EDUCATION FINANCE WATCH 2023 Introduction The Education Finance Watch (EFW ) is a collaborative effort between the World Bank (WB), the Global Education Monitoring (GEM) Report, and the UNESCO Institute for Statistics (UIS). The EFW aims to provide an analysis of trends, patterns, and issues in education financing around the world. The EFW uses various sources of education, economic, and financial data from the World Bank, UIS, the International Monetary Fund (IMF), and the Organisation for Economic Co-operation and Development (OECD). The first volume of the EFW report (EFW2021) documented continuously increased global education spending in absolute terms over the decade but indicated that the COVID-19 pandemic would interrupt this trend. EFW2022 shed light on the impact of COVID-19 on global education spending in 2020, the first year of the pandemic, and revealed that half the analyzed sample of countries reduced their annual education spending, in real terms.1 This year, the EFW2023 updates analyses on trends and patterns of education spending for the past 10 years, up to 2021, the second fiscal year after COVID-19. As a special theme of this year’s volume, the EFW2023 sheds light on changes in the school-age population and projects its fiscal implications for the upcoming ten years for selected countries. In 2021, © Vincent Tremeau | World Bank | 2020 low-income countries (LIC) increased year-on-year total education spending (a total of government, households, and development aid) in real terms. This increase was driven by an Although the increase was notable, it was insufficient to close increase in government spending, which reached 50 percent of the learning gap sustained during the pandemic. Indeed, around total education spending, while official development assistance the world countries of all income levels are grappling with (ODA) to LIC decreased both in absolute and relative terms. pandemic-induced learning loss. 1 To account for inflation, we use real terms throughout the EFW2023. EDUCATION FINANCE WATCH 2023 © Vincent Tremeau | World Bank | 2020 The pandemic caused significant learning losses that will average) of the generation hit by the pandemic (Psacharopoulos harm the future of people and economies alike if the right et al. 2021). investments in education – in terms of amount, efficiency, and equity – are not implemented right now. School closures are Learning loss is significantly higher in middle-income countries expected to reduce the learning-adjusted years of education (MIC) and LIC, which already come from a low learning base. across developing regions by roughly a third to a full year This situation positions around 86 percent of the world’s current (Azevedo et al. 2021). This, combined with deskilling due to student population at risk of encountering lower future earnings prolonged unemployment, will likely lower future earnings within countries with tighter economic restrictions. Education and dent human capital (Fasih et al. 2020; Fuchs-Schündeln was hit the hardest in MIC. These countries account for 76 et al. 2022). Overall, the COVID-19 pandemic could reduce percent of the world’s student population and face a full year of developing regions’ potential growth by a further 0.6 percentage lost learning, which will likely contract future annual earnings points, to 3.4 percent over the next decade, absent reforms to by 9 percent and annual economic growth by 0.1 percent. Future boost underlying drivers of long-term growth (World Bank economic growth in LIC is likely to suffer the most because of 2021). Increasing investment in human capital can help reverse pandemic-induced learning loss. LIC learning loss is equivalent to the losses caused by overlapping adverse shocks of recent years 0.7 years of education, which is likely to lead to a reduction of 7.4 (Schady et al. 2023; World Bank 2023). percent in annual earnings which will translate into a 7.5 percent decrease in their annual economic growth. Larger learning losses Less learning translates into lower earnings over the course and tighter economic restrictions put LIC and MIC countries of an individual’s lifetime. While more years of education at high risk of falling into a vicious cycle where low education are associated with higher earnings, every standard deviation spending produces less learning in the aftermath of the pandemic (SD) increase (decrease) in cognitive skills is associated with a leading to lower economic growth, and lower economic growth substantial increase (decrease) in earnings. The global learning produces even lower investment in education, and so on. loss is equivalent to 0.7 years of lost learning (0.2 SD), which could translate into an annual reduction of 6.5 percent in the High-income countries (HIC) investing heavily in education future earnings of current students once engaged in a job, as a every year managed to minimize their learning loss. HIC result of lower productivity due to fewer cognitive skills. This encountered a learning loss equivalent to 0.4 years of schooling, reduction in earnings prospects could contract national income which could result in a decrease of 4.2 percent in future annual growth by 2.2 percent each year of working life (45 years on earnings and a contraction of 0.7 percent in national economic growth. 2 EDUCATION FINANCE WATCH 2023 Key findings: 1. The pandemic exacerbated the global learning crisis and government education spending is insufficient to close the learning gap. Along with a slight rise in annual real spending on education, government per-capita education spending increased in 2021. This is striking, given that it occurred during the height of the pandemic. However, spending as a percentage of gross domestic product (GDP) declined in all country income groups, except LIC. Government spending on education has historically accounted for less than half of combined education spending in LIC but reached 50 percent in 2021. In countries of all income levels, particularly LIC and MIC, spending increases were far from sufficient to even make a dent in the large learning gap. 2. Now that external support is needed most, ODA is falling – and could be spent more efficiently. Aid to education fell by 7 percent, from US$19.3 billion in 2020 to US$17.8 billion in 2021, because of the reduction in general budget support, which returned to pre-COVID levels. ODA for education continues to be important for LIC, accounting for 13 percent of their total education expenditure; nevertheless, the proportion of ODA devoted to education is declining. Only 30 percent of direct aid to education among the ten largest donors to sub-Saharan Africa (SSA) goes straight to © Ed Wray | World Bank | 2012 recipient countries; the remainder is funneled through donors’ aid agencies, international and domestic non-governmental organizations (NGOs), and multilateral organizations. Since experiencing or will soon face a decline in per-capita public 2017, there has also been a significant discrepancy between aid education expenditures partially caused by a growing school- commitments and disbursements among multilateral donors, age population, while others are seeing school-age population totaling US$1.7 billion in unused commitments over the five stagnation or decline and could free up fiscal space to increase years. per-capita education expenditure. 3. Families spend significant portions of their funds on 5. Closing the learning gap will require more efficient and education; more than one-third of total education spending equitable financing. In all countries, particularly those with in LIC and lower-middle-income countries (LMIC) is from relatively lower income, the level, efficiency, and equity of households. Within countries, household spending varies education spending is often inadequate to reach learning goals. considerably by household characteristics. This includes socio- More money will only partially solve it. In a current climate economic status, household location, education levels, and the of increasing inflation, high debt-to-GDP ratios in many type of school children are (or are not) enrolled in. Overall, countries, and falling ODA, particularly to LIC, spending households spend more money on non-state (private) than smarter is the imperative and urgent next step. Doing so can state (public) schooling. In selected African countries, it costs mitigate pandemic-related learning loss, helping to develop families 1.5 to 5 times more to send a child to a private rather the foundational skills needed to grow human capital and than public school. sustain economies into the future. 4. Demographic changes present challenges and opportunities 6. Education spending data availability has improved for education financing. While more education spending considerably, yet gaps remain. The actual availability of does not necessarily lead to better education outcomes, the latest data for the key education finance indicator, learning outcomes are lowest in countries spending the “governmental spending on education as a percentage of least per school-age child. LIC and LMIC exhibit a striking GDP,” fell from 76 percent of countries reviewed in July 2022 variation in demographic change: some countries are to just 66 percent in July 2023. 3 1 Global education spending trends This section puts observed education spending trends during the onset of COVID-19 into a longer-term perspective and analyzes changes in spending distribution by source. Education spending data availability has improved considerably, yet gaps remain (see Section 7). The EFW uses a coherent methodological approach to deal with missing education spending data by imputation Figure 1. Returns to education (percent) Country group Level Low income High income Tertiary © Atet Dwi Pramadia | World Bank | 2018 Primary to estimate total education spending for the past 10 years. The specific methodology and rationale are described in detail in the Education Investment accompanying Technical Note. Stock market Female The economic and social implications Gender Male of spending well on education 0.0% 5.0% 10.0% 15.0% 20.0% Return (%) Education is a key investment in human capital that translates Source: Author compilation based on various sources into economic growth. Education raises productivity (Sianesi and van Reenen 2003; Hanushek and Woessmann 2010; Hall and Jones 1999; Schoellman 2012; Hendricks 2002). According to harmonized learning outcomes (HLO) database, The longer a child stays in school, the greater the return. a change of 1 percent in learning is associated with a change Returns to education are highest at the tertiary level across of 7.2 percent in annual growth (Angrist et al. 2021). It is the world, even in LIC where there is a scarcity of skilled labor. an economically and socially productive investment partially This fact has been characterized as a race between education because of its positive association with earnings (Card 2018). and technology (Goldin and Katz 2009; Tinbergen 1974), One additional year of schooling increases earnings by 10 and shows that, while the demand for skills is increasing due percent (Psacharopoulos and Patrinos 2018; Montenegro and to a technological change that puts a premium on higher Patrinos 2021), which is greater than stock market returns order competencies, globally the supply of schooling is not (Heckman 2008; Arias and McMahon 2001). In almost every keeping up with demand for such skills. Further, workers country and at all levels of schooling, returns to education with higher education are better at learning new skills and are higher for women (10 percent) than for men (8 percent) adapting to changing conditions (Schultz 1975), for which (Figure 1). the labor market also pays a premium. Also, during crises, the employment prospects of tertiary education graduates LIC have the most to gain by investing in education. The social are protected by their skills, while the earnings of those with rate of return to human capital – compared with investment in secondary education or less are more vulnerable, which likely physical capital – is higher in LIC than in countries with greater contributes to persistent wage inequality. The insulating income levels (Psacharopoulos 1973; Psacharopoulos et al. 2017). effect of higher education was observed during the financial Also, the average payoff for one more year of schooling is higher crisis of the 1990s and the late 2000s recession (Cholezas et in LIC than in MIC and HIC. This also happens by education al. 2013; Fasih et al. 2021; Fiszbein et al. 2007; Patrinos and level. The average returns to primary, secondary, and tertiary Sakellariou 2006; Psacharopoulos et al. 1996), as well as during education are higher in LIC. the COVID-19 pandemic. Analysis of the returns to education 4 EDUCATION FINANCE WATCH 2023 Figure 2. Total real spending on education began to increase in 2021 following a brief period of stagnation due to the COVID-19 pandemic a. Government, ODA, and household spending on education, b. Evolution of total real education spending (all sources), by country constant 2021 US$, trillion, 2012 - 2021 income group 2012 – 2021 (2012 = 100) 6 170 LIC: Low-income countries 5.4 LMIC: Lower-middle-income countries 5.3 5.3 5.1 5.2 UMIC: Upper-middle-income countries 5 4.9 4.8 4.9 160 4.6 4.7 HIC: High-income countries LIC, $29 bil 150 LMIC, $454 bil 4 140 LIC, $18 bil Trillion $ LMIC, $305 bil UMIC, $1.58 tri 3 UMIC, $1.15 tri 130 HIC, $3.17 tri 2 120 1 110 HIC, $3.37 tri 100 0 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 90 Government Development Assistance Household Total 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Note: LIC: Low-income countries, LMIC: Lower-middle-income countries, UMIC: Upper-middle-income countries, HIC: High-income countries, follows the World Bank country income classification published in 2022 (https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups, accessed on 28 June 2023). Interpolation was used to fill in missing data and ensure a comparable sample of countries in all periods. Considering the change in data availability in the EFW2023 database post-imputation, there is a slight increase in UMIC from 2019 to 2020, reversing a decrease initially reported in EFW2022. This shift is largely due to data changes. Specifically, 