EDUCATION FINANCE WATCH 2021 EDUCATION FINANCE WATCH 2021 Introduction To achieve national and international education goals, many countries will need to invest more in their education systems. During the last decade, government education spending has increased steadily, but the Covid-19 pandemic has impacted public finances dramatically, and the prospects for maintaining these increases have deterio- rated. But the education finance challenge is not only about mobilizing resources but also about improving the effective- ness of funding. Unfortunately, recent increases in public education spending have been associated with relatively small improvements in education outcomes. Although access to education has improved, 53 percent of ten-year- olds in low- and middle-income countries are unable to read and understand a short age-appropriate text (World Bank 2019).1 Tackling the large spending inefficiencies and World Bank / Ousmane Traore (MAKAVELI) inequalities common to many education systems will be vital in order to make better use of resources and strengthen the link between spending and education outcomes. development assistance data at the end of the year. Each EFW will follow a similar format and explore annual This Education Finance Watch (EFW) is a collabora- changes in the main sources of education funding tive effort between the World Bank and UNESCO’s (namely governments, households and development Global Education Monitoring (GEM) Report. It assistance).2 Each year the EFW also aims to shine a draws together and summarizes the available information spotlight on an important education financing issue. on patterns and trends in education financing around The spotlight for this year’s EFW is on the impact of the world. The EFW will be published every year soon Covid-19 and provides a snapshot of how education after the main annual release of public expenditure and budgets are changing in response to the pandemic.3 1 Projections suggest that Covid-19 related school closures are likely to increase this share to between 56 and 63 percent depending on the effective- ness of mitigation measures (World Bank 2020c). 2 As this is the first EFW, it takes a retrospective view of spending trends over the past decade, but future EFWs will focus on annual spending changes. 3 Since the Covid-19 pandemic affects all sources of education funding, the assessment of its impact is reported separately for each source. Two-thirds of low- and lower-middle-income coun- tries, included in the data collected for EFW, have cut their education budgets since the onset of the Covid-19 pandemic. In comparison, only a third of upper-middle and high-income countries have reduced their budgets. These budget cuts have been relatively small thus far, but there is a danger that future cuts will be larger, as the pandemic continues to take its economic toll, and fiscal positions worsen. These differing trends, imply a signif- icant widening of the already large spending disparities seen between low- and high-income countries. The quality and coverage of education finance data is low. As Section 3 shows, the availability of good quality data on core spending indicators is limited. This can make it difficult to track overall levels of funding and how these funds are utilized. Over the past three years, for example, fewer than a fifth of countries reported how much they spend on primary, secondary and tertiary education to UNESCO or the IMF. Real-time infor- mation on budget changes, so important in tracking and responding to crises, is not systematically available. EFW will also monitor changes in the quality and coverage of international education spending data and highlight efforts to improve systems and approaches for collecting and using spending data. The data and analysis in each EFW are publicly available, and feedback from readers ILO/Minette Rimando is welcome, with a view to future improvements.4 UNICEF Rwanda/2020/Saleh 4 A technical note is also available, explaining the sources of data, how they have been combined and approaches to estimation. 2 1 EDUCATION FINANCE WATCH 2021 Mobilizing resources for education GPE/Tabassy Baro 1.1. How has global education slightly increased their spending over the 10-year period (Figure 1). By contrast, rates of growth were much faster spending changed in the last ten in low- and middle-income countries, where spending years? on education rose by 5.9 percent a year, such that by the Prior to the Covid-19 pandemic, global education end of the period it was more than 77 percent higher in spending grew steadily, with low- and middle-income real terms than at the start. The contribution of house- countries registering the fastest growth. Global educa- holds, governments and development assistance to global tion spending grew annually in real terms by 2.6 percent education spending has remained relatively constant over per year between 2009 and 2019.5 This rate of increase is time, with governments contributing around 82 percent, slower than global economic growth rates and masks two households 17 percent, and development assistance less diverging trends. High-income countries, which account than 1 percent.6 for about two-thirds of global education spending, only Figure 1: Real spending on education has grown steadily over the last 10 years Government, household and official development Growth in real education spending, by income level assistance spending on education, constant 2018 (2009=100) US dollars (trillions), 2009–2019 Source: EFW team estimates using data from OECD, UIS, IMF and World Bank online databases. Notes: In order to estimate global spending, estimates are used for missing country level data (see EFW technical note). World Bank income group classifications in 2020 are used to group countries: LIC = low-income country, LMIC = lower-middle-income country, UMIC = upper-middle-income country, and HIC = high-income country. Global education spending refers to expenditure on education services by governments, households and donors and follows UNESCO Institute of 5  Statistics definitions. The share of development assistance in total education spending is 18 percent in low-income countries, 2 percent in lower-middle-income countries 6  and 0.3 percent in upper-middle-income countries. 3 Figure 2: Except in low-income countries, governments account for the bulk of education spending Distribution of total education spending by source, constant 2018 US dollars (billions), 2018–2019 Source: EFW team estimates using data from OECD, UIS, IMF and World Bank online databases. Notes: In order to estimate global spending, estimates are used for missing country level data (see EFW technical note). 