E L P EARLY LEARNING PARTNERSHIP E L P EARLY LEARNING PARTNERSHIP Over-Enrollment in the Early Grades EARLY LEARNING PARTNERSHIP In this short note, we summarize findings from a KEY MESSAGES forthcoming analysis for 39 countries with unusually large over-enrollment in the early primary grades. We n There is a substantial problem of investigate the scope of the problem, the underlying unusually large over-enrollment across many low-income countries. causes and some potential solutions. Our analysis reveals that this over-enrollment is explained n This over-enrollment is primarily primarily by excessive, sustained repetition in the caused by repetition in the early grades early grades. This pattern results in substantial of primary school (which is under- reported within official statistics in system inefficiencies and wastage. We discuss policy many countries). The large gap between implications related to expansion of pre-primary enrollment in Grade 1 and 2 is often education, improvement of quality in the early grades attributed to dropping out, when in and allocation of resources. most cases it is just differential over- enrollment. The world is facing a learning crisis. n This excessive repetition results in substantial system inefficiencies and As countries around the world have abolished school fees, wastage, with estimated costs for some millions of new children have entered education systems. Many countries of 1.2 extra years of education of these children, however, cannot read, write, or do basic math, per child and an estimated 5% to 10% of even after several years of primary school. Recent estimates the education budget wasted. suggest more than 600 million children are not reaching basic proficiency.i n There is evidence to suggest that increasing access to pre-primary This learning crisis has its roots in children’s earliest years enrollment could help alleviate this when we fail to invest in quality early learning experiences over-enrollment. to build strong foundations. n Many countries are already paying for There seems to be a problem in the “Foundational First Fiveii” by an inefficient version of early childhood which we mean the two years before primary school and the education because parents are first three years of primary school. Too many children enter choosing to enroll children in primary school without the preparation they need to succeed, and then school early with the expectation that enter classrooms that are overcrowded and low quality. Without they will repeat early primary grades. strong foundations, children risk churning through a cycle of underperformance, repetition, and, eventually, drop-out. n Countries could potentially afford In many low-income countries, approximately half of the expansion of pre-primary education children in Grade 2 cannot read a single word in a simple using the resources that are currently sentence. Figure 1 shows the distribution of these students wasted on repetition and over- who are unable to read across a set of 53 country/language enrollment in the early grades. combinations. This brief summarizes findings from a longer forthcoming report “Over-Enrollment in the Early Grades,” prepared jointly by teams at RTI and the World Bank. Lead authors for RTI are Luis Crouch and Katherine Merseth. Lead author for the World Bank is Amanda Devercelli, with Min Ju Choi, Amina Denboba and Anita Gurgel contributing. 1 Figure 1: Students who cannot read a single word within a simple sentence1 100% 2.5 2.3 80% 2.1 1.9 60% 1.7 40% 1.5 1.3 20% 1.1 0.9 0 0.7 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 0.5 E1/P E2/P Arbitrary country/region/language ID The education system is dramatically overcrowded Figure 2: Trends in over-enrollment in primary school NEPAL MADAGASGAR in the early primary grades. from 39 countries E6/P11 E6/P12 Nearly 40 countries in the world have enrollment rates 100% Enrollment Enro in Grade 1 that exceed the population of children of E5/P10 2.5 E5/P9 Pop Population Median corresponding grade-for-age E4/P9 by 30% or more. Figure 2 2.3 E4/P8 presents the progression of enrollment as children proceed 80% Ideal E3/P8 2.1 E3/P7 through the education system in the 39 countries which 1.9 we have analyzed. EachE2/P7line represents data for one 60% E2/P6 1.7 country and the teal line represents the median. The blue E1/P6 E1/P5 line represents the hypothetical 0 line where the number 1M 200,000 400,000 600,000 800,000 of 1.2M 1.5 1.4M 1.6M 1.8M 40% 0 200,000 400,000 600,000 800,000 1M children enrolled would match the number of children at 1.3 the appropriate age. In the earlier grades (grades 1-3), there 1.1 is substantial over-enrollment; this decreases over time 20% 0.9 so that by the end of primary school there is substantial 0.7 7 9 11 13 15 17 19under-enrollment. 