COVID-19 disruptions and education in South Africa: Two years of evidence

ABSTRACT

This paper provides an overview of learning losses and altered schooling patterns in South Africa during the COVID-19 pandemic (2020–2021). Five major trends emerge from a review of the evidence. These include significant learning losses (38–118% of a year of learning), widened learning inequality, lowered grade repetition rates, increased secondary school enrolments and an unprecedented rise in candidates writing and passing the National Senior Certificate (NSC) examination. School completion significantly increased in 2021 and 2022, spurred by COVID-19 adjusted assessment and promotion practices in Grades 10 and 11. Larger numbers of youth also achieved a NSC pass or Bachelor's pass enabling access to university. With twin pandemic shocks of learning losses and secondary school enrolment increases, remediating losses and realigning progression rules to effective assessment practices should be prioritised.

KEYWORDS:

• Learning losses
• dropout
• early grade Reading
• COVID-19
• South Africa

JEL:

• I21
• I24
• I28

1. Introduction

COVID-19 lockdowns and school closures have had a profound global impact on education. In 19 out of every 20 countries around the world, schools completely shut for a median of 17 weeks from the start of the pandemic until the beginning of 2022 (UNESCO, 2021). In South Africa, school closures in 2020 resulted in an average loss of 54% of in-person learning time in that year (Department of Basic Education (DBE), 2022a:5). Even after schools reopened in mid-2020 via a staggered grade return, the majority of schools implemented rotating attendance schedules for over 18 months, leading to further reductions in face-to-face teaching. For instance, in the latter half of 2021, Grade 3 students lost 22% of their in-person class time due to rotations and irregular absenteeism, although this average hides inequalities across the system (Gustafsson, 2022a).¹ It was not until February 2022 that daily school attendance fully resumed. Despite substantial losses in face-to-face teaching time, opportunities for remote learning were extremely limited. Nationally, only 11% of youth aged 5–24 years in educational institutions participated in remote instruction in 2020 (Statistics South Africa, 2022:10).

While school attendance and economic activity have returned to a more usual state of functioning, the disruptions caused by the pandemic continue to affect education. This paper’s primary objective is to provide an overview of what we know about learning losses and other schooling impacts in South Africa during two pandemic years (2020 and 2021). In addition to summarising existing South African evidence on learning losses (Ardington et al., 2021; Kotze et al., 2022; Van der Berg et al., 2022; Mullis et al., 2023), we collate evidence on dropout, repetition, school enrolment and school completion patterns during the pandemic. This investigation primarily draws on existing reports and published studies, although it briefly augments existing evidence with updated descriptive analysis of schooling datasets.

In the following section, international research on COVID-19 learning losses and dropout is discussed. This helps to contextualise the South African situation on a global scale. Subsequently, South African evidence of COVID-19 impacts on learning and schooling is presented in sections 3 and 4. Our findings are organised in two parts: firstly, we discuss evidence on learning losses and declines in curriculum coverage (section 3) and then we explore how COVID-19 has affected student progression through the education system as reflected in enrolment, dropout and retention patterns and school completion (section 4). The South African evidence reveals five key findings: significant learning losses, disparities in learning losses, declining secondary school dropout, lowered repetition rates and this in turn has encouraged higher rates of school completion. In the discussion section, we consider the South African response to the learning crisis and best practices for recovery. Finally, section 6 concludes, emphasising the need for a comprehensive statewide remediation campaign.

2. Global evidence on COVID-19 disruptions to schooling and learning

Even though three years have elapsed since the beginning of the pandemic, learning setbacks and their impact on human capital continue to be a major global concern. A recent analysis of reading comprehension scores from the 2021 Progress in International Reading Literacy Study (PIRLS) by Jakubowski and colleagues (2023) underscores this concern. The global findings reveal significant declines in Grade 4 reading comprehension scores. Unadjusted results suggest an average reading decline of 15 points across countries on the PIRLS scale, amounting to 18% of a standard deviation (SD).² However, after adjusting for age and grade differences across countries, the decline increases to 27 points, equating to 33% of a SD which equates to a loss of over a year's worth of schooling (Jakubowski et al., 2023:2). Notably, students who typically achieve lower scores experienced even greater declines.

Before the release of 2021 PIRLS results, Patrinos et al. (2022) reviewed 35 robust studies of learning losses. Of the 35 studies, 32 found evidence of learning losses averaging 0.17 SD, representing roughly half a school year of learning. Although just 4 of these 25 studies are from low- to middle-income countries (LMICs), larger losses compared with higher-income countries are apparent in these LMICs, including Mexico (0.55 SD loss), Brazil (0.32 SD loss) and China (0.22 SD loss) (Patrinos et al., 2022).

