This paper was written by Martin Gustafsson
The United Nations Secretary-General, António Guterres, echoed the concerns of people and organisations around the world when he recently referred to the impact of the COVID-19 pandemic on schooling as a ‘generational catastrophe’. Children and youths are falling behind in their learning, and this is expected to have an impact lasting decades, especially if longer term effects on economic development and future earnings are taken into account.
This report focusses on the impacts of the pandemic on learning proficiency, specifically as measured by Sustainable Development Goal (SDG) Indicator 4.1.1. Over the last couple of decades, there has been a growing awareness of how crucial learning proficiency, especially that of younger children, is for human development. The evidence is clear that improvements in proficiency underpin future economic development, and the building of more cohesive and equal societies. The indicators on learning proficiency are among the most discussed indicators within the SDG framework.
There have been a number of attempts to understand and quantify the learning losses caused by the pandemic, with a view to shaping the necessary mitigation strategies. The current report represents one such attempt. What was clear around the end of 2020, when the pandemic was still far from over, is that the effects of the pandemic on schooling and learning were large, yet it was still too early to gauge precise effects. Moreover, while education actors around the world have responded to the crisis in often heroic and innovative ways, the optimal approaches to mitigating long-term impacts remain unclear. The pandemic’s threats to education are unprecedented in their nature and magnitude. While a wealth of evidence on how learning occurs, and what improves educational quality, is enormously helpful in charting the way forward, a greater understanding about the specificities of the pandemic and schooling is needed.
This report brings certain important specificities to the fore. This is done in a manner which emphasizes issues education planners would be familiar with, and need to grapple with. Though the model developed for the current report uses country-level data, the aim is not to provide guidance to individual countries. Rather, this report aims to provide global projections, and to identify dynamics which planners must focus on. These include: the magnitude and nature of the pandemic-related disruptions, not just to schools, but also pre-school institutions; the relationship between disruptions in the contact time of learners and losses in learning proficiency; the movement of age cohorts through the schooling system, and what this means for future proficiency levels and recovery strategies; what recovery means in terms of accelerating learning, and the point at which one can expect a return to trajectories envisaged before the pandemic.
The model producing the projections, in an Excel file, uses as its point of departure a projection model published by the UNESCO Institute for Statistics (UIS) shortly before the start of the pandemic. A key input in the new model is the contact time with teachers that children have lost, per country, from February to November 2020, expressed as a percentage of the regular school year. These statistics take into account partial closures, including situations where schools are open but attendance occurs on a rotational basis. By 11 November, the average child had lost 54% of a year’s contact time. Time lost is then converted to a fraction of a year of learning lost. Evidence, both from before and during the pandemic, point to an important effect: interruptions in contact time lead to learning losses which are larger than what is suggested by the actual time lost. This is because learners tend to forget skills acquired even before the interruption. A ‘forgetting ratio’ of 2.0 is used in the model: for every month of contact time lost, two months of learning are assumed to have been lost. A value of 2.0 is in line with the limited evidence we have on the magnitude of the ratio. Thus, if on average 54% of the school year has been lost, just over a year’s learning will have been lost on average. The model takes into account the fact that a year’s learning is not the same across the world: countries which perform relatively poorly in internationally comparable assessments, do so because the amount of learning occurring between one grade and the next is lower. The model assumes that learning losses in each country are coupled with worsening inequality: learners who performed well previously, and would often be socio-economically advantaged, experience smaller learning losses than learners who did not perform well previously.
Beyond 2020, the model assumes learners moving into, for instance, Grade 6 will continue to be less proficient than what could be expected without the pandemic, because these students lost learning in a previous year. In fact, without any remedial acceleration, or catching up, Grade 6 learners up to 2025 would all be equally behind – in 2025 children who were in Grade 1 in 2020 would reach Grade 6. But the model assumes that even beyond 2025, Grade 6 children would display the effects of the 2020 disruptions, because these disruptions affected pre-schools too. Though the data on pre-school disruptions in 2020 are very limited, it appears pre-schooling was as disrupted as schooling in Grade 1 and above. The model draws from UIS data on pre-school participation to gauge the probability within each country of pre-school disruption effects having been felt when children enter school.
What is also taken into account is the possibility that children who were in utero during 2020 could experience exceptional cognitive development difficulties. This draws from evidence that a social and economic shock such as the pandemic can have a lasting impact which is especially large for children who were in utero during the shock. While in many countries these effects may not endure beyond a few years, it nonetheless seems important to bear in mind within any projections which age cohorts of children were in utero during the onset of the pandemic.
The trajectory which has just been described is illustrated by the red curve in the following graph, which focuses on Grade 3. The red curve represents the world’s children in Grade 3, drawing from the data of individual countries, and assumptions such as the 2.0 forgetting ratio discussed above.
In 2019, around 59% of the world’s approximately 132 million children who should be in Grade 3 were proficient in reading. Thus, 54 million children in 2019 were not reading as they should. The 54 million includes roughly 12 million children who were not attending any form of schooling in 2019, who are all considered non-proficient for the purposes of this analysis. The figures for proficiency in numeracy would be similar. It is estimated that the learning losses associated with the pandemic would reduce the percentage of proficient children at the Grade 3 level to 49%. This means the number of non-proficient children at this age would increase from 54 million to 68 million – the pandemic would push 14 million children at just the Grade 3 level below the proficiency threshold.
