1 Introduction

Although the move from Mode 1 to Mode 2 knowledge production was meant to herald a new era of multidisciplinarity in understanding, studying and dealing with social problems, the experience of the ways expert knowledge is produced and governed is far from the ideal evolutions that such schematic representations tend to offer. Indeed, despite the intractability of the problems at hand, global education governance has not been informed by the vision and multiplicity of perspectives that arise when different forms of knowledge come together. Instead, the datafication of education problems severely restricted the scope of knowledge production. Even worse, the quantification of education problems quickly led to their economisation, as economists of education rose to positions of authority in major IOs and hence their concerns around issues of input/output mechanisms, efficiency of education systems and cost/benefit analyses began to dominate the debate of how to bring improvement and reduce inequities. Thus, instead of Mode 2’s proclaimed multidisciplinary epistemological perspective, what is to be observed in the field of education—and arguably more broadly—is that the discipline of economics has emerged as the great unifier that brings together actors, narratives and policy solutions.

Therefore, this chapter will argue that the complexity of policy problems led not necessarily to the rise of interdisciplinarity, but rather in mono-disciplinarity; as a result, certain kinds of knowledge were privileged, while others were silenced in the process. Since the 1990s, there has been a slow but concerted effort to use quantification as an instrument of the economisation of education discourse and practice as the single, universal language of global education policy: as we will see, the economisation of global education policy, with its emphasis on comparability, efficiency and cost-effectiveness, enabled the communication between different disparate groups and became the language and policy of choice for education policymakers around the world.

The chapter uses the notion of epistemic infrastructures (Tichenor et al., 2022) in order to empirically chart this process as having happened at three levels: (1) economisation that occurred through the central positioning of economists within the education inequity debate; (2) through the expansion of an economic/instrumental way of thinking about education; (3) finally, economisation occurred through the increasingly central role of international organisations whose primary remit centres around economic growth concerns (such as the OECD and the World Bank). What we observe, as a result, is the construction of economic knowledge in education, at the expense of other perspectives, and thus, as we will see, the simultaneous production of knowledge and non-knowledge as part of the process. This dichotomy, namely between the production of certain types of data at the expense of others will be discussed at the final sections of the chapter, in an effort to understand the effects of mono-disciplinarity in global education governance.

2 The Contours of SDG4: Theorising with Epistemic Infrastructures

As already discussed in the previous chapter, the World Education Forum (WEF) was celebrated in Incheon, the Republic of Korea, in May 2015, with the participation of over 1500 people, including 120 Ministers of Education and representatives from a wide range of international governmental and non-governmental organisations. The event at Incheon represented a milestone in the history of UNESCO summitry, a long trajectory of large education conferences that demanded fair, free and quality education for all. Similar to others prior to it, the main product of WEF 2015 was the so-called Incheon Declaration, along with the Framework for Action adopted by UNESCO Member States a few months later, in November 2015. In conjunction, both documents established an ambitious and highly aspirational education agenda for the period 2015–2030 and condensed in the overarching goal to ‘ensure inclusive and equitable quality education and promote lifelong learning opportunities for all’ and a number of associated targets; this is the Sustainable Development Goal 4 (SDG4) (UNESCO, 2016).

Indeed, the SDG4 is one of the 17 Sustainable Development Goals (SDGs) that are ‘integrated and indivisible and balance the three dimensions of sustainable development: the economic, social and environmental’ (UN, 2015, p. 5). According to the UN, ‘they result from what is arguably the most inclusive process of consultation in the history of the United Nations, reflective of substantive input from all sectors of society, all actors of the international community and all parts of the world’ (UNESCO, 2017, p. 4). Indeed, as will be shown here, it is precisely this inclusive and participatory governance model that became key in the formation of many aspects of the SDG4 agenda and its implementation. As a programmatic document oriented at nurturing and securing a form of collective commitment towards a shared set of aspirations, the new agenda builds on a well-established tradition of consultation and collaboration that has come to be recognised as a characteristic of the UN system. What is interesting—and will be discussed later in this chapter—is that, despite the proclaimed collective and broad set of aspirations, certain kinds of data production for specific indicators remain dominant, at the expense of a focus on others.

