID-19

The threat of COVID-19 on Ethiopia's recent gains in preprimary education

Girls' education and COVID-19. What past shocks can teach us about mitigating the impacts of the pandemic. Malala Fund

The effect of school closures on standardised student test outcomes. Discussion Paper Series DPS20

GAGE virtual research toolkit: qualitative research

We've got through hard times before": Acute mental distress and coping among disadvantaged groups during COVID-19 lockdown in North India -A qualitative study

Pivoting to Inclusion: Leveraging Lessons from the COVID-19 Crisis for Learners with Disabilities

Keeping the world's children learning through COVID-19

Ministry of Education, Government of the Federal Democratic Republic of Ethiopia. Ministry of Education. (2020b). Environmental and Social Management Framework (ESMF) for Ethiopia COVID-19 Education Emergency Response Project

Risk and resilience: From good idea to good practice. OECD Development Co-operation Working Paper No

Teacher value-added using Young Lives Ethiopia school survey data -Diagnosing the learning crisis: Potential applications of value-added analysis

Interrupted education in Ethiopia: Support for students during COVID-19 school closures

Ebola: Beyond the health emergency. Summary of research into the consequences of the Ebola outbreak for children and communities in Liberia and Sierra Leone

A lost year of learning for girls in Ethiopia: Evidence from the Young Lives at work COVID-19 phone survey

My husband can go to work and I will go to my school

The education (negative) twin shocks, and the opportunity they bring. Education for Global Development

The effect of the Ebola crisis on the education system's contribution to post-conflict sustainable peacebuilding in Liberia

COVID-19 school closures may further widen the inequality gaps between the advantaged and the disadvantaged in Ethiopia. Cambridge Partnership for Education

Disability and Learning in Ethiopia: What Has Changed as a Result of the COVID-19 Pandemic? RISE

United Nations Children's Fund

Continuing crisis for children in Tigray, Ethiopia, amid reports of atrocities. Statement by UNICEF Executive Director Henrietta Fore

UNESCO COVID-19 education response: How many students are at risk of not returning to school? United Nations Educational, Scientific and Cultural Organization

Socio-economic impact of Ebola virus disease in West African countries: A call for national and regional containment, recovery and prevention

Save our education. Protect every child's right to learn in the COVID-19 response and recovery

Accepted Article This article is protected by copyright. All rights reserved World Bank (2019a) Current Education Expenditure

The COVID-19 pandemic: Shocks to education and policy responses

The World Bank's Education Response to Covid-19

The effect of Covid-19 on primary education in Ethiopia: Perspectives of school principals and teachers. Research and Policy Paper 20/10. University of Cambridge and Research for Equitable Access and Learning

The authors wish to thank Kathryn O'Neill for editorial support and Anna Andreoli for design support.

This article is protected by copyright. All rights reserved schools build back better after the pandemic. In this way, building resilient education systems will allow all adolescents to realize their right to equitable and quality education in times of stability as well as crisis.

Nicola Jones ODI, London, United Kingdom. Email: n.jones@odi.org.uk

The data will be archived in 2022 in the UK Data Archive service as part of the GAGE project's commitment to the use of its data as a public good. For further queries please contact gage@odi.org.uk.

This article is protected by copyright. All rights reserved

This article is protected by copyright. All rights reserved

This article is protected by copyright. All rights reserved Jones, N., Baird, S., & Lunin, L. (2018) . GAGE research design, sample and methodology. Gender and Adolescence: Global Evidence.

This article is protected by copyright. All rights reserved

This article is protected by copyright. All rights reserved https://blogs.worldbank.org/education/education-negative-twin-shocks-and-opportunitythey-bring