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 00.5 1 3E1/P5 7 11 9 E2/P 13 15 E3/P 17 19 21 23 25 E4/P 27 29 E5/P31 33 35 37 E6/P 39 41 43 45 47 49 Arbitrary country/region/language ID This unusually large over-enrollment is largely the Arbitrary country/region/language ID result of persistent repetition in the early primary grades. Figure 3. Enrollment compared to population of children NEPAL MADAGASGAR Figure 3shows enrollment trends in Madagascar as an at appropriate age in Madagascar 160 example. The red triangle highlights the E6/P12 unusually large Enrollment MADAGASGAR 140 over-enrollment in the early grades. In Madagascar, there Enrollment Population E5/P9 120 is extreme over-enrollment Population in Grade 1 with more than 1.5 E6/P11 E6/P12 100 million children enrolled, compared with E4/P8only 680,000 Percentage Enrollment E5/P10 80 E5/P9 children at the appropriate age. The over-enrollment E3/P7 60 Population continues to exist but drastically decreases E2/P6 in Grade 2 and E4/P9 E4/P8 40 3, before evening out in Grade 4 and then inverting by E1/P5 E3/P8 20 E3/P7 Grade 5, so that there are fewer children enrolled than in 0 0,000 400,000 600,000 800,000 1M 1.2M 1.4M 1.6M 1.8M the appropriately-aged population. 0 200,000 400,000 E2/P7 600,000 800,000 1M 1.2M Bo om Quintile 2nd Quintile Middle Quintile E2/P6 4th Q E1/P6 G1 to 7 G2 to G1 GREPRE E1/P5 0 200,000 400,000 600,000 800,000 1M 1.2M 1.4M 1.6M 1.8M This analysis is based on data from real countries, but we are not using individual countries’ name. 1 2 1.3 20% 1.1 0.9 0 0.7 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 0.5 E1/P E2/P E3/P E4/P Arbitrary country/region/language ID Figure 4 shows the enrollment trends in Nepal. The Figure 4: Enrollment compared to population of yellow triangle for Nepal highlights the larger size of over- children at appropriate age in Nepal enrollment in earlier grades. In Grade 1, there are 1.1 million children enrolled, despite there being less than 630,000 NEPAL MADAGASGAR children of the appropriate age. Over-enrollment decreases 160 2.5 E6/P11 throughout the grades but still continues to exist in Grade E6/P12 Median Enrollment 140 Enrollment E5/P10 5, with more than 2.3 730,000 enrolled children compared to E5/P9 Population 120 Population Ideal just over 660,000 children 2.1 of the appropriate age. 100 E4/P9 E4/P8 Percentage In Malawi, children in Grade 2 are, on average, 1.7 80 1.9 E3/P8 years older than children in Grade 1 . This cannot happen E3/P7 60 through late enrollment, 1.7 as even late enrollees into Grade E2/P7 E2/P6 40 1 should age only 1 1.5year, if they do not repeat. The fact 20 E1/P6 that children “actively” age 1.7 years in Grade 1 suggests E1/P5 1.3 0 0 large400,000 200,000 though scale,600,000 1M informal, perhaps 800,000 1.2M repetition. 1.4M 1.6M The 1.8M 0 200,000 400,000 600,000 800,000 1M 1.2M 1.1 same survey confirms that there are about twice as many children in Grade 0.9 1 as there are in the population of appropriate age. In 0.7 Burundi, according to the most recent 43 45 47 49 51 53MICS survey, children age three years in the first two 0.5 grades. Similar to the caseE1/P of Malawi,E2/P there are about E3/P twice E4/P E5/P E6/P as many children in Grade 1 as in the population. Low levels of pre-primary education enrollment, Figure 5: Relationship between foundational year permanently-inflated Grade 1 intake rates, over- variables and GDP per capita. NEPAL enrollment and repetition in Grade 1, and a large drop- off between Grades 1 and 2 are tightly associated. 160 E6/P12 Enrollment 140 Table 1 presents a summary of our analysis for the 39 Population 120 E5/P9 countries, primarily using four indicators: E4/P8 1. Grade 1 enrollment to population of appropriate age 100 Percentage (column 3) 80 E3/P7 2. Grade 2 enrollment to Grade 1 enrollment (column 4) 60 E2/P6 3. Gross intake into primary schooling (column 5) 40 4. Pre-primary gross enrollment ratio (column 6) 20 E1/P5 We used these four variables to create an index or 0 0 200,000 of 400,000 indicator foundation years dysfunctionality 600,000 800,000 1M all low, for 1.2M Bo om Quintile 2nd Quintile Middle Quintile 4th Quintile Top Quintile lower-middle, and upper-middle income countries. The G1 to 7 G2 to G1 GREPRE INTAKE 39 countries with the highest scores (indicating the most severe over-enrollment and other related variables) are Source: Calculated from downloaded World Bank EdStats database. presented in Table 1. The variables shown in Table 1 are Legend:  G1to7 = Grade 1 enrollment to population of age 7 averages of the