Moscoviz and Evans (2022) identified 40 empirical studies that estimated student learning losses and dropout rates across countries with varying income levels. Even in high-income countries like the Netherlands, Belgium, Germany and Italy, where distance learning was feasible, learning outcomes fell short of their usual standards (Contini et al., 2021; Engzell et al., 2021; Maldonado & De Witte, 2022; Schult et al., 2022). In the United States, where assessment data has been recorded as far back as 1992, fourth graders’ average reading and mathematics scores during the pandemic years were lower than all previous assessment years dating back to 2005 or 2003 (NAEP, 2022).

The COVID-19 pandemic has exacerbated educational inequalities in numerous countries. Even in instances where no overall average learning losses are observed, students with lower socioeconomic status experience higher learning setbacks (Moscoviz & Evans, 2022). Across studies, Patrinos et al. (2022) find greater learning loss among students or schools with lower socio-economic status, a conclusion reinforced by Betthäuser et al. (2023). Another noteworthy pattern is that the longer schools remained closed, the greater the learning losses. For every week schools were closed on average, learning declined by 0.01 standard deviations (Patrinos et al., 2022). Jakubowski et al. (2023) find that 10 weeks of full and partial closures resulted in an 8-point reading decline in PIRLS scores (equivalent to a decline of 9% of a standard deviation), whereas 25 weeks of closures led to a 20-point decline in PIRLS scores (equivalent to a decline of 24% of a standard deviation).

Studies have also explored the impact of COVID-19 on school dropout rates (Moscoviz & Evans, 2022). In LMICs, dropout rates after school reopened had typically risen relative to pre-pandemic dropout rates although dropout rates remained constant in some countries such as Senegal and Ghana. By contrast, evidence from South Africa indicates a decline in dropout rates between 2020 and 2021 among youths aged 15–19 years (DBE, 2022a, 2023a). Along with patterns seen in Hungary (Hermann, 2022), this presents another exception to the broader global trend.

3. South African findings: COVID-19 impacts on learning

3.1. Large learning losses in language and mathematics

Various studies have identified substantial learning losses in South Africa due to COVID-19 (Ardington et al., 2021; Kotze et al., 2022; Van der Berg et al., 2022; Mullis et al., 2023; Spaull, 2023). Findings are summarised in Table 1. Collectively, these studies offer compelling evidence of learning losses spanning various subject areas and grade levels. When comparing the Progress in International Reading Literacy Study (PIRLS) results from 2016 to 2021, South Africa experienced some of the most significant learning losses when expressed in PIRLS points across all 32 countries that participated in both years. South Africa’s PIRLS mean score declined by 31.4 points from 320 in 2016 to 288 in 2021 (Mullis et al., 2023), following two consecutive rounds of improvement in PIRLS scores from 2006 to 2011 and 2011 to 2016. Relative to South Africa’s 2016 PIRLS standard deviation (SD) of 106.5, 31.4 points translates into 30% of a SD but relative to Jakubowski et al.’s (2023) within-country global average SD of 82.4, a decline of this magnitude is 38% of a SD. The 2016 to 2021 PIRLS loss in South Africa of 31.4 corresponds to almost 50% to 60% of a year of learning³ on average (where a year of learning in South Africa in PIRLS points is about 55–60 points) (Böhmer & Wills, 2023). However, it is noted that a decline of 31.4 points could be an underestimate of the actual losses in PIRLS. Accounting for differences in the age and composition of the South African PIRLS sample from 2016 to 2021 suggests larger declines (Böhmer & Wills, 2023).

Table 1. Summary of average schooling days lost and learning losses in South Africa from available studies.

