1 - The Learning Crisis

Summary

This chapter introduces donor financing of education reform in developing countries and the global compacts in the sector. There is an assumption that learning was not part of these compacts and if only it were part of the discourse and learning was consistently measured, this crisis would not be upon us. However, the first global event in Jomtien, Thailand in 1990 included a significant and urgent obligation of the donor community to address learning. This concerted focus disappeared in following global events. Nonetheless, donors influenced by the shared passion for learning at Jomtein were consistently involved with measuring learning, helping to keep an eye on the unfolding learning deficit. Three examples are offered to illustrate this commitment – in a project (the Indian District Primary Education), across regions and during the primary early grades. These examples shine a spotlight on the tried and tested assessments of learning. In spite of donor focus on this important aspect of education, there is a learning crisis across developing countries.

The international community makes an enormous investment in schooling around the globe. Global compacts, especially Education for All (EFA), helped establish a collective approach, shared commitments, and common agendas for the education sector among international development finance institutions (IDFIs). Historically, the approach to working in developing countries was fragmented but evolved over the years, becoming a shared response to the global learning crisis. The coherent response reflects donors’ collective attentiveness to building effective school systems. Considering the size of financial outlays by IDFIs in developing countries, it is essential to understand what was in these global compacts and what was achieved vis-à-vis these agreed global outcomes. This chapter will show how, despite this level of support for over thirty years, the progress made on agreed global outcomes is limited. This global journey in education reform serves as a backdrop to the discussion of the interventions, the evidence base, and finally the reimagining of what needs to be done to resolve the learning crisis.

Considerable soul-searching has taken place regarding the limited progress made vis-à-vis the EFA global outcomes. Accompanying this is a widely shared assumption that the learning crisis would not have happened if learning was part of the global compacts and, thereby, systematically monitored through student assessments. Take, for example, the assumption inherent in the first recommendation in the recent World Development Report on Education (World Bank 2018a):¹

Assess learning – to make it a serious goal. Measure and track learning better; use the results to guide action … What gets measured gets managed. Lack of measurement makes it hard to know where things are, where they are going, and what actions are making any difference. Knowing these things can provide focus and stimulate action … Use measurement to shine a light on learning. The first step to improving systemwide learning is to put in place good metrics for monitoring whether programs and policies are delivering learning.

(p. 17)

So, the argument put forward is that if only international donors facilitated the proper measurement of learning and understood the abysmal performance in learning, reform programs would have addressed the constraints unheeded in the past (Burdett 2016). On the contrary, this chapter shows how IDFIs have assessed learning levels for a long time, and the assumption that learning was not center stage is incorrect.² However, this measurement has done little to improve donor programming that can improve learning. International development finance institutions were always keen on understanding the learning landscape and finding solutions to the bottlenecks to quality education in the places where they worked. Though IDFIs rejected examinations for determining educational achievement, learning assessments were considered the gold standard. And accordingly, IDFI-financed student assessments were key to observing the stark deficits in learning.³ Quality education itself was an attempt to address the learning crisis.

To further the claim that IDFIs were keen on measuring learning from the very beginning, student learning assessments in three settings are examined: first, within a project; second, across geographical regions; and finally, during the early grades of schooling. These three settings represent the different ways in which IDFIs financed the testing of student learning. For learning within a project, I examine the detailed and extensive learning assessments undertaken in one of the early and large multidonor programs, the Indian District Primary Education Program. To examine the measurement of learning across geographies, I look at three regionally organized learning assessments, two in sub-Saharan Africa and one in the Pacific region. An examination of early grade assessment across the world, which owes its origins entirely to the IDFIs, completes the picture.

It was the IDFIs’ reporting of learning levels over the years that helped the development community name this scenario the “learning crisis.” The work done with learning assessments helped to capture a wide-angle portrait of the learning crisis across developing countries. The World Bank report Ending Learning Poverty: What Will It Take? presented at the fall meetings of the World Bank and International Monetary Fund (IMF) in October 2019 specifically highlights the crisis, characteristically rechristening it in economic terms as “learning poverty” (World Bank 2019). Convinced of the deficit in learning, the global development community has made the decision to cut learning poverty (as an indicator of the learning crisis) by half by 2030. Before debunking the notion that assessment of learning was not part of the global education agenda, let me sketch a brief summary of IDFIs’ funding for education reform.

Funding for Education Reform

Since the 1960s, IDFIs have supported education in developing countries. Though many of the multilateral and bilateral finance institutions were established soon after World War II, funding for education reform only began later in the 1960s (Jones 2007). International entities finance education in two ways.⁴ Budget support, which subsidizes a country’s education budget, is used mainly for recurrent expenses such as teacher salaries. Direct aid, on the other hand, supports specific projects. Projects fund predefined or predesigned activities implemented over a specified period. The detailed design could take a few months to a year to develop. Since the early 2000s, UNESCO’s Global Education Monitoring Reports in an annex titled “Aid Tables” list annual IDFI support for education under two headings – total and direct aid.⁵ Total aid to education includes both budget support and direct aid. Both total and direct aid could be designated for a subsector (such as basic, secondary, etc.) or remain unspecified.

Aid for education is provided by IDFIs in the context of bilateral or multilateral relationships. Aid provided by all the countries that are members of the OECD Development Assistance Committee is listed by UNESCO.⁶ These countries maintain a bilateral relationship with various developing countries. The multilateral group consists of international agencies (mainly banks) and United Nations (UN) organizations. Over the years, additional countries have joined the Development Assistance Committee, and the number of multilateral entities supporting education has also expanded. Initially, only twenty-two countries and about ten multilateral entities financed education in developing countries; in 2019 there were thirty-five countries and sixteen multilaterals.

The swelling in the number of countries and multilateral organizations also resulted in a steady increase in education funding over the years (Table 1.1). Overall, annual aid to all levels of education increased from USD 6.9 billion in 1999 to USD 13.5 billion in 2016. Since this book is about learning, which must start with basic education (grades 1 to 8), keeping track of total and direct aid to basic education is particularly relevant. Total aid to basic education doubled from USD 2.7 billion in 1999 to USD 5.9 billion in 2016. Direct aid to basic education more than doubled, from USD 1.5 billion to USD 4 billion during the same period. The bilateral direct aid to developing countries almost tripled between 1999 and 2016 (USD 1 billion to USD 2.8 billion). Though according to the Global Education Monitoring Report (2018), basic education aid allocated in 2016 was the lowest on record.