2020 data were added for three more countries on net government education spending (net of ODA) in EFW2023, while the UIS reported seven fewer countries for actual spending. Source: Author estimates using the EFW2023 database. in fourteen countries for which we have data before, during, 15.0 percent during the pandemic and persisted above their and after the pandemic (Argentina, Bolivia, Brazil, Colombia, pre-pandemic level afterwards, at 14.7 percent. In contrast, Costa Rica, Ecuador, Egypt, Panama, Paraguay, Peru, Rwanda, the average returns to an additional year of schooling declined El Salvador, Thailand, and the United States), shows that during the pandemic (from 8.9 to 8.8 percent) and continued the average returns to higher education grew from 14.1 to their downward trend afterwards (to 8.5 percent). © Charlotte Kesl | World Bank | 2010 5 EDUCATION FINANCE WATCH 2023 How has total education spending Figure 3. In 2021, for the first time since 2015, changed over the last ten years, government spending made up 50 percent of total education expenditure in LIC particularly since the COVID-19 pandemic? Distribution of total education spending by source, year, and country income group, percentage, and billions US$ 2015 11 3 8 At the beginning of the pandemic, total global education 2019 13 4 10 income spending stagnated, but it increased again in 2021. The global Low 2020 13 5 10 total real expenditure on education, combining spending by 2021 15 4 11 governments, ODA, and households, inched upwards in 2021 2015 219 8 130 middle income to US$5.4 trillion after two consecutive years at US$5.3 trillion2 2019 257 8 162 Lower 2020 260 10 155 (Figure 2.a). This is a noteworthy development given that it Billion US$ 2021 283 9 164 occurred during the height of the pandemic. It is also slightly 2015 925 3 389 surprising since early reports warned of declines in spending Upper middle 2019 1,016 4 468 income (Alam and Tiwari 2021; Riggall et al. 2021; UN 2020). The 2020 1,028 5 466 overall increase (US$0.1 trillion) is driven by an increase in 2021 1,085 4 493 2015 2,594 0 486 government spending in all country income groups. The increase 2019 2,820 0 528 income in 2021 was modest in HIC, and more substantial in LMIC, High 2020 2,832 0 503 with a total year-on-year increase of 7.6 percent (Figure 2.b). 2021 2,834 0 533 This difference highlights how various country income groups 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% have reacted differently and recovered to pre-COVID-19 levels Government Development Assistance Household of education spending at diverse rates. Note: Interpolation used to fill in missing data and ensure a comparable sample of coun- tries in all periods. As in EFW2022, 218 countries and territories were included in the EFW2023 database. To avoid double-counting, government expenditure nets out part of Government resources remain the main education funding the ODA received by countries. source in all country income groups. More than three-quarters Source: Author estimates using the EFW2023 database. (78 percent) of global education spending in 2021 stemmed from government spending net of ODA contributions,3 which increased by 2 percent from 2020 in real terms (Figure 2.a). Households account for a sizable share of total spending in LIC expenditures within their countries, with households and donors and LMIC, accounting for 37 and 36 percent of the total, in playing a larger funding role than in countries with more income. 2021. After a 15 percent increase from 2019 to 2020, ODA to By contrast, UMIC and HIC mostly rely on governments for education decreased by 7 percent in 2021. their financial sources for education (around 70 percent and 85 percent, respectively, in 2021). Despite the proportional increase In LIC, government spending reached 50 percent of the in LIC government spending on education, ODA for education combined education expenditure in 2021. Historically, LIC continues to be important in LIC, accounting for 13 percent of governments have funded less than half of total education the total (Figure 3). 2 EFW2023 revised the estimates of total global education spending upward by approximately 8 percent over the past 10 years (total education spending in 2020 was estimated at US$4.9 trillion in EFW2022, but new estimates in EFW2023 is US$5.3 trillion). This upward revision was due to a combination of various factors, including the replacement of imputed data last year with newly available data this year, revisions of reported data in the past year, and a rebasing year for constant values from 2020 to 2021. 3 To avoid double-counting, aid received by countries is deducted from government expenditure. 6 2 Government spending Government funding of education makes sense because education and its returns constitute a global public good. Education is a basic human service, a human right, and optimal investment is often thwarted by market failures, borrowing constraints, and differences in household incomes and preferences, among others (Haveman and Wolfe 1984; Figure 4. Many LIC and LMIC fell short of either one or both international public education spending benchmarks in 2021 Education as a share of total government expenditure and as a share of GDP (%), 2021 25 © Deshan Tennekoon | World Bank | 2010 Oreopoulos and Salvanes 2011). Schooling is also a mechanism PRY for enhancing social cohesion and nation-building, producing GTM Combinations equal to public numerous positive externalities. These are some of the reasons ed. spend of 6% of GDP governments do and should continue to invest in education, even 20 in the face of economic, social, climate, and public health-related Education as share of TGE (%) LIC crises. Education builds human capital and human capital is the PSE LMI backbone of every country, strengthening its ability to withstand CIV PER PHL UMI and overcome, innovating into the future. TCD MLI KHM RWA BTW HIC 15 BRB NPLPSE SLE How does fiscal space vary across countries for mobilizing greater LAO TCA ECU funding for education? 10 BGD MDV TTOJOR MRT Combinations equal to 4% of GDP Governments are facing challenges in achieving the minimum LBR TLS KNA recommended international benchmarks on public education MCO spending. To ensure countries have enough financial resources 5 10 15 20 25 30 35 40 45 50 to provide education for all, UNESCO and its partner agencies Govt. spending as a share of GDP (%) established international benchmarks, namely that governments N=27 for all countries (LIC & LMIC = 16, UMIC & HIC = 11) spend at least 4-6 percent of their gross domestic product Note: Data presented here do not include interpolated values. Source: UIS database, accessed April 2023. (GDP) and/or 15-20 percent of total government spending on education (Incheon Declaration Education 2030). Of 27 countries with available data in 2021,4 we can distinguish three groups of countries: (1) countries that met both targets (four education spending exceeds 4 percent of GDP and 15 percent countries); (2) countries that achieved one or the other (16); and of total government expenditure. To meet recommended (3) countries that met neither (seven) (Figure 4). Interestingly, international benchmarks of over 4 percent of GDP, some our analysis reveals no definitive correlation between a country’s countries would need some combination of expanding the income level and its adherence to international benchmarks for share of government spending in the economy and increasing public education spending. Examples of the first group include the share of education in total spending. For instance, for Bhutan, Mali, Rwanda, and the West Bank and Gaza5, where Mauritania to reach 4 percent of GDP, either government 4 See Appendix for the full list of countries. 5 World Bank denomination. The denomination used by UNESCO is the State of Palestine. 7 EDUCATION FINANCE WATCH 2023 Figure 5. Comparative increase in per-capita government education spending as a share of GDP per capita in LMIC (2018-19 vs 2020-21) a. Changes in government education spending per capita (constant b. Changes in government education spending per capita as a share 2021 US dollars) by income group; two-year averages: 2012- of GDP per capita by income group; two-year averages: 2012-13, 13, 2018-19, 2020-21 2018-19, 2020-21 9,000 8,205 8,457 127 8,000 7,564 116 7,000 115 6,000 5,000 US$ 85 80 % 4,000 74 78 3,000 67 70 2,000 1,126 1,165 1,000 941 340 50 301 337 44 54 0 12-13 18-19 20-21 12-13 18-19 20-21 LIC LMIC UMIC HIC LMIC UMIC HIC Note: Estimates including interpolated values of government education spending are used in place of missing country-level data. The composition of countries is the same in the two time periods (2012-13, 2018-19, and 2020-21). Per-capita figures use total pre-primary to tertiary school-age population in UIS and UNDESA. Due to an excessive amount of missing data on government spending as a percentage of GDP per capita throughout the observed year, Figure 5.b is limited to only three country income groups (LMIC, UMIC, and HIC). Source: Author estimates using EFW2023 database. spending as a share of GDP would need to significantly increase When viewed per capita, government spending on education (from 19 to 45 percent) or the country would need to almost since 2012 increased in countries of all income levels, with HIC triple its share of education spending in total public spending improving the most (by US$1,008) and LIC improving the least (from 9 to 21 percent) while maintaining its current government (by US$14). The amount a country spends on education per child spending on GDP. is the most direct measure to assess whether sufficient resources are devoted to education. While it is difficult to establish a In the following subsections, we look at government education benchmark for the cost of ensuring quality education in different spending over the past decade, particularly during and after the countries and contexts, comparisons are informative. UMIC COVID-19 pandemic. and LMIC were in the middle, increasing government per capita spending on education over the past decade by US$276 and US$45, respectively (Figure 6.b). Changes in per capita How has government education education spending over the past decade are influenced by overall spending changed in the last decade? decreases in HIC countries and increases in LIC countries’ school-aged populations (see Section 6 for more on demographic Since 2012, government spending on education relative to shifts). GDP has increased the most in LIC followed by UMIC, while remaining stable in HIC and decreasing in LMIC. Zooming in on changes before and after COVID-19 onset, More recently, in HIC, government spending declined (0.2 per-capita government spending decreased in LMIC, and percentage points) from 4.9 percent (2020) to 4.7 percent rose in HIC and, to a lesser degree, in LIC. Upon evaluating (2021), and in UMIC (0.1 percentage points) from 4.9 the change in per-capita government spending across income percent (2020) to 4.8 percent (2021). In LIC, government categories, a downturn was detected in LMIC for 2020-21 education spending as a share of GDP has increased steadily (US$337) from 2018-19 (US$340), in contrast to continuing since 2018 (3.2 percent), reaching an average of 3.6 percent in growth in other income groups ( Figure 5 ). Only HIC 2021 (Figure 6.a). However, it is well below the international significantly increased expenditures, surpassing pre-COVID benchmark of 4-6 percent of GDP (Education 2030 levels. LIC have also increased spending, but to a far lesser degree Framework for Action). and from a very low base. 8 EDUCATION FINANCE WATCH 2023 Figure 6. In 2021, government education spending as a share of GDP increased only in LIC a. Government education spending as a percentage of GDP by b. Government education spending per capita (constant 2021 US income group, 2012-2021 dollars) by income group, 2012-2021 5.5 7,507 8,515, HIC LIMC, 4.8% UMIC, 4.8% HIC, 4.7% HIC, 4.7% 4.5 LMIC, 4.6% UMIC, 4.5% US$ % 1,195, UMIC LIC, 3.6% 919 3.5 LIC, 3.1% 341, LMIC 296 42 56, LIC 2.5 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Note: Estimates on spending as a percentage of GDP include interpolated values. Interpolation was done to fill in missing data and ensure a comparable sample of countries in all periods. Source: Author estimates using the EFW2023 database. Following the COVID-19 pandemic, overall, global income groups (Figure 6.a). Per capita education spending in government education spending per capita increased in 2021, LMIC increased from US$332 in 2020 to US$341 in 2021. after a downturn in 2020. In HIC, despite a discernible decrease Similarly, UMIC increased its per capita spending in education in the government share of education spending as a percentage on average from US$1,135 in 2020 to US$1,195 in 2021.8 Yet, the of GDP and while total spending on education stayed the same, global economy tentatively recovered in 2021 (IMF 2022). This there was a 2 percent increase in education spending per capita suggests that at a time when education investment should have (up from US$8,340 in 2020 to US$8,515 in 2021) (Figure increased to offset learning loss due to pandemic-related school 6.b). In LIC, though experiencing an increase in the proportion closures, the growth in education spending lagged economic of education spending compared to GDP and an expansion recovery rates in LMIC and UMIC. in overall educational expenditure, a comparatively marginal absolute rise in per-capita education spending (from US$52 in 2020 to US$56 in 2021, or by 8 percent in relative numbers) How has education spending changed was registered. over the past decade, by region? Despite these increases, many countries are not spending Over the past decade, government spending on education (per enough on education, particularly in LIC. In 2021, government capita) increased until the COVID-19 pandemic. After that per-capita spending on education was on average 152 times6 (2018/2019-2020/2021), all regions – except LAC – continued higher in HIC than in LIC.7 The amounts spent vary widely: to grow, though at a lower rate. Average per-capita government expenditures per child per year in LIC are on average US$56. In education spending rose in all regions between 2012-2013 and UMIC, US$1,195, and HIC, US$8,515. With this information, 2018-2019 (see all lines between the beginning and midpoints it is important to distinguish between a country’s commitment in Figure 7.a). Following the onset of the pandemic in 2018- to education and the resources it has available to invest. 