Thus, this chapter traces the development of the epistemic infrastructure of the SDG4 in order to show the ways that the incremental build-up of the discourse, technical expertise and, given this apparent universality of the SDG agenda, the fragile but necessary actor alliances facilitated a paradigmatic policy shift in the field of education: this is the move from the measurement of schooling (Barro & Lee, 1996) to the measurement of learning. The shift entailed the prioritisation of an emphasis on learning outcomes, skills and competencies, measured through what children ‘can do’ with the knowledge they acquire at school. In other words, instead of the traditional education statistics that measured inputs such as education expenditure, teacher salaries or length of the school year, the pendulum shifted to a greater interest in decontextualised, applied knowledge, measured in real-life contexts. Although the work around the construction of the SDG4 (both prior to and after 2015) is not the only process that facilitated this shift (indeed its origins lie in New Public Management and the economisation of education discourse in the 1980s and early 1990s—see Gunter et al., 2016; Ozga et al., 2009), the global nature of the SDG4 process and the active involvement of most key education actors in its production led to a concerted effort to devise global learning metrics (Crouch & Montoya, 2019). Thus, alongside other key venues (one of them being OECD’s PISA, as will be discussed further on) the SDG4 became a prime site of the production of this radical reconceptualization of education measurement and policy with implications across the world.

Indeed, the complexity and length of the SDG4 process render the painting of a comprehensive picture of all related events and actors as a futile endeavour. A focused analysis of the production of the SDG4, viewed through the lens of the notion of ‘epistemic infrastructures’, allows for a close-up on the interdependency of materialities, technologies, individual actors and organisations that participated in its making. Indeed, the paper adopts the definition of an epistemic infrastructure as the ‘complex interplay of material, techno-political and organisational structures within which (statistical) knowledge is produced, disseminated and translated into global public policy’ (Tichenor et al., 2022).

Earlier literature on infrastructure studies (Star, 1999; Winner, 1986) highlighted their invisibility; infrastructures were seen as comprised by social, material and technological elements that are interdependent and flow seamlessly into one another, facilitating the unobstructed move of numbers, people, goods and ideas in the production of new ways of measuring, viewing and living in this world. However, in contrast to the neat accounts of global education reforms flowing top-down, the SDG4 has never been the perfect invisible infrastructure, moving ideas and practices from some imaginary ‘centre of calculation’ (Latour, 1987) to the periphery. Instead, long before its inception, it has been a site of conflict and contestation, a space where relationships break-down and—more often than not—metrics fail. Since the idea of metric ‘failure’ might have normative connotations, it needs to be clarified that I examine ‘failing metrics’ as those that lose their policy momentum, by increasingly being perceived by the policy, expert and professional communities as irrelevant or even misleading; ultimately, their continued measurement is seen as having detrimental, rather than positive effects on the policy arenas they are meant to contribute. Such failings can be either real or manufactured, yet the outcome is the same: the failure of achieving global goals (irrespective of whether they are misplaced or, in fact, unattainable in the first place) sparks quests for improved metrics that will excite, persuade and ‘stick’ anew (Bandola-Gill, 2020). Yet, despite such perceived failures, it is the infrastructure’s break-down that fuels its growth and expansion. As this chapter will show, the paradigmatic shift from the policy focus on schooling to learning happened through the concerted efforts to discredit certain kinds of knowledge production, in favour of others that were seen as linking education a lot closer to economic prosperity; that is, economic knowledge.

The policy prioritisation of learning and its associated outcomes is not a novel topic in education research. Although there has been scholarship on the discursive expansion of the language of learning outcomes and skills (Klees et al., 2019), as well as some critical literature on the validity and robustness of the new learning metrics (Benavot & Smith, 2020), and on their effects on global education policy reforms (Mundy et al., 2016), the chapter discusses the entanglement of materialities, discourses, ideas and practices into the building of a new epistemic infrastructure that has prioritised the dominance of the discipline of economics in education governance globally.

Indeed, these entanglements have allowed a plethora of contestations to unfold: one of the most prominent ones is the emphasis on some indicators versus others, as well as the issue of the democratic decision-making process. After a brief overview of the intellectual terrain on infrastructures and some methodological considerations (Sect. 2), the following section (Sect. 3) will discuss the history of the shift of education discourse from the measurement of inputs to the dominance of measuring skills and outcomes. In particular, I will discuss the ways in which some powerful actors prepared the ground for a move away from the measurement of schooling (through measuring access and completion) to learning (through the measurement of literacy skills). The primary means of facilitating this change was through presenting the MDG education targets as misleading and thus as ‘failing’ metrics; the mobilisation of new evidence and a ‘killer’ number (Stevens, 2011) was used in order to create the space for contestation and change. The building of a discourse of the economic versus the wider social benefits of education, alongside the production of new metrics to replace the old ones, became a vital mix and thus the building block in the construction of the infrastructure of the SDG4. In addition, I will highlight the importance of the temporal dimension in the building of epistemic infrastructures, in terms of first, their temporal discursive framing of ‘past failures/ current crisis/ future projections’, as well as in relation to the slow, step-by-step build-up of the measurement infrastructure in order to gather steam, create the evidence, build a support base and thus have greater policy influence.