Table 1.1 Annual aid to the education sector in USD billion

	Total aid to education		Total aid to basic education		Direct aid to basic education
	Annual average 1999–2000*	2016#	Annual average 1999–2000*	2016#	Annual average 1999–2000*	2016#
Bilateral	5.2	9.5	1.8	4	1	2.8
Multilateral	1.7	3.8	0.9	1.9	0.5	1.2
Total	6.9	13.4	2.7	5.9	1.5	4

Source: Global Education Monitoring Reports aid tables 2008 to 2018.

* Constant 2005 USD millions

^# Constant 2016 USD millions

Global Compacts in Education

So, what are the moments and events that knit the passion and resolve of the international donor community together in their efforts to make a difference in education around the world?⁷ A series of international conferences enabled dialogue among partners and an impressive consensus emerged (Lewin 2011). These events were part of the global EFA movement.⁸ What is notable is that, initially, learning was held up as the primary and most important end for monitoring progress in the education sector. After the first event, learning went backstage and was no longer writ large in these conferences. Before describing the three settings where IDFIs continued to measure learning in the next section, the achievements vis-à-vis the EFA indicators at the end of the Millennium Development Goals era are listed.

The first EFA event took place in Jomtien, Thailand, in March 1990. Hosted by UNESCO, UNDP, UNICEF, and the World Bank (called the “Inter-agency Commission”), the participants included all the multilateral and bilateral agencies, 155 country governments (38 from North America and Europe), and 125 NGOs. The Jomtien conference was an important event as it brought together for the first time representatives from IDFIs, recipients of IDFI funding, and members representing civil society. It created the joint milieu for the commitment to learning. Rosa Maria Torres (former Minister of Education and Culture, Ecuador), who attended this conference, captures the enthusiasm and anticipation: “Jomtien succeeded in creating a spirit of a fresh start, of hope, of ‘this time it’ll work’ … In Jomtien, anything seemed possible, [and] the future looked promising, quality and equity were somewhat new words, and made for credible goals” (Torres 2001).

The Jomtien conference adopted the “World Declaration on Education for All” (Inter-Agency Commission 1990), which included six specific outcomes.⁹ The third outcome is critically important to understand why IDFIs failed with learning. This outcome, which centered on actual learning, states that there will be an “improvement in learning achievement such that an agreed percentage of an appropriate age cohort (e.g., 80 percent of 14-year-olds) attains or surpasses a defined level of necessary learning achievement” (Inter-Agency Commission 1990, p. 53). Again, the report on this conference, written about three years later (UNESCO 1993), has this to say about learning and the measurement of academic achievement:

Sending a child to school is of little benefit if the child does not learn something useful there. Unfortunately, this is the case in too many classrooms, especially in the developing world.

The World Declaration on Education for All [1990] specifically addressed the problem of learning achievement: “whether or not expanded educational opportunities will translate into meaningful development for – an individual or for society – depends ultimately on whether people learn as a result of those opportunities, i.e., whether they incorporate useful knowledge, reasoning ability, skills, and values.”

(UNESCO 1993, p. 36)

The other outcomes focus on early childhood, universal primary school access and completion, expansion of basic education and training, and adult education.

A second EFA conference was held in April 2000 in Dakar, Senegal (UNESCO 2000; King 2011).¹⁰ Similar to Jomtien, attendance was widespread and included all the principal constituencies. According to Torres, who also attended this event, there was a palpable sense of failure and an acknowledgment of an unfinished task: “Jomtien was an invitation to create, to invent and to dream. Dakar, faced with the crass contrast between rhetoric and reality, between documents and facts, between goals and achievements, restrained the imagination, encouraged excuses and self-justification, and provided a temptation to inflate numbers and to blur realities” (Torres 2001). The learning goal, so clearly listed in Jomtien’s World Declaration on Education for All, was redefined as “programming for learning,” and the notion that quality education will automatically lead to learning took hold. The goal states that the focus will be on “improving every aspect of the quality of education and ensuring their excellence so that recognized and measurable learning outcomes are achieved by all, especially in literacy, numeracy and essential life skills” (UNESCO 2000, p. 17). According to Al-Samarrai and colleagues: “the emphasis on educational quality was based on a relatively simple conception of educational reform in developing countries, namely that quality improvements could of themselves lead to significant increases in both enrolments and attainment levels. Insufficient attention was given to social, economic, political and cultural factors that determined the demand for primary education” (Al-Samarrai et al. 2002, p. 2). To add to what Al-Samarrai is alluding to is the bypassing of institutions that contain the politics, culture, and so on operating in a country. The mindset that external agencies can single-handedly introduce, substitute, or motivate reform from the classroom becomes a common sentiment among IDFIs. The last EFA global event took place in Incheon, Korea, in May 2015. Like the other events, more than 1,000 participants attended, representing the same constituencies as the previous conferences. Again, the matter of education quality steering the way to learning was forcefully addressed: “We commit to quality education and to improving learning outcomes, which requires strengthening inputs, processes, and evaluation of outcomes and mechanisms to measure progress” (UNESCO 2015a, p. 68).

Some of this transnational and multi-stakeholder discussion on keeping an eye on learning spilled over into the United Nations’ envisioning of development beyond the Millennial Development Goals. The United Nations General Assembly on September 25, 2015 established the Sustainable Development Goals to continue the emphasis on global development associated with people, planet, prosperity, peace, and partnerships (UNESCO 2015a). Sustainable Development Goal four pertains to education, and ensures “that all girls and boys complete free, equitable and quality primary and secondary education leading to relevant and effective learning outcomes” (United Nations 2015b, p. 20).¹¹

Education Sector Achievements

Before delving into what happened with learning, it is important to summarize IDFI achievements in the area of primary education. Though the learning goals receded by Dakar, the global community monitored the progress and achievements made on six goals and associated eighteen indicators published each year in UNESCO’s Global Education Monitoring Reports.¹² Except for Early Childhood Care and Education (ECCE), UNESCO’s Institute of Statistics is responsible for collecting and reporting on all the outcomes listed in the global compact. Responsibility for early childhood lies with UNICEF.

Table 1.2 shows the progress made by countries between 2000 and 2015 (UNESCO 2018).¹³ Pre-schooling or ECCE may be relevant as it prepares children for first grade academic expectations. It is especially beneficial in contexts with high adult illiteracy and few opportunities for cognitive nurturing available for children before commencing formal school, which is often the case in developing countries. The ECCE goal has two indicators, enrollment in ECCE and students’ transition from ECCE to grade 1. Based on data that is not exhaustive or consistent, the gross enrollment ratio (GER) for pre-primary increased from 32 percent in 2000 to 47 percent in 2015.¹⁴ A Multiple Indicator Cluster Survey was recently introduced by UNICEF in thirty-one low- and middle-income countries to address data issues. Based on this survey, the median ratio for the transition from ECCE to grade 1 was 36 percent, which is low.