2019 and until 2020-2021, government spending slowed around the world, dropping by US$88 in Latin America and the Per-capita spending on education grew year-on-year in 2021 in Caribbean (LAC), and rising at a more modest rate elsewhere real terms for LMIC and UMIC (Figure 6.b), but government (see decline in grey line and less-steep inclines in all other lines education spending as a percentage of GDP declined in these from the middle to endpoints in Figure 7.a). Overall, Europe 6 This is slightly lower than the 162 times noted in 2020 (World Bank and UNESCO 2022). 7 Within country income groups in 2021, per capita government education spending also varied significantly, from US$17 in the Central African Republic to US$123 in Tajikistan (both LIC), and from US$71 in Pakistan to US$1,076 in Moldova (both LMIC). 8 We use GDP deflator. 9 EDUCATION FINANCE WATCH 2023 Figure 7. Comparative decrease in per-capita government education spending (2012-13 vs 2018-19 vs 2020-21) a. Changes in government education spending per capita (constant b. Changes in government education spending per capita as a share 2021 US dollars) by region of GDP per capita by region 8,000 7,128 7,372 127 7,000 6,552 116 6,000 115 5,000 US$ 4,000 % 3,113 80 3,284 3,000 2,988 81 2,688 2,501 2,691 79 73 75 2,000 1,912 1,864 2,000 71 1,000 254 368 277 391 282 0 234 12-13 18-19 20-21 12-13 18-19 20-21 Sub-Saharan Africa South Asia Latin America & Caribbean Middle East & North Africa East Asia & Pacific Europe & Central Asia Latin America & Caribbean East Asia & Pacific Europe & Central Asia Note: Estimates including interpolated values are used in place of missing country-level data. The composition of countries is the same in the three time periods (2012-13, 2018-19, and 2020- 21). Per-capita estimates use total pre-primary to tertiary school-age population in UIS and UNDESA. Due to an excessive amount of missing data on government spending as a percentage of GDP per capita throughout the observed year, Figure 7.b is limited to only three regions (LAC, EAP, and ECA). Source: Author estimates using EFW2023 database. and Central Asia (ECA) registered the largest absolute increase EAP saw an increase in education spending per capita as well in the level of per-capita government spending on education as an increase in the proportion of economic output allocated (US$820) between 2012 and 2021. In relative terms, South to education spending per person between 2018-19 and 2020- Asia showed the largest increase with per-capita government 21, indicating average government education spending in those spending going up 54 percent over the decade. two regions grew between pre- and post-COVID-19 onset even while their economies were impacted by the pandemic. Government education spending per capita as a share of GDP per capita has not recovered. Educational spending per capita What is being spent is not always spent well. Achieving better as a share of GDP per capita shows the proportion of economic learning outcomes requires efficient use of adequate resources. output that is allocated to education spending per person (Figure There are wide variations in efficiency in spending between 7.b).6 In LAC countries, average education spending per capita countries: South Africa, an upper middle-income country, invests decreased between 2018-19 and 2020-21. At the same time, US$1,400 per child. However, it struggles with a persistently high education spending per capita as a share of GDP per capita learning poverty rate of 79 percent (2016 PIRLS). This level of increased, indicating the decrease is led by the overall economic learning poverty is like that of Guinea, a much poorer country, downturn experienced in LAC.9 By contrast, both ECA and which invests about US$50 per child.10 9 LAC region faced a significant economic downturn in 2020 (negative 6.5 percent annual GDP growth from 2019, which is more than double of world average of negative 3.1 percent that year). WDI. https://data.worldbank.org/indicator/NY.GDP.MKTP.KD.ZG, accessed on July 27, 2023. 10 Here it is interesting to note that the variance in human resources costs between South Africa (more) and Guinea (less) does not fully explain the large difference in their returns on education investment. 10 3 Education aid How has aid to education changed? Aid to education has declined. Over the period of a year, international aid to the education sector declined by almost US$2 billion, or by 7 percent in real terms (Figure 9).11 It fell © Irina Oleinik | World Bank | 2012 earmarked to education is dominant (over 85 percent) in total external finance flows (sum of ODA, Other Official Flows (OOF), private development finance, and equity investment) to education in 2021. In LICs, earmarked aid to education was particularly high (almost 95 percent) in 2021. The share of ODA in total external from US$19.3 billion in 2020, to US$17.8 billion in 2021.12 Total finance flows in the education sector is higher than in other sectors aid to education includes two factors: (1) direct aid from donors such as Transport and Energy, whose share is about 55 percent of that are specifically earmarked for education, and (2) a 20 percent total external finance flow (Figure 8). A large part of a reduction estimation of general budget support, which is aid provided to in aid can be attributed to a decline in general budget support. governments without being earmarked for specific projects or Excluding global budget support, direct aid to education has sectors.13 Direct aid (ODA grants and ODA loans) specifically remained stagnant since 2018 at around US$15 billion. Figure 8. Education receives most of its total external finance flows in the form of grants a. Total external finance flows, by sector and source, 2021 b. Total external finance flows to education, by source and income group, 2021 100 Equity investment, 0% 100 Private development finance, 4% 80 80 Other official flows, 10% ODA loans, 15% 60 60 % % 40 40 84 63 67 ODA grants, 71% 61 47 20 20 20 8 0 0 Transport Energy Agriculture Health Education Upper middle Lower middle Low Note: Definition: ‘Other official flows’ are bilateral other official flows (e.g., non-concessional developmental loans) except grants and loans for commercial purposes, and all grants and loans by multilateral development institutions. ‘Private development finance’ is defined as private investment mobilized by official interventions. OECD estimates the volume of private flows that are mobilized by guarantees, syndicated loans, and collective investment vehicles (OECD). Source: Author estimates based on the OECD CRS database (2023). 11 This includes only ODA grants and loans for education; it does not include other official flows (OOF) such as non-concessional loans (e.g., IBRD loans) or private development finance. 12 The most recent year with sector and recipient level data. 13 For this report, we estimate that 20 percent of direct budget support to governments is diverted to education. 11 EDUCATION FINANCE WATCH 2023 Figure 9. Aid to education fell by 7 percent from 2020 to 2021 Total aid to education disbursements, by education level, 2010-2021 20 18 16 Constant 2021 USD billion 14 12 10 8 6 4 2 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Basic Secondary Post-secondary Note: Amount is disbursement base. Source: Author estimates based on the OECD CRS database (2023). The largest brunt of the decline is being felt by LICs, and specifically in basic education.14 The OECD’s Creditor Reporting System (CRS), which tracks international aid flows, shows that © Dana Smillie | World Bank | 2010 almost 85 percent of aid to education goes to individual recipient countries, as opposed to regional or broader cross-country Most of the aid to education is from a handful of donors. More programs. Of that 85 percent, LICs received 22 percent of all aid than 50 percent comes from 5 donors – European Union, France, directed to education. Basic education receives approximately 39 Germany, Saudi Arabia, and the World Bank combined. Including percent of total education aid, secondary education receives 21 the next 3 top donors – Japan, the United Kingdom, and the percent, and tertiary education receives 40 percent. Since 2010, United States, we reach 70 percent of all education aid donations. the share of aid to basic education has fallen by 5 percentage Between 2019 and 2021, Germany was recorded as the largest points, while the share of secondary education has increased by donor to international education aid, giving an average of US$3.3 roughly 4 percentage points, showing a slight movement. Since billion a year. However, it is important to note that Germany, as 2015 there has been a notable increase in aid to basic education, well as other countries like France, allocate 60 percent of their specifically targeted at LICs. In absolute terms, the amount they aid to education at the post-secondary level within their own received almost doubled from US$1.1 billion in 2015 to US$2.0 countries. This is because these countries count disbursements billion in 2020, before falling to US$1.7 billion in 2021. that go largely to scholarships and imputed student fees to international students in their ODA total. These waived tuition Sub-Saharan Africa (SSA) receives just over one-third of all aid expenses are estimated and recorded as aid. Japan also distributes for basic education, and MENA’s share of the total is increasing. more than one-third of its aid to scholarships and imputed student Sub-Saharan Africa has over half of the world’s primary-age fees. The United States, by contrast, does not include international out-of-school children,15 received approximately US$2 billion of student scholarships as a form of education aid. total basic education in 2021, which has stayed at the same level as it was in 2010 (US$1.9 billion). By contrast, the amount of aid to If scholarships and imputed student fees are excluded from basic education for countries in the Middle East and North Africa all grant donations, the World Bank is the largest education (MENA) has more than doubled – increasing from US$0.8 billion aid donor. The World Bank roughly finances US$1.8 billion in 2010 to US$1.9 billion in 2021 in real terms. In large part, this in education aid a year through International Development is due to the recent influx of aid for the Syrian refugee crisis. Association (IDA). This is followed by Germany and the United 14 Per OECD definition, basic education covers "primary education, basic life skills for youth and adults, and early childhood education". 15 Approximately 67 million primary education age children are estimated to be out-of-school in 2021, of which 36 million children are in SSA (GEMR and UIS 2022). 12 EDUCATION FINANCE WATCH 2023 Figure 10. European Union, Germany, and World Bank/ Figure 11. Education takes a back seat: Widening IDA have been increasing aid to education disparity in funding priority with health Three-year average of aid to education, excluding scholarships and Share of education sector in comparison to health in sector-specific imputed student fees, seven largest donors, 2004-06, 2009-11, ODA, 2002-2021 2014-16, and 2019-21 25.0 Health 2.0 World Bank / 20.0 IDA, 1.8 1.5 Germany, 1.4 15.0 Constant 2021 USD billion United States, 1.4 % 1.0 European Union, 1.0 10.0 Education United Kingdom, 0.7 France, 0.6 5.0 0.5 Japan, 0.5 0.0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 0.0 2004-06 2009-11 2014-16 2019-21 Note: Amount is disbursement base. Note: Amount is disbursement base. Source: Author estimates based on the OECD CRS database (2023). Source: Author estimates based on the OECD CRS database (2023). States (US$1.4 billion each roughly), and the European Union In Sub-Saharan Africa, donors disbursed a third of their aid (EU) (US$1 billion). Aid from the United States has remained to education in 2021 through non-government organizations constant recently, while other donors such as Germany, the EU, and universities. France and Germany channel large parts of France, Japan, and the World Bank show slight increases in aid through their own development agencies. The World Bank/ recent years. However, in contrast, education aid from the United IDA only funds recipient countries directly; by contrast, the EU, Kingdom has dropped significantly by 39 percent between 2014 Germany, the United Kingdom, and the United States distribute and 2021 (Figure 10). no more than 20 percent of their aid directly to recipient countries. © Sarah Farhat | World Bank | 2017 13 EDUCATION FINANCE WATCH 2023 Figure 12. Multilateral commitments to education exceeded disbursements in recent years Total aid to education disbursements and unused commitments, by type of donor, 2002–2021 12.0 Bilateral Multilateral 10.0 Constant 2021 USD billion 8.0 6.0 4.0 2.0 0.0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Disbursement Unspent commitment Source: Author estimates based on the OECD CRS database (2023). The share of education aid compared to other sectors has fallen in recent years. Despite international calls for more education financing, the portion of international aid allocated to education has decreased slightly since 2002 and has remained stagnant in recent years. It reached the lowest point of 9.7 percent in 2013 and 2015 (Figure 11). Although there were some signs soon after that the share might recover, as it reached 10.9 percent in 2019, it fell back to 9.7 percent in 2020–2021, absorbing the impact of the shift in donor focus on health during the pandemic, which saw an increase from 16.6 percent in 2019 to 19.5 percent in 2020 and 23.3 percent in 2021. There is a lag among donors between commitments to education and disbursements allocated in the same year. It is not straightforward to compare the two: commitments predate disbursements by some years. However, it is worth noting that in education, there has been a larger discrepancy between commitments and disbursements in recent years among multilateral donors. As we see from the figure below (Figure 12), the discrepancy was first seen in 2017–18. The figure shows that © Nafise Motlaq | World Bank | 2013 for multilaterals, on average, US$1.7 billion has gone unspent since 2017. In the education sector, bilateral aid has been 4 percent unspent since 2002, while multilateral aid is on average only 4 percent of bilateral aid and 3 percent of multilateral aid 21 percent unspent. In contrast, the health sector spends all or has gone unspent each year. More research to inform policy uses more aid than it commits each year: since 2002, on average, could be helpful. 