60% of aid expenditure is considered reflected in public budgets (therefore subtracted from government expenditure), with the remaining 40% allocated through other channels. Governments, households, and development part- 1.2. How has government ners—in that order—are the main funders of educa- education spending changed over tion. Their contributions differ significantly across the last ten years? country income groups (Figure 2). Governments are the largest funders of education in all income groups, Government funding for education as a share of while the direct contribution of households to educa- national income has not changed significantly over the tion spending tends to be greater in poorer countries. last decade for any income grouping as a whole.7 For In 2018–19, for instance, households in high-income example, between 2010–11 and 2018–19 government countries accounted for only 16 percent of total educa- education spending as a percentage of GDP remained tion spending compared to 38 percent in low-income at 4.3 percent in lower-middle-income countries, and countries. Household spending on education is several increased only slightly from 3.2 to 3.5 percent of GDP in times larger than official development assistance in both low-income countries (Figure 3). However, these income low- and lower-middle-income countries. The impor- group averages mask differences in trends between indi- tance of aid for education funding is also very different vidual countries. In the low-income country grouping, among income groups. In low-income countries, aid government spending as a share of GDP increased from to education represents 18 percent of total education 3.9 to 5.5 percent in Burkina Faso, for example, between spending while in lower-middle-income countries it 2014–15 and 2018–19, but declined from 5.2 to 4.4 only accounts for 2 percent. The next section looks at how percent in Malawi over the same period. funding from governments, households and development assistance has changed over the last ten years and how Current levels of government spending in low- and the Covid-19 pandemic is affecting current and future lower-middle-income countries fall short of the funding for education from these three sources. levels required to achieve the SDGs. Governments in richer countries tend to devote a greater share of national income to education than low-income countries Given data limitations, the EFW does not report government spending by levels of education (see Figure 11). However, broad patterns show that 7  about 40 percent of total government spending is devoted to primary education, 40 percent to secondary, and 20 percent to tertiary. 4 EDUCATION FINANCE WATCH 2021 Figure 3: Government education spending as a share of GDP has remained flat in the past 10 years Government education spending as % of GDP by Total public education spending per capita income group (constant 2018 US dollars) Source: EFW team estimates using data from OECD, UIS, IMF and World Bank online databases. Note: In order to estimate global spending, estimates are used for missing country level data (see EFW technical note). Per-capita figures use total pre-primary to tertiary school-age population. (Figure 3). In 2018–19, public education spending in approximately US$ 48 per school-aged child compared high-income countries was 4.7 percent, compared to to US$ 8,501 in high-income countries. Moreover, 3.5 percent in low-income countries. As populations average trends in per-capita spending in low-income and in low- and middle-income countries are much larger, lower-middle-income countries are diverging in relative the share of GDP spent on education is insufficient to and absolute terms from spending in upper-middle- and achieve the improved quality of education—and access to high-income countries. For example, the gap in spending it—envisaged by the SDGs. A study undertaken by the per capita between low- and upper-middle-income coun- Global Education Monitoring Report in 2015 estimated tries increased from US$ 813 to US$ 1,045 in real terms that education spending in low- and lower-middle-in- between 2010–11 and 2018–19 (Figure 3). Differences come countries would need to increase from 3.5 percent in the growth of education spending per child or young to 6.3 percent of GDP between 2012 and 2030 to deliver person are partly the result of differences in population universal pre-primary, primary and secondary education growth rates between low-income and high-income coun- (UNESCO 2015). tries. Between 2009 and 2019, public education spending doubled in real terms in low-income countries (Figure 1) Disparities in spending on education per child or young but this translated into an increase in per-capita spending person between rich and poor countries are large and of only 30 percent (from US$ 37 in 2010-11 to US$ 48 in have continued to widen.8 Levels of spending vary enor- 2018-19). By contrast, spending in high-income countries mously between countries in different income groups. only increased by about 16 percent but spending per-capita For example, in 2018–19, governments in upper-mid- increased by 13 percent (from US$ 7,544 to US$ 8,501). dle-income countries spent 20 times more per child or young person than low-income countries and four times Regional and income group averages mask large as much as lower-middle-income countries. The differ- country differences in spending trends (Figure 4). In ences are even starker between low- and high-income roughly 40 percent of low-income countries, spending countries. In 2018–19 low-income countries spent per child or young person has fallen; in Chad, for instance, Trends in spending are tracked on the basis of overall government education spending per child or young person (defined as the pre-primary, pri- 8  mary, secondary and tertiary school-age population). This helps to compare levels of funding between countries or groups of countries, accounts for differences in the size and growth of child and youth populations across countries, and also provides an assessment of the adequacy of funding for all children rather than merely those who are able to attend schools, universities and other educational institutions. 5 Figure 4: Regional and income group averages mask wide disparities in country spending trends Changes in government education spending per capita (constant 2017 PPP US dollars), 2013–14 and 2017–2018 Low- and lower-middle-income countries Upper-middle-income countries Source: UIS, IMF and World Bank online databases. Note: Per-capita figures use total pre-primary to tertiary school-age population. The blue arrows show a rise, while the orange arrows show a fall. Countries are ordered in terms of magnitude of change. See EFW technical note for information on compilation of dataset. 