Table 1.2 IDFIs’ achievements on education outcomes

Outcome	Indicators	2000	2015
Expand and improve ECCE especially for the most vulnerable and disadvantaged children	GER in ECCE	32%	47%
	Transition from ECCE centers to grade 1		36%
Universal access to and completion of primary education for all children	Gross intake ratio to grade 1	106%	106%
	Net intake ratio to grade 1	65%	69%
	Primary GER	99%	103%
	Primary net enrolment ratio	83%	89%
	Decrease in out-of-school children (in millions)	101	62
	Gross intake ratio in the last year of primary school	82%	90%
Ensure learning needs for youth and adults	GER for lower secondary education	72%	85%
Increases (50%) in adult literacy	Youth literacy rate	87%	91%
	Adult literacy rate	81%	86%
	Youth literacy gender parity index	0.93	0.97
	Adult literacy gender parity index	0.88	0.92
Eliminate gender disparities	GPI in primary GERs	0.92	1
	GPI for secondary GERs	0.92	0.99
	GPIs in tertiary	0.99	1.12
Improving education quality	Teachers trained*		97%
	National standards met with training*	85%	93%
	Pupil–teacher ratios#	26:1	23:1
	Primary repetition rate	5%	1.8%
	Survival rate to the last grade of primary	76%	80%
	Coefficient of efficiency (years to complete primary)	No data
	Methodology for monitoring learning	No consensus

Source: UNESCO 2018.

Note: * % of the teacher workforce; # students per teacher

A second critical goal for the IDFIs was to achieve universal primary education. Several indicators are used to monitor progress on this goal. The gross intake ratio in grade 1 has remained constant at 106 percent throughout this period. The GER, however, is one area where IDFIs had some success. Gross enrollment increased from 99 percent in 2000 to 103 percent in 2015.¹⁵ The World Development Report (World Bank 2018a) lauds the dramatic increase in enrollment, which contrasts with that of developed countries. For example, the United States took forty years “to increase girls’ enrolment from 57 to 88 percent” (World Bank 2018b, p. 4).¹⁶ The net intake and net enrollment ratios offer more fine-grained information on student participation in the school system.¹⁷ However, net intake and net enrollment data are questionable estimates. Reliability depends on whether there is clear and consistent issuance of birth certificates in the country in which a child is born. Many developing countries have not reached this stage yet.¹⁸ For whatever it is worth, the estimate for net intake rates has increased from 65 percent in 2005 to 69 percent in 2015. During the same period, net enrollment rose from 83 percent to 89 percent. The gross intake ratio for the last grade of primary school increased from 82 percent in 2000 to 90 percent in 2015.¹⁹ Overall, the universal primary education data does not align with household surveys (UNESCO 2019), which show that a much higher number had not completed primary school.

The flip image of the success in enrollment is a reduction in children not attending school. The number of out-of-school children supposedly decreased from 101 million in 2000 to 62 million in 2015. These numbers claim a steady decrease since 2008, after which progress in reducing out-of-school children stalled. It is not clear whether the out-of-school measure represents a child’s temporary absence or continuous absence (never attended school). If it is the former it is unclear how many days it takes for a child to be considered a dropout (in a country). In a developed country, these global numbers would be corroborated with daily student attendance data maintained by the teacher, which often does not happen in developing countries. This fuzziness in out-of-school information contributes to the lack of consistency in reporting of out-of-school numbers. For example, in the Millennium Development Goals final report in 2015 (United Nations 2015a), the number of out-of-school children of primary school age had fallen by almost half, to an estimated 57 million in 2015, down from 100 million in 2000. Yet in the 2020 Global Education Monitoring Report (UNESCO 2019), this number had increased to 59 million.

A final EFA goal has to do with the quality of education. The way it was reconceptualized at Dakar reveals that it has less to do with learning outcomes than with inputs such as teacher training and providing instructional materials.²⁰ Consistent information was available for three indicators used to monitor quality (UNESCO 2018). The first one was the number of teachers trained. In 2015, 97 percent of the teachers were academically qualified and 93 percent had met national standards regarding these qualifications. The second indicator was pupil–teacher ratios, which declined during the same period from twenty-six students to a teacher in 2000 to twenty-three students to a teacher by 2015. The survival rate to the last grade of primary school increased from 76 percent in 2000 to 80 percent in 2015.

Though significant progress has been made overall on the EFA/Millennium Development Goals, the achievements lack texture and detail thirty-one years after Jomtien. If learning had remained the priority as advocated by the Jomtien global compact and had remained as an EFA goal, IDFIs would have had to address this vagueness and lack of clarity. As early as 2000, UNESCO raised the alarm by stating, “some countries have shown that a sizeable percentage of children is acquiring only a fraction of the knowledge and skills they are expected to master. What students are meant to learn has often not been clearly defined, well-taught, or accurately assessed” (UNESCO 2000, p. 17).

With learning no longer center stage, the goals, indicators, and results sideline and confuse the meaning and purpose of education. The measure for enrollment and completion illustrates these deficiencies. Even though enrollment in primary schools is reaching near-universal levels, these numbers do not capture the loss of instructional time due to the significant student dropout and teacher absenteeism during the school year.²¹ Early Global Education monitoring reports included tables with student dropout rates, which many countries did not report on. This situation may have caused the UNESCO team to abandon reporting on dropout entirely.²² Student absence compounded with teacher absence (discussed in more detail in Chapter 2) would have had a severe effect on learning in a classroom. Therefore, only measuring enrollment has no meaning without consistent data on the regularity and effectiveness of teaching and learning in the classroom.

Another example that shows how nebulous the goals and indicators are, is the data on the high levels of teachers trained, which is perceived to impact education quality and learning positively. Information on the trained, untrained, or partially trained teacher workforce as representing quality is an inappropriate measure of an effective education system. These categorizations do not reflect the considerable skills teachers must demonstrate and the support that teachers require to function effectively. For example, training does not reflect whether teachers’ knowledge of the subject and the different areas within a subject are adequate to teach that subject and help children become proficient learners. If the IDFIs had seriously grappled with the complexities of producing learning, there would have been red flags on their support to teacher training long before this crisis. I will take this up in detail in Chapter 2.