14 4 Household spending on education Economic shocks often decrease household spending on education (Read 2020), as families are less able to pay for education until the economy recovers (World Bank 2020). EFW2021 and EFW2022 reviewed trends and patterns of household spending in education over the past decade, using the global database. EFW2023 advanced this analysis by accessing © Dominic Chavez | World Bank | 2016 schools and may pay private school tuition for some families. However, school fees go beyond tuition and often include books, uniforms, etc. Even when governments proclaim ‘fee-free’ education, households spend money to send their children to school. This constitutes a regressive tax when education is compulsory, and often excludes poor children from schooling household survey data and country case studies on household (Hamillman and Jenker 2014; UNESCO 2021). education spending, such as by education level, type of schools, and education items that households pay for. International remittances increase household expenditures on education and other essential goods and services, particularly in LIC and LMIC. Recent estimates (Ratha et al. 2023) show How important is household that globally, remittance flow to LIC and LMIC was limited education spending globally? in 2020 (1.4 percent decrease from 2019), after which the flow surged and continues growing (Ajefu and Ogebe 2021). The Household spending as a share of total education spending is impact in SSA LIC and UMIC lagged, with countries registering highest in LIC and MIC. In 2021, households accounted for a large decrease in 2020 (13 percent down from 2019) before 32 percent of total education spending in LIC and MIC, while recovery in 2021 (16.3 percent up from 2020). Families relying on household contributions to education in HIC were around 16 remittances to fund their children’s education likely experienced percent (Figure 3). income loss due to the pandemic’s onset (Al-Samarrai 2020), and thus their education spending in 2020 may have decreased Household spending levels are different from country to country in response. and within a country. In LIC and MIC, household spending as a proportion of total education spending ranges from 5 to Household spending on education is elastic. This is based on the 80 percent. This wide difference is seen even within the same household’s socioeconomic status and the specific characteristics country’s income group. For instance, in 18 LIC household (line items) of public investment. When government spending education spending accounts for between 5 and 73 percent of on education increases as an overall share of GDP, household total education financing. In general, richer households spend spending on education also increases (Naurin and Pourpourides a higher proportion of their income on education than poorer 2023). However, within certain groups, households spend less households (Foko et al. 2012, World Bank and UNESCO 2022). on education when governments spend more. For example, However, the financial burden is higher for poorer households in a household with few resources, if the government begins who are less able to afford schooling for their children than for providing educational materials like textbooks, a parent may richer families (UNESCO 2021). reallocate education resources to other household needs. In India and Zambia, when the government provided grants to schools, Education spending is costly for households, especially poor parents reduced investments in their children’s schooling (World households. Within countries, governments provide public Bank 2018). 15 EDUCATION FINANCE WATCH 2023 Figure 13. The share of education in total household consumption varies by economic status and location Share of household spending on education as a share of total household consumption, %, various years 10 Wave 1 Wave 2 8 6 % 4 2 0 Rural Total Urban Poorest Poorer Middle Richer Richest Rural Total Urban Poorest Poorer Middle Richer Richest Rural Total Urban Poorest Poorer Middle Richer Richest Rural Total Urban Poorest Poorer Middle Richer Richest Rural Total Urban Poorest Poorer Middle Richer Richest Rural Total Urban Poorest Poorer Middle Richer Richest Location Wealth Location Wealth Location Wealth Location Wealth Location Wealth Location Wealth Burkina Faso, 2014-19 Côte d'Ivoire, 2015-19 Ethiopia, 2019-22 Tanzania, 2015-21 Nigeria, 2016-19 Ghana, 2013-17 Source: Author calculation using household survey datasets https://www.worldbank.org/en/programs/lsms. How much do households in LIC and household consumption expenditure from 2.2 percent in 2015 to LMIC spend on education and on what 2.0 percent in 2019. The decrease was stronger in two countries items? where spending was compared before and after COVID-19. Households in Ethiopia (from 1.7 percent in 2019 to 1.3 percent Regardless of economic status or school type, households bear in 2022) and Tanzania (from 3.1 percent in 2015 to 1.1 percent significant costs when sending their children to school. This is in 2021), reduced the share of education in their consumption. In the case for all six LIC and LMIC analyzed in EFW2023.16 For contrast, Burkina Faso (from 1.9 percent in 2014 to 2.3 percent in example, in Ethiopia, households spent on average Ethiopian 2019), Ghana (from 5.9 percent in 2013 to 8.2 percent in 2017), Birr 2,342 (approximately US$43) per primary student, which and Nigeria (from 4.8 percent in 2016 and 5.9 percent in 2019) was equivalent to 4 percent of the country’s GDP per capita in increased the share. 2022.17 It is noteworthy that households have maintained and even increased their absolute levels of spending on education While wealthier families are often seen to allocate a higher share despite the collapse of incomes over 2020 and 2021 and the against their total household consumption for education, this significant amount of time children were not enrolled in school. is not always the case. For instance, poorer families allocated a larger proportion of their budget for education in Burkina Faso In some LIC and LMIC countries, households have decreased (2014) and Côte d’Ivoire (2019) than did wealthier families. the share of their budget allocated to education. Ivorian Also, urban households spent a larger share of their budget families slightly decreased the share of education in their total on education than did rural households. For instance, urban 16 To analyze household spending we give examples from six LIC and LMIC countries: Burkina Faso, Côte d’Ivoire, Ethiopia, Ghana, Nigeria, and Tanzania. The data come from household surveys that were conducted at least twice in the past 10 years. Household education spending was analyzed in terms of its share of total household consumption for a particular year, by the type of educational institutions (state or non-state schools), household residential areas (urban or rural), household wealth quintiles, and education levels (pre-primary, primary, secondary, and higher education). 17 Own calculation, using Ethiopia Socio Economic Survey 2022 dataset for education spending per capita and IMF WEO database for GDP per capita (accessed on May 30, 2023). 16 EDUCATION FINANCE WATCH 2023 Figure 14. Most household spending on education went to primary and secondary education Distribution of household education spending, total (state and non-state), % 100% 9% 7% 12% 8% 15% 13% 14% 18% 15% 90% 24% 29% 80% 32% 33% 70% 59% 47% 37% 41% 35% 44% 29% 51% 44% 60% 50% 40% 40% 54% 51% 30% 37% 38% 32% 34% 42% 34% 36% 37% 20% 35% 28% 10% 10% 10% 13% 12% 9% 4% 4% 2% 4% 3% 5% 8% 0% 2014 2019 2019 2022 2015 2019 2013 2017 2016 2019 2015 2021 Burkina Faso Ethiopia Côte d'Ivoire Ghana Nigeria Tanzania LIC LMIC Pre-primary Primary Secondary Tertiary Source: Author calculation using household survey datasets. households in Burkina Faso spent 3.5 percent of their budget cost in private schools is 1.5 to 5 times higher than in public on education while their rural peers spent 1.7 percent in 2019 schools; in Ethiopia and Tanzania, it is 10 times higher. While (Figure 13). some private schools cost a lot, others supply their education service at low cost (UNESCO 2021). Private costs are especially Household spending is largely concentrated in primary and high at the pre-primary and primary education levels, compared secondary education. EFW2023 looked at which education to public schools. For instance, in 2018 Ethiopian households levels households spent most in each country. Among the needed to pay more than 20 times to send their children to countries analyzed, families spent over 80 percent of their non-state pre-primary schools compared to their peers who sent education funds on primary and secondary education, and just children to public pre-primary schools. Tanzanian families spent 20 percent on tertiary. Households in Burkina Faso allocated the 19 times more on non-state primary schools in 2021, while the most (up to 96 percent in 2019), while Ethiopian households ratio in secondary education was relatively lower (5 times). In allocated the least (66 percent in 2019) to primary and secondary private primary schools, fees are key spending items (the largest education. This pattern among countries remained constant item in four countries and the second largest in two). In public (Figure 14). primary schools, fees are not among the top three items in any country. This pattern is observed in other countries in Africa. Not surprisingly, households spend more money on private Overall, the share of school fees is lower when children attend schools than on public. In most cases, the per-student education public schools (except in Burkina Faso in 2014). 17 5 Equity and efficiency Will increasing spending be enough to close the learning gap? School closures during COVID-19 worsened the global learning crisis. Before the COVID-19 pandemic, six out of ten Figure 15. Increasing education spending would lead to decreased learning loss 0.24 0.20 © Nafise Motlaq | World Bank | 2013 students in LIC and MIC could not read and understand simple 0.16 texts by age 10 (World Bank et al. 2022). Upon the pandemic’s Learning Loss (SD) onset, learning losses were projected (Azevedo et al. 2021). Actual 0.12 learning losses are much worse than projected: now seven out of ten cannot read (World Bank et al. 2022). On average, one month 0.08 of school closures led to one month of lost learning (Schady et If per-capita education spending is al. 2023). Learning loss is significantly larger for students whose increased by 30% (US$56 to 0.04 US$73), LIC would mitigate schools faced relatively longer closures (on average of 33 percent learning loss equivalent for 0.36 of a SD – roughly equivalent to one year of schooling – in 65 year of schooling 0 countries globally, and 20 percent of a SD in ECA). Further, $56 (0%) $70 (25%) $73 (30%) $76 (35%) $78 (40%) $81 (45%) $84 (50%) $87 (55%) lower-achieving students before COVID-19 have experienced Education spending per child, US$, 2021 constant, % increase from $56 much larger learning losses than higher achievers ( Jakubowski Learning loss (SD) Learning loss (SD) longer school closure et al. 2023; Patrinos et al. 2023). Source: Author estimates based on Patrinos (2023) data and EFW2023 database. Without adequate remediation, learning loss will likely translate into a huge negative impact on the global economy - up to a (2023), EFW2023 found schools in LIC closed for an average 0.68 percentage point reduction of GDP growth, totaling a of 30 weeks and learning outcomes were significantly lower than global loss of more than 80 trillion (analysis using PIRLS 2021 global benchmarks (e.g., harmonized learning outcomes (HLO): data) ( Jakubowski et al. 2023). Because economic growth is a 329 in LIC vs 462 in global samples). If government per-capita main driver of increased education spending, the global income education spending in LIC were increased by 30 percent (US$56 loss could in turn substantially decrease future public education to US$73), LIC could mitigate learning loss equivalent to 0.36 spending (Al-Samarrai et al. 2019). In this negative scenario, a years of schooling. If LIC schools closed longer (33 weeks, which negative cycle perpetuates – lower-skilled individuals earn less, is the average for MIC), LIC would need to increase education countries experience GDP erosion as a result and consequently spending more to reduce learning poverty at the same level (0.24 invest less in education, ad infinitum. to 0.12 SD), at least $76 (Figure 15). Countries with lower education spending and efficiency Increasing funding for education has a significant, though are expected to see the largest learning gains from increased relatively weak, impact on student learning. Al-Samarrai education financing. While in general there is no direct correlation and Lewis’s (2021) econometric analysis in Brazil, Colombia, between education spending and student learning outcomes, Indonesia, and Uganda shows that both government transfers and there are empirical suggestions that colinear relationships exist subnational education spending have a positive and significant between the two in countries where education spending is still impact on student achievement. For example, in Indonesia, a 10 low or relatively low (Al-Samarrai et al. 2019; Vegas and Coffin percent increase in subnational per capita education spending 2015). Using global data from 41 countries compiled by Patrinos increased students’ test scores by 0.6 percent. However, although 18 EDUCATION FINANCE WATCH 2023 © Khasar Sandag | World Bank | 2013 the relationship is significant, it is not very strong, and there is Education financing is an important factor: Often, countries considerable variation in the effectiveness of subnational entities are either spending too little, the spending is inefficient, or a in translating funding into outcomes. combination of both (Al-Samarrai and Lewis 2021). What governments spend on, and how, are key to recovering learning Increasing education spending is necessary, but not sufficient to loss and avoiding the ‘negative cycle,’ mentioned above. If halt learning loss and overcome the learning crisis. If countries spending is inefficient, meaning misallocated and misaligned, increase education spending by one percent from 2020-2021 and/or does not target what works, it will not translate into levels, learning loss could be mitigated by 0.4 percent in SD, student learning. Further, education funding must be equitable equivalent to 0.01 years of schooling. However, learning loss to bring and keep all children in school and learning. because of school closures is larger than this (author estimates using data in Patrinos 2023). It is estimated that global education Key learning outcomes are lowest in countries spending the expenditure is over US$5 trillion. Public expenditure on least per school-aged child. EFW2023 updates the analysis education is significant; on average, countries spend about 4.5 to done by World Bank and UNESCO (2021) and Arias and 5 percent of GDP on education, except for low-income countries Kheyfets (2023), using the latest available public education (3 to 3.5 percent, on average). spending data.20 Figure 16 plots the relationship between LAYS21 and spending for each country. LIC – with few exceptions – Despite increases in education spending and schooling achieves very low LAYS with very low levels of spending per worldwide,18 learning poverty has reached high levels.19 child (blue in Figure 16.b). These countries could benefit from 18 Primary education completion rate (both sex) in LIC increased from 46.6 percent in 2012 to 55.8 percent in 2021 (SDG4 Indicator Dashboard http://sdg4-data.uis.unesco.org/, accessed on August 21, 2023). 