Learning loss from Covid in Sub-Saharan Africa: Evidence from Malawi

This page in:

Learning to count. Photo: Shutterstock
Learning to count. Photo: Shutterstock

Emerging evidence, including from a recent review of 29 studies from 17 countries, demonstrates that the Covid-19 pandemic and associated closures of schools have been correlated with substantial loss in learning. In South Africa, Grade 2 students lost between 57% and 70% of a year of learning compared to their pre-pandemic peers, and Grade 4 students lost 62% to 81% of a year of learning. In Uganda, where schools were closed for two full years, the share of young students who were able to read and comprehend a story did not reduce between 2018 and 2021, but the share who could not even read letters of the alphabet doubled in size. However, other studies suggest much lower levels of lost learning. A particular challenge in sub-Saharan Africa is the lack of availability of representative, longitudinal datasets on dynamics of students’ learning that can provide valuable insights on where the students are on the learning curve when they return to schools.

Lessons from Malawi

In Malawi, the Government closed all public schools for a total of 7 months. The Malawi Longitudinal School Survey (MLSS) implemented by the World Bank in partnership with Government, conducts regular learning assessments with a representative, longitudinal sample of students who were in Grade 4 in 2016-18. This rich data offers the sharpest available lens on learning trajectories for a nationally representative sample of students in sub-Saharan Africa. Employing data from the three rounds of the survey on the same students, twice prior to school closures, and once after reopening of schools, we produced the first comprehensive picture of students’ learning profiles before and after Covid-imposed school closures in a low-income country.  We estimated the learning gain for 100 days of additional schooling before Covid and compared learning levels in 2021 with those we would expect if the pandemic had not occurred.

We found that, on average across English, Math and Chichewa (the main language of instruction), students’ learning was 97 points (0.8 standard deviations, s.d.) below where we would project if the pandemic had not taken place (on a difficulty-adjusted scale with 500 as the mean). This is the equivalent to around two years of lost learning in total at pre-pandemic levels.

Did loss in learning stem only from the closure of schools, or also other factors?

Many studies have attributed the loss of learning from Covid explicitly to the lost teaching time from closure of schools. In Malawi, however, students have not only lost learning from the lack of physical schooling time due to school closures, but also suffered a one-off loss in foundational knowledge on concepts which they had previously mastered. Adjusting our estimates to account for the difference in time between the last pre-Covid assessment, the reopening of schools, and our post-Covid assessments, we have estimated the share of learning loss observed that stems from each dimension of lost learning.

Of the total 97 points of learning loss, 40 (0.3 s.d., almost one year’s learning) points occurred during the closure of schools. In other words, an individual student returned to schools following the reopening with, on average, 40 points less on scaled test-scores than they achieved in previous assessments. Assuming that students learned nothing during the physical closure of schools (the reach of remote learning interventions appears to have been limited), we would only expect the seven-month closure to translate into 14 points (0.12 s.d.) of lost learning. This suggests that not only did students not learn during the closure of schools, but there was a one-off reduction in knowledge from the Covid shock, amounting to an average of 26 points (0.22 s.d., around six months’ learning), which is not accounted for by the lost schooling time.

Once students returned to school, learning did not return to the pre-Covid pace

The learning loss from the closure of schools is less than half the total average learning loss from Covid; the remainder – 57 points (0.48 s.d., more than one year’s learning) – is the result of a slow-down in learning after schools reopened. Prior to the pandemic, students gained an average of 13.4 points of learning for every 100 days of schooling. Post-pandemic, factoring in the closure of schools, students gained only 6.9 points of learning for every 100 days of schooling, halving the pace of learning (Table 1). This suggests that schools have not successfully adjusted their teaching to support students to catch up lost learning. If this trend continues, we may see a growing gap in learning over time with students affected by Covid falling further behind their expected trajectories of learning. In Malawi, where learning levels were already low before the pandemic, this will have severe consequences for human capital development.

Figure 1
Figure 1. Learning levels and trajectories pre- and post-Covid closure

In Figure 1 the blue line shows how learning would have progressed if the pre-Covid learning trajectory had continued. The red line shows the actual learning trajectory post-Covid. The gap between the blue and red lines at the far left shows the one-off impact of the closure of schools (the gap between the red and green lines shows the explained difference in learning (14 points) as a result of the physical time lost from closure of schools. The gap between the green and blue lines on the y-axis shows the unexplained drop in learning (26 points) that may reflect students losing previous mastery over foundation concepts). The green line shows the learning trajectory post-Covid, adjusted for the 14-point expected loss of learning from the closure.

Table 1 summarizes the key findings. We regressed differences on differences in test scores over three points of time for a longitudinal sample of the same students. The overall learning loss is recovered from a dummy for post-Covid, adjusted for time. The overall gain from midline to endline (after Covid closure) was 73 points less than between baseline and midline; adjusted for time this is 97 points. The slowdown of learning is derived from the coefficient on the interaction of time exposure with the post-Covid dummy.

Table 1. Impact of Covid on learning

Overall learning loss  One-off Covid shock Slowdown in learning
97.3 40.4 56.9
Of which loss of schooling time: 14.1
Learning per 100 days Pre-Covid Post-Covid
13.4 6.9

The findings suggest that more research is needed to estimate not only the one-off learning loss from the closure of schools, but the change in pace of learning in schools after they reopened.  

Figure 2. Learning levels and trajectories pre- and post-Covid closure: English and Math
Figure 2. Learning levels and trajectories pre- and post-Covid closure: English and Math
The dynamics of learning loss appear to play out differently across subjects (Figure 2). In English and Chichewa, we see a similar pattern to that observed in total scores, with a one-off drop in learning associated with the closure of schools exacerbated by a post-closure slowdown in learning over time. However, in Math, we observe a modest (and statistically insignificant) one-off drop in learning due to the closure of schools, but a much larger slowdown in learning after schools reopened.

Lessons learned

Developing countries need to pivot their education systems to mitigate the slowdown by adapting teaching practices and curriculum. A business-as-usual approach which simply resumes pre-pandemic instruction is unlikely to respond to the needs of students following the reopening of schools. Instead, there is a need for a concerted focus on remedial learning, targeted to the new learning level of students, if countries are to restore pre-pandemic trajectories of learning – let alone catch up the learning that has been lost.


Ravinder Gera

Consultant, World Bank Group

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000