While the impact of aid on education outcomes is limited, there was a decided movement toward donor coordination in the Fast-Track Initiative (Bermingham 2011; UN 2015b) renamed the Global Partnership for Education (GPE).²³ There is now increased harmonization across the work done by donors and an avoidance of duplication in programming. Donors are working more systematically together collaborating with each other as they work with the country governments that receive financial assistance for education reform. A demonstrable result of this partnership is the support that is provided to a country in the development of plans that cover the entire sector. Though not in detail or directly relevant to improving learning, these plans are based on a situation analysis that captures sector challenges, and proposes actions to address them (Clarke 2017). The analysis is mostly confined to large-scale data and rarely tries to capture and integrate an investigation of meso-level institutions. There are sector strategic plans of over seventy developing countries publicly available on the Global Partnership for Education’s website. The status and links between subsectors are highlighted in these plans, providing an overview of what is going on in a country. Donor transparency and accountability have increased with the establishment of clear baselines and public reporting on the progress made in their programs.²⁴ Harmonization among donors, along with effective collaboration with local governments, can be a potentially powerful and effective instrument for education reform.

To conclude this discussion, the development community’s frustration over the last few decades, with regards to schools not being able to fulfill their fundamental purpose of cultivating children’s minds and building knowledge and skill to navigate the world, is reinserted in the Sustainable Development Goals. Through the Sustainable Development Goals, finally, what a student can understand and use is now upheld as the ultimate objective of an education system, reclaiming the original intentions of the first global compact in education. The next section deals with measuring learning and the IDFIs’ involvement with student assessments that could have been used more effectively and comprehensively in understanding the impact of funding for basic education. It also highlights the futility of believing that if the developing country or an IDFI knew how bad learning is something would or could be done without an in-depth analysis of what was causing the problems at the micro and meso levels.

Measuring Learning

Before discussing how much the donor community was involved with tracking student learning within nations, it would be useful here to outline the kinds of testing of student learning that takes place in developing countries. Though the measurement of learning is not enough to propel systems to address the learning crisis, it is a necessity for many aspects of education. Most importantly, for diagnostic purposes, evaluation of how much children are learning lifts the curtain on how dire the situation is. At the country level, data on learning guides the process of teaching and learning; it helps to review the performance of the teacher; it is crucial for the future of the individual student; and it assists with fine-tuning a system in terms of monitoring and accountability.

There are different ways in which students are tested in school systems across the world. Let me point to the dominant two: formative assessment and large-scale testing. Formative assessment refers to the ongoing, and regular, teacher testing of students’ mastery of the curriculum throughout the school year. Formative assessments are non-negotiable for improving learning as they guide the teacher in what is not learned and the existing levels of learning to build on. Testing is done in classrooms and records are usually maintained at the classroom and school levels. Aggregating this data at the national level is not done as it is difficult to ensure that the circumstances under which the tests were administered are common across schools.²⁵ Even so, for a system to know how formative assessment takes place in classrooms is critical information.²⁶

The learning crisis is based on the results of large-scale testing of student learning at the country level, which includes examinations and summative assessments (Greaney and Kellaghan 2008; Clarke 2012; Kellaghan and Greaney 2019) administered to students in different grades and disciplines. The pronouncement of a learning crisis is not based on the results of examinations, which are referred to as norm-referenced in that the performance of a student is evaluated in the context of the performance of other students taking the same test. The content of examinations is closely aligned to the curriculum representing the norms or body of knowledge on which the student is tested. All developing countries, especially those with colonial histories, conduct examinations and this is the mechanism in place for reviewing student performance.

The learning crisis is based on the results of large-scale assessments, which are criterion-referenced, evaluating whether a population has mastered what is appropriate for an age group. What is tested thus is not closely aligned with a curriculum but generic, allowing for comparisons across countries and subnational entities. Summative assessments claim to help review the effectiveness of the education system and thus help to guide reform. Aligned to the IDFI concern with learning and the limitations with examinations, donors have upheld, on the one hand, assessments in projects and regions for the last thirty years and on the other hand, financed the development and administration of assessments itself.

Several criticisms are leveled against measuring learning only with numbers represented by test scores (Hamilton et al. 2015). Three of these come to the fore. First, focusing on test taking and numbers is argued to be reductionist; thus, unable to capture complex learning. Second, it also tends toward lowering the value of other forms of knowing other than test taking that might be proof of unmeasurable knowledge being mastered. Finally, there is a concern that such measuring of learning can be used negatively by policymakers against teachers and educational administrators. To counter these valid criticisms, measuring learning numerically could be considered a partial but necessary reflection that needs to be adequately supported with other types of information such as the formative assessments of student learning in the classroom and the testing of multiple intelligences.²⁷ It is also critical to understand and monitor how large-scale test results are used to improve learning in the school system.

IDFI Support for Learning Assessments

Let me turn to three instances of IDFI commitment to monitoring learning and their financing of summative learning assessments that helped the donor community realize the weaknesses in learning levels. The first instance is a large-scale and well-funded project – the Indian District Primary Education Project. This project, negotiated in the early 1990s for several billion dollars over a period of about twelve years, shows very early donor support for learning assessments.²⁸ Choosing this project in India is also important as a number of research studies were undertaken in this country using data generated from the states where the project was being implemented. The second instance focuses on direct support for three sets of cross-national assessments in primary education. Two of these are in the Africa region in English and French, respectively, and the third is in the Pacific region. The third instance targets assessments in the early grades in primary school (early grade reading and early grade mathematics). The rationale for early grade assessment is to understand the impact of interventions in the early grades and when the decline in quality education begins resulting in weak learning. These three instances, on the one hand, represent the range of IDFI involvement with measuring learning and, on the other hand, also reveal the depth and spread of the crisis in learning, beginning with a single large country (India) before moving to groups of countries in two regions (Africa and the Pacific). This discussion would not be complete without including the most recent influential IDFI initiative, early grade assessments.

Early Within-Project Assessments

Soon after Jomtien, in the early 1990s, World Bank staff began discussing the receipt of IDA funds with the Indian Secretary of Education (the highest administrative authority for this sector in the Indian government). These conversations coincided with India’s increasing concerns with the sector’s status, resulting in the establishment in November 1994 of the Indian District Primary Education Program (DPEP).²⁹ In collaboration with donors, the Indian government led the design of this program (Abadzi 2002). Beginning with implementation in 42 districts in 7 states, DPEP expanded to cover 272 districts in 18 states.