19 For instance, learning poverty are 99 percent for Zambia, 98 percent for Lao PDR, 97 percent for Democratic Republic of Congo and 91 percent for Philippines. (World Bank and UIS. June 2022. Learning Poverty Global Database, https://datacatalog.worldbank.org/int/search/dataset/0038947, accessed on August 22, 2023). 20 As at writing EFW2023, 2020 is still the latest for LAYS data. 21 The LAYS metric combines quantity (expected years of schooling) and quality (harmonized learning outcomes). 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 see Filmer et al. (2020); Kraay (2019); Patrinos and Angrist (2018). 19 EDUCATION FINANCE WATCH 2023 Figure 16: Countries differ in how effectively they translate funding into outcomes a. Mean per-capita education expenditure and LAYS, all countries, b. Mean per-capita education expenditure and LAYS, LIC, and LMIC, 2020 2020 12 14 12 10 10 8 8 LAYS LAYS 6 6 4 4 2 2 0 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 200 400 600 800 1,000 1,200 1,400 US$ US$ Note: Spending per capita is computed as total public education divided by the school-age population. Estimates on spending per capita include interpolated values. Source: Author calculations based on the Human Capital Index (HCI) and EFW database. more spending to adequately fund essential inputs to ensure Improving efficiency effective delivery of education services, and from cost-effectively investing to maximize impact on learning (Banerjee et al. 2023). Global evidence highlights myriad ways to improve education It is important to note that the correlation between education spending efficiency. This includes reducing repetition, drop- spending and outcomes is weaker at higher levels of spending: out, and delayed entry, deploying teachers efficiently, avoiding some countries have education outcomes that are far worse than teacher absenteeism, promoting teaching effectiveness, procuring those of other countries with similar spending levels per child (see textbooks sustainably, paying teacher salaries, transferring school LMIC in yellow in Figure 16.b). Education spending efficiency grants promptly, and managing fiduciary risks effectively, all of can be improved by: (i) prioritizing universal foundational which evidence shows can raise educational outcomes (Arias and learning, and (ii) spending a greater portion of government Kheyfets 2023). Here we present an example about how evidence expenditure on high-efficiency educational investments rather can be used to improve the efficiency of education spending. We than low-return investments (Psacharopoulos et al. 1986; draw on the approaches identified in the literature (Banerjee et International Commission on Financing Global Education al. 2023). Opportunity 2016; World Bank 2018, Angrist et al. 2023). One promising approach is targeting instruction by learning Increasing the cost-effectiveness of education spending could level rather than age or grade. This approach requires grouping maximize learning returns on education investments. Even children by learning level and providing each with teaching before the pandemic, actual learning in LIC was very low. On specifically tailored to their learning needs (Angrist et al. 2023). It average, a child living in a low-income country expected 7.6 years can be accomplished with the use of technology (software) or with of schooling, but only 4.3 years of actual learning – as measured teachers and teacher assistants alone. On average, this approach by LAYS – with a government education expenditure of US$51.8 could provide a full extra LAYS at a unit cost of US$33.3. per student a year. In contrast, a child living in a high-income Teaching according to learning level with technology support country was expected to attain 13.1 years of schooling, of which for one year can improve learning by 0.27 SD, which has the 10.3 years would be of a learning level comparable to Singapore, potential to increase students’ future earnings by 5.5 percent while with an annual government education per capita expenditure of yielding US$1,724 in future benefits per beneficiary at a student US$8,400. cost per year of US$26.6. This translates into a Benefit-Cost Ratio 20 EDUCATION FINANCE WATCH 2023 (BCR) of 65. The rise in earnings associated with this approach will represent US$261 more in earnings each year per beneficiary in LIC, and its Net Present Value (NPV) over an average working life of 45 years would amount to US$6,390 per beneficiary assuming a discount rate of 3 percent. If every primary student in LIC profits from this intervention, the total labor earnings in these countries will increase by US$28,048 million over 45 years, which represents 4.8 percent of the current year GDP. The non-technology implementation of this approach could increase learning by 0.15 SD in one year, potentially transforming into 3 percent higher future earnings, and future benefits amounting to US$940 per student for each dollar invested. In this case, each beneficiary would receive $142 more in earnings each year for an accumulated NPV of $3,485 over their working life. If all primary students in LIC were to receive this intervention, total labor earnings might increase by US$15,299 million in LIC over 45 years, representing 2.25 percent of current year GDP. Structured pedagogy aims to improve classroom instruction © Dominic Chavez | World Bank | 2016 with a set of coordinated inputs. This includes, for example, lesson plans, materials, teacher training, etc.), so teachers can give classes with clarity – even if they have limited training, best school available. Informing parents and students about the minimizing students’ misunderstanding. On average, this returns to education increases LAYS by 0.08 (Angrist et al. 2023) approach improves learning by 0.13 SD per year of intervention, in LIC, which is in line with a moderate typical effect of 0.1 SD which would reflect on a 2.6 percent increase in future earnings (Evans and Yuan 2022). Although the effect is moderate, it is of beneficiary students, namely US$123 more earnings each year. cost-effective because information can be shared at a very low The NPV of the additional earnings that each beneficiary might cost. Considering that 1 SD increase in learning correlates with receive over their working life amounts to US$3,021. Again, if a 20 percent increase in individual earnings (Angrist et al. 2021), all primary students in LIC were to benefit from this approach, this modest learning improvement would increase students’ future total labor income would likely increase by US$13,259 million earnings by 2 percent. In this case, each beneficiary student could over the working life of beneficiary students, equivalent to receive US$95 more each year in earnings, which over a working 2.25 percent of current year GDP. Also, it is estimated that its life of 45 years, amounts to a NPV of US$2,323. If all primary successful implementation would yield US$105 in benefits per students in LIC were to receive this intervention, total labor each US$1 invested (Angrist et al. 2023). One extra LAYS using income in LIC could increase by US$10,199 million, namely 1.7 this approach would require an investment of $33.3 per student. percent of current year GDP, over the working life of beneficiary students. It is estimated that successful implementation of this Providing information to parents has been proven to propel approach could produce US$21 in benefits per each US$1 learning when it shifts beliefs around the benefits of education, invested (Angrist et al. 2023). Table 1 summarizes the three increasing enrollment and compelling parents to choose the outlined interventions. Table 1. Learning gain (SD) by strategy implemented Learning gain Earnings increase Initial spending Learning gain with with per capita due to learning gain Strategy per capita initial spending spending of US$50 in previous column (US$) (SD) (SD) (%) Structured pedagogy 7.8 0.13 0.84 16.77 Teaching according to learning level with technology use 26.6 0.275 0.52 10.32 Teaching according to learning level without technology 19.5 0.15 0.38 7.68 Source: Author elaboration based on Angrist et al. (2023) data. 21 EDUCATION FINANCE WATCH 2023 Closing the learning gap widened by the pandemic will require pre-primary education: wealthier families benefited more in Côte more funding for education and investing each additional dollar d’Ivoire, while poorer families benefited more in Ghana. This better to get its maximum benefit. According to the regression finding fits with current global trends as countries recognize the we used to estimate the learning loss recovery trajectory for importance of early childhood development to decrease delayed LIC (Figure 15), improving learning by 0.1 SD would require entry and increase learning over the life cycle, and countries begin an additional investment of US$14 per student. However, the to fund this priority, given its high return on investment and correlation between education expenditure and actual learning positive impact on human capital development. is historically weak as governments mostly invest in inputs (building more schools, lowering student-teacher ratios, hiring Sometimes governments have chosen to disproportionately more teachers) without regard to the actual learning occurring in subsidize higher education. Typically, expenditure on tertiary the classroom (Angrist et al. 2023). One of the cheapest options education is skewed toward the wealthiest as those with access would be providing information related to education where this to higher education mainly come from the richest households. is not already widespread as the cost of providing information is Such inequities in spending result in large variances in the very low and can yield a 0.1 SD increase in learning outcomes. quantity and quality of education received by different groups. But if low-income countries were to invest US$50 more per This, in turn, contributes to greater inequality in potential future student – at least in the most underserved – in cost-effective earnings, and persistent socio-economic disparities. Reallocating approaches any of the other two smart buys approaches (such as education budgets away from tertiary to basic education can structured pedagogy or teaching according to learning levels with make education financing more equitable. or without technology), they would be able to improve learning between 0.38 and 0.84 SD (equivalent to 1.15 and 2.54 years At the same time, higher education financing can be made more of additional schooling) while increasing future earnings of the efficient by tapping new sources of funding. Doing so means most disadvantaged between 7.7 and 16.8 percent (see Table 1). using future earnings to finance contemporary education. In higher education, this can be done with income contingent loan repayments (Barr 2014; Chapman 2016). Over 30 years later, Equity can be improved income-contingent loans (ICL) exist in different forms in 10 countries, although scheme design, eligibility, interest rates, and Education is critical to equalizing opportunities and providing debt forgiveness regimes differ widely between systems, and have each child with the skills to achieve their full potential. Children changed over time within jurisdictions (Chapman and Dearden in disadvantaged and vulnerable situations face barriers to school 2022). access and learning related to household income and location, gender, ethnicity, and disability, among others (UNICEF Basic literacy, including reading with comprehension, and 2023). We examine the distribution of public spending to numeracy support individuals in reaching their full potential, education across wealth income quintiles at each education level while contributing to sustainable development, inclusive growth, (pre-primary, primary, secondary, and tertiary) in two countries and gender equality, among others (UNICEF 2022). Primary (Côte d’Ivoire 2015 and 2019; and Ghana 2013 and 2017), using education is the key stage for children to acquire foundational household surveys and published data in government budget and skills and yields the highest social returns among all education expenditure reports. levels across the globe, especially in LICs (Psacharopoulos and Patrinos 2018). Government spending for primary education