6 EDUCATION FINANCE WATCH 2021 this spending fell from PPP$ 104 to PPP$ 79 between between low-income and high-income countries are 2013–14 and 2017–18. Meanwhile, in most low-income not attributable to differences in the priority accorded to countries spending per child or young person increased education in their budgets. Total government spending modestly over this period, with some countries regis- as a share of GDP is much higher in high-income than tering large improvements. In Sierra Leone, for example, in low-income countries. In 2018–19, for example, in government education spending per child or young low-income countries the share of education spending person rose from PPP$ 99 to PPP$ 207 between 2013–14 in the total budget was 15 percent while total public and 2017–18. Over the same period, spending per child spending as a share of GDP amounted to 22 percent. or young person fell—albeit generally only slightly—in By contrast, high-income countries devote a slightly fewer than a quarter of lower-middle-income countries, smaller share of their budget to education (13 percent) mostly in Africa. In Ghana, for instance, government but total public spending as a share of GDP was much spending per capita fell from PPP$ 500 to PPP$ 467 larger (30 percent). That is, the main difference between over that period. All countries in South Asia and many high- and low-income countries stem from the differ- in East Asia, Europe, and Central Asia recorded increases ences in the overall size of the public sector, not differ- in real spending per capita, such as in Sri Lanka, where ences in how education is prioritized. Moreover, larger government education spending per capita rose from GDP and smaller cohorts of school age children and PPP$ 620 to PPP$ 1,013 over that period. youth translate into much higher per-capita spending on education in high-income countries compared to Large differences in education spending as a share of low-income countries. GDP between low-income and high-income countries are not due to differences in the priority accorded to Mobilizing more government funding is often education in government budgets but due to differ- constrained by limited revenues. Figure 5 shows the ences in the share of overall government spending in magnitude of government spending as a proportion of GDP. Government education spending as a share of GDP, and the share of this spending devoted to educa- GDP is the product of the share of total government tion. Some countries (such as Uganda or Papua New spending in GDP and the share of total government Guinea) fall below income group averages for both spending devoted to education. The large differences indicators while other countries (such as Honduras or Figure 5: Fiscal space for mobilizing greater funding for education varies considerably across countries Education as a share of total government budget, and government spending as a share of GDP in low- and lower-middle-income countries (%), 2017–18 Source: EFW dataset using UIS, IMF and World Bank online databases. See EFW technical note for list of ISO codes. 7 Uzbekistan) are well above. The dashed lines in Figure (if these are included in the education budget), but 5 plot the combinations of the two spending indicators exclude any education spending that local governments to mark out the zone between two benchmarks—of make from their own resources.10 Second, budget plans 4 and 6 percent of GDP respectively—associated are used, rather than actual spending or midyear allo- with adequate government spending on education cations (except in the case of Indonesia) which may be in successful countries. For many countries reaching important in some countries. However, the budget plans such levels would require increases in overall domestic provide clear signals of the prioritization of spending on resource mobilization (total government spending as education and thus can be taken as an early indicator of a share of GDP) in addition to increasing the propor- changes in education spending trends.11 tion of the budget devoted to education. For example, given current levels of total government spending, A simple comparison of growth in education budgets Uganda would have to devote more than a quarter pre- and post-Covid shows a positive trend: on average of its overall budget to education to reach even the annual budgets increased by 4 percent post-Covid lower benchmark of 4 percent. Given the competing compared to 1.1 percent pre-Covid. However, the priorities for government spending, it seems unlikely share of countries that reduced their education budget that significant increases in education funding will be increased from 45 to 52 percent. The magnitude of the realized with just greater prioritization of education changes was also generally positive: average decreases in in the government budget. In many countries, it will countries that reduced their budgets were smaller (-8.3 also require broader domestic resource mobilization compared to -4.6 percent) and average increases slightly efforts to increase government revenues. larger (8.7 percent compared to 13.2 percent). However, breaking the sample down by income group shows some The effects of Covid-19 on government important differences behind these headline figures. education budgets In order to understand the short-term impact of the Education budgets declined after the onset of the Covid-19 pandemic on education budgets, informa- Covid-19 pandemic in 65 percent of low- and lower- tion was collected for a sample of 29 countries across middle- income countries compared to only 33 percent all regions.9 The sample represents about 54 percent of high- and upper-middle-income countries (Table 1). of the world’s school and university aged population. Despite more low- and lower-middle-income coun- The information collected was then verified with World tries decreasing their budget, the average reductions Bank country teams. The dataset includes budget data were smaller post-Covid than pre-Covid (-9.1 versus for two years before the onset of the Covid-19 crisis and -3.6 percent). Average budget increases for these coun- the first budget after the Covid-19 crisis, allowing for a tries were slightly higher post-Covid (12.4 versus 8.7 comparison of trends before and after the crisis began. percent). However, the trends for higher income coun- This represents a first effort to provide a broad picture tries were more positive. Not only did a higher share of trends in education budgets post-Covid-19: it will of upper-middle and high-income countries increase need to be expanded and refined in future, as it has a their budgets, but the average increase was higher than few drawbacks. First, the data include the resources of a in low- and lower-middle-income countries (13.8 central government, including local government transfers percent versus 12.4 percent). These divergent trends 9 The sample includes three low-income countries (Afghanistan, Ethiopia, Uganda); 14 lower-middle-income (Bangladesh, Egypt, India, Kenya, Kyrgyz Republic, Morocco, Myanmar, Nepal, Nigeria, Pakistan, Philippines, Tanzania, Ukraine, Uzbekistan); 10 upper-middle-income (Argentina, Brazil, Colombia, Jordan, Indonesia, Kazakhstan, Mexico, Peru, Russia, Turkey); and two high-income countries (Chile, Panama). Except for Ethiopia and Kazakhstan, for which expenditure at the local level was available for 2021. The following countries have education shares 10  below 10 percent and therefore are likely to have other main financing sources besides budget assigned by the central government: Argentina, Brazil, Egypt, India, Myanmar, Nigeria, Pakistan, and Russia. The timing of the fiscal year of the country and the date of the approval of the budget is used to define pre- and post-Covid periods. Pre-Covid 11  budgets are defined as those approved before March 2020, before the onset of the Covid-19 crisis. This is important because the fiscal year varies for countries in the sample and means that the first post-Covid budgets were approved at different times in different countries: February 2021 for those with fiscal years ranging from April to March, July/August 2020 for those with fiscal years ranging from October to September, April/May 2020 for those with fiscal years ranging from July to June, and late 2020 for those with fiscal years ranging from January to December. 8 EDUCATION FINANCE WATCH 2021 Table 1: Change in education budgets before and after Covid-19 (%) All countries High- and Low- and upper-middle- lower-middle- income income Pre- Post- Pre- Post- Pre- Post- Covid Covid Covid Covid Covid Covid Total education budget Change in education budget 1.1 4.0 2.2 6.7 0.3 2.0 Share of countries decreasing education budget 44.8 51.7 41.7 33.3 47.1 64.7 Average decrease in countries that decreased their budget -8.3 -4.6 -6.9 -7.4 -9.1 -3.6 Average increase in countries that increased their budget 8.7 13.2 8.8 13.8 8.7 12.4 Education’s share of the total budget Average change in percentage points -0.5 0.1 -0.3 -0.1 -0.6 0.2 Average share 13.4 13.5 13.7 13.6 13.2 13.4 Source: Country budget documents for selected countries. Note: Budget changes are expressed in real terms. will exacerbate pre-existing inequalities between low- affecting their education budgets post-Covid. It is and lower-middle-income countries and high- and therefore to be expected that if budgets were decreasing upper-middle-income countries in education spending pre-Covid, the trend would continue, in the absence of (Figure 3). radical policy changes. On the other hand, a positive pre-Covid trend is less obviously predictive of the post- The share of the total budget going to education did not Covid outcome, which would likely be determined by change very much. Taken together with the magnitude the impact of the crisis on the country. A summary of of the changes explained above, this indicates that total the changes in trends is given in Table 2. budgets grew faster in upper-middle and high-income countries after the crisis (6.7 percent versus 2.0 percent Analysis of the trends in individual countries reveals a in low- and lower-middle-income countries), and despite very mixed picture. Table 2 shows the share of countries a smaller share of that growth going to education, the that increased or decreased their budget post-Covid, resulting increase was higher than for lower-income according to their trends pre-Covid. For the overall countries (Table 1). Again, this signals a widening of sample, it shows that just above half of the countries spending inequality between country income levels. (52 percent) continued the previous trend (24 percent of them a decreasing trend and 28 percent an increasing Understanding the broader determinants of these trend). The rest of the countries shifted trends: 21 percent changes in the budget is important, but beyond the of the countries with decreasing budgets increased their scope of this year’s EFW. However, a first step is to look budgets post-Covid, and 28 percent of countries with at the correlation between the trends pre-Covid and the increasing budgets decreased their budgets post-Covid. trend post-Covid for individual countries. Research has shown that the growth in education spending globally Lower-income countries are more likely to continue a is explained largely by economic growth and overall decreasing trend in their education budgets or to shift budget growth, rather than changes in the share of from a positive to a negative trend after Covid. The the budget going to education (Al-Samarrai, Cerdan- differences, again, suggest that inequalities in education Infantes, and Lehe 2019). Thus, countries with poor spending are likely to increase. While only 8 percent of macroeconomic performance pre-Covid are likely higher income countries continued a decreasing trend, to have much worse performance post-Covid, likely 35 percent of lower income countries did. Furthermore, 9 Table 2: Change in education budgets after Covid-19 by earlier status (%) All countries High- and upper- Low- and lower- middle-income middle-income Countries with decreasing Decreased post-Covid 24 8 35 budgets pre-Covid Increased post-Covid 21 33 12 Countries with increasing Decreased post-Covid 28 25 29 budgets pre-Covid Increased post-Covid 28 33 24 Total 100 100 100 Source: Country budget documents for selected countries. while only 25 percent of higher income countries Responding to the Covid crisis requires additional flipped from positive to negative growth in the budget, spending to adapt schools for compliance with the 29 percent of lower income countries did. Conversely, necessary measures to control contagion and to while 33 percent of higher income countries flipped fund programs to make up for the losses in learning from a negative to a positive trend, only 12 percent of students experienced while schools were closed lower income countries did. Also, a greater share of (Tanaka 2020). It is not clear that countries that have higher income countries with a positive trend pre-Covid seen a decline in their education budget will be able continued that trend afterwards than did lower income to cover these costs alongside the regular increases in countries (33 percent versus 24 percent). funding needed to support growing school-age popu- lations. Despite the urgent need for adequate funding When comparing budget changes for countries by to allow school systems to reopen safely, about half of their pre-Covid levels of learning adjusted years of the countries in the sample cut their education budgets. schooling (LAYS), countries starting from a higher level This scarcely bodes well for the future, when macroeco- pre-Covid saw their budget increasing more than those nomic conditions are expected to worsen. starting