New literacy data shines a spotlight on the learning crisis

Girl reading New data show the impact of the COVID-19 pandemic on students’ reading proficiency. Copyright: Khasar Sandag / World Bank

Note: This blog was updated on July 26, 2023, following a closer scrutiny of data, ensuring that the differences reported are directly comparable and statistically significant.

Since 2018, the World Bank has warned early and frequently about a silent learning crisis brewing for a generation of children. This crisis has deepened with the COVID-19 pandemic, which triggered school closures for 1.5 billion students globally. Since then, researchers have undertaken painstaking empirical work to detail overviews of learning and enrollment losses. Data that reveals the extent of the crisis is of crucial importance to catalyze bold investments and reforms of education systems.

New data highlights low reading outcomes

Today marks the release of the Progress in International Reading Literacy Study (PIRLS) 2021 survey, the first internationally comparable assessment of reading outcomes since the start of the pandemic. The new PIRLS data allow us to analyze reading outcomes of 4th graders in 57 countries and territories, including 45 countries for which data are available from both before (2016) and well into the pandemic (2021).

Generated by a long-running, rigorous international assessment program on primary-school reading comprehension, these data provide a strong basis for analyzing time trends and supportive factors linked to reading outcomes over two decades.

We hope that these data get the attention they deserve and hope to see four key headlines on the study.

1. You cannot hide learning losses

Between 2016 and 2021, average achievement scores declined in 20 out of 31 (or 65%) countries and territories for which directly comparable data are available (and these differences are statistically distinguishable from zero). Average achievement scores went down in countries in very different contexts, including in high income countries. In two countries, average PIRLS scores decreased by more than 30 points: Azerbaijan (-32) and South Africa (-32). In simple terms, an average student in these countries may have fallen behind almost one year in reading progress, if we apply the assumption that children gain about 44 points in a school year. Beyond this, PIRLS data do not report on school dropouts, which also accelerated during the school closure period in many countries.

2. A larger share of students are poor readers

PIRLS provides information on the share of students achieving at least minimum reading proficiency (a score of 400), where children below this proficiency level are those unable to locate and retrieve explicit information from texts or make straightforward inferences from written text. The share of children below minimum proficiency (which IEA calls the "low international benchmark”) is used by countries to report on SDG 4.1.1, and makes up one component of “Learning Poverty” (the other component being the share of children out of school).

Out of 31 countries and territories for which PIRLS data is comparable between 2016 and 2021, 22 countries (or 71%) show that in 2021 more children were below minimum reading proficiency than in 2016. The biggest differences are again visible in Azerbaijan and South Africa. These countries show an increase (i.e., a deterioration) in the share of children not meeting minimum proficiency of, respectively, 7 percentage points (from 2% to 9%), and 3 percentage points (from 78% to 81%).

The share of children below the minimum proficiency level are affected by a reduction in average scores and changes in inequality--in some countries, the pandemic is likely to have affected poor readers.

3. Some countries show improvements.

Some countries did show learning gains between 2016 and 2021, particularly in the Middle East and North Africa region. For instance, 5 out of 6 comparable countries showed improvements, and Egypt improved in average achievement of 46 PIRLS points. However, the shares of poor readers remain high: in Egypt, the share of children below minimum proficiency dropped from 69 to 55. The share of children below minimum proficiency is 53% in Jordan, 41% in Iran, Islamic Republic, and 41% in Oman.

While these positive trends matter and are a good start, MENA countries still need to do more to catch up: a staggering one in three students in the MENA region overall does not meet minimum proficiency in reading by the age of 10, figures that are low given the income levels of the gulf countries. The region has also one of the largest gender gaps in learning (in favor of girls).

4. The new data are a powerful reminder that the world is learning crisis, which has been made worse by the pandemic.

Around the time of the previous PIRLS wave in 2016, countries committed under SDG4 (in the Incheon Declaration) to ensuring that all children receive a quality primary education by 2030. A low bar to measure “quality education” is that all children read with comprehension. To achieve SDG4, we still need a dramatic acceleration in the quality of education and learning levels, hence a result where many countries show stagnation or increases in learning poverty reveal an extremely worrying situation.

Unfortunately, the PIRLS data show that many countries are further away from this goal than they were before. We recognize that teachers, schools, and communities have made many efforts to remediate losses. The data show that bold and transformational investments and reforms are still urgently needed to address recent learning losses, as well as the learning crisis before the pandemic. During this period of learning recovery, these reforms could follow the RAPID framework, a menu of policies compiled by the World Bank and its partners from available evidence: Reach all children; assess learning, using a diverse set of learning assessments; prioritize the fundamentals; increase the efficiency of instruction; and develop children’s overall wellbeing. Many of these policies would help countries both recover from the pandemic, as well as with low learning levels in general.

Deeper analysis still needed

There are many factors to understand better, such as the long-term changes that go beyond the school closures and the pandemic effect alone. Likely, the data presented above understate the losses due to the pandemic—given that there could have been pre-pandemic learning improvements between 2016 and 2020—and we expect to see deeper analyses that control for other factors. Some countries had to postpone fieldwork to the start of Grade 5 due to the extraordinary circumstances during which data have been collected.

The data will keep us busy for some time to come. For example, we now have more information on the vast underperformance of boys in reading as compared to girls, the amount of reading instruction in an average school week, teachers’ perspectives, and children’s and parents’ reading behavior and reading enjoyment outside of the school. This wealth of data can help us improve education systems in many corners of the globe. High-quality data such as these are a global public good, and we can only hope that they will be used as much as possible by researchers and policymakers around the world.


Jaime Saavedra

Human Development Director for Latin America and the Caribbean at the World Bank

Sheena Fazili

Junior Professional Associate

Yi Ning Wong

Junior Professional Associate at the World Bank's Education Practice

Join the Conversation

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