Section 4 will then move on to the analysis of the workings of the Technical Advisory (later Cooperation) Group, in charge of the development of some of the SDG4 indicators. The section will show how the TAG/TCG began its work in 2014 primarily as a group of expert IO statisticians and later expanded into a much larger—and with a different function—grouping that included country and civil society representatives, all in the name of democratising the measurement agenda and process. Thus, beginning with the small, highly technical and elitist group in 2014, we observe how the slow building of a much larger infrastructure of actors and materialities came together to support, prop up and legitimise the work of the production of numbers. Thus, this section will focus more on the spatial features of the infrastructure, as it expanded across contexts and fields of practice, to include a much wider actor membership and achieve greater coordination across the local, national and global levels.

Section 5 will discuss the infrastructural qualities of meetings of the SDG4, by showing how, instead of a seamless flow of coordination and cooperation, it was failing metrics and the continued break-down of the proceedings that both acted as generative forces that ensured its continuity and growth. These meetings that bring together a range of actors, from the local to the international levels are, as I will show below, those slow and convoluted processes that ‘wicked’ problems (Guy Peters, 2017) are discussed and a range of possible monitoring solutions agreed upon. As this chapter shows, the process of the rise of mono-disciplinarity in education requires not only the co-construction of specific kinds of knowledge by the relevant IOs, but also another significant function of theirs: that of the making of ignorance, or as we prefer to call it, non-knowledge. The social production of non-knowledge is a necessary pre-condition for reaching agreement about what kind of knowledge will be pursued in order to achieve a minimum consensus, so as to ensure ‘buy-in’ but also maintain actors’ own interests, values and positionings intact (Grek, 2020). Thus, the construction of non-knowledge is an essential part of the measurement process: rather than the opposite of knowledge, however, or its reading as a binary, here it is viewed as a symbiotic relationship, necessary for balancing out and achieving some kind of constant equilibrium—and hence movement—of the metrological field.

Finally, in terms of methods, the chapter is built on three main sources of data: first, the discourse analysis of documents relating to the production of the SDG4, as well as materials that predated it. CDA is a particularly apt method for the analysis of the making of infrastructures because it sees text as a key aspect of how certain understandings of the world are shaped and perpetuated by practices of infrastructuring (Meyer, 2001; Wodak & Fairclough, 1995). Hence, the analysis of these documents is useful for, on the one hand, showing what is technically possible, while on the other, explaining what the principles and perspectives of those participating in the production of the infrastructure are.

Second, the chapter’s empirical analysis is based on twenty in-depth interviews with key actors of international organisations and the civil society. Finally, the social network analysis component focused on an exploration of the role of the SDG4 meetings and the alliances and connections they generated. The combination of these methods allowed for a study of the discursive meaning produced by relevant IO and research reports. Interviews gave me an insight into the experience, views, positionings and choices of the key actors that participated in the infrastructure. Lastly, social network analysis, focusing on the two main indicator technical groups, explored their meetings as the key stabilising moment when negotiations achieved the desired pax romana before disagreement and conflict unravelled again. Thus, the research design offered the capacity to study different elements of the infrastructure, their entanglements, effects and the ways certain kinds of knowledge production dominated over others.

3 The Rise of Infrastructures: Vogue, Vague or ‘Really Useful Knowledge’?

The term ‘really useful knowledge’ is derived from radical education thought of the nineteenth century; it was supporting a critical understanding of self and society; it was knowledge meant ‘to set you free’.

‘Infrastructures are conceptually unruly’ (2013, p. 329), Brian Larkin wrote, and there could not have been a more accurate description for the varied application of the term. In fact, it is precisely the conceptual plasticity and the focus on materiality that has made infrastructures such a popular concept in social theory. Nonetheless, they have not always been as vogue as they are today: in fact, it was only in the mid-1990s when Geoffrey Bowker (1995) first pointed towards the materiality of infrastructures as a way of understanding their function and effects. Bowker saw infrastructures as largely invisible backdrops to social action and thus analytically not penetrable; he therefore proposed the notion of ‘infrastructural inversion’, as a way of breaking the invisibility and flow of the infrastructure. Infrastructural inversion (Bowker, 1995) was about making the invisible visible, through a focus on material relations and the ways they reconfigure how we know and live in the world.