				Percent of a year of learning lost / Percent of a standard deviation
				Oral readng assessment			Written assessment
Study	Grade	Sample / Data	Average school days lost **	Alphabetic knowledge	Home Language (HL) reading	English first additional (EFAL) reading	Reading comprehension (Usually in home language)	Language	Mathematics
	1st year of pandemic (2020)
DBE (2022a)	All grades	Public schools	54% (2020 vs. 2019)
Ardington et al. (2021)	Grade 2	Eastern Cape Funda Wande no-fee school sample	60% of a normal year (2020 vs. 2019)	70% of a year of learning	57% of a year of learning	–
Ardington et al. (2021)	Grade 4	Mpumalanga EGRS II no-fee school sample	56% (2020 vs. 2019)	–	81% of a year of learning	62% of a year of learning
	Both years of the pandemic (2020 and 2021)
Böhmer & Wills, (2023); Mullis et al. (2023); Spaull (2023)	Grade 4	PIRLS – nationally representative sample						31.4 point decline; 50-60% of a year of learning assuming 55–60 points is a year of learning; 30% of a SD assuming 2016 PIRLS SD of 106.5
Van der Berg et al. (2022)	Grade 3	Western Cape – almost all public schools	155 days lost across 2020 & 2021					38% of a year of learning* (15% of a SD)	90% of a year of learning* (36% of a SD)
Van der Berg et al. (2022)	Grade 6	Western Cape – almost all public schools	155 days lost across 2020 & 2021					67% of a year of learning* (27% of a SD)	98% of a year of learning* (39% of a SD)
Van der Berg et al. (2022)	Grade 9	Western Cape – almost all public schools	155 days lost across 2020 & 2021					46% of a year of learning* (14% of a SD)	106% of a year of learning* (32% of a SD)
Kotze et al. (2022) and own calculations	Grade 4	North West EGRS I no-fee school sample	37% (2021 vs. 2019)^ 56% (2020 vs. 2019)^	–	54% -118% of a year of learning	46% of a year of learning
Gustafsson (2022a)	Grade 3	SASAMS – almost all public schools in 7 provinces, applying NECT rotation information	Lost 22% of a normal term (measured in term 3, 2021)

Notes: SD = Standard Deviation. ^From Bisgard et al. (2022) where school days lost account for COVID-19 related school lockdowns, not attending school due to rotational schedules, and other discretionary school closures. *Van der Berg et al. (2022) using Table ES2, assumes that a year of learning in primary schools is 0.4 SD and in secondary schools 0.3 SD. **Expressed as a percentage of days lost relative to a normal school year unless otherwise stated.

In terms of reaching the low PIRLS international benchmark – a signal of being able to read for meaning – 81% of Grade 4 learners in South Africa could not reach this benchmark in 2021, an increase from 78% in 2016. It is important to note that PIRLS only assesses reading comprehension and at a specific grade level. For evidence of both mathematics and language learning losses at three other grade levels, we turn to the Western Cape province. The Western Cape is the only province of nine in South Africa that systematically evaluates learning through standardised tests. In the Western Cape, approximately 155 school days were lost over the course of 2020 and 2021 (Van der Berg et al., 2022).

Against this backdrop, Van der Berg et al. (2022) conducted a comparative analysis of Grade 3, 6, and 9 cohorts in 2021 and 2019 from the same schools, assessing their performance on the same questions in each year using the Western Cape Systemic Tests. Figure 1 illustrates Van der Berg et al.’s (2022) estimates of year-on-year comparisons of average language and mathematics performance at the end of the year. In comparison to cohorts assessed in 2019, those assessed in 2021 were found to be 38–67% of a school year behind in language and 90–106% of a school year behind in mathematics (Van der Berg et al., 2022:5).

Figure 1. Performance declines in the Western Cape Systemic Tests (2019–2021).

Assuming the same learning loss in Grade 6 mathematics performance in the Western Cape extended to the entire country, the percentage of students failing to reach the low international benchmark of 400 points in the Trends in International Mathematics and Science Study at the Grade 5 level (TIMSS-N) would increase from 64% to 76% (Van der Berg et al., 2022:45).

Declines in average learner performance on the Western Cape Systemic Tests appear to be larger in primary grades (compared to Grade 9) and in mathematics compared to language, confirming earlier expectations of large losses in mathematics (Soudien et al., 2022). A conservative estimate indicates that Grade 3 mathematics proficiency has fallen by nearly a full year of learning (Van der Berg et al., 2022). Assuming the typical learning progress over a year in primary schools is 0.40 standard deviations (SD), and 0.30 SD in secondary schools, the losses in the Western Cape Systemic Tests are as follows: 0.15 and 0.36 SD in Grade 3 mathematics and language; 0.39 and 0.27 SD in Grade 6 mathematics and language; and 0.14 and 0.32 SD in Grade 9 language and mathematics (Van der Berg et al., 2022).⁴

3.2. Unequal learning losses in language and mathematics

Using the nationally representative 2016 and 2021 PIRLS results, stark differences in learning losses are observed across language groups and by school socio-economic status (Böhmer & Wills, 2023; Spaull, 2023) (see Figure 2). Where language correlates strongly to socio-economic status in South Africa, the much larger reading declines in 2021 PIRLS scores for children assessed in African languages as opposed to English or Afrikaans present a concern in a system characterised by significant pre-pandemic learning inequalities (see Figure 2).

Figure 2. PIRLS average scores by language group in South Africa, 2016 vs 2021.

