First estimates of global learning loss in student achievement using comparable reading scores

At the World Bank, and with partners, we have been tracking the learning losses from COVID-19 for the past three years. At the start of the pandemic in 2020, we predicted learning losses and then in 2021 we reported on actual losses from several countries. But our latest analysis published in the journal Economics Letters shows that the losses are broader and deeper than we knew and confirms our commitment to helping countries recover the losses and reduce learning poverty.

This latest analysis reveals that school closures led to student learning losses and that those losses are widespread. We found that international reading scores declined from 2016 to 2021 by more than a year of schooling. Students in schools that faced longer closures and lower-achieving students suffered larger declines.

This analysis is more precise than previous estimates. We compared grade 4 score data from assessments of the Progress in International Reading Literacy Study (PIRLS) in 2001, 2006, 2011, 2016 and 2021 in 55 countries. This allowed us to see the precise evolution of reading scores before and after COVID-19, building on an earlier analysis by our colleagues.

Merging this data with information from UNESCO on the number of weeks that schools were fully or partially closed, we modeled the effect of those full or partial closures on achievement.

What became clear is that the decline in achievement was significantly larger for students in schools that faced longer closures. In schools affected for more than 8 weeks, actual achievement was lower than expected by 34 points—more than a year’s worth of schooling. In schools that were unaffected or affected for less than two weeks, achievement declined by only 13 points, less than half a year’s worth of schooling.

Furthermore, countries with no closures in 2021 achieved results as predicted based on previous achievement. With every additional week of school closures, the results declined by 0.8 points. Thus, 25 weeks of closure resulted in decline of 20 points and 50 weeks of school closures resulted in a decline of 40 points—more than a year’s worth of schooling. These numbers are consistent with reviews of learning losses using national data sources.

We found no differences for boys and girls. However, as expected, lower-achieving students experienced much larger losses. The lowest achieving students in schools affected for more than 8 weeks suffer a loss of 36 points (more than a year’s worth of schooling), compared to 24 points loss for the top students (about ¾ of a year of schooling). In countries with 10 weeks of full or partial closures, the best students experienced no loss. The weakest students lost 12 points. However, in countries with 50 weeks of full or partial closures, the best students lost 28 points, while low achievers lost 37 points. These changes suggest significantly larger learning loss among the lowest-performing students, consistent with other studies of the impact of duration on learning loss.

The persistence of learning gaps three years after COVID-19 is concerning to say the least. And evidence of massive catch up does not yet exist. Learning recovery should be accelerated now with well-designed policy initiatives, especially in those countries where there are no plans yet in place.

The evidence is unequivocal: a crisis in education is still unfolding and students from the poorest backgrounds have borne the brunt. We owe it to them to stem the damage before it’s too late. This will require more effort and spending on what works recover learning losses and to further improve learning outcomes.