Similarly, in 1996, Susan Leigh Star and Karen Ruhleder saw invisibility as a key quality of infrastructural systems. Nonetheless, they also identified the seamless flow of the infrastructure as a fragile achievement that was prone to break-down and failure (Star & Ruhleder, 1996). The invisibility/visibility conundrum was further discussed by Larkin (2013), who suggested that infrastructures can be invisible but can also become a spectacle, and thus depend on their visibility for their success. However, following Larkin, even when an infrastructure is open, visible and ready to be experienced, what is there to see? According to Harvey et al.,

Provisionally, and minimally, we might say that we are dealing with technologically mediated, dynamic forms that continuously produce and transform sociotechnical relations. That is, infrastructures are extended material assemblages that generate effects and structure social relations, either through engineered (i.e. planned and purposefully crafted) or non-engineered (i.e. unplanned and emergent) activities. (2017, p. 5)

This analysis contributes to the literature on infrastructures, by showing the particularities of the mix of materials, practices and meanings in the making of measurement agendas, such as the SDGs. Given the centrality of knowledge and data production in global governance, the concept of ‘epistemic infrastructures’ (Tichenor et al., 2022) is particularly apt, since it advances the analytical purchase of the—STS-primarily informed—concept to bring it much closer to policy theory and practice. In particular, as the chapter shows, both the flow and the failures, the unlikely alliances and the clashes, did not only facilitate the production of a system of measurement and a particular way of naming and understanding educational realities in the twenty-first century. They also brought about a much more fundamental and—as it appears—permanent policy shift: this was the dominance of the economic paradigm in education measurement, practice and values. The move away from the measurement and thus prioritisation of educational inputs (numbers of teachers, school facilities, financial support and others) towards the measurement of outputs (learning outcomes, test results, skills and competencies) did not merely take place at the discursive level, or the measurement one. Neither has it only been circulated and promoted among organisations and actors, experts and professionals, that work in the field of education. Rather, it produced a monodisciplinary dominance in the education policy field that has had dramatic consequences in the way education policies at the country level are made (Verger et al., 2019). The intention of the chapter is to utilise the three orders of the epistemic infrastructure (the materialities, the interdependencies and the paradigmatic shifts) in order to place emphasis on the role of the discipline of economics for producing knowledge for policy.

4 From Schooling to Learning: The Incremental Building of an Infrastructural Base, 2006–2013

The discursive and logical shift moved the measurement agenda from a focus on schooling to learning began as early as 2000s. On the one hand, the OECD PISA, although measuring the skills and competencies of 15-year-olds in the global North (at least in the first rounds of the learning assessment and before its expansion in 2012 and 2015), received unprecedented media and policy attention worldwide; this was due to PISA’s ranking of countries according to their education performance. PISA and subsequently the OECD prided itself in decontextualizing education by focusing global, comparative testing not on the knowledge that students acquire at school (thus moving away from traditional ways of approaching schooling and curricula) but on what students can do with this knowledge. The OECD made direct links between countries’ future competitiveness to how well schools prepare students to enter the labour market. PISA results were announced at the end of each testing cycle (every 3 years) and caused ‘shock and awe’ to many European countries in particular (and increasingly globally) including the ‘education catastrophe’ that hit Germany, or the ‘education miracle’ that turned Finland into an education tourist hotspot for education ministers and experts from around the world (Grek, 2009, 2013). In many senses, OECD PISA became the flagship international comparative test that shifted the focus of education policymakers to outputs, rather than inputs, and to learning rather than schooling. The significance of PISA data is undisputable, given that European education governance became dependent on it, in order to—for the first time ever—create indicators and benchmarks to measure education performance in EU member states—what was called the Lisbon agenda (Lawn & Grek, 2012).