Several donors funded the program, but in different states.³⁰ The World Bank resourced the project in eighteen Indian states, the UK Department for International Development (DFID)³¹ in four states, the European Commission and the Netherlands in one state each, and UNICEF funded specific components in the state of Bihar.³² Most importantly, the Indian government and donors jointly supervised the program across the country. Since improving education quality was critical to the project, expenditure on any construction was limited to 24 percent and management 6 percent. States contributed 15 percent of the total budget.³³

Student assessments were critical and closely monitored in this project. The objectives of the project included (Hirshberg 2002):

• Providing access for all children to primary school or equivalent non-formal education.

• Reducing overall dropout rates.

• Increasing average learning achievement levels by 25 percent in language and mathematics, and to ensure a minimum score of 40 percent in other subjects.

• Reducing gaps in enrollments, dropout, and learning among gender and social groups.

• Establishing capacity at the district, state, and national levels to plan, manage, and monitor the program.

From the commencement of the project in 1994, DPEP undertook baseline language and mathematics assessments of learning for grade 1 and grade 3/grade 4 students. Midterm learning assessments took place in 1997 and terminal assessments in 2002. Using multilevel statistical methods, the World Bank staff designed student evaluations for the National Council of Educational Research and Training. In the design, students from different backgrounds and schools nested within villages, clusters, blocks, districts, and states were tested. While the data was collected appropriately, the Council staff did not have the background to work with the statistical method appropriate for analyzing assessment data and explanations were also insufficient to enable staff to learn the statistical methods and analyze the information. At midterm and end-term tests, it was just simple descriptive statistics familiar to the staff that was in evidence. Even with this, more could have been done. For example, averages across social groups and regions were analyzed without distribution and ranges in test results.³⁴

Notwithstanding weaknesses with the above assessments, in 2004, Pratham, an NGO, introduced a method of trying to understand children’s learning in the Annual Status of Education Reports (ASER). They began by going into rural India and testing young children in the villages, reporting each year on the results. Over the years, the methods used and the assessment framework have become increasingly sophisticated, and grade-specific information is also collected. From 2016, samples have been census-based, and the age group tested is from five to sixteen-year-olds. Early ASER detailed the dismally low levels of learning in the sampled DPEP states and districts. Learning in the Sarva Shiksha Abhiyan (SSA) program fared no better.³⁵ According to a recent Pratham report (2019), in language, there was a decrease in students in grade 5 who could read a grade 2 text from 53 percent in 2008 to 44 percent in 2018. The average proportion per state of grade 5 students who could read a grade 2 text ranged from 29 percent to 73 percent across Indian states. The number of students in grade 8 who could read a grade 2 text was 69 percent. Grade 5 students who could do division decreased from 34 percent in 2008 to 23 percent in 2018, and the range was 14 percent to 52 percent across Indian states. The number who could do division in grade 8 had only increased to 40 percent.³⁶

Based on extensive school visits during program supervision, donors encountered first-hand non-learning in the classroom. Program review or supervision missions took place twice a year, and each of these visits included representation from donor agencies and the Indian government. Each review had a clear Terms of Reference agreed to by the federal governments and IDFIs. The review team would meet at the capital city, New Delhi (capital of India and headquarters of the Ministry of Human Resource Development) and then pairs made up of review members from different agencies would visit a state for a week. During the week, two to three days would be spent visiting schools and observing teachers teach. These visits would also include extended conversations with education administrators, teachers, community leaders, and parents. On one of these visits about six years into the project, the IDFIs decided to informally administer tests to students in the observed schools. The tallied results of the informal tests were exceedingly low. Most of the children were performing well below grade level.³⁷ While these informal tests gave donors a clear sense of the inadequacies in learning, the government was not amenable to reviewing the results as the process had not been stated and cleared in the Terms of Reference. Moreover, it had been agreed at the start of DPEP that baseline, midterm and end-term achievement tests would be the only means for IDFIs to monitor learning.

Recently, DFID (National Council for Education Research and Training 2020) financed the Australian Council for Education Research, well known for its skills in assessment, to work with national institutions in the development and administration of the India’s National Achievement Survey (NAS), which tested the learning of students in grade 3, grade 5, and grade 8.³⁸ In the SSA, the first NAS was conducted in 2012 for a grade 5 sample of students, which continued to highlight weak learning (UNESCO 2015c). The most recent National Achievement Survey report (National Council for Educational Research and Training 2020) shows some improvement but significant learning challenges remain, especially in the variation across states. In some states, at least a third of the students struggle with reaching minimum proficiency levels in both language and mathematics. Fewer states did well in the grade 8 assessments compared to grade 3, and students found it a challenge to handle complexity. Furthermore, the NAS data does not include students that have dropped out of school, which is around 50 percent (UNESCO 2015c). The Annual Status of Education Reports do test children in this category, perhaps providing a more comprehensive picture of the learning crisis in India.

Albeit insufficient, based on what took place in the DPEP, learning was important for the IDFIs and there was consistent and ongoing monitoring of student achievement. Jomtien’s strong focus on learning translated directly into the Indian context. In fact, when I joined this project in the late 1990s, only learning was being monitored and there was no real monitoring of enrollments. It was much later after Dakar that the attention shifted and IDFI staff in DPEP requested regular reporting on enrollment and completion from the Indian government.

Regional Assessments

UNESCO-UNICEF’s Monitoring Learning Achievement program was one of the earliest attempts to introduce international learning assessments in developing countries (Chinapah 2003).³⁹ This survey, launched in 1992, was administered twice for grades 4 and 5 (Monitoring Learning Achievement 1) and grade 8 (Monitoring Learning Achievement II). Forty-eight countries participated in Monitoring Learning Achievement 1 and II and only twenty-four countries in the second round. The program developed and strengthened capacities in these countries for rolling out learning assessments. According to Gove and colleagues (2015):

Monitoring Learning Achievement turned out to be much more costly and time-consuming than originally expected, however, and the results were discouraging rather than motivating. For example, none of the nine sub-Saharan African countries achieved the stated goal of 80 percent of minimum levels of learning, per the Education for All … In some cases, results were so poor as to suggest that students were simply guessing the multiple-choice responses. Few countries pursued the Monitoring Learning Achievement process after the Dakar meeting.

(p. 5)

Around the same time, two landmark learning assessment programs were rolled out in sub-Saharan Africa.

The Programme d’Analyse des Systèmes Educatifs de la Confemen (PASEC) was established in 1991 by the Conference of Ministers of Education of Francophone Countries. Based in Dakar, Senegal, assessments are administered to grade 2 (early primary) and grade 6 students (late primary) in language and mathematics.⁴⁰ The program also builds capacity for conducting learning assessments, with a view to facilitate better planning in education. France is the primary financier of this initiative.⁴¹ The first PASEC assessment took place in 2014 (PASEC 2015) on a representative sample of about 900 early primary school students and 3,000 end of primary students in each of the ten participating African countries. The tests were also administered in 2019 (Le Nestour 2021) and early results are not significantly encouraging.⁴² The PASEC tests are based on a competency scale for each subject. There are five levels in language and four in mathematics. The potential for mastery and non-mastery is determined through a “sufficiency level” - mastery is above level 2 in language and above level 1 in mathematics.