The red curve points to a return to the original proficiency trajectory for Grade 3 only in 2030. The grey and black curves reflect scenarios where there is successful catching up, or remediation, in the form of more learning than usual. For instance, the grey curve represents an acceleration of 10% a year. This means learners would need to learn 10% more than a normal year’s worth of learning, in order to catch up to where they would have been without the pandemic. In that scenario, a return to the previously envisaged trajectory would occur earlier, in 2027. Accelerating learning, especially in developing countries, has been on the agenda for many years, and there is now considerable research to inform optimal strategies. At the same time, it should be kept in mind that acceleration to take a schooling system to new levels of proficiency is not the same as acceleration to recover from a loss in historical levels of proficiency. The latter is likely to be easier to achieve as teachers and administrators are familiar with the desired end result, and are very likely to see it as desirable and achievable.
In the original trajectory, represented by the green line in the graph, gradual but ongoing improvement was envisaged. This is based on earlier projections released by the UIS, which draw from evidence of learning gains seen in international testing programmes. The graph serves as a reminder of an important matter. Countries which were experiencing ongoing improvement before the pandemic need to ask themselves what factors were driving this. These factors, which are likely to be linked to the quality of training of new teachers, support to all teachers, and accountability systems, should continue to receive attention. Planners need to balance the focus on remediation programmes aimed at addressing the learning losses, against a continued focus on other drivers of long-term development. Put differently, while addressing the pandemic-related learning losses, countries should also strive to ensure that new learners entering school, who would not have experienced disruptions, though they may have experienced pre-school disruptions, reach levels of proficiency close to those seen before the pandemic or, even better, in line with a country’s previously envisaged improvement trajectory.
Projections for the end of primary and end of lower secondary levels are also provided in the report. At these levels, similar dynamics apply, but a return to a pre-existing trajectory occurs around three years later for the end of primary and five years later for the end of lower secondary. These lags can be shortened with the right learning acceleration.
Projections from the three levels analysed – Grades 3, 6 and 8 – permit an estimation of how many of the 1.06 billion children across eight age cohorts, corresponding to Grades 1 to 8, would move below the proficiency threshold as a result of the pandemic. The number of children of these ages falling below the threshold would increase from 483 million to 581 million in 2020. The pandemic would push just under 100 million children below the proficiency threshold. This number excludes children who would carry learning losses with them into Grade 1 as a result of disruptions to pre-schooling, and adolescents in schools and post-school institutions above Grade 8 who would suffer the educational effects of the pandemic.
There are key challenges which are not captured in the above graph. One is that education budgets are expected to decline as a result of the economic effects of the pandemic. This will compound the problems, especially if teachers feel they are bearing more than their fair share of the budget cuts, and if spending on teachers puts pressure on spending on educational materials. Reductions in spending on school meal programmes could have very serious negative consequences for the physical and cognitive development of children from poor households. The report discusses these matters, which are to some extent within the control of education planners.
What education planners have little control over is the economic effects of the pandemic on households, the most serious effect being a worsening of poverty. One result of this could be an increase in the percentage of children not in school. Little is known at this stage about the likely magnitude of this. While poverty may make it more difficult for households to send children to school, reductions in child labour, the abolition of school fees in many countries and increases in the coverage of school meal programmes in the last two decades are all factors which would work against higher levels of dropping out, especially for younger learners.
Tragically, increased dropping out is unlikely to affect the SDG proficiency indicators to a large degree. This is because those countries where more dropping out is most likely are also countries where children had low levels of proficiency even before the pandemic. In sub-Saharan Africa, in particular, only 20% of lower primary children are proficient in reading, yet 81% of primary-aged children are in school. Given the strong links between socio-economic background and learning, one can roughly say that outside the middle class, few learners in this region become proficient. And given that the poor are most likely to drop out of school, the result would be more non-proficient children outside school and fewer non-proficient learners in school. Clearly, even if more dropping out does not have an impact on the SDG proficiency statistics, the matter is a serious one in terms of, for instance, child nutrition and psychological well-being. Moreover, there are degrees of ‘non-proficiency’. Children should be as close as possible to the level of proficiency they should ideally be at.
Of the previously mentioned figure of 100 million children across eight age cohorts who would move below the proficiency threshold, 34 million would be children in Central and Southern Asia, while 29 million would be in Eastern and South-eastern Asia. These would be the two worst affected regions in absolute terms. In terms of percentage point changes in the percentage of proficient children, the largest decline is seen in Latin America and the Caribbean – from 70% to 51% in Grade 3, for example. Sub-Saharan Africa sees rather small declines. This is because even in 2019, the percentage of children who were proficient was low – for instance 20% at the Grade 3 level. Much of the learning losses occurring in this region would occur among children already below the level of proficiency. Put differently, the SDG indicators on learning proficiency provide a rather limited picture of the impacts of the pandemic on learning in sub-Saharan Africa.
Countries with effective programmes to monitor progress in, for instance, early grade reading will be in a good position to compare likely future trends without the pandemic, to actual outcomes with the pandemic, of the kind presented in the current report. Such comparisons will assist in determining what the effect of the pandemic has been on learning outcomes, and what remediation seems best. Countries which do not have these monitoring programmes will find it harder to interpret what lies behind the unusual trends which can be expected in the coming years. In particular, such countries may find it difficult to determine exactly how large the initial learning losses of 2020 were. The shock to learning brought about by the pandemic should be a catalyst for ‘building back better’, specifically improving the monitoring of learning, and taking teacher training, support to schools, and school accountability systems to new levels.