from a lower level. The sample was divided into “high LAYS” and “low LAYS” pre-Covid using the median 1.3. How has aid to education value of the Learning Adjusted Years of Schooling Index in the sample countries.12 Low LAYS countries increased changed? their budgets by 2.5 percent on average, compared with Aid has remained at about 0.3 percent of donor coun- 5.4 percent of high LAYS countries. About 47 percent of tries’ gross national income for the last 15 years. In high LAYS countries decreased their education budgets, 2005, European Union countries committed to allocate compared with 57 percent of low LAYS countries.13 0.7 percent of gross national income (GNI) to official Although this analysis is from a limited sample of coun- development assistance (ODA). While other countries tries, and the relevant trends will thus need to be followed have not committed to this target, it is a useful bench- up closely, these early trends indicate that inequality in mark to assess the priority they accord to ODA. The education spending and outcomes across countries is likely OECD Development Assistance Committee (DAC) to increase as a result of the Covid crisis. countries increased ODA from 0.21 percent to 0.30 percent of GNI between 2001 and 2019 (though most of that increase was in the early 2000s); and levels have remained around 0.3 percent since 2005. Of the 30 DAC countries, only five met the 0.7 percent target in 2019. The median LAYS in our sample is slightly lower than the global median (7.6 versus 7.9). When using mean LAYS (7.4 in our sample) to categorize 12  countries instead of the median, the results do not differ qualitatively. The average post-Covid decrease in countries that decreased their budget was much larger in high LAYS countries (-7.0 percent) than in low LAYS 13  countries (-2.6 percent). 10 EDUCATION FINANCE WATCH 2021 Since 2009, total aid as a share of the national income The share of aid allocated to different education levels of recipient countries has increased for low-income has remained relatively stable with basic education countries but fallen for lower-middle-income coun- receiving 43 percent of total aid between 2009 and 2019. tries. Between 2009 and 2019, total ODA as a share of Over the same period, secondary education received 19 GNI fell from 1.2 percent to 0.7 percent in lower-mid- percent of total aid to education and post-secondary the dle-income countries. However, in low-income countries, remaining 39 percent. A large share of aid allocated to it fell initially from 9.6 percent in 2009 to 7.9 percent in post-secondary education are scholarships and student 2015 but has since increased to 10.8 percent in 2019. It imputed costs for higher education in donor countries. For has exceeded 20 percent, on average, in recent years in the example US$ 4 billion (61 percent) of total aid to post-sec- Central African Republic, Liberia, Malawi and Somalia. ondary education in 2019 was used for these purposes. Aid for education has increased by 21 percent over the Since 2009, the Middle East and North Africa region last 10 years which translated into an annual increase exhibited the largest increase in aid to basic education. of 2 percent. Disbursements had increased rapidly in the Between 2009 and 2019, aid to basic education in the 2000s, more than doubling from US$ 6 billion in 2002 region increased from US$ 798 to US$ 1,214 million to reach a peak of US$ 13.3 billion in 2010. However, (Figure 7). Its share in overall ODA to basic education also over the next five years, aid to education stagnated, before increased from 18 percent in 2009 to 24 percent in 2019, picking up again in 2016 (Figure 6). Total ODA declined largely in response to emergencies, initially in Iraq and by 9 percent between 2010 and 2014 in the aftermath then in the Syrian Arab Republic and Yemen. Meanwhile of the great financial crisis. However, since 2014 aid aid to education in Sub-Saharan Africa has been on an to education has increased by 30 percent reaching its upward trend over the last five years. Between 2014 and highest recorded level of US$ 15.9 billion in 2019. 2019 aid to Sub-Saharan Africa increased from US$ 1.4 to US$ 1.7 billion. This is likely to underestimate the total allocated to Sub-Saharan Africa because a share of aid that Figure 6: Aid has increased by 21 percent over is not allocated to countries (unspecified aid) in the OECD the last 10 years database is likely to flow to the region.14 With respect to the Total aid to education disbursements, by education level, 2009–19 (constant 2018 US dollars: billions) distribution of ODA by country income group, low-income countries saw aid to basic education fall between 2009 and 2015. Since 2015, aid to basic education in low-income countries has been increasing and reached US$ 1.6 billion in 2019. Allocations to lower-middle-income countries have also fluctuated but have fallen from US$ 2.6 billion in 2016 to US$ 2.2 billion in 2019 (Figure 7). In per-capita terms, aid to basic education in low-in- come countries is higher than in lower-middle-income countries. Assuming that all unspecified recipients of aid to basic and secondary education are low- and lower-mid- dle-income countries, the total amount of aid received for those purposes by those countries, in 2019 was US$ 7.6 billion. These aggregate levels of aid translate into spending of US$ 14 per primary school-age child in low- and US$ 8 in lower-middle-income countries (constant 2018 dollars). Source: GEM Report team analysis based on OECD CRS (2021). In secondary education, it corresponds to US$ 7 and US$ 3, respectively, per secondary school-age adolescent. Between 2009 and 2019, unspecified aid averaged $1 billion per year of which approximately $370 million were disbursements by the Global Part- 14  nership for Education. About 79 percent of GPE disbursements went to Sub-Saharan Africa (UNESCO 2020). GPE disbursements to recipient countries are not recorded in the OECD DAC-CRS and are therefore included in the unspecified category in Figure 7. 