Nonetheless, perhaps more so than the OECD, it was the work of the World Bank that shifted the education debate, given the Bank’s influence in the Global South (Prada-Uribe, 2012). The World Bank opposed the MDG emphasis on access to education, suggesting that lack of education had never been only a matter of whether children are in school or not; instead, it was suggested that the focus should be on what children achieve at school when they are there. The work was undertaken by senior economists at the World Bank and the links to improved national economic growth were explicit from the start: in two seminal research reports (Glewwe, 2002; Hanushek & Kimko, 2000), it was suggested that individual mobility and better economic outcomes were achieved in countries that focused on knowledge and skills acquired in primary schools, rather than those systems that merely aimed to increase access. In 2006, another World Bank report became a milestone moment for education measurement, as it shifted the debate not only in education policy circles but also in development ones. The report, provocatively entitled ‘From Schooling to Learning’ (IEG-WB, 2006), was written by the Independent Evaluation Group and created a polemical discourse against the MDGs’ focus on access and completion: it suggested that the current emphasis was misplaced and that much more attention should be given to the improvement of skills and competencies, as it is the latter that lead to economic prosperity and better outcomes. As a consequence, the Center for Global Development appointed three World Bank economists to further explore the issue; their report, A Millennium Learning Goal: Measuring Real Progress in Education (Filmer et al., 2006), unequivocally suggested that there was no evidence that showed that completion of primary school guaranteed the achievement of minimal levels of literacy and numeracy and that a re-think was long overdue. The example of the failed MDGs is an excellent illustration of the core argument of this paper in regard to the power of metrics not only to influence policy direction, but also in fact to be the space where policy work is done: it was the production of new metrics by education economists that pushed for the idea that previous metrics had failed. And it is precisely the perceived failure of the MDGs that created the new space for contestation around which new metrics (and thus policy priorities) should replace them. The materiality of data, reports and meetings intersected with the work of specific expert organisations and actors and led to a substantial policy shift. As I discussed, these expert actors were international organisations with a very explicit mission and objective: that is, to increase economic growth and development.

Indeed, the arguments developed by education economists at the OECD and the World Bank had far more purchase in the development community groups, rather than in education (at least at the start). Both DFID (the UK’s former Department for International Development) and USAID (the United States Agency for International Development) produced new strategies in the period of 2010–2015 that identified the measurement of learning outcomes as an institutional priority and consequently channelled their education investments accordingly. Although there were a number of voices from academia that suggested that a singular focus on learning outcomes would take the attention away from other important pedagogical aspects (Barrett, 2011; Tikly, 2015), their commentary remained ‘academic’; they had little policy influence and impact. Yet, there were still quite a few voices in education, especially those from UNESCO and the civil society, that were worried about the new trend and the misplacement, as they saw it, of education and schooling measures with those of outputs. Once again, the two functions of education, the humanistic and the economic one, were pitted against one another. The result was the slow emergence of ‘a divide between those emphasizing quality and those primarily concerned with learning outcomes…Even if the differences between the two approaches were originally a matter of nuance or emphasis, they ended up forming two distinct communities of understanding, informed by different sets of ideas’ (Fontdevila, 2021, p. 177).

Indeed, as the decade progressed and the end of the MDG timeframe was drawing to a close, we can observe a much more concerted effort to change not only the discourse (that had already been achieved) but also to start building an infrastructure for the establishment of a new measurement agenda, one in which learning, skills and competencies would be centre stage and would replace the previous targets. The key protagonist in this new era was not the World Bank (though it was always supporting at the background) but a new initiative, the Global Compact for Learning (GCL), which was launched in 2011 by the Brookings Institute Center for Universal Education. GCL quickly became an advocacy tool; through its reports, it created a sense of urgency, putting forward the idea that there was a learning crisis that was ‘hitting the poorest, most marginalized and the youth particularly hard’ (CUE, 2011). Just a year later, UNESCO in conjunction with the Global Education Monitoring Report (GEMR, 2012) published an estimate of the number of children not achieving basic literacy skills as reaching 250 million. The shocking figure became further ammunition not only for those economists that were pushing for the learning turn, but also for those who were suggesting the benefits of international learning assessments; without them, there would have been no evidence of this crisis. Thus, the crisis discourse had created a sense of urgency and would quickly turn into the need for action. Not only was it obvious that the MDG targets, set in 2000, were not going to be met, but also it had become evident—to some, at least—that these targets were ill-defined and misplaced and thus were failing economies and millions of children around the world.

Crucially, GCL prepared the ground for the launch of another key initiative: the Learning Metrics Task Force (LMTF) was established in 2013 with the aim to ‘catalyze a shift in the global conversation on education from a focus on access to access plus learning’ (UIS/CUE, 2013, emphasis mine). This was a subtle, yet fundamental change and an open invitation to the two measurement camps to come together in search of the post-2015 agenda. Brookings invited the UNESCO Institute of Statistics (UIS) to head the task force, an important gesture towards an actor that appeared more trustworthy (to teacher organisations and civil society, at least) than the World Bank. More crucially, this was not an elite exercise; rather, LMTF was a very diverse organization that included a wide range of actors not only from the international organisations’ expert world, but also regional organisations, donors, governments, statistical agencies and civil society. The pluralistic nature of the membership coupled with its UIS leadership and the timing (the preparations for the post-2015 agenda had already begun) made the LMTF the perfect opportunity to build the measurement infrastructure not only up but wide; it also offered a way to break away from the dominance of economic thinking and allow a broader conversation. This was the moment when the build-up of the new measurement agenda was to stretch across contexts and organisations to expand spatially, too. Essentially, the establishment of the LMTF became the foundation for building—what would later be called—the SDG4.