benefits poorer households. The analyses for different data points for the two countries suggest that each year, government expenditure Using transfers to make education distribution for secondary and tertiary education levels was spending more effective skewed to wealthier households and this was especially prominent at the tertiary level. However, spending on primary education was Because intergovernmental fiscal transfers comprise a large pro-poor: the Lorenz curves show the distribution of expenditure share of subnational spending in decentralized countries, their by consumption quintile above the diagonal line (45-degree black design and implementation might improve education financing line in Figure 17), which means that the poorer received a higher and outcomes. Recent estimates indicate that 84 percent of the share of government spending. Results are mixed in the case of world’s children live in countries where subnational authorities 22 EDUCATION FINANCE WATCH 2023 Figure 17. Government education spending is skewed towards richer households, especially at the tertiary level Lorenz curves of public education spending by education level Côte d’Ivoire 2015 Côte d’Ivoire 2019 Lorenz curve for public spending on education by level, Côte d'Ivoire 2015 Lorenz curve for public spending on education by level, Côte d'Ivoire 2019 1 1 Cumulative percentage of government education spending Cumulative percentage of government education spending 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cumulative percentage of population, poorest to richest Cumulative percentage of population, poorest to richest Pre-primary Primary Secondary Tertiary Line of perfect equality Ghana 2013 Ghana 2017 Lorenz curve for public spending on education by level, Ghana 2013 Lorenz curve for public spending on education by level, Ghana 2017 1 1 Cumulative percentage of government education spending Cumulative percentage of government education spending 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cumulative percentage of population, poorest to richest Cumulative percentage of population, poorest to richest Pre-primary Primary Secondary Tertiary Line of perfect equality Source: Author calculations using household surveys, UIS, and government budget reports. govern.22 Over the last three decades, many have introduced If the focus is narrowed to the basic level, this share would be reforms to decentralize education, particularly basic education.23 significantly higher. In Uganda, subnational governments account In most decentralized countries, subnational governments for over 80 percent of government spending on primary and account for over 50 percent of total public education spending. secondary education (Al-Samarrai and Lewis 2021). 22 The estimates are based on the Fiscal Decentralization Dataset, International Monetary Fund, Washington, DC, (accessed May 8, 2020) https://data.imf.org/?sk=1C28EBFB-62B3-4B0C- AED3-048EEEBB684F and OECD (2019). 23 See Dyer and Rose (2005) and Channa and Faguet (2016) for discussions of the potential benefits and actual effects of decentralization in the education sector. 23 EDUCATION FINANCE WATCH 2023 Box 1. Using intergovernmental transfers to improve education in Brazil Using existing revenue sources, a performance- based general transfer improved education spending efficiency, and equity in Ceará, Brazil (Loureiro and Cruz 2020; Wetzel and Viñuela 2020). In 2008, an innovative reform of the fiscal transfer system in Ceará linked an important general transfer to learning outcomes. In Brazil, states are obligated to transfer 25 percent of consumption tax revenues to their municipalities as a general-purpose transfer. States have discretion over how they transfer a quarter of the total transfer and since 2008, Ceará began allocating 72 percent of these discretionary funds to municipalities based on their education performance. © Conor Ashleigh | World Bank | 2015 The discretionary transfer amount is determined by a primary “education quality index” that is designed Most subnational governments receive a combination of both to improve performance and to increase equity. general- and specific-purpose transfers. General-purpose A comprehensive census-based learning assessment is transfers are typically unconditional and can be allocated across used to calculate the index. The assessment consists of subnational government responsibilities, including education. indicators on early grade literacy, learning measured Central governments often complement these general transfers at the end of primary school, and the proportion of with conditional or specific-purpose transfers targeted and tied children transitioning to the next grade. Municipalities to providing certain inputs or improvements in education outputs are allocated transfer resources based on their scores on or outcomes. Beyond core funding, transfer systems can also these indicators as well as on the magnitude of their provide an effective system for channeling funds to education educational improvements over the preceding year. systems during times of crisis. For example, in the United States, Overall, performance-based transfers are a big revenue federal stimulus packages used existing transfer mechanisms to source for municipalities, representing as much as provide additional financial support to local education systems one-third of all revenue for poorer municipalities in during the COVID-19 crisis. While all transfers provide Ceará (Loureiro and Cruz 2020). subnational governments with additional resources, relatively large conditional (specific-purpose or performance-based) Rigorous evaluations show the performance-based transfers may improve education services most (Al-Samarrai reform to the fiscal transfer program improved learning and Lewis 2021). For example, in Brazil and China, the estimated outcomes in most municipalities. Even though the marginal effect for education-specific conditional transfers was transfer was not a specific-purpose transfer, evidence higher than that for unconditional transfers (Box 1). indicates it led municipalities to increase their spending on basic education and narrowed per capita differences Thoughtfully designed fiscal transfers could improve equity. in transfers between municipalities (Franca 2014). The World Bank public expenditure reviews and other studies have transfers also narrowed learning gaps between poor and shown that, in approximately one-half of developing countries wealthy municipalities (Brandão 2014). Evidence indicates with available data, there was a negative and statistically their effectiveness is linked to (1) careful design to avoid significant relationship between subnational poverty rates and any negative consequences - such as the risk of inter- education spending (Manuel et al. 2019).24 When education governmental transfers widening existing inequalities, transfers include a per-student element, subnational governments (2) solid and comprehensive information systems, and know that if they expand access, they will receive funding from (3) good implementation capabilities of subnational the central government to help cover the costs of providing governments, down to the local (municipal) level. more school places. This has had the effect of reducing the cost 24 There are relatively few countries that show an opposite and statistically significant relationship. 24 EDUCATION FINANCE WATCH 2023 burden on subnational governments and households while narrowing access inequalities. China’s New Mechanism to Guarantee Rural Compulsory Education Financing, introduced in 2006, strengthened incentives for provincial governments to increase basic education access at the same time school fees were abolished. The New Mechanism introduced a specific-purpose transfer allocated to provinces on a per-student basis designed to cover elements of non-salary funding and to compensate subnational governments for the revenue lost by the abolishment of school fees. The share of per-student funding covered by the transfer was linked to the socioeconomic characteristics of each province.25 The New mechanism increased primary and secondary enrollment and narrowed enrollment outcomes between provinces (and counties). It also improved attainment and learning outcomes, and these effects were larger for students from disadvantaged backgrounds (Ha and Yan 2018; Xiao, Li, and Zhao 2017). Ding et al. (2020) found that transfers did not lead to any significant increase in total education spending because they substituted for other “off-budget” spending, including © Gerhard Jörén | World Bank | 2012 household tuition fees. This may also help to explain their positive impact on outcomes since the burden of funding shifted away from households to governments, removing household cost estimates of subnational fiscal gaps, which measure the gaps constraints associated with sending children, particularly from between subnational revenues and expenditure obligations, as low-income families to school. well as population size, and levels of economic development (Al-Samarrai and Lewis 2021). Others use specific-purpose Subnational transfers can enhance education spending transfers: In Brazil, the Fund for the Development of Basic efficiency, especially when designed on a per capita by child Education (FUNDEB) addresses equity issues by guaranteeing basis. In Bulgaria, the introduction of per capita financing led to minimum levels of education spending among municipalities. the merging or closing of some schools, significantly increasing Before the program began in 1996 with FUNDEB’s predecessor, efficiency. Although some spending inefficiencies remain, per the Fund for the Development of Primary and Lower Secondary capita funding formulas act as automatic stabilizers that adjust Education (FUNDEF), there were large differences in education financing mechanisms immediately in response to demographic spending between municipalities, driven by poorer municipalities’ shifts. Conversely, fiscal transfers that are not per capita by limited revenues. Before the program started, the wealthier South, child can sometimes inadvertently drive inefficiencies in public Southeast, and Central West regions in Brazil were spending education spending. In Indonesia, the formula for the largest almost twice as much per student as the poorer regions in the general transfers includes incentives for district governments to North and Northeast (Gordon and Vegas 2005). These spending spend more on hiring civil servants than on non-salary spending. disparities led to significant differences in education outcomes In education, this has resulted in a tendency to hire more teachers and exacerbated more general socio-economic inequalities than is required to comply with minimum service standards and between regions. FUNDEF and FUNDEB narrowed spending maximum class sizes and is an important driver of inefficiency inequalities by redistributing a portion of federal, state, and (Lewis and Smoke 2017; World Bank 2012). municipal tax revenues among all municipalities to guarantee a minimum level of spending per student while increasing To address inequity through per-student transfers at education funding in the poorest states (Cruz and Silva 2020; sub-regional levels, some countries employ general-purpose Gordon and Vegas 2005). Per child funding – particularly transfers, while others use specific-purpose transfers. In China, considering demographic shifts could be considered in education equalization general-purpose transfers make up approximately funding mechanisms. (For further reading on intergovernmental two-thirds of all transfers to provinces and counties. These transfers to improve education financing, see Al-Samarrai and transfers are allocated according to formulas that include Lewis 2021.) 25 The central government transfers covered 80 percent of the administrative expenses and lost school fee income of the least developed provinces in the west and 60 percent of those of the central provinces. Initially, the wealthier eastern provinces did not receive any funds through the compulsory specific-purpose education transfer, but this changed in 2015 when they received transfers amounting to 50 percent of their total expenses. 25 6 Spotlight: Demographic shifts Around the world, school-aged populations are changing LIC and LMIC school aged populations are expanding rapidly. In 2020, LIC and LMIC in SSA were home to 492 million children and youth, from 5 to 24 years old. By 2030, this © Arne Hoel | World Bank | 2016 For the majority of LIC and LMIC analyzed, government per-capita education spending is projected to increase while the school-age population is projected to grow. Their education spending will grow faster than the school-age population. For instance, the school-age population in Burkina Faso is estimated to increase by nearly 25 percent in ten years. If number will reach 620 million.26 The financial pressure imposed Burkina Faso increases its total government expenditure by working to provide education to all children and recover on education at the pace it took during the last decade, its COVID-19 learning losses (Sánchez et al. 2023a; Jakubowski education spending will grow much faster than its population et al. 2023) will be even higher, especially for LIC and LMIC. growth; government education spending will increase more We categorize LIC and LMIC based on their school-age than 220 percent over the next ten years. This will lead to higher population growth and total public education expenditure in per-capita education spending for the coming years in Burkina the past decade (2010-2019), grouping countries into three Faso (red circle in Figure 18). Some regions, including SSA, categories (Figure 18). are projected to increase their school-age population in the Figure 18. Projected changes in school-age population and total government spending 250 Projected total government education expenditure change, 2020-30 (%) Burkina Faso 200 150 Sri Lanka Education spending will grow India faster than the population (per-capita spending will 100 increase) 50 0 Sudan Population will grow faster -50 than education spending (per-capita spending will decrease) -100 -30 -20 -10 0 10 20 30 40 50 Projected school-age population change, 2020-30 (%) Note: We use UN population projection data to estimate school-age population changes and used the past 10-year trend in government education expenditures (constant 2021 USD) to project changes between 2020 and 2030. Scatter plots (countries) above the dotted line: school-age population growth rate is higher than the total government education spending growth rate. Scatter plots (countries) below the dotted line: school-age population growth rate is lower than total government education spending growth rate. Source: Author estimates based on UN-DESA, World Bank Databank, and EFW2023 database. 