11 Figure 7: Aid to basic education has increased the The effects of Covid-19 on aid to education most in the Middle East and North Africa Aid to education is likely to contract as a result of Aid disbursements to basic education (constant 2018 Covid-19. A joint UNESCO-UNICEF-World Bank US dollars: billions), 2009–2019 survey of national education responses suggested that 88 By region percent of low- and 76 percent of lower-middle-income countries had received development assistance to cover Covid-related costs in education (UNESCO, UNICEF, and World Bank 2020). Some of this aid may have been repurposed to respond to the emergency. For example, lower than expected GPE disbursements in 2018–19 mean that GPE was well positioned to respond flexibly to the pandemic by providing more than US$ 500 million in 2020 to help countries mitigate the impact of school closures. However, this does not mean that the prospect for aid is positive. Donor countries are likely—and some have already begun—to shift their budget away from aid to domestic priorities related to unemployment and support packages to businesses. Donor priorities may By country income group shift to health or other emergencies, as a survey of DAC members suggests (OECD 2020). Moreover, interna- tional student mobility—which accounts for US$ 3.1 billion of current aid allocations to education—has been curtailed as a result of the pandemic, so this is likely in effect to lead to a drop in aid to education since it accounts for US$3.1 billion of current aid allocations to education. Overall, some estimates predict that aid to education may fall by US$ 2 billion from its peak in 2020 and not return to 2018 levels for another six years (UNESCO 2020). Source: GEM Report team analysis based on OECD CRS 1.4. How has household education (2020). spending changed? Households in low- and lower-middle-income coun- tries tend to contribute a greater share of total education spending than those in upper-middle and high-income countries. As a share of national income, household spending on education in low-income countries tends to be approximately six times greater than spending by households in high-income countries (Figure 8). The greater spending by households in low- and lower-middle- income countries is also reflected in their share of overall education spending. For example, in 2018–19, households in low-income countries accounted for 43 percent of total education spending, compared to only 16 percent for households in high-income countries (Figure 2). 12 EDUCATION FINANCE WATCH 2021 Figure 8. Households contribute much more The effects of Covid-19 on household to education in low- and lower-middle-income education spending countries than households in upper-middle- The pandemic has resulted in a large and negative and high-income countries income and health shock for many households. Global Total household education expenditure, % of GDP poverty rates are expected to rise for the first time since the 1998 Asian financial crisis. Forecasts suggest that as many as 93 million additional people will fall into poverty between 2020 and 2021 as a result of the pandemic (World Bank 2020d).16 The largest increases in poverty are expected to be in South Asia and Sub-Saharan Africa, where education systems rely heavily on household financial contributions (see Figure 8). The impact of the pandemic on household incomes is likely to reduce their ability to support the costs of education. Unless these declines are actively addressed, they will most probably lead to higher school drop-out rates. Research on previous economic shocks in low-income countries demonstrated the resulting reduced incomes, lowered household educa- Source: EFW calculations based on EFW database. tion spending, and lowered rates of school participation Note: Total household spending as percentage of GDP is (World Bank 2020a). These effects are likely to be more computed by aggregating all expenditures by level (namely acute in the current crisis since education systems are also primary, secondary and tertiary) and dividing the total by the faced with additional costs associated with reopening country’s GDP. See EFW technical note for details on the calculations. LIC data is unavailable for the years 2016–2019. schools safely, and there is emerging evidence to show that in some countries households are expected to bear While data is limited, household education spending some of these additional Covid-related costs. Although as a share of GDP has increased in low-income coun- social protection programs have been stepped up in many tries and households still contribute significantly countries to support households through the pandemic, to the costs of education. Constructing averages for the response in low-income countries has been fairly low-income countries is only possible prior to 2016 small. For example, additional social protection measures but this shows that household spending as a share of in low-income countries averaged around US$ 6 per national income increased from 2.8 to 3.6 percent capita compared to US$ 28 and US$ 56 in lower- and between 2006–10 and 2011–15 (Figure 8).15 There has upper-middle-income countries respectively (Gentilini, been a decline, albeit modest, in education spending as Almenfi, and Dale 2020). a share of GDP by households in middle-income coun- tries since 2011–15. In contrast, household spending in high-income countries has increased steadily from 0.5 to 0.7 percent of GDP between 2006–10 and 2016–19. The bulk of total household education spending tends to support children in primary and secondary education. For example, data for 2012 from Ethiopia shows that 72 percent of total spending by households is on primary and secondary education. In 2006–2010, only 35 percent of the low-income countries in the EFW database had data on household spending as a share of GDP. In 2011–2015, 15  that proportion fell to 17 percent. Only three countries have data for both periods. Forecasts based on the US$ 1.90 poverty line. The global poverty rate is forecast to increase from 8.4 percent in 2019 to between 8.9 and 9.4 percent 16  in 2021. 13 2 Using funds equitably and efficiently Aulia Erlangga/CIFOR Public spending on education can be highly unequal, Figure 9: Public education spending is unequal with wealthier groups often capturing a greater share Distribution of total public education funding by of the available resources. Inequalities tend to be highest income quintile in poorer countries where differences in enrollment patterns by income quintile tend to be most pronounced and can result in significant inequalities in public educa- tion funding across the income distribution (Figure 9). In LICs, for instance, 40 percent of total public education funding benefits the wealthiest quintile, and only 10 percent the poorest. These inequalities can be exacerbated by subnational differences in education spending. It is very common for a child living in one part of a country to attend a school that receives several times as much funding as a roughly comparable school elsewhere in the same country. For example, in Sudan, spending per child is approximately six times as much in the highest Multiple between highest and lowest-spending regions spending region as it is in the lowest spending region (based on per-capita education spending) (Figure 9). Subnational public spending differences tend to reinforce existing patterns of poverty and disadvan- tage. In many countries, per-capita education spending is significantly lower in poorer regions than in wealthier regions. For