5 From IOs’ Advisory to Cooperative Role: Brokerage and Collaboration

LMTF brought together a vast array of actors and organisations in its efforts to offer legitimacy to the task of shifting the debate and subsequently the post-2015 goals for education. As the previous section showed, it approached the contentious topic of the prioritisation of metrics and goals diplomatically, suggesting that they were interested in exploring ‘access plus learning’ metrics. Thus, economists extended an olive branch to academics, the civil society and professional organisations that perceived the learning focus as reductionist and as reflecting merely the economistic lens of the Bank’s ideological positioning. Additionally, UIS’ leadership (and not the World Bank’s, for example) gave the project not only credibility but also a ticket to move away from merely debating over priorities (the 250 million failing children was an alarm that kept on ringing) towards trying to find practical measurement solutions for their aims—in light of PISA and other regional, cross-national tests, the attention turned to the production of learning assessments, which, as it happened, have become the key data production machines for the SDG4 agenda (Fontdevila, 2021).

Despite the seemingly celebratory and ambitious language, the work of the LMTF was challenging, given that consensus had to be found not only on the aims themselves but also in relation to how these aims would translate into measurable indicators, as well as which spaces of deliberation would constitute the legitimate decision-making venues for making these choices. This is due to the fact that the efforts to devise the SDG4 indicator framework did not start by the UN Statistical Commission, but dated back to the establishment of an inter-agency, ad hoc platform known as the Technical Advisory Group (TAG). Originally, the TAG was established by UNESCO in 2014 and recruited experts from UNESCO itself, but also from the GMR, the OECD, UNICEF and the World Bank. In many senses, while after 2014 LMTF 2.0—as the version came to be called—continued the debate at country level (Anderson, 2014), TAG adopted the work of the original LMTF with its focus on ‘seven learning domains, and recommendations for global measurement areas’ (Anderson, 2014). Chaired by the UNESCO Institute of Statistics, TAG was a much smaller grouping, with its membership limited to IO experts, and with the task to devise the ‘post-2015’ indicator agenda.

From March 2014 to May 2015, the TAG embarked on the process of mapping existing and potential education indicators, taking into consideration both their alignment with the (anticipated) targets and questions of data availability. Importantly, the work of the TAG benefitted from the input of a global consultation process, running from November 2014 to January 2015. In May 2015, the group’s proposal was incorporated to the Framework for Action at the WEF in Incheon. That was a pivotal moment for the group’s continuity, since the WEF recommended that the TAG is expanded, in order to include civil society and UNESCO member states organisations’ representatives. It was partly the distrust towards the IOs leading the measurement agenda by the EFA actors, and partly the universalistic and participatory agenda of the SDGs that had brought this significant change, which also led to the renaming of TAG as the ‘Extended TAG’. Subsequently, the Extended TAG conducted on-going open consultations led by regional leaders. Very quickly, what was a small, rather swift and efficient technical team of IO education economists (with their own of course internal conflicts and competitions) had suddenly opened up to a much larger governing structure that required coordination, continuity, funding, support, meaning and a sense of purpose and unity: in other words, it became a complex infrastructure, ever expanding and changing, but always propping up and pushing the work of numbers.

Areas of concern for ETAG related to the issue of whether ‘temporary placeholder’ indicators should be devised, especially in relation to the lack of a universally comparable metric for learning outcomes. Above all, a major qualitative difference had already taken place in comparison to the previous education MDGs: five of the seven SDG4 targets now focused on learning outcomes and skills, a major departure from previous targets which focused on access and completion. In 2016, with the new SDG4 agenda formally adopted, the ETAG shifted again, giving rise to the Technical Cooperation Group (TCG), with the same broad membership (UIS, 2017) and remaining operative to date.

Additionally, in parallel to the TCG, another group came into existence, following on the footsteps from the LMTF: this was the ‘Global Alliance for Monitoring Learning’ (or GAML in short), the successor of the LMTF. Also created in 2016, GAML was originally defined as an ‘umbrella initiative to monitor and track progress towards all learning-related Education 2030 targets’ (UIS, 2016, p. 49) and was tasked with the development of tools, methodologies and shared standards to measure learning outcomes in the context of SDG4. Following the TCG, its membership is open to any individual or organisation willing to contribute to the work of GAML and includes IOs, civil society organisations, a variety of technical partners and assessment organisations, and representatives of United Nations (UN) Member States.