There is also evidence of widening inequality in learning across wealthier schools (proxied by Quintile 4–5 schools) and poorer schools (proxied by Quintile 1–3)⁵ in the Western Cape Systemic Tests. Figure 3A and B show the percentage of Grade 3 and 6 learners not achieving a pass mark (50%) in language and mathematics. Worse performance is seen for learners in both Quintile 1 and 5 schools in 2021 compared to 2019. In Quintile 5 schools, 16% of Grade 3s in 2019 did not achieve a pass in mathematics, increasing to 25% in 2021 (Van der Berg et al., 2022). Greater losses are evident in Quintile 1 schools, even though they started from a much lower performance level. In 2019, 48% of Grade 3 students in these schools did not achieve a pass in mathematics, which rose to 66% in 2021.

Figure 3. Percentage of learners in quintile 1 and 5 schools not achieving a pass mark (50%), Western Cape Systemic Tests.

3.3. Large losses in foundational reading skills

The language assessments of PIRLS and the Western Cape Systemic tests focus on higher-level reading skills, such as written comprehension. They are not designed to identify deficiencies in foundational reading skills like alphabetic knowledge, oral reading fluency, or oral language, which are essential for reading comprehension (Zuilkowski et al., 2019). As a result, these tests may not capture lower-level reading skills, leading to test floor effects. Analysis of early grade reading assessments from no-fee school samples in three provinces (Mpumulanga, North West and the Eastern Cape), however, shed light on COVID-19 setbacks to the acquisition of foundational reading skills, as discussed in Ardington et al. (2021) and Kotze et al. (2022).

Following a period of progress in reading development in South Africa (DBE, 2020), the COVID-19 pandemic significantly hindered early grade reading. In 2020, Foundation Phase learners assessed in 57 Eastern Cape and 180 Mpumalanga no-fee charging schools lost 56–60% of the number of school days in a normal school year (see Table 1). This resulted in a 57–81% reduction in reading development in Nguni Home Languages and a 62% reduction in reading development in English First Additional language (EFAL) among Grade 2 and 4 students compared to a normal year (Ardington et al., 2021). In a sample of 202 North West province schools, students in Grade 3 experienced a 37% loss of school days in 2021 due to rotational schedules after losing more than half of a typical school year in 2020. By the third term of 2021, Grade 4 learners in these North West schools had lost 54–118% of a year’s worth of learning in foundational Home Language reading skills.

For the same samples, Figure 4 shows how students’ oral reading fluency, or word reading, develops over a regular school year and how this development stagnated during two years of the pandemic. In a typical Grade 4 year, learners improve by about 12 words per minute on average when reading in Setswana. However, by the third term of 2021, Grade 4 students during the pandemic were reading roughly the same number of words or even fewer in a minute than a comparative group of Grade 3s in 2018 (Kotze et al., 2022). English reading trajectories have also worsened. In a normal year, Grade 4 learners in the Mpumalanga and North West province samples typically improve by 11–14 words correct per minute. During the pandemic, this declined to 4–6 additional words per minute.

Figure 4. Deterioration in reading development over a year in no-fee school samples in three provinces.

The decline in early grade reading has significant implications for children’s development and their ability to achieve a crucial Medium-Term Strategic Framework education goal: that 10-year-old students can read with comprehension (DBE, 2022c:6). Failing to reach fundamental milestones in decoding skills and reading fluency during Grades 1–3 significantly jeopardises their capacity to read with understanding by the time they complete primary school (Wills et al., 2022).

3.4. Reduced contact time and reductions in workbook coverage

Learners fell behind during the pandemic because they received far less instruction than in a normal year. Table 1 summarises what we know about the amount of face-to-face contact time lost as seen in different studies. Reduced contact time translated into reduced curriculum coverage – a key mechanism by which learning losses likely occurred. For example, drawing on analysis of children’s writing in workbooks in Bisgard et al. (2022), there is clear evidence that less content was covered during the pandemic.

In South Africa, every child in primary grades is meant to receive workbooks from the Department of Basic Education (DBE). In term 3 of 2018 and 2021, Grade 3 students’ workbooks (as selected as being the best by the teacher) were assessed in the Early Grade Reading Study (EGRS I) in North West province (Bisgard et al., 2022). The number of pages with evidence of any work done or writing can be compared across 2018 and 2021 workbook samples from the same schools. The number of pages showing any work done declined in Setswana Home Language workbooks (22 to 17 pages) and in English First Additional Language (EFAL) workbooks (21 to 18 pages). Evidence of any writing of at least one full sentence across pages in Home Language workbooks also dropped (12 to 7 pages). In EFAL workbooks, there was a notable decrease in the number of pages with any full-sentence writing (12 to 7 pages) and writing of paragraphs (6 to 3 pages). These reductions in workbook coverage occurred despite workbooks being the primary method for non-contact instruction during school closures and rotational scheduling (Bisgard et al., 2022). Less writing evidence in lower grade levels may explain the significant negative impact of COVID-19 on language results observed in the Western Cape Systemic Tests and in PIRLS (Van der Berg et al., 2022:12; Mullis et al., 2023).