The 2014 assessments (PASEC 2015) show that 70 percent of beginning primary pupils were not reaching the sufficiency level in language, and over 50 percent not doing so in mathematics. By late primary or grade 6, it was about 60 percent for both subjects. Table 1.3 shows the performance of countries in grade 6. Except for Burundi, countries perform relatively better in language than in mathematics. Burundi has performed significantly better than other countries in both subjects with more than 50 percent of students in language and more than 90 percent in mathematics performing above proficiency levels. Chad and Niger have done particularly badly, with less than 20 percent of students reaching threshold levels.

Table 1.3 PASEC scores (late primary): francophone sub-Saharan countries

		Language
		Level <1	Level 1	Level 2	Level 3	Level 4
Benin		5	18	26	29	23
Burkina Faso		2	11	30	36	21
Burundi			5	39	49	7
Cameroon		6	20	25	25	24
Congo		5	23	32	24	17
Côte d’Ivoire		5	19	28	26	22
Niger		32	42	18	6	2
Senegal		4	14	21	26	35
Chad		20	37	27	13	3
Togo		6	24	32	23	16
Average		8	21	28	26	17
			Mathematics
			Level <1	Level 1	Level 2	Level 3
Benin	Math		25	36	29	11
Burkina Faso	Math		13	29	37	22
Burundi	Math		1	12	47	40
Cameroon	Math		30	35	24	12
Congo	Math		28	43	23	6
Cote d’Ivoire	Math		29	44	24	3
Niger	Math		68	24	6	1
Senegal	Math		15	27	30	29
Chad	Math		44	37	16	3
Togo	Math		21	32	28	20
Average	Math		27	32	26	15

Level <1 is the lowest for language and for math. Source: PASEC 2015.

Ministers from southern and eastern Africa established the Southern and Eastern Africa Consortium for Monitoring Educational Quality (SACMEQ) in 1995 with long-term support from the government of the Netherlands. In addition to financial support, practical and hands-on capacity building training was offered, rather than the traditional university training focused on theory and method. The Netherlands and UNESCO-IIEP continue to support the entity.⁴³

Assessments here are undertaken for grade 6 students. Only five countries participated in SACMEQ I, which took place in the 1990s. Ten more countries joined for SACMEQ II, which was carried out between 2000 and 2004.⁴⁴ Zimbabwe did not participate in SACMEQ II but returned in SACMEQ III, which took place between 2006 and 2011. A total of 61,000 students were assessed in this program. While SACMEQ IV took place between 2012 and 2014, no reports have been published for it except country reports for South Africa, Mauritius, Botswana, and Namibia. Angola participated in SACMEQ IV as an observer. Students were tested on an eight-level scale in language and mathematics. In language, it ranges from prereading to critical reading and in mathematics from pre-numeracy to abstract problem-solving.

Scores for SACMEQ III are available for fifteen countries (Hungi et al. 2010). Excluding the countries that did not receive significant aid as they were high (Seychelles) and upper-middle income (Botswana, Namibia, South Africa, and Mauritius) countries, average SACMEQ scores in language ranged from 434 in Zambia to 578 in Tanzania. In mathematics, the range is from 435 in Zambia to 557 in Kenya. There is considerable variation across countries and within countries. The averages are not as revealing as the disaggregated data (Table 1.4). In language 38 percent of students are very weak (Level 1–3) and in mathematics, the figure is 66 percent. The size of the group that has performed well is high for Malawi, Lesotho, and Zambia. In addition, scores were much lower for students that came from a low socioeconomic background.

Table 1.4 SACMEQ distribution of students (full sample and low socioeconomic status (SES)) across assessment levels: grade 6 language and mathematics

	Full sample (in percent)				Low SES (in percent)
	Language		Mathematics		Language		Mathematics
	Level		Level		Level		Level
	1–3	4–8	1–3	4–8	1–3	4–8	1–3	4–8
Kenya	20	80	38	62	24	76	43	57
Lesotho	53	48	81	19	64	36	89	12
Malawi	73	27	92	8	76	24	92	8
Mozambique	44	47	69	26	57	43	79	21
Swaziland	7	93	44	56	10	90	50	50
Tanzania	10	89	43	57	14	86	49	51
Uganda	46	54	75	25	56	44	82	18
Zambia	73	27	92	8	35	65	95	5
Zanzibar	21	79	73	27	50	50	86	15
Zimbabwe	36	64	57	43	47	53	70	30
Average	38	62	67	33	43	57	73	27

Source: SACMEQ III – Project results. Pupil achievement levels in reading and mathematics (2010).

From another region in the world, the Pacific Island Literacy and Numeracy Assessment (PILNA), which was established in 1980, is managed by the Education Quality and Assessment Program (EQAP). This international development organization made up of twenty-six representative governments and territories is supported by Australia, New Zealand, and UNESCO. The Pacific Island Literacy and Numeracy Assessment was first administered in 2012 to fourteen countries and then again in 2015 to thirteen countries. In 2018 PILNA was administered to students with four and six years of formal schooling.⁴⁵ The predominant emphasis in the PILNA reading and writing tests are to empower a person to be able to communicate and comprehend the world. Four areas are assessed in the PILNA mathematics tests: numbers, operations, measurement and geometry, data and chance. These four areas enable a person to use these math skills in everyday life.⁴⁶ Like SACMEQ, there are eight levels of proficiency for language and mathematics plus a Level 0. Minimum proficiency for four years and six years of schooling are Level 3 and Level 5, respectively. There is an intent in the proficiency levels to identify misconceptions and errors in student learning.

The Pacific Island Literacy and Numeracy Assessment report for 2018, available to the public, is not disaggregated by country performance (EQAP 2019).⁴⁷ Only five of the fifteen countries that participated are low income or low-middle income (Micronesia, Papua New Guinea, Kiribati, Tokelau, and Solomon Islands). The remaining countries are high middle-income countries and therefore do not receive significant aid. Based on aggregate numbers that include all the fourteen countries, literacy and numeracy skills of Pacific Island students had improved, but there is a significant proportion of students at the lower end of the proficiency scale that struggle with fundamental skills. Forty-seven percent of the Year 4 students and 37 percent of the Year 6 students did not achieve minimum levels in language. This proportion is lower for mathematics where 17 percent each in Year 4 and Year 6 did not reach minimum levels. The 2013 PILNA report for Papua New Guinea Department of Education (2013) is available and shows that after six years of formal schooling about 82 percent of students in language and 63 percent in mathematics are not performing at minimum levels of proficiency. Annex 1, Table 1.5 includes a summary of the areas included in PASEC, SACMEQ, and PILNA tests.