Therefore, the political game of numbers became too high stakes to leave it to the technical experts only. Wider legitimacy was sought and gained through the expansion of the measurement infrastructure into an epistemic one: one that became legitimate and dominant through its active involvement of actors from across sectors and countries. Even though the involvement of the majority of these actors was generally passive, the language of the new indicators became the new episteme: that is, a way of knowing, describing and communicating about the world that was not merely about the craft of numbers but involved the production of a new governing paradigm: that of the dominance of the mono-disciplinarity of economics in global education governance.

6 Constructing Non-knowledge: Mono-disciplinarity and the Silencing of Alternative Perspectives

Indeed, as we saw in the previous section, the open, inclusive and participatory nature of the consultative process facilitated by UNESCO and the EFA architecture was in many ways unprecedented, and the openly negotiated and improvisatory character of the SDG debate contrasted with the technocratic origins of the MDGs (cf. Fukuda-Parr & McNeill, 2019).

In many ways, it is precisely this open debate and the participatory nature of the SDG governing architecture that has allowed a plethora of contestations to unfold: one of the most prominent ones is the large emphasis on some indicators (especially those that measure performance in literacy and mathematics) that comprise goal 4 versus others. Table 1 offers a useful overview of the different indicators in goal 4.

Table 1 The SDG4 indicators are as follows: 4.1.1 on reading and maths proficiency; 4.2.1/2 on early childhood; 4.3.1 on VET; 4.4.1 on ICT skills; 4.5.1 on gender equality; 4.6.1 on adult literacy and numeracy; 4.7.1 on global citizenship and sustainable development. Available at http://uis.unesco.org/sites/default/files/documents/11-global-indicators-sdg4-cheat-sheet-2018-en.pdf
Full size table

Although the development of SDG4 has been described as ‘arguably the most inclusive process of consultation in the history of the United Nations’ (Naidoo, 2016), this was not matched by the making of the relevant indicators to measure the ambitions (McGrath & Nolan, 2016; Smith, 2019). As discussed in previous sections, the process became quite technical from the start. Statisticians and their considerations for valid and robust data took hold of the process and most non-statistical knowledge was excluded. This was not however the only omission; perhaps the more significant one took place when there was an early decision upon some indicators which would be considered ‘global’ versus those that were relegated to the description of ‘thematic’ (Smith, 2019). This was a key moment, since,

While global indicators are universally applied and expected to be reported by all countries, thematic indicators are considered voluntary. Therefore, the majority of resources in indicator creation, monitoring, reporting and state action will focus on the global indicators while thematic indicators are not taken into account in the UN’s annual SDG report. (Smith, 2019, p. 3)

The pendulum had already swung. Although target 4.1 was promising that ‘by 2030, ensure that all girls and boys complete free, equitable and quality primary education leading to relevant and effective learning outcomes’, the 4.1.1 global indicator that came to be associated with it was much more limited in reporting on ‘quality’ only, whereas free and equitable education were downgraded to ‘thematic’, if they were even considered at all (King, 2017). In other words, the production of certain knowledge was privileged over others; this of course was done (and is always done) on the basis of the methodological robustness and validity of the exercise.

Indeed, many of the interviewees that METRO examined, suggested that the fundamental problem of the SDGs lies in the fact that it began the process by setting the ambitions and establishing the goals, rather than checking whether there was enough data or the right methodologies to monitor them. Nevertheless, the (limited arguably) resources that were put in the process were invested in indicators that were already backed up with significant statistical evidence. The strategic choice to construct non-knowledge by emphasising some indicators versus others becomes even more evident in the tensions that the negotiations around indicator 4.7 as created. As Antonia Wulff, from Education International, contends,

The expert group in charge of the SDG indicators rejected the proposed measurement strategy for target 4.7 on education for sustainable development, human rights and global citizenship… Education International is generally concerned about the slow progress made on key indicators and, importantly, the large disparity in the time, effort and resources put into developing 4.7 indicators as opposed to the learning outcomes under target 4.1. We are impatient to move forward. (Wulff, 2018)

Although limitations of space in the present chapter do not allow for a more extensive empirical analysis of the privileging of certain kinds of education data production over others, the above example serves as a useful illustration of the making of ‘non-knowledge’; rather than simply an ‘inability-to-know’, strategic decisions were made in relation to which disciplinary perspective was prioritised and took hold. The sociology of quantification has already persuasively discussed how quantification creates visibility, in antithesis to aspects of social life less easy to count. Although a collectively agreed ambition, indicator 4.7 on global citizenship, unless prioritised, measured and backed up with data, will remain a tokenistic representation of those ambitions that turned into ‘goals’ but were then strategically silenced in the process. Thus, at least in the field of transnational performance measurement agendas, the making of any knowledge implies simultaneously the omission of other routes to knowledge, or, in other words, the active production of non-knowledge.