Apart from the loss of instructional time, children may have experienced negative psychological effects in an uncertain environment (Bisgard et al., 2022) and due to COVID-related disruptions in their home life. The impacts of these other factors on learning have yet to be disentangled from instructional disruptions in South Africa.

4. South African findings: School enrolment, dropout, repetition and school completion during COVID-19

4.1. Declining dropout among adolescents and rising enrolment

There were substantially high levels of non-participation in school during the pandemic (Gustafsson, 2021; Shepherd & Mohohlwane, 2022). Initially, non-participation was framed as school ‘dropout’ but then was subsequently attributed to ‘extended absenteeism’ in the context of disrupted school schedules. Gustafsson (2021) suggests that as many as 910 000, predominately younger learners, were extended absentees. Relative to a pre-pandemic situation, updated analysis of the National Income Dynamics Survey – Coronavirus Rapid Mobile Survey (NIDS-CRAM) data revealed that the return of learners to Grades 10–12 eventually improved, surpassing pre-pandemic levels (Shepherd & Mohohlwane, 2022:770).

Contrary to initial expectations and early evidence from NIDS-CRAM (Shepherd & Mohohlwane, 2021), South African public schools experienced an overall decrease in dropout during the pandemic and a rise in enrolment at the secondary level (Gustafsson, 2022b; DBE, 2022d, 2023a). Enrolment among students aged 15 and above, beyond compulsory school age, increased across all provinces from 2020 to 2021, with increases ranging from 0.5% in the Northern Cape to 3.1% in the Western Cape (DBE, 2022d). Additionally, Grade 12 enrolment saw a 20% rise from 2020 to 2021 (DBE, 2022a:12). The entire basic education system experienced a substantial enrolment increase from 2019 to 2021, surging by half a million students primarily due to reduced dropout (DBE 2022a:2), bringing total enrolment to approximately 13.4 million in 2021. The growth in enrolments started before the pandemic and did not abate (DBE, 2022b:23).

An analysis of learner-unit level administrative data from the public education system (DBE, 2023a) confirms the decline in dropout rates during the pandemic at higher grade levels (See Figure 5). For example, average learner departure rates from the public education system at the Grade 10 level declined from about 10% in 2018 to 8% in 2020.

Figure 5. Percentage of learners departing at the end of each grade, national rates from learner unit records.

Where dropout did occur between 2020 and 2021, it was in the initial school grades, not in higher grades (DBE, 2022d). About 2.3% fewer children than expected (roughly 27 000 children) did not enrol as first-time learners in 2021. Additionally, up to 19 000 learners of compulsory school-going age (under 15 years) dropped out. The slight decline in enrolment of younger learners was attributed to the reluctance of parents to enrol learners in either Grade R or Grade 1 in a pandemic-disrupted school environment or to concerns about contracting the virus (DBE, 2022d:1).

Yet international reviews (Moscoviz & Evans, 2022) included estimates that school dropout in South Africa increased, drawing largely on preliminary participation rate estimates from NIDS-CRAM (Shepherd & Mohohlwane, 2021). Confusion has arisen due to the challenges of interpreting telephonic survey responses on school attendance or non-return of learners to school during disrupted school schedules.

4.2. Declines in repetition rates

With very large reductions in teaching time and changes to assessment requirements during 2020 and 2021, typical learner progression patterns have been disrupted. A key trend observed has been large and sustained system-wide reductions in repetition rates at all grade levels, but particularly in Grades 10 and 11 at the end of 2020 (see Figure 6).

Figure 6. Repetition rates by grade in South Africa, 2018–2020, national rates from learner unit records.

Reduced repetition rates (and rising promotion) occurred largely due to the decoupling of progression decisions from assessment and from altered progression rules. In 2020 and 2021 assessments were reduced; examinations (except the National Senior Certificate) were cancelled; moderation processes were eased and the contribution of school-based assessments (SBAs) to learners’ final marks was raised significantly in all grades. The SBA component of the promotion requirements for Grades 10 and 11 was increased from 25% to 60% (Hoadley, 2020:15, 2023) despite concerns about SBA reliability (Van der Berg & Shepherd, 2015). This led to more learners being progressed through the system, reflected in reduced repetition rates at all grade levels but particularly at the higher grades (see Figure 6). The lowered repetition rates in 2020 in Grades 10 and 11 are unprecedented in South Africa.