Early Grade Reading and Mathematics Assessment across the Globe

In the review of USAID support from 1990 to 2005, there is consistent attention given to the assessment of learning in projects across the globe (Chapman and Quijada 2009). The World Bank and USAID commissioned the Research Triangle International (RTI) to develop an assessment tool to help capture the “extent to which early grade primary school children are learning to read with an acceptable degree of comprehension and at an acceptable rate of fluency” (Gove et al. 2015, p. 8). This endeavor resulted in the production of the Early Grade Reading Assessment (EGRA) and Early Grade Mathematics Assessment (EGMA) focused on evaluating foundational skills (Gove and Wattenberg 2011; Kim et al. 2016). Early grade assessments assume that by tackling weakness in reading and mathematics when a child enters school in the initial grades of primary education, learning will be sustained. These tools, developed mainly from the US experience in testing children and research into early grade student assessments, was also informed by the work of Helen Abadzi (2006) on cognition and language.⁴⁸ The EGRA was introduced as a pilot in 2007 in Gambia, Kenya, Nicaragua, Senegal, and South Africa. This has expanded to about sixty countries. Development of EGMA started in 2008 (Platas et al. 2016), focusing on evaluating a sense of numbers, progress in mathematics skills, and problem-solving.⁴⁹ According to the Australian Council of Education Research (2020), about twenty-two countries have either administered or planned to administer EGMA. Annex 1, Table 1.6 includes the content areas in EGRA and EGMA.

Supported by donors, EGRA captured early grade reading levels in several countries. The results from Guyana, Iraq, and Jordan are not included as these are upper middle-income countries. The results for some of the sub-Saharan African countries are very weak.⁵⁰ For example, in 2013, EGRA tests were administered to children across twelve mother tongue languages and English in Ghana. The range for the proportion of children with a zero reading fluency score in grade 2 was 58 percent to 92 percent across the languages. Taking Malawi specifically, grade 2 Chichewa language fluency tested in 2010 was 95 percent and in grade 4, 55 percent with a zero reading score, only decreasing to 90 percent and 39 percent in 2012.

More recently, RTI claims (Gove et al. 2017) to have achieved some improvement in Kenya and Uganda at the end of several years of intensive programming. In Kenya, there was a reduction in grade 2 of non-readers, and an increase in pupils reading fluently. Corresponding to this finding was a similar decline for mother tongue (Kiswahili) instruction where non-readers declined and fluent readers increased. More details on the Kenya performance are provided in Chapter 4. In Uganda, 95 percent of children were reading during classroom observation in the schools receiving support as opposed to only 11 percent in the control schools. About 40 percent of the trained teachers were helping students pronounce letters while none of the teachers were doing this in the schools not receiving training support. These results are small considering the years of programming in these two countries. Furthermore, the improvements were not expressed according to reading fluency or words per minute. Results in the other countries (Ethiopia, Liberia, Malawi, and Tanzania) included in this report were weak to mildly encouraging. Moreover, the results were not expressed according to reading fluency but words per minute. For example, in Liberia correct words per minute increased from 4.8 to 14.2 on oral reading fluency relative to the comparator group which showed no improvement.

Both EGRA and EGMA are stated to be flexible frameworks that can be adapted to language and context (Dubeck and Gove 2015). There has been much criticism of EGRA, significantly of the focus on decoding rather than comprehension, the singular emphasis on reading rather than other skills, and the lack of consideration given to the specific orthographies of different languages. This is especially relevant to establishing the benchmark of eighty words per minute, which may not be appropriate for all languages. The creators would argue that phonics is a stepping-stone and that EGRA is now used with EGMA. Kim and colleagues (2016) conducted a meta review of early grade literacy performance and any evidence for more than foundational literacy was weak. The issue that also needs to be flagged concerns the tacit assumption of easy and germane transferability of assessment tools between donor and recipient contexts. After all, research instruments that circulate internationally are themselves embedded in specific cultural worldviews and operational systems.

Global Status of Learning

Drawing from much of the measurement of learning discussed in previous sections (within-a-project, within geographical regions, and globally at early grades of schooling), recent seminal reports on education summarize the scale of the learning crisis. Global Education Monitoring Reports produced annually and the World Development Report 2018 capture overall data on learning in developing countries. Both reports summarize the PASEC, SACMEQ, PILNA, and EGRA data in addition to other regional and large-scale assessments conducted by NGOs. The consensus is unambiguous: Though there are significant improvements in student enrollment, overall learning levels are alarmingly low. The World Development Report states, “Children learn very little in many education systems around the world: even after several years in school, millions of children lack basic literacy and numeracy skills” (World Bank 2018a, p. 5). All the data sets previously described show significant weaknesses in learning across the developing countries.

Without a thorough critical evaluation of IDFIs’ attempts at establishing assessment systems, which were envisaged to make a difference to learning, attention has again shifted to instituting international and comparable standards for measuring learning. Thus, for example, UNESCO’s Institute of Statistics established the Global Alliance to Monitor Learning to help countries develop learning assessments that enable comparability across countries so progress toward Sustainable Development Goal (SDG) 4 can be monitored. Alternative linking methodologies (statistical and non-statistical) are being pursued (UNESCO 2018) by the Global Alliance to Monitor Learning.⁵¹ Let me start with the most widespread. Using statistical techniques, the Global Education Monitoring Report (UNESCO 2019) summarizes learning levels by linking proficiency scales. Overall:

387 million or 56% of children of primary school age did not reach the minimum proficiency level in reading. This was the case with 81% of children in Central and Southern Asia and 87% of children in sub-Saharan Africa but only 7% of children in Europe and Northern America. In addition, 230 million or 61% of adolescents of lower secondary school age did not reach the minimum proficiency level in reading. Similar estimates apply for minimum learning proficiency in arithmetic.

(p. 125)

The Institute of Statistics database reports on the proportion of children reaching minimum proficiency in grade 2/3 and at the end of primary. There are no ranges provided and minimum proficiency could be considered a minimalist approach to understanding learning. Despite the efforts to find common ground, it will not be easy to accomplish a joint approach to assessment. Though donors supported the development and administration of student assessment, there was a mistaken fear of setting the bar too high for the countries that were receiving funding. Furthermore, setting common standards is not easy as each developing country and region has its own system of testing learning, often established by their colonial predecessors. The progression in learning from grade to grade may also differ from country to country making it difficult to have any shared expectations or rubric for measurement.

