COVID-19 learning losses: early grade reading in South Africa

Image: Kimberly Farmer on Unsplash

A year after the World Health Organization declared the outbreak of COVID-19 a pandemic, around half of the world’s students were still experiencing complete or partial school closures (UNESCO 2021). The length of school closures varies by the level of economic development, with children from poorer countries missing substantially more classroom instruction time than children from high income countries (United Nations 2020). Every day that children are out of school, they risk falling further behind with potential long-term consequences for their future well-being. With prolonged closure, learning losses can be expected to exceed what is suggested by actual days of school lost as children forget skills acquired before the closure (Gustaffason and Nuga 2020). School closures may also increase the risk of dropout for vulnerable children (Smith 2020).

Not only is the average number of schooling days lost higher in developing countries, the ability to respond to school closures and support remote learning depends crucially on home learning environments, parental ability to support learning, connectivity, and digital skills; all attributes along which there is a great divide between richer and poorer countries.  Moreover, developing countries were facing a learning crisis prior to the pandemic with children already battling to keep up with curricula demands and classes characterised by high variability in learning levels (Pritchett 2020). School closures will likely amplify that variation and schools that are able to provide effective remediation and pivot to more targeted individualised learning will be in a better position to mitigate the impact. Disparities along all of these dimensions will only exacerbate learning inequality between high and low income countries.

Governments, international organisations and education policy researchers urgently need accurate information on the costs of school closure if they are to optimally manage responses to the ongoing pandemic and design recovery strategies. Evidence on actual learning losses due to COVID-19 is only starting to emerge and comes almost exclusively from very high income countries, where school closures were fairly short and schooling systems are highly effective. This article contributes to the literature on COVID-19 learning losses by providing one of the first pieces of evidence on the impact of COVID-19 on learning from a developing country. We compare early grade reading gains for different cohorts of learners at the same schools in South Africa.

Eastern Cape early grade reading study

The study is set in the three urban and peri-urban districts in the Eastern Cape (Nelson Mandela Bay, Sarah Baartman and Buffalo City) and includes 57 no-fee primary schools with isiXhosa as the language of learning and teaching. The study has followed two cohorts of learners over three waves of data collection (see Table 1). We are able to exploit this longitudinal data to compare progress during the grade 2 year in 2020 (cohort 1) against progress in 2019 (cohort 2).

Table 1. Learner cohorts by data collection wave

Fieldwork wave Cohort 1 Cohort 2
Wave 1 (Jan/Feb 2019) Grade 1 Term 1


Grade 2 Term 1


Wave 2 (Oct/Nov 2019) Grade 1 Term 4


Grade 2 Term 4


Wave 3 (Feb/Mar 2021) Grade 3 Term 1


Grade 4 Term 1


To measure the impact of the pandemic on learning we need a plausible counterfactual against which to compare learner outcomes. Cohort 2 provides such a credible counterfactual for cohort 1 for the following reasons. First, the two groups of learners were randomly selected from two adjacent grades (grades 1 and 2) in the same schools at the same time. There is no reason to believe that there would be systematic differences between grade 1 and 2 learners in the same school. Indeed, both groups are shown to have very similar baseline characteristics. Second, we see evidence of parallel learning trends in the two cohorts pre-pandemic. Finally, the cohort means are very close at similar points (e.g. end of grade 1 and start of grade 2) pre-pandemic. All of these support the argument that the cohorts are inherently similar.

Schooling losses

The average grade 2 learner in our sample was able to come to school in 2020 for only about 40 percent of the days of the previous school years.

Moreover, the usual school day was slightly shorter and absenteeism was higher than pre-pandemic. Around one in ten learners never returned to school at all during 2020.

Learning losses

We are able to measure learning losses on four reading skills – letter sound knowledge, complex consonant sound knowledge, familiar word reading and reading connected text (oral reading fluency). The red line in Figure 1 illustrates the performance on the letter sound task of an average learner in cohort 1, who were in grade 2 in 2020. The dashed line shows the performance of the counterfactual, learners in grade 2 in 2019. This represents the expected performance of cohort I in the absence of the pandemic. The difference between the red and dashed line then provides a measure of the impact of the COVID-19 school closure.

Figure 1. Correct letter sounds per minute – cohort 1 versus cohort II trajectory

There were significant learning losses across all four reading skills. Figure 2 shows annual achievement gains in each sub-task separately by cohort.

Cohort 1 learners gained around 16 fewer letter sounds and 6 to 8 words less than cohort 2 learners over the grade 2 year. Across the reading tasks, learning losses were between 57 and 70 percent of a normal grade 2 year.

Figure 2.  Reading losses

We find no gender differences in the impact of COVID-19 school closures. The pandemic had the most severe impact on the least proficient learners and more muted effects on those with higher initial reading proficiency.


The opening of schools in South Africa for the 2021 academic year was delayed by a month in response to the second wave of COVID-19 infections and on return, social distancing requirements were still in force. This means that currently the learners in this sample are only attending school every second day.

While it is almost certain that the deficits reported here will continue to increase without full time attendance, longer-term it is unclear whether these gaps will remain static, grow or narrow over time. Andrabi et al. (2020) and Pritchett (2020) warn that the short term losses are likely to be a lower bound if pedagogy continues as usual in line with curriculum demands. As children are moved up grade levels, those who are behind will continue to learn less each year.

Read the full report on these learning losses here.



Andrabi, T., Daniels, B. & Das, J. (2020). Human capital accumulation and disasters: Evidence from the Pakistan earthquake of 2005. Oxford: RISE.

Gustafsson, M. and Nuga, C. (2020). How is the COVID-19 pandemic affecting educational quality in South Africa? Evidence to date and future risks. NIDS-CRAM Insight Brief I.

Pritchett, L. (2020) Developing country schools need to reopen with different teaching. Oxford: Research on Improving Systems of Education.

Smith, W. (2021) Consequences of school closure on access to education: Lessons from the 2013–2016 Ebola pandemic. International Review of Education.

UNESCO (2021). One year into COVID-19 education disruption: Where do we stand? Retrieved 11 May 2021.

United Nations (2020). Policy Brief: Education during COVID-19 and beyond. Washington.