The sustained reduction in repetition rates post-Covid is implied in the analysis of repetition data from two provinces. End of year (2020 and 2021) repetition rates for a balanced sample of schools in two provinces – Gauteng and the Eastern Cape – are shown in Figure 7. Across primary school grades (1–7) in the Eastern Cape, a four-percentage point decline in repetition rates in 2020 is observed relative to pre-pandemic (2016) levels. The decline remains relatively entrenched at the end of 2021. A similar pattern is observed in Gauteng. Historically high repetition rates in Grade 1 (18% in the Eastern Cape and 12% in Gauteng in 2016) had declined to 2016 pre-pandemic Grade 2 repetition levels (12% in the Eastern Cape and 9% in Gauteng) by the end of 2021. This may have implications for the acquisition of foundational reading and numeracy skills taught in Grade 1 where pre-pandemic patterns of ‘holding back’ children in Grade 1 are related to issues of school readiness.

Figure 7. Reduction in repetition rates during the pandemic. Gauteng and the Eastern Cape province.

At the Further Education and Training (FET) level (Grades 10 to 12), repetition rate reductions are particularly notable. In 2016 almost a third of Grade 10s in the Eastern Cape and Gauteng were not progressed to Grade 11. From 2016 to 2020, Grade 10 repetition rates declined from 32% to 18% in the Eastern Cape and from 33% to 17% in Gauteng before stabilising slightly in 2021 to 23% in the Eastern Cape and 24% in Gauteng.

With much larger numbers of learners pushed through the system, this in turn has implications for raising class sizes in higher grades while reducing class sizes in lower grades. Historically, repeaters have been more likely to drop out in later grades (Van der Berg et al. 2021), so lower dropout rates are expected, in turn raising enrolment levels in the highest grades.

4.3. Increases in school completion

The National Senior Certificate (NSC) (commonly referred to as the ‘matriculation’ examination) is South Africa’s flagship certification programme in the final year of school (Grade 12). Hoadley (2020:16) regards it as ‘a stellar achievement’ that the DBE was able to retain the NSC examination in 2020 despite pandemic-related disruptions, considering that this is ‘the only high-stakes examination with consequences for learners’ future academic and work opportunities’. Despite the importance of the NSC as a certification system, it is not designed to measure the overall quality of the education system. Large percentages of any Grade 1 cohort do not go on to sit the examination 12 years later, despite improvements over time in ‘survival’ to Grade 12 (DBE, 2020). Those that do sit the NSC examination are a select group who did not drop out. For this reason alone, the NSC results are not a good metric to track how the education system has been affected by COVID-19 disruptions. Yet, the NSC has historically been the key education performance metric of concern for provincial administrations and in public debate.

The NSC results overall did not reflect any notable negative COVID-19 impacts. The normal pass rate declined only in 2020, increasing slightly in 2021 and approached the pre-pandemic peak in 2022. Relative to pre-pandemic years 2018–2019, there were no major variations in Bachelors-level pass rates (an important signal for acceptance into university) from 2020 to 2021. Counterintuitively, the Bachelor pass rate reached its highest-ever level in 2022 (see Figure 8). The number of Bachelor passes achieved in 2022 was also nearly 50% higher than in 2019. However, as has been experienced in a tertiary context in South Africa (Whitelaw et al., 2022), ‘improved’ or consistent results during COVID-19 may simply not reflect true learning gains or sustained learning.

Figure 8. National Senior Certificate (NSC) results 2013–2022.

On the one hand, less severe learning losses are expected in higher grades. Grade 12s were favoured in the staggered return of learners to school. Fewer days were lost in Grade 12 in 2020 at 22% of school days, compared to 42% of school days lost by Grade 4s and 9s in 2020 (before rotations) as seen in Appendix Table A1 (Kotze, 2021). Historically, vacation and after-school programmes have also been used to prepare Grade 12 learners for the NSC.

On the other hand, buoyant NSC results in 2020–2022 seem counterintuitive. It seems unlikely that vacation programmes could have entirely made up for losses experienced not only during the Grade 12 year, but in Grade 11 – an important preparation year for the NSC. Furthermore, there were many more candidates sitting the 2021 examination at 704 000 candidates. Then the year 2022 saw an unprecedented number of candidates write the NSC, 725 000 candidates – 30% more than the 2013–2019 average. There was a concurrent and unprecedented number of candidates achieving a Bachelors-level pass in 2021 and 2022 at about 256 000 and 279 000 (see Figure 9). The increase in candidates writing was largely attributed to very large increases in learners being promoted to Grade 12 in 2021 and 2022 (Gustafsson, 2022c) and thus much higher ‘survival’ to matric. Two additional contributing reasons include a larger population⁶ of 18-year-olds in 2021 and 2022 than previous years, and the possibility that more candidates were registered as full-time candidates with the removal of the Multiple Examinations Opportunity⁷ in 2020.