most recent 5-year period, 2011 to 2015.2 In G2toG1 = Grade 2 enrollment to Grade 1 enrollment GERPRE = Pre-primary gross enrollment ratio addition, the table presents countries’ values on a summary INTAKE = Gross intake ratio into Grade 1 indicator of foundation years dysfunctionality (column 2). Figure 5 shows that these variables move with each other and do so in a manner that is related to the overall level of development of the countries. Over-enrollment in Grade 1 and the gross intake into Grade 1 both decrease with GDP per capita, and the ratio of Grade 2 to Grade 1 enrollment and the pre-primary gross enrollment ratio both increase with GDP per capita. However, these factors are generally le Quintile 4th Quintile Top Quintile correlated with each other more strongly than they are GREPRE correlated with GDP per capita. Thus, the solution cannot INTAKE simply be “let general development solve the problem.” 2 An average was used because not all countries report data for all years and averaging allows smoothing of unstable numbers and inclusion of countries which report for different, but close years. In a few cases the most recent available data are not for the period 2011-2015 but earlier (Guinea-Bissau and Nicaragua). 3 TABLE 1: DYSFUNCTIONALITY IN THE EARLY YEARS Ratio of Ratio of Index of Ratio of Ratio of Gross Gross Change Gross Grade 1 Grade 1 foundation Grade 1 Grade 2 Enrollment Intake 2006-2010 Intake Ratio Country enrollment enrollment years issues enrollment to to Grade 1 Ratio pre- Ratio to to 2011- into Grade 1 to population to population (2011-2015) population enrollment primary Grade 1 2015 2001-2005 2006-2010 2001-2005 Madagascar 98 2.25 0.70 15 179 2.04 2.02 -0.21 144 Rwanda 91 2.11 0.76 15 167 2.37 2.27 0.26 164 Guinea-Bissau 86 1.74 0.71 6 154 1.74 NA 0.00 NA Sierra Leone 84 1.83 0.77 9 153 2.10 1.84 0.26 NA Benin 81 1.54 0.88 20 152 1.40 1.18 -0.14 103 Ethiopia 79 1.61 0.69 19 148 1.40 1.07 -0.21 96 Uganda 77 1.54 0.69 11 141 1.78 2.05 0.24 177 Angola 76 1.78 0.94 79 169 1.53 NA -0.24 NA Togo 71 1.54 0.85 13 132 1.53 1.40 -0.01 102 DRC 71 1.38 0.74 4 127 1.31 0.81 -0.07 69 Chad 70 1.59 0.68 1 125 1.30 1.08 -0.29 84 Nicaragua 70 1.78 0.77 58 148 1.99 1.84 0.21 155 Malawi 70 1.99 0.76 82 158 2.01 2.21 0.02 177 Burundi 69 1.78 0.74 9 126 1.92 1.04 0.13 80 Nepal 69 2.04 0.73 83 158 2.11 1.87 0.07 114 Cambodia 68 1.39 0.87 15 127 1.76 2.07 0.37 157 Yemen 64 1.21 0.81 1 114 1.13 1.09 -0.08 104 Cameroon 63 1.44 0.77 32 126 1.34 1.36 -0.10 98 Timor-Leste 63 1.79 0.84 17 118 1.94 2.36 0.15 107 Syrian Arab 62 1.32 0.86 10 114 1.33 1.31 0.00 117 Republic Myanmar 62 1.23 0.86 23 120 1.14 1.22 -0.10 125 Mauritania 61 1.10 0.97 7 112 1.07 1.15 -0.03 104 Bangladesh 59 1.27 1.03 29 117 1.28 1.33 0.01 119 LPDR 59 1.45 0.82 28 116 1.86 1.81 0.41 116 Botswana 58 1.19 0.94 17 110 1.18 1.20 -0.01 113 Mozambique NA 1.59 0.87 NA 154 1.54 1.49 -0.05 119 Vanuatu NA 1.45 0.84 99 NA 1.43 1.41 -0.02 NA Namibia 55 1.31 0.85 18 106 1.23 1.23 -0.08 104 Solomon 44 1.32 0.92 97 120 1.31 1.18 -0.01 NA Islands Liberia 34 1.35 0.81 157 128 1.22 NA -0.13 NA 4 How much is unusually large over-enrollment costing countries? The following country examples illustrate that countries are spending almost twice as much as they could to educate students in the early grades because many children repeat the early grades at least once. To illustrate the potential costs to systems, we present here three scenarios, all representing real data from real countries. All three of these cost-efficiency analysis approaches confirm that countries are losing significant resources due to wastage during the Foundational First Five years. 1 Unusually large over- 2 Over-enrollment, mostly due 3 A combination of the factors enrollment in the early to repetition, costs education during the Foundational grades can cost 5% of the systems the equivalent of First Five years of education annual education budget approximately 1.2 grades of explains 34% of the variance of a country each year. extra education per student. in completion rates across How did we calculate this? A second approach offers a countries, which is more way to double-check the above than expenditure per student • Enrollment in Grade 1 calculations by approximating the explains, signaling that these accounts for 28% of total costs in similar countries that do primary enrollment in factors account for a lot of not repeat students. For example, Rwanda. the identified countries with high the internal inefficiency of • Rwanda spends about 34% over-enrollment had a median these systems. of its education budget on share of primary schooling in In order to determine how these primary education. total education