26 Author estimates using UN-DESA and World Bank Databank data. 26 EDUCATION FINANCE WATCH 2023 © Ollivier Girard | World Bank | 2019 next decades. Such increases require more financial resources reallocated to support retired and therefore economically less to expand access and ensure the quality of education (Sánchez productive older generations. et al. 2023b; UN 2023). For instance, SSA countries will need to recruit 15 million more primary and secondary teachers By contrast, per-capita education spending is projected to (8.7 million new teaching posts and 6.3 million teachers for decline in more than half of the countries we examined (10 those who leave the profession) to meet SDG4 targets by 2030 countries) if they maintain downward trends in government (Teacher Task Force 2021). education spending observed during the last decade. For instance, Sudan – its school-age population will go up by Countries with estimated declines in their school-age nearly 30 percent in 10 years – will likely face a challenge in a population (the so-called ‘demographic dividend’) may continuous decrease of per-capita education spending if they do benefit from less costly public education. This could lead to not increase education allocations to keep up with school-age an increase in per-capita government education spending, population growth rate (green circle in Figure 18). Given this, benefiting students (Miller and Mauricio 2011) (yellow circle it may be prudent if: (1) countries maintain efforts to increase in Figure 18). Sri Lanka and India are among such countries. government education spending and plan and act to realize better Staying on their past efforts and commitments, public spending learning outcomes, using available financial resources effectively; on education in these two countries could more than double (2) countries with expanding school-age populations accelerate in a decade. In addition, the UN estimates that the school- efforts to allocate more financial resources to education to age populations in these two countries will decline by 6 outpace their demographic shifts, increasing education spending percent in the next ten years. These two factors could drive per capita; and (3) countries consider demographic impacts on higher per-capita education spending in the future. However, education financing, the magnitude of which varies significantly considering the fiscal implications of aging populations in across countries. As school-aged populations continue to shift, countries with fewer youth entering the labor market, education governments might increase efforts to address inequities in financing is less elastic to school-age population shifts (Grob spending, with particular attention to poorer geographic areas, and Woler 2007). This might be because money will need to be and disadvantaged children. 27 7 Data Spotlight: Monitoring education spending Education spending data availability has improved considerably, yet gaps remain. UIS collects and reports education spending data based on country reporting. As of May 2023, 18 percent of countries had not reported 2017 data on education spending as a share of GDP, and 34 percent had not reported 2021 data (Figure 19.a). The share of education spending in total government spending is a global SDG 4 © Bart Verweij | World Bank | 2012 The timely availability of updated education finance data has decreased compared to EFW2022. Data availability has increased for the indicator of education spending (as a percentage of GDP or of total government spending) after imputation. However, compared to 76 percent and 73 percent actual data availability in EFW2022 for government education expenditure as a share of GDP and government education expenditure as a share of total monitoring indicator. There is a notable gap, with 50 percent government expenditure, data availability in EFW2023 has fallen of countries yet to submit 2021 data for this indicator (Figure to 66 percent and 50 percent, respectively. This is partially because 19.b). some countries did not report on time (Figure 20.b). Figure 19: The gap in education expenditure data is still severe. a. Data availability for government education expenditure as a share b. Data availability for government education expenditure as a share of GDP of total government expenditure 100% 100% 90% 18% 20% 22% 21% 90% 34% 30% 33% 80% 4% 4% 80% 39% 42% 70% 6% 70% 50% 25% 8% 60% 60% 9% 50% 25% 50% 15% 40% 78% 76% 40% 43% 72% 30% 55% 30% 61% 58% 47% 38% 20% 41% 20% 10% 10% 15% 12% 0% 0% 2017 2018 2019 2020 2021 2017 2018 2019 2020 2021 UIS Others Not reporting Note: Others come from BOOST, WDI, Publication Expenditure Reviews, and the IMF’s GFS. The overall number of countries and territories is 218. Source: Author estimates using the EFW2023 database. July 2023. Figure 20: Latest data availability decreased in EFW2023 a. 2020 data in EFW2022 vs EFW2023 b. 2019 data in EFW2021 vs 2020 data in EFW2022 vs 2021 data in 100% 100% EFW2023 24% 27% 80% 80% 34% 50% 58% 60% 76% 74% 73% 60% 80% 83% 40% 40% 76% 73% 66% 50% 20% 42% 20% 24% 26% 27% 20% 17% 0% 0% 2020 data 2020 data 2020 data 2020 data As of As of As of As of As of As of in EFW2022 in EFW 2023 in EFW2022 in EFW 2023 Apr 2021 Apr 2022 May 2023 Apr 2021 Apr 2022 May 2023 Education as % of GDP Education as % of Total Education as % of GDP Education as % of Total Governmental Expenditure Governmental Expenditure Not reporting With data Source: Author estimates using EFW2021, EFW2022 and EFW2023 database. July 2023. 28 EDUCATION FINANCE WATCH 2023 Conclusion There is a strong commitment to education. Total government spending as a share of total education spending reached 50 percent in low-income countries. Poor people in lower-income countries spend large portions of their income on education. While there was a small increase in annual real per capita spending on education in 2021, the percentage of GDP declined in all country income groups, except low income. Furthermore, aid decreased in real terms. While more spending does not necessarily lead to better learning outcomes, a minimal level of spending is needed. Moreover, public resources for higher levels of education are skewed to wealthier households, while government spending for basic education benefits poorer households. © Arne Hoel | World Bank | 2010 29 EDUCATION FINANCE WATCH 2023 Appendix Government spending as a share of GDP, education spending as a share of total government expenditure, and education spending as a percentage of GDP, 2021 Govt. spending as a Govt. education spending Govt. education spending Country share of GDP (%) as a share of TGE (%) as % of GDP Bangladesh 13.0 10.2 1.8 Guatemala 13.4 23.0 3.1 Lao PDR 16.3 10.8 1.9 Chad 18.4 15.7 2.9 Mauritania 19.2 9.1 1.7 Côte d’Ivoire 20.3 16.6 3.5 Peru 23.5 16.7 4.0 Monaco 24.2 5.9 1.4 Paraguay 24.8 23.6 3.5 Mali 26.4 16.2 4.4 Philippines 27.3 16.6 3.9 Nepal 27.7 14.4 4.0 Sierra Leone 28.3 14.1 3.3 Cambodia 28.6 15.7 1.7 Turks and Caicos Islands 29.0 11.2 3.3 India 29.3 14.6 4.6 Liberia 29.7 7.4 2.5 West Bank and Gaza27 30.4 17.9 5.5 Trinidad and Tobago 31.1 9.8 3.0 Rwanda 31.6 15.5 5.6 Jordan 32.8 9.6 3.2 Barbados 33.8 14.8 5.0 Ecuador 35.8 10.4 3.7 Bhutan 39.3 15.7 7.0 Maldives 39.5 10.0 5.0 Saint Kitts and Nevis 44.8 7.2 2.8 Timor-Leste 44.8 7.5 5.5 Source: UIS database, accessed April 2023. 27 World Bank denomination. The denomination used by UNESCO is the State of Palestine. 30 EDUCATION FINANCE WATCH 2023 References Ajefu, J.B., Ogebe J.O. 2021. The effects of international remittances on Channa, A, J-P Faguet. 2016. Decentralization of Health and Education expenditure patterns of the left-behind households in Sub-Saharan in Developing Countries: A Quality-Adjusted Review of the Africa. Reviews of Development Economics 25: 405–429. Empirical Literature. World Bank Research Observer 31 (2): 199–241 Alam, A, Tiwari, P. 2021. Implications of COVID-19 for low-cost Chapman, B. 2016. Income contingent loans in higher education private schools. UNICEF Issue Brief 8. financing. IZA World of Labor 2016. 227 Al-Samarrai, S. 2020. The Impact of the COVID-19 Pandemic on Chapman, B, Dearden, L. 2022. Income-contingent loans in higher Education Finance. World Bank. education financing. IZA World of Labor 2022, No 227v2 Al-Samarrai, S., Cerdan-Infantes, P., Lehe, J. 2019. Mobilizing Cholezas, I., Kanellopoulos, N.C., Mitrakos, T., Tsakloglou, P. 2013. The Resources for Education and Improving Spending Effectiveness: Impact of the Current Crisis on Private Returns to Education in Establishing Realistic Benchmarks Based on Past Trends. World Greece. Economic Bulletin 38: 33-63. Bank Policy Research Working Paper 8773. Cruz, T., Silva, T. 2020. Minimum Spending in Education and the Al-Samarrai, S., B. Lewis. 2021. The Role of Intergovernmental Fiscal Flypaper Effect. Economics of Education Review 77 (C): 102012. Transfers in Improving Education Outcomes.  International Ding, Y., Lu, F., Ye, X. 2020. “Intergovernmental Transfer under Development in Focus. Washington, DC: World Bank. Heterogeneous Accountabilities: The Effects of the 2006 Chinese Angrist, N., Djankov, S., Goldberg, P., Patrinos, H. 2021. Measuring Education Finance Reform.” Economics of Education Review 77 human capital using global learning data. Nature 592(7854): 403-408. (C): 101985. Angrist, N., Evans, D., Filmer, D., Glennerster, R., Rogers, F. H., Dyer, C., Rose, P. 2005. Decentralisation for Educational Development? Sabarwal, S. 2023. How to Improve Education Outcomes Most An Editorial Introduction. Compare 35 (2): 105–13. Efficiently? A Comparison of 150 Interventions Using the New Evans, D., Yuan, F. 2022. How Big are Effect Sizes in International Learning-Adjusted Years of Schooling Metric. World Bank Policy Education Studies? Educational Evaluation and Policy Analysis 44 Research Working Paper 9450. (3): 532-40. Angrist, N., Aurino, E., Patrinos, H., Psacharopoulos, G., Vegas, E., Fasih, T., H.A, Patrinos, and M.J. Shafiq. 2020. The Impact of COVID- Nordjo, R. Wong, B. 2023. Improving Learning in Low- and 19 on Labor Market Outcomes: Lessons from Past Economic Lower-Middle-Income Countries. Journal of Benefit-Cost Analysis. Crises. Education for Global Development (blog). Arias, O., Kheyfets, I. 2023. The Adequacy of Public Expenditure on Fasih, T., Patrinos, H., Shafiq, M. 2021. Economic crises and returns to Education and the Needs. World Bank, Washington, DC. university education in middle-income countries: stylized facts and Arias, O. and McMahon, W.W. 2001. Dynamic Rates of Return to COVID-19 projections. Current Issues in Comparative Education 23(1). Education in the US. Economics of Education Review 20(2): 121-138. Filmer, D., Rogers, H., Angrist, N., Sabarwal, S. 2020. Learning- Azevedo, J.P., Hasan, A., Goldemberg, D., Geven, K. and Iqbal, S.A. adjusted years of schooling (LAYS): Defining a new macro measure 2021. Simulating the potential impacts of COVID-19 school of education. Economics of Education Review 77. closures on schooling and learning outcomes: A set of global Fiszbein, A., Giovagnoli, P.I., Patrinos, H.A. 2007. Estimating the estimates. World Bank Research Observer 36(1): 1-40. Returns to Education in Argentina Using Quantile Regression Banerjee, A., Andrabi, T., Banerji, R., Dynarski, S., Glennerster, R., Analysis: 1992-2002. Economia 53(1-2): 53-72. Grantham-Mcgregor, S., Muralidharan, K., Piper, B., Saavedra, Foko, B, BK Tiyab, G Husson. 2012. Household Education Spending: J., Yoshikawa, H., Ruto, S., Schmelkes, S. 2023. Cost-effective An Analytical and Comparative Perspective for 15 African Approaches to Improve Global Learning-What does Recent Evidence Countries. IIEP - Pôle de Dakar, UNESCO. Tell Us are Smart Buys for Improving Learning in Low-and Middle- Franca, E.M. 2014. Repasse da cota-parte do ICMS aos municípios income Countries? World Bank, Washington. cearenses: Avaliação das mudanças ocorridas no período de 2009 a Barr, N. 2014. Income Contingent Loans and Higher Education 2011. Universidade Federal do Ceará. Financing: Theory and Practice. In Income Contingent Loans: Theory, Fuchs-Schündeln, N., Krueger, D., Ludwig, A., Popova, I. 2022. The Practice and Prospects. Palgrave Macmillan, London. long-term distributional and welfare effects of Covid-19 school Brandão, JB. 2014. O rateio de ICmS por desempenho de municípios closures. Economic Journal 132(645): 1647-1683. no Ceará e seu impacto em indicadores do sistema de avaliação da Goldin, C., Katz, L.F. 2009. The Race between Education and Technology. educação. Fundação Getulio Vargas, School of Public and Business Harvard UP. Administration, Rio de Janeiro. Grob, U., S.C. Wolter. 2007. Demographic Change and Public Card, D. 2018. Returns to Schooling. In The New Palgrave Dictionary of Education Spending: A Conflict between Young and Old? Economics. Palgrave Macmillan, London. Education Economics 15(3): 277-292. 31 EDUCATION FINANCE WATCH 2023 Gordon, N., Vegas, E. 2005. Educational Finance Equalization, OECD. 2019. 2019 Report of the World Observatory on Subnational Spending, Teacher Quality, and Student Outcomes: The Case of Government Finance and Investment—Key Findings and Country Brazil’s FUNDEF. In Incentives to Improve Teaching: Lessons from Profiles. Paris: OECD. Latin America, 151–82. Washington, DC: World Bank. Oreopoulos, P., Salvanes, K.G. 2011. "Priceless: The Nonpecuniary Benefits Ha, W., Yan, F. 2018. Does money matter? The Effects of Block of Schooling." Journal of Economic Perspectives, 25 (1): 159-84. Grants on Education Attainment in Rural China: Evidence Patrinos, H. A., Angrist, N. 2018. “Global Dataset on Education Quality: A from Intercensal Population Survey 2015. International Journal of Review and Update (2000–2017).” World Bank, Washington, DC.