Figure 9. Full-time candidates that wrote the National Senior Certificate, passed or obtained a Bachelor-level pass, 2013–2022.

As an unexpected outcome, ‘the pandemic may have pushed the system onto a new level at which far more youths obtain the NSC for many years into the future’ (DBE, 2022a:4), with more potentially qualifying for entry into tertiary studies. Pre-pandemic, of roughly 1 million learners entering Grade 1 in a school year, about 450 000 would leave school before reaching Grade 12 (DBE, 2022a:1), implying only 45% of every age cohort among the youth population obtained the NSC in Grade 12 or very soon thereafter. The increase in Grade 12 learners in 2021, for instance, implies that the number of learners annually leaving before matric declined from around 460 000 before the pandemic to perhaps as low as 200 000 (DBE, 2022a:2).⁸

Little time has been given for universities and training colleges to prepare for this. The tertiary system will have to accommodate disruptions to learning in school, adding bridging course content to make up for topics and skills not covered in Grade 11 and/or 12 in pandemic years. This is imperative where evidence already suggests widening achievement gaps between students funded by the National Student Financial Aid Scheme (NSFAS) and students not funded by NSFAS in 2021 (Whitelaw et al., 2022).

5. Discussion

It will take time to gauge the persistence of learning losses, disruptions to learner flows and the extent of recovery that will be seen. How well learning losses are mitigated will depend on the effectiveness of well-designed remediation programmes and continued tracking of learning and retention patterns.

A state-wide remediation programme, with allotted budgets, was not prioritised timeously as a response to COVID-19 learning losses in South Africa. The dominant COVID-19 ‘recovery’ plan in eight of nine South African provinces involved revising Annual Teaching Plans in delivering national curricula and allowing for an interim deviation from the original curriculum. ‘Remote’ support programmes were also set-up for Grade 12s. The Tswelopele Campaign was developed to support Grades R-9 learners with support content to catch up on learning losses through television, mobile chat platforms and YouTube channels (DBE, 2022e:7). Accessing this content, however, depended on individualised learner decisions (and the availability of home resources). As Hoadley (2023:13) reflects, ‘An emphasis on self-directed learning and digital offerings shifted responsibility for addressing learning losses away from the DBE to the vagaries of learners’ access to social and educational supports’.

It was only after three years, just before the release of PIRLS 2021 results in 2023, that a national Learning Recovery Program (LRP) was introduced to address these issues. However, the LRP’s guidelines mainly focused on planning for learning recovery and delegated decision-making to teachers without allocating additional time or resources for significant changes beyond ‘business-as-usual’ teaching (Hoadley, 2023). This is unlikely to bring about substantial changes in schools to address learning gaps, posing a strategic challenge for South Africa’s pandemic response, with potentially long-term implications for human capital development (Kaffenberger, 2021).

By exception, the Western Cape province adjusted their Foundation Phase (Grades 1–3) instructional time to give more emphasis to core subjects (specifically mathematics and language) from mid-2022 (Makwala King, 2022). This was a direct response to evidence of learning losses in the province (Van der Berg et al., 2022). In further announcements in 2023, the Western Cape committed budget and resources to the #BackonTrack project, a learner recovery programme for Grades 1–12, including more time for core subjects, after-school and holiday classes, training teachers on the science of reading and skills training for parents to support learning at home (WCED, 2023).

Despite a limited national COVID-19 recovery response, new measures can still be introduced to address learning deficits going forward. Some countries have demonstrated that with the right support, significant recovery is possible and learning losses are avoidable (Patrinos et al., 2022). For instance, Singh et al. (2022) found that by the end of 2021, children aged 5–7 years in rural Tamil Nadu, India, had experienced significant learning losses, equivalent to 1–2 years of schooling (0.7 SD in mathematics and 0.34 SD in language). Two-thirds of this learning loss was recovered in the first six months after schools reopened due to catch-up efforts, teachers’ and schools’ compensatory actions, and a state-wide after-school remedial instruction programme called ‘Education at Doorstep.’ The programme featured after-school remedial camps led by around 200 000 locally hired volunteers. Almost a quarter of the cohort-level learning recovery is attributed to this government-run programme. Less advantaged students were more likely to attend these after-school sessions leading to progressive recovery (Singh et al., 2022:3).