expenditure variables affect the production at 46%. Over-enrollment in efficiency of primary school • 56% of Grade 1 enrollment is the first two grades in these completers, we created a summary over-enrollment. countries represented some 20% variable proxying for “foundation • Therefore, reducing Grade 1 years issues” -- a simple of enrollment. If this could be over-enrollment could free eliminated, the median share combination of the four variables up about 5% of the primary of primary schooling in total indicated earlier (Pre-primary education budget. education expenditure could gross enrollment ratio; Gross Rwanda spends only about 0.2% be reduced to 37%. In addition, intake into primary schooling; of its education budget on pre- we looked at a set of countries Grade 1 enrollment to population primary education. By addressing with similar GDP per capita to aged 7; Grade 2 enrollment to wastage due to Grade 1 over- those with unusually large over- Grade 1 enrollment). As a measure enrollment, Rwanda could free up enrollment, which managed to of expenditure, we used public 5% of the education budget, which enroll all their children without per-pupil spending in primary could be used to dramatically creating over-enrollment. For school as a proportion of GDP increase spending on pre-primary these countries, the share of per capita. We observed very little (even more resources would be primary schooling in total relationship between primary available if over-enrollment in education spending was 38%. school completion and per Grades 2 and 3 was addressed as These comparisons suggest that, pupil expenditure as a share of well). with efficiency savings, spending GDP per capita. In contrast, our on primary education could be summary indicator explains 34% reduced somewhat. of the variance in the completion rate. Thus, a combination of the “Foundational First Five” factors explains completion more effectively than expenditure per student does, signaling that these factors account for substantial internal inefficiency within these systems. 5 What are the implications for policy, planning, and financing for education? 1 Families are choosing 3 Even those who enter at the 6 Countries can promote to enroll their children right age do not learn learning by having young in primary school early at adequate levels. children attend school in (possibly due the absence For many of the countries developmentally appropriate, of affordable pre-primary examined, children in Grades child-centered classrooms 2 and 3 also continue to age education). with same-age peers. more than one year per grade, Some parents are expecting their If children enter primary indicating substantial repetition children not to learn enough in school at the right age, with in Grades 2 and 3. It is likely that their first “run-through” of Grade better foundations, they are the continued over-enrollment in 1. A recent household survey in more likely to progress through Grades 2 and 3 is a consequence Uganda indicated that 28% of primary school on time, with less of the lack of preparation for parents sent their children to repetition and less drop-out. In schooling and the low levels of primary school “early” because Uganda, for example, a recent learning in Grade 1. it was free (unlike pre-primary in-depth analysis and household school) and that 56% of these survey revealed that attending parents did so understanding 4 Some countries are already pre-primary school reduced the that their children might have to risk of repetition of Grade 1 by spending substantial repeat Grade 1.v approximately half. resources on early childhood education because of either official or (more likely) 7 Progress is possible 2 Current official statistics may drastically underestimate unofficial repetition. through improved the actual repetition rate and Data suggest that, by having quantity and quality early thus insufficiently account ratios of enrollment to the childhood education and appropriately aged population improved attention to the for it during education sector of 1.50 and 1.28 in Grades 1 and planning. first few grades. 