7 Discussion: Mono-disciplinarity and Quantification

This chapter focused on an analysis of the ways that developments in global education governance since at least the start of the century propped up and legitimised the rise of economics as the dominant epistemology and method for perceiving and solving educational issues of inequity and performance management. Conflicting ideas and interests in this field reveal how epistemic infrastructures, rather than being monolithic blocks, remain fragile and, despite their claims to data and objectivity, are still open to plenty of epistemological and methodological contestation. In the case of the SDG4, it is evident that such contestation and the perceived failure of previous metrics emboldened education economists of major IOs to shift the agenda and move it along. The chapter showed how the perceived failing of the education MDGs (with the use of flagship numbers of emergency, such as the 250 million children not having basic literacy skills) was used as a vehicle to slowly build the dominance of education economics that, although having made plenty of ‘concessions’, is now perceived as the dominant disciplinary regime in global education reforms. There were plentiful of circumstances that the disagreement was such, that a possible break-down seemed almost unavoidable: for example, the reason of the compromise in the drawing of the main parameters of the SDG4 was the real possibility of the exclusion of an education-focused goal, due to the polarisation of the two ‘camps’. Yet, it is precisely the diversity and entanglement of the infrastructure’s social, technical and political elements that sustained and even strengthened the dependence on a single disciplinary field, that of economics, as the only robust and efficient way to measure and evaluate education performance.

Thus, the chapter focuses on the incompleteness and fragility of the infrastructure, alongside the generative power of failing metrics to provide fertile ground for more—and allegedly more precise and truthful—production of data, informed as we have seen both methodologically and epistemologically from a very specific disciplinary perspective. Here the chapter’s focus aligns with Calkins and Rottenburg (2017) in their engagement with ‘infrastructuring as a material-semiotic practice’: although the stable materiality and the techno-scientific dimensions of infrastructural work remain in place, the term is meant to denote the on-going, continuous nature of infrastructuring as practice rather than as a solid, stable space of production. Quantification in epistemic infrastructures becomes the fuel and language of practice, as it brings together ideas and objectivity in one entangled mix. In addition, as we have seen, quantification also lends to the dominance of specific methodological considerations, or at least path dependency with previous data collections, so that there are either crisis calls for needing to change the measurement agenda (as the calls for moving away from education inputs showed) or the dominance of the status quo and the strategic and systematic ignorance of certain types of knowledge. This level of confidence and self-belief, as Fourcade and her colleagues observed, has been a strong characteristic of economists, and perhaps the key quality that distinguishes them from other social scientists:

The fact is that -in some ways true to its philosophical origins- economics is a very moral science after all. Unlike atoms and molecules, the ‘objects’ upon which economists seek to act have a perspective on the world, too. Human life is messy, never to be grasped in its full complexity or shaped according to plan: people act in unanticipated ways; politics makes its own demands; cultures (which economists do not understand well) resist. Thus, the very real success of economists in establishing their professional dominion also inevitably throws the into the rough and tumble of democratic politics and into a hazardous intimacy with economic, political and administrative power. It takes a lot of self-confidence to put forward decisive expert claims in this context. That confidence is perhaps the greatest achievement of the economics profession -but it is also its most vulnerable trait, its Achilles’ heel. (Fourcade et al., 2015, p. 111)

Indeed, one of the main findings of the METRO project (within which the case of the SDG4 was studied) is the changing role of international organisations and their increased confidence to produce not only data but also expert advice on policy directions. Specifically, rather than assuming the expert role of the data producers (therefore asserting their credibility through the production of scientific truth), they have taken a new, brokerage role (Bandola-Gill, 2020; Bandola-Gill et al., 2021; Grek, 2020), working across different institutions and actors and pushing for narratives that link education (alongside other areas of sustainable development) with economic growth and prosperity. Such links are of course not new; since the 1960s and the establishment of many IOs, education was seen as the policy field where interventions and reforms would bring it closer to labour market needs and the production of ‘manpower’. Nonetheless, through the increase of the number of actors involved in global governance arrangements, what some have called the ‘stakeholderization’ of governance, IOs (and especially those who remit involves economic development, such as the World Bank and the OECD) have acquired new prominence and power through monitoring exercises, like the SDGs: as the example of the SDG4 showed the latter create zones of visibility and intervention, while simultaneously produce areas of opaqueness and invisibility.