–– Educational Development 62: 174–83. Patrinos, H., Sakellariou, C. 2006. Economic Volatility and Returns to Hall, R.E., Jones, C.I. 1999. Why do Some Countries Produce so Education in Venezuela: 1992–2002. Applied Economics 38(17): Much More Output per Worker than Others? Quarterly Journal of 1991-2005. Economics 114: 83–116. Patrinos, H. A., Jakubowski, M., Gajderowicz, T. 2023. Evaluation of Hanushek, E.A. and Woessmann, L. 2010. Education and Economic Educational Loss in Europe and Central Asia. Policy Research Growth. Economics of Education Review 60(67): 1. Working Papers; 10542. Haveman, R.H., Wolfe, B.L. 1984. “Schooling and Economic Well- Psacharopoulos, G., Velez, E., Panagides, A., Yang, H. 1996. Returns to Being: The Role of Nonmarket Effects.” Journal of Human education during economic boom and recession: Mexico 1984, 1989 Resources 19 (4): 378–407. and 1992. Education Economics 4(3): 219-230. Heckman, J.J. 2008. Schools, Skills, and Synapses.  Economic Psacharopoulos, G. 1973. Returns to Education: An International Inquiry 46(3): 289-324. Comparison. Elsevier, Amsterdam. Hendricks, L. 2002. How Important is Human Capital for Psacharopoulos, G., Montenegro, C., Patrinos, H.A. 2017. “Education Development? Evidence from Immigrant Earnings.  American Financing Priorities in Developing Countries.” Journal of Economic Review 92: 198–219. Educational Planning and Administration 31(1): 5-16.” Hillman, A.L., Jenkner, E. 2004. Educating Children in Poor Countries. Psacharopoulos, G., Tan, J.P., Jimenez, E. 1986. Financing Education in Economic Issues No 33. International Monetary Fund. Developing Countries: An Exploration of Policy Options. World Bank, IMF. 2022. World Economic Outlook Update: Gloomy and More Uncertain. Washington, DC. IMF, Washington, Psacharopoulos, G., Patrinos, H.A. 2018. Returns to investment in International Commission on Financing Global Education Opportunity. education: a decennial review of the global literature. Education 2016. The Learning Generation: Investing in Education for a Changing Economics 26(5): 445-458. World. International Commission on Financing Global Education Psacharopoulos, G., Patrinos, H. 2018. Returns to Investment in Opportunity, New York. Education: A Decennial Review of the Global Literature. World Jakubowski, M., Gajderowicz, T., Patrinos, H. 2023. Global Learning Bank PRWP 8402. Loss in Student Achievement: First Estimates Using Comparable Psacharopoulos, G., Collis, V., Patrinos, H., and Vegas, E. 2021. The Reading Scores. Economics Letters. COVID-19 Cost of School Closures in Earnings and Income across Kraay, A. 2019. The World Bank Human Capital Index: A Guide. World the World. Comparative Education Review, Vol. 65. No. 2. Bank Research Observer 34(1): 1-33. Ratha, D., Plaza, S., Kim, E.J., Chandra, V., Kurasha, N., Pradhan, B. 2023. Lewis, B.D., Smoke, P. 2017. Intergovernmental Fiscal Transfers and Migration and Development Brief 38: Remittances Remain Resilient Local Incentives and Responses: The Case of Indonesia. Fiscal but Are Slowing. KNOMAD–World Bank, Washington, DC. Studies 38(1): 111–39. Read, L. 2020. COVID-19 and options for financing education. Loureiro, A., Cruz, L. 2020. Achieving World-Class Education in Adverse Background paper prepared for the Save Our Future white paper Socioeconomic Conditions: The Case of Sobral in Brazil. Washington, Averting an Education Catastrophe for the World’s Children. Save DC: World Bank. Our Future. Manuel, M., Coppard, D., Dodd, A., Desai, H., Watts, R., Christensen, Riggall, A., Kashefpakdel, E., Mullan, J., Rajagopalan, K., Sutoris, P., Z., Manea, S. 2019. Subnational Investment in Human Capital. Korin, A. 2021. COVID-19 and the Non-State Education Sector. Development Initiatives, Bristol, UK. Education Development Trust. Montenegro, C.E. and Patrinos, H.A. 2021. A Data Set of Comparable Sánchez, A., Gregory, L., Crawford, M., Oviedo, M., Herman, R., Estimates of the Private Rate of Return to Schooling in the World, Ahlgren, E. 2023a. Learning Recovery to Acceleration: A Global 1970–2014. International Journal of Manpower. Update on Country Efforts to Improve Learning and Reduce Miller, T., Mauricio, H. 2011. The Fiscal Impact of Demographic Change in Inequalities. World Bank, Washington DC. Ten Latin American Countries in Population Aging Is Latin America Sánchez, J.B., M.S. Lai, C. Sawyer, P. Gerland. 2023b. Population, Ready? World Bank. Education and Sustainable Development: Interlinkages and Select Naurin, A., Pourpourides, P.M. 2023. On the Causality Between Policy Implications. Policy Brief No. 152. UN DESA, New York. Household and Government Spending on Education: Evidence Schoellman, T. 2012. Education Quality and Development from a Panel of 40 Countries. Empirical Economics 65: 567-585. Accounting. Review of Economic Studies 79: 388–417. 32 EDUCATION FINANCE WATCH 2023 Schady, N., Holla, A., Sabarwal, S., Silva, J. 2023. Collapse and Recovery: Wetzel, D.L., Viñuela, L. 2020. Intergovernmental Fiscal Transfers and How the COVID-19 Pandemic Eroded Human Capital and What Performance Grants in Brazil. In Intergovernmental Transfers in to Do about It. World Bank. Federations, 204–23. Edward Elgar Publishing. Sianesi, B. and Reenen, J.V. 2003. The Returns to Education: World Bank. 2012. Making Better Use of Teachers: Strengthening Macroeconomics. Journal of Economic Surveys 17(2): 157-200. Teacher Management to Improve the Efficiency and Equity of Schultz, T. 1975. The Value of the Ability to Deal with Disequilibria. Public Spending. Jakarta: World Bank. Journal of Economic Literature 13(3): 827-846. World Bank. 2018. World Development Report 2018: Learning to Realize Teacher Task Force. 2021. Closing the gap: ensuring there are enough Education’s Promise. World Bank, Washington, DC. qualified and supported teachers in sub-Saharan Africa. UNESCO. World Bank. 2020. The COVID-19 Pandemic: Shocks to Education Tinbergen, J. 1974. Substitution of Graduate by other Labour. Kyklos and Policy Responses. World Bank, Washington, DC. 27(2): 217-226. World Bank. 2021. Global Economist Prospects. World Bank, United Nations. 2020. Policy brief: Education during COVID-19 and Washington, DC. beyond. UN Policy Briefs. World Bank. 2023. Global Economist Prospects. World Bank, United Nations. 2023. Population, Education and Sustainable Washington, DC. Development: Report of the Secretary-General. E/CN.9/2023/2. World Bank and UNESCO. 2021. Education Finance Watch 2021. World United Nations. Bank, Washington DC. UNESCO. 2021. Global Education Monitoring Report, 2021/2: Non-state World Bank and UNESCO. 2022. Education Finance Watch 2022. World Actors in Education: Who Chooses? Who Loses? UNESCO, Paris. Bank, Washington, DC. UNICEF. 2022. Commitment to Action on Foundational Learning: Ensure World Bank, UNESCO, UNICEF, USAID, FCDO, and Bill and Foundational Learning as a Key Element to Transform Education. Melinda Gates Foundation. 2022. “The State of Global Learning UNICEF. 2023. Transforming Education with Equitable Financing. Poverty: 2022 Update.” Conference Education, June 23, 2022. UNICEF, New York. Xiao, Y., Li, L., and Zhao, L. 2017. “Education on the Cheap: The Long- Vegas, E, Coffin, C. 2015. When Education Expenditure Matters: An Run Effects of a Free Compulsory Education Reform in Rural Empirical Analysis of Recent International Data. Comparative China.” Journal of Comparative Economics 45 (3): 544–62. Education Review 59(2): 199-398. © Nafise Motlaq | World Bank | 2013 33 EDUCATION FINANCE WATCH 2023 Acknowledgments The EFW2023 was prepared by May Bend, Yitong Hu, Yilin Pan, Harry Anthony Patrinos, Thomas Poulsen, Angelica Rivera-Olvera, Nobuyuki Tanaka (World Bank), Manos Antoninis and Yuki Murakami (GEM Report). It was prepared under the overall guidance of Luis Benveniste, Jaime Saavedra and Halil Dundar (World Bank). It benefited from UIS data access and comments from Silvia Montoya (UIS), inputs from Omoniyi Babatunde Alimi, Kebede Feda, and Neha Devi Tengur (World Bank), and comments from Samer Al-Samarrai, Raja Bentaouet Kattan, Myra Murad Khan, and Norbert Rudiger Schady (World Bank). The team benefited from valuable supports from Patrick A. Biribonwa, Stefano De Cupis, Raiden C. Dillard, Tania Fragnaud, Massimo Mastruzzi, Kristyn Schrader-King (World Bank) and Katherine Redman (GEM Report). © 2023 International Bank for Reconstruction and Development / The World Bank and UNESCO solely with you. If you wish to re-use a component of the work, it is your responsibility to determine This work is a co-publication of The World Bank and UNESCO. whether permission is needed for that re-use and to obtain permission from the copyright owner. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the Examples of components can include, but are not limited to, tables, figures, or images. views of The World Bank, its Board of Executive Directors, or the governments they represent, or those of UNESCO. All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; The World Bank and UNESCO do not guarantee the accuracy, completeness, or currency of the data e-mail: pubrights@worldbank.org. included in this work and do not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, Designed by alejandro espinosa/sonideas.com. processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank and Photography credits— 1: © Vincent Tremeau | World Bank | 2020. Used with the permission of UNESCO concerning the legal status of any territory or the endorsement or acceptance of such Vincent Tremeau | World Bank | 2020. Further permission required for reuse. 2: © Vincent Tremeau boundaries. | World Bank | 2020. Used with the permission of Vincent Tremeau | World Bank | 2020. Further permission required for reuse. 3: © Ed Wray | World Bank | 2012. Used with the permission of Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of Ed Wray | World Bank | 2012. Further permission required for reuse. 4: © Atet Dwi Pramadia | the privileges and immunities of The World Bank, all of which are specifically reserved. World Bank | 2018. Used with the permission of Atet Dwi Pramadia | World Bank | 2018. Further permission required for reuse. 5: © Charlotte Kesl | World Bank | 2010. Used with the permission of Rights and Permissions Charlotte Kesl | World Bank | 2010. Further permission required for reuse. 6: © Deshan Tennekoon | World Bank | 2010. Used with the permission of Deshan Tennekoon | World Bank | 2010. Further permission required for reuse. 7: © Irina Oleinik | World Bank | 2012. Used with the permission of This publication is available in Open Access under the Creative Commons Attribution-ShareAlike Irina Oleinik | World Bank | 2012. Further permission required for reuse. 8: © Dana Smillie | World 3.0 IGO (CC-BY-SA 3.0 IGO) license (http://creativecommons.org/licenses/by-sa/3.0/igo/). Under Bank | 2010. Used with the permission of Dana Smillie | World Bank | 2010. Further permission this license, you are free to copy, distribute, transmit, and adapt this work, including for commercial required for reuse. 9: © Sarah Farhat | World Bank | 2017. Used with the permission of Sarah Farhat purposes, under the following conditions: | World Bank | 2017. Further permission required for reuse. 10: © Nafise Motlaq | World Bank | 2013. Used with the permission of Nafise Motlaq | World Bank | 2013. Further permission required for Attribution—Please cite this publication as follows: The World Bank and UNESCO (2023). Education reuse. 11: © Dominic Chavez | World Bank | 2016. Used with the permission of Dominic Chavez | Finance Watch 2023. Washington D.C., Paris: The World Bank and UNESCO. World Bank | 2016. Further permission required for reuse. 12: © Nafise Motlaq | World Bank | 2013. Share Alike — If you remix, transform, or build upon the material, you must distribute your Used with the permission of Nafise Motlaq | World Bank | 2013. Further permission required for contributions under the same license as the original. reuse. 13: © Khasar Sandag | World Bank | 2013. Used with the permission of Khasar Sandag | World Translations—If you create a translation of this work, please add the following disclaimer along with Bank | 2013. Further permission required for reuse. 14: © Dominic Chavez | World Bank | 2016. the attribution: This translation was not created by The World Bank or UNESCO and should not be Used with the permission of Dominic Chavez | World Bank | 2016. Further permission required considered an official World Bank or UNESCO translation. The World Bank and UNESCO shall not for reuse. 15: © Conor Ashleigh | World Bank | 2015. Used with the permission of Conor Ashleigh | be liable for any content or error in this translation. World Bank | 2015. Further permission required for reuse. 16: © Gerhard Jörén | World Bank | 2012. Used with the permission of Gerhard Jörén | World Bank | 2012. Further permission required for Adaptations—If you create an adaptation of this work, please add the following disclaimer along reuse. 17: © Arne Hoel | World Bank | 2016. Used with the permission of Arne Hoel | World Bank | with the attribution: This is an adaptation of an original work by The World Bank and UNESCO. 2016. Further permission required for reuse. 18: © Ollivier Girard | World Bank | 2019. Used with the Views and opinions expressed in the adaptation are the sole responsibility of the author or authors permission of Ollivier Girard | World Bank | 2019. Further permission required for reuse. 19: © Bart of the adaptation and are not endorsed by The World Bank and UNESCO. Verweij | World Bank | 2012. Used with the permission of Bart Verweij | World Bank | 2012. Further permission required for reuse. 20: © Arne Hoel | World Bank | 2010. Used with the permission of Third-party content—The World Bank and UNESCO do not necessarily own each component of Arne Hoel | World Bank | 2010. Further permission required for reuse. 21: © Nafise Motlaq | World the content contained within the work. The World Bank and UNESCO therefore do not warrant Bank | 2013. Used with the permission of Nafise Motlaq | World Bank | 2013. Further permission that the use of any third-party-owned individual component or part contained in the work will not required for reuse. infringe on the rights of those third parties. The risk of claims resulting from such infringement rests DOI: https://doi.org/10.54676/XTON4555