example, in Uganda, the relationship between district per-capita spending on education and levels of poverty is negative and statistically significant (Manuel et al. 2019). There are large differences among countries in terms of translating government education spending into years of schooling and learning outcomes. On average, richer countries tend to have higher spending and better learning outcomes. However, some countries at similar Source: Left-hand panel: UNICEF (2020). Right-hand panel: levels of per-capita spending achieve significantly better Manuel et al. (2019) and various World Bank Public Expenditure outcomes. For example, as shown in the right-hand Reviews. panel of Figure 10, countries like Chad and Niger spend Note: Left-hand panel: black vertical lines represent 20 percent of public spending—if public spending were equally distributed, the amounts similar to those of Malawi, Afghanistan and poorest and wealthiest quintiles would each receive 20 percent of it. 14 EDUCATION FINANCE WATCH 2021 Sierra Leone, but achieve less than half of the learning Increases in spending per-capita over the last ten years adjusted years of schooling. Clearly many factors drive have not always improved education outcomes. A these differences in efficiency. Some factors relate to recent study showed that average increases in per-capita characteristics of service delivery that are difficult to education spending increased education outcomes by change. It is generally cheaper, for instance, to provide a relatively small amount. It showed that the spending education services in densely populated and more elasticity for cases where spending had increased was urbanized countries than their more sparsely populated only 0.08, meaning that, for every 10 percent increase counterparts. However, many education systems also in per-child spending, outcomes improved by only suffer from spending inefficiencies that are the result of 0.8 percent (Al-Samarrai, Cerdan-Infantes, and Lehe suboptimal spending decisions, limited accountability, 2019).17 Using the new data on learning adjusted years and the diversion of education funds for other uses. of schooling it is possible to look at how changes in spending per capita have affected education outcomes over the last ten years. Although only a small group of countries have available data, this shows that only roughly three-quarters of countries that increased spending per capita registered improvements in outcomes. Figure 10: Countries differ in how effectively they translate funding into outcomes Expenditure Per Child and Learning-Adjusted Years of Schooling (LAYS), 2020 All countries Countries with education spending below US$ 3,000 per-child (constant 2017 PPP dollars) Source: EFW calculations based on Human Capital Index (HCI) and EFW database and following an approach outlined in Al-Samarrai et al. (2019). Note: Spending per child is computed as total public education spending on primary and secondary education divided by the school-age population. The curved line is an estimated stochastic frontier. The further below the frontier a country lies the less efficient it is. The learning adjusted years of schooling (LAYS) is a metric that combines components of quantity (expected years of schooling) 17  and quality (harmonized learning outcomes). The expected years of schooling measures the number of years of school a child born today can expect to obtain by age 18. It is based on age-specific enrollment rates between ages 4 and 17 and has a maximum value of 14. Meanwhile, harmonized learning outcomes are calculated using a conversion factor. For more details on the methodology, see Filmer et al. (2018); Kraay (2018); Patrinos and Angrist (2018). 15 3 Monitoring and reporting education spending UNICEF Rwanda/2020/Saleh Progress in the education sector must be monitored Despite the importance of data for monitoring, compa- on the basis of information on funding levels and how rable information on core education finance indica- these funds are used. This is also central to accountability tors is only available for a relatively small number of and ensuring that commitments to improve education countries.18 The EFW has combined information on outcomes are backed up by adequate funding. Without government spending on education from three main this information, it is impossible to track trends in educa- sources: UNESCO Institute of Statistics (UIS), the tion funding and to tackle spending inequalities and International Monetary Fund (IMF) and the World inefficiencies outlined in EFW. Bank’s EdStats database. Even using all three of these Figure 11: Information on core education spending indicators is available for fewer than two-thirds of countries Number of countries with data on core education spending indicators, 2016–2019 Education spending as a share Education spending as a share of Spending by level of education of GDP government spending Source: EFW dataset using UIS, IMF and World Bank online databases. Note: The total number of countries covered in the EFW database is 218. For data reporting on spending by level, only those with complete data on primary, secondary and tertiary level are counted as ‘complete’; those with at least one data point on any of the levels are categorized as ‘incomplete’. 18 There are no large or consistent differences in data reporting rates for countries in different income groups. 16 EDUCATION FINANCE WATCH 2021 international data sources, only 61 percent of countries value of education spending as a percentage of GDP reported government education as a percentage of GDP for countries showing information in more than one in 2018. Information on spending at different education dataset. Reducing these differences, by trying to align levels was available in only 17 percent of countries, definitions and approaches, could yield markedly better much of that incomplete and sometimes inconsistent data coverage and quality. The Covid-19 section of EFW (Figure 11). While the UIS database provides the also shows that although information on spending is majority of data used in EFW (79 percent), other frequently available online it is buried in government sources tend to provide more information for the most budget documentation that is generally neither easy to recent years. For the year 2018, for example, it provided find nor to interpret. Working with countries to develop only 45 percent. This suggests that although a time lag protocols to use readily available information on govern- is inevitable in the reporting of cross-country data it is ment budgets and spending could also improve the possible to reduce it. quality and timeliness of spending data. Finally, there has been significant progress on developing better tools for Improved quality and coverage of data will necessitate