2, respectively, these 39 observed Many countries in Latin America Or, they may interpret the countries may be, to some extent, demonstrated the same patterns enrollment drop-off between already paying for pre-primary we have described in this paper Grades 1 and 2 as dropping out, education, either explicitly (when during the 1980s but have when it is most likely just higher funding formulas are driven by since made progress. In Peru, repetition in Grade 1, and would enrollment numbers) or implicitly for example, a combination have vastly different policy and (in the form of inefficiency and of increased access to early expenditure implications. wastage). childhood education, reforms to improve basic education quality 5 As suggested by other and targeting of early education to those most in need helped research and buttressed contribute to a reduction in by the research presented Grade 1 enrollment from 165% here, programs to target of the population in 1995 to an the children most in need almost even 100% today. of stronger foundations are likely to yield high returns, not just via better social indicators in the long term, but in terms of short-run fiscal savings. 6 Could expansion of pre-primary education help reduce unusually large over-enrollment? Countries with substantial over-enrollment in Pre-primary education could be part of the solution, Grade 1 (and to some degree Grade 2 as well) largely but only if it is high quality. coincide with those having very low pre-primary Expanding pre-primary education without addressing grade enrollment ratios. repetition and classroom quality risks further bloating a For the 39 countries with Grade 1 over-enrollment system that is already over-paying. In fact there is some exceeding 30%, the pre-primary school enrollment ratio evidence that simply adding one more grade to inefficient was 24% on average. Low levels of pre-primary service systems just transfers the over-enrollment problem one provision are not surprising, because the poorest countries grade down. Stimulating and supportive interactions have the least resources to devote to education services between teachers and students and effective use of age- that have long been considered “nonessential.” appropriate curriculum are essential elements of quality; governments must develop their capacity to ensure that these conditions of quality can be met. The cost of improving the Foundational First Five years in many countries is already being paid, It is important to admit that shifting resources from either in terms of direct fiscal costs or in terms of wastage in the early primary grades to quality pre- wastage and inefficiency, but without good results. primary programs might not be cost-neutral on a Expansion of pre-primary education is one potential per child basis. solution to alleviate the current bulge in enrollment Expansion of pre-primary might cost more per child if the in the early grades. Countries could open pre-primary current over-enrollment is absorbed through crowding. classrooms by shifting resources currently spent on over- But if pre-primary expansion and quality improvements crowded classrooms and repetition in Grade 1 and Grade in the first few grades are paired together, the cost per 2. In the countries identified as having unusually large primary school completer is likely to come down. In over-enrollment, pre-primary enrollment is associated other words, many systems can expect some primary- with a reduction of the over-enrollment and repetition in cycle cost-savings through improving system efficiency, Grade 1. An increase of one standard deviation in pre- and reducing the cost per primary school completer (i.e., primary enrollment is associated with a reduction of 0.27 reducing the “cycle cost per completer”), which could be standard deviations of over-enrollment in Grade 1, and used to fund pre-primary expansion, which in turn would increasing from none to complete pre-primary enrollment further reduce wastage during the primary cycle possible is associated with a 21% reduction in the Grade 1 over- (assuming quality design and implementation). enrollment. 7 RESOURCES UNESCO. Counting the Number of Children Not Learning: Methodology for a Global Composite Indicator for i Education. UIS Information Paper 47 (2017) http://uis.unesco.org/sites/default/files/documents/ip47-counting-number- children-not-learning-methodology-2017-en.pdf Crouch, L. & Merseth, K.A. Stumbling at the First Step: Efficiency implications of poor performance in the foundational ii first five years. Prospects (2017). doi:10.1007/s11125-017-9401-1. iii Malawi National Statistical Office and UNICEF 2008, p. 313. iv Burundi Institut National de Statistiques et d’Études Économiques 2008, Tables DQ1 and DQ8 Weatherholt, T., Crouch, L., Pressley, J., Healey, H., Jordan, R., Merseth, K., and Dombrowski, E. (March, 2018). Uganda v Early Years Study. Report prepared by RTI International for the UK’s Department for International Development, East Africa Research Hub. Schiefelbein, E., & Wolff, L. (1993). Repetition and inadequate achievement in primary schools in Latin America: vi Magnitudes, causes, relationships and strategies. Santiago, Chile: UNESCO Bulletin of the Principal Project in Education. 8