Ways in which the information collected by the Institute of Statistics database can be used is unclear. Of the 246 countries in the database from across the globe, only a small proportion of countries report learning levels for grade 2/3, and an even smaller group report on proficiency levels at the end of primary. The three figures (Figure 1.1, Figure 1.2, and Figure 1.3) capture information collected by the Institute of Statistics for developing countries in South Asia and sub-Saharan Africa. There are few patterns evident in the data. Four countries in South Asia are included in Figure 1.1. Nepal reports achievement only at the end of primary. Reading in Bangladesh and India is slightly higher than mathematics whereas in Pakistan, reading is substantially higher than mathematics performance.

Figure 1.1 South Asia: percent of students achieving minimum proficiency in reading and mathematics

Source: Institute of Statistics Data provided between 2014 and 2018.

Figure 1.2 Sub-Saharan Africa: percent achieving minimum proficiency in francophone countries

Source: Institute of Statistics Data provided between 2014–2018.

Figure 1.3 Sub-Saharan Africa: percent achieving minimum proficiency in anglophone countries

Source: UNESCO Institute of Statistics Data provided between 2014–2018.

At the end of primary, reading performance is better than mathematics across countries. There are no connections between early grade and end primary learning levels. Low mathematics performance in Pakistan suddenly increases substantially in end primary. Whereas in the other countries there is a decline in end primary levels when compared to early grades. In Nepal, achievements are higher in end primary than the other countries.

Figure 1.2 and Figure 1.3 summarize learning in a group of anglophone and francophone countries in sub-Saharan Africa. Reading performance is lower than mathematics in all the francophone countries, which is the opposite in English-speaking countries. Except for Senegal, all the other countries report on performance in end primary. There are no observable trends between minimum proficiency in grade 2/3 and end primary. There is decline in end primary in several French-speaking countries, especially Burundi. In contrast, in the only two countries – Uganda and Nigeria – that report on end primary there is an increase in achievement in both reading and mathematics.

According to the World Development Report (World Bank 2018a), as expected, low levels of learning vary according to household income. For example, in Togo, more than 75 percent of children from poor households did not reach proficiency levels compared to about 25 percent from rich households. This difference can also be visualized at a global level. In high-income countries in Europe and Central Asia, the figure is over 80 percent and in low-income countries in sub-Saharan Africa, less than 20 percent. The contrast is also evident in UNICEF’s Multiple Indicator Cluster survey. Results for learning levels in two countries from this cluster survey are reported in the Global Education Monitoring Report (UNESCO 2018), North Korea, and Sierra Leone. In North Korea, 95 percent of students in the 7–14 age group were able to read and 82 percent could do basic mathematics. In Sierra Leone, the figure was 39 percent and 34 percent, respectively. According to both the Global Education Monitoring Report and World Development Report the low levels of learning are further exacerbated as students transition through the system. To conclude this section, the global status of learning illustrates the reality and far-reaching nature of this crisis.

Conclusion

This chapter juxtaposes the amount of money spent by IDFIs on improving education and the actual outcomes achieved. As Elizabeth King (2011), a leader holding for a time the highest position for education in the World Bank, bemoans: “How did we come to dissociate going to schools from learning-in-school. As they say in the Philippines: a person who does not look back to where he/she started will not reach his/her destination (“ang hindi lumingon sa pinanggalingan, hindi mararating ang paroroonan”). Let’s not forget what brought us together in 1990 as we look to the future.” The verdict is categorical: donor funding was not able to bring about the desired progress in student learning. The most prevalent response to this crisis is to assume that this was because learning was not part of the EFA/Millennium Development Goals and, therefore, not measured.

In contrast to such a supposition that learning assessments were not taking place, this review has shown the IDFIs’ work in the education sector started with a deep and profound commitment to improving learning in the developing world. Though this commitment was made at Jomtien, a decade later the proclamation on learning was substituted with the importance of providing quality education at the global level. Despite this modification in the global compact, donor agencies continued to finance learning assessments as they wanted to know the status of learning. International Development Finance Institutions were involved with the measurement of learning within-a-project (DPEP), at regional levels (PASEC, SACMEQ, and PILNA), and in early grade assessment. This continued emphasis shows IDFIs’ serious attempt to understand the impact of their programming on learning.

The dirge of the learning crisis was playing along in the background of IDFI efforts all this time. Through the support for learning assessments, donors, on the one hand, were intent on understanding how much students were learning and on the other hand, expected to gauge the effectiveness of their interventions in the education sector in developing countries based on these results. So, the massive deficit in learning among developing countries, despite large increases in aid for education, should not be simplistically attributed to an absence in large-scale student testing.

International assessments seem to suggest that while this data is helpful to compare performance across developing countries, it is the processes that are internal to a country that drive sector reform and improved learning (Clarke 2017). No doubt the Global Alliance for Monitoring Learning is rightfully pursuing alternative linking methodologies that are crucial for addressing different populations in a diverse world. However, these instruments for measuring learning have to be embedded organically within the institutional ecology of the educational system in a country. External assessments rarely influence the complex and extensive change management needed for student learning to happen.

The latest initiative reflecting this focus on learning assessments and further measurement is the project entitled Assessment for Learning (A4L), piloting yet another diagnostic tool to assess learning. Three countries are participating in the pilot – Ethiopia, Vietnam, and Mauritania (Read and Andersen 2021).⁵² Evaluation of this initiative found limited commitment in developing countries for another learning assessment initiative. Instead of assuming the learning crisis is because it was not measured, I suggest that the learning crisis stems from the use of assessment results in a messy and transitory manner in the complex and opaque institutional landscape of educational systems in developing countries. I will show how critical it will be to unveil this landscape, first identifying elements that represents this opacity and second, specifying the dimensions that the IDFIs can contribute to make learning happen. This is the challenge of the rest of the book.

This chapter has shown how ineffectual learning is a serious problem across developing countries despite significant amounts of financing allocated to the education sector. It sets the stage for the following chapters, which examine why IDFI interventions failed to mitigate this situation of weak learning and the research and the evidence base, which failed to redirect the work in developing countries. Stitching together donor programming with the learning assessment experience at the end will highlight the need for considering a different approach to education reform that works organically within the history, milieus, and mindsets of micro (school and classroom) and meso (governance departments) institutions across the world.