collecting information on education spending through, ensuring the comparability of existing data sources for example, utilizing a National Education Accounts and adopting new approaches. The EFW has drawn approach and strengthening education modules in house- together the three main sources of government spending hold expenditure surveys conducted regularly by national data, but differences in approaches to measurement mean statistics offices (IIEP, UIS, and Pôle de Dakar 2016, that different sources of data provide different magni- Oseni et al. 2018). What is now most clearly needed tudes for core education indicators. For example, there is to encourage wider adoption of these approaches by was an absolute difference of 0.7–0.8 percentage points providing technical support to statistical and planning between the World Bank, IMF and UIS datasets in the agencies at the country level. GPE/ Kelley Lynch 17 4 Summary UNICEF/Kanobana Since 2010, funding for education has grown most rapidly in low- and lower-middle-income countries, where the gaps Global spending on between the funding needed to achieve the SDGs and current allocations are the widest. The EFW has highlighted the education has increased pandemic’s impact on household income and donor funding, continuously in absolute and this is likely to reduce education funding. The impact of terms over the last 10 years these reductions is likely to be felt most acutely in low- and lower-middle-income countries, where these sources of funding but the signs are that the make up a large share of total education spending. Although pandemic may interrupt the data collected for EFW on government education budgets paints a mixed picture, the deterioration in government finances this upward trend. over the medium term suggests that without concerted efforts to prioritize education, the outlook for mobilizing the required resources for education will worsen. The severity of the health and economic shock caused by the The policies that countries pandemic will determine to a large extent the policies required to finance the response in a given country, and thereby minimize adopt to protect and disruptions to the development of education there (World Bank increase education funding 2020b). However, EFW has highlighted that education systems in many countries face significant challenges in using funding will differ, but most can effectively. The pandemic has brought these spending inequalities make better use of the funds and inefficiencies into sharp focus, and over the medium term it allocated for education. is vital that spending be aligned more tightly to improvements in education outcomes. Tackling the global learning crisis and monitoring the impacts The pandemic also of the pandemic will require better information on how well education systems are functioning. This includes better infor- highlights the critical mation on the levels and sources of funding and how these importance of monitoring funds are used to ensure that education is available to all. As EFW has shown, more can be done with existing data sources to the patterns and trends in sharpen the picture of education financing. But efforts to build funding of education. capacity and systems to collect and track education spending are also needed to improve both the quality and coverage of existing sources. Next year’s EFW will report back on progress. 18 EDUCATION FINANCE WATCH 2021 References Al-Samarrai, Samer, Pedro Cerdan-Infantes, and UNESCO. 2015. “Pricing the Right to Education: the Jonathan David Lehe. 2019. “Mobilizing Resources Cost of Reaching New Targets by 2030.” Policy for Education and Improving Spending Effectiveness: Paper 18, UNESCO. Global Education Monitoring Establishing Realistic Benchmarks Based on Past Report, Paris. Trends.” World Bank Policy Research Working Paper 8773, World Bank, Washington D.C. UNESCO. 2020. “ COVID-19 is a serious threat to aid to education recovery.” Policy Paper 41, UNESCO. Filmer, Deon, Halsey Rogers, Noam Angrist, and Global Education Monitoring Report, Paris. Shwetlena Sabarwal. 2018. Learning-Adjusted Years of Schooling (LAYS): Defining a New Macro Measure UNESCO, UNICEF, and World Bank. 2020. What of Education. Washington, DC: The World Bank. have we learnt? Overview of findings from a survey of ministries of education on national responses to Gentilini, U., M. Almenfi, and P.Dale. 2020. “Social COVID-19. Paris, New York, Washington D.C: Protection and Jobs Responses to COVID-19: A UNESCO, UNICEF, World Bank. Real-Time Review of Country Measures.” 14, World Bank, Washington D.C. UNICEF. 2020. “Addressing the learning crisis: An urgent need to better finance education for poorest IIEP, UIS, and Pôle de Dakar. 2016. Methodology of children.” New York, USA: UNICEF. National Education Accounts. Paris: International Institute for Educational Planning, UNESCO. World Bank. 2019. Ending Learning Poverty: What Will It Take? Washington D.C. Kraay, Aart C. 2018. Methodology for a World Bank Human Capital Index. Washington, DC: World Bank. World Bank. 2020a. THE COVID-19 pandemic: Shocks to Education and Policy Responses. Washington D.C. Manuel, Marcus, Dan Coppard, Amy Dodd, Harsh Desai, Richard Watts, Zach Christensen, and World Bank. 2020b. The Impact of the COVID-19 Stephanie Manea. 2019. “Subnational investment in Pandemic on Education Financing. Washington D.C. human capital ”. ODI, London. World Bank. 2020c. Learning Poverty in the Time of OECD. 2020. “The impact of the coronavirus ( COVID- Covid-19: A crisis within a crisis. Washington D.C. 19) crisis on development finance.” OECD, Paris. World Bank. 2020d. Poverty and Shared Prosperity Oseni, Gbemisola, Friedrich Huebler, Kevin Robert 2020: Reversals of Fortune. Washington D.C. Mcgee, Akuffo Amankwah, Elise Legault, and Andonirina Rakotonarivo. 2018. Measuring house- hold expenditure on education: A Guidebook for designing household survey questionnaires. The World Bank. Patrinos, Harry Anthony, and Noam Angrist. 2018. Global Dataset on Education Quality: A Review and Update (2000–2017). Washington, DC: World Bank. Tanaka, N. 2020. “Additional Activities and Associated Costs for School Re-Opening under Covid-19 Pandemic – quick glance based on available infor- mation.” The World Bank, Washington D.C. 19 Acknowledgments EFW was prepared by Samer Al-Samarrai, Pedro Cerdan-Infantes, Aliya Bigarinova, Juanita Bodmer and Marianne Vital (World Bank) and Manos Antoninis, Bilal Barakat and Yuki Murakami (UNESCO Global Education Monitoring Report). It was prepared under the guidance of Omar Arias and Jaime Saavedra and benefitted from comments from Cristian Aedo, Jose Cuesta, Peter Darvas, Amer Hassan, Gabriela Inchauste, Silvia Montoya, and Lars Sondergaard. The brief was edited by John Steinhardt and laid out by Danielle Willis. 20