South African schools could consider implementing a large-scale remediation programme for two key reasons. First, South Africa has deployed 320 000 teacher assistants to schools as part of the Presidential Youth Employment Initiative (PYEI). This teacher assistant programme could be leveraged for remediation, building on the significant early grade learning gains identified during the pandemic through the deployment of teacher assistants in a Funda Wande programme (Ardington & Henry, 2021). Qualitative insights suggest that teacher assistants aid in classroom management and provide additional tutoring through facilitating clubs (DBE, 2022f). Second, catch-up programmes are common for the National Senior Certificate (NSC) (DBE 2022e:27–28). If vacation programmes are widely promoted to prepare learners for the NSC, why cannot this be done in earlier grades using teacher assistants?

International evidence strongly supports in-person remediation as the preferred approach to address learning losses (Singh et al., 2022). Remote instruction has proven to be less effective in remediating learning gaps (Muñoz-Najar et al., 2021) and it is often inaccessible for most learners and teachers in LMICs. In 2020, for instance, only 11% of South African youth aged 5–24 years attending an educational institution engaged in remote instruction (Statistics South Africa, 2022). Recovery effects from using remote instruction or private tutoring also tend to exhibit a positive socio-economic gradient in terms of access and impact.

Nevertheless, remote learning options could be better deployed in future lockdowns. Some learning has been promoted through a combination of phone-based tutoring and text messaging in LMICs (Hassan et al., 2021; Angrist et al., 2022; Lichand et al., 2022). Where almost 60% of South African youth aged 5–24 years had a smartphone in 2020 (Statistics South Africa, 2022:17), more opportunity existed for digital learning opportunities than was fully leveraged.

6. Conclusion

This paper confirms significant pandemic-related learning losses in South Africa, particularly in lower grades, necessitating policy and instructional adjustments. If learning losses are not addressed, they will compound, especially for the most disadvantaged learners (Kaffenberger, 2021).

Furthermore, aside from learning losses, COVID-19 has disrupted educational progression, placing new demands on the basic education system and the tertiary education system. Enrolment patterns have shifted significantly, with reduced dropout in the FET Phase and improved survival rates into higher grades driven by more lenient pandemic-related progression. The basic education system must realign progression to achievement standards. As Whitelaw et al. (2022:3) reflect, ‘spuriously improved or even constant academic performance during the pandemic could unintentionally impact academic performance in the long term if students proceed to higher levels without sufficient baseline knowledge and competencies’.

Learning losses and secondary school enrolment increases have occurred as twin shocks in a context of education budget cuts in the basic education system. Even in a financially constrained environment, prioritising remediation of learning losses is paramount. This requires more than mere adjustments to Annual Teaching Plans or generic recovery guidelines. The long-term human development consequences of inaction will surpass the short-to-medium-term costs of effective intervention to address these losses.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Allan and Gill Gray Philanthropies .

Notes

1 For approximately one-third of schools implementing rotations, Grade 3 attendance was as low as 45% in term 3 of 2021 (DBE, 2022a:5).

2 The average within-country SD of reading achievement in reference is 82.4 points (Jakubowski et al., 2023).

3 Previous reports have referred to a 2016 to 2021 learning loss in PIRLS of 80% of a year of learning (see Spaull, 2023), where a year of learning is incorrectly assumed to be 40 points. However, as Spaull and Pretorius (2019:153) note ‘the oft-cited 40-point figure for a year of learning is based on three Nordic countries’ rather than South Africa.

4 The pandemic-related learning declines in the Western Cape should be viewed in relation to the already low performance that predated COVID-19 and that the Western Cape is the best performing province in South Africa along with Gauteng.

5 Official school ‘Quintile’ classifications distinguish schools along the lines of community wealth. Generally, Quintile 1–3 schools are not allowed to charge fees while Quintile 4–5 schools are allowed to charge fees.

6 In the period 2003–2005, South Africa’s birth rates unexpectedly rose by 13% (Gustafsson, 2018).

7 The Multiple Examination Opportunity is a programme to support learners who had failed a grade at least twice but are then progressed to the next grade. They are given the option to split their subjects, for the final NSC examination, over the final exam period and the following year’s June exam period.

8 The percentage of youths successfully completing Grade 12 has increased from about 45% in 2005, to around 62% in 2021 drawing on General Household Survey data (DBE, 2023b).

Appendix

Table A1. Days of school lost due to official school closures in 2020.

	Maximum possible school days per DBE regulations	% of 2019 days (199) lost due to school closures only	Weeks of loss due to official school closures only
Grade 7 & 12	155	22%	8.8
Grade R, 6 & 11	136	32%	12.6
Grade 3 & 10	131	34%	13.6
Grade 1 & 2	126	37%	14.6
Grade 4 & 9	116	42%	16.6
Grade 5 & 8	102	49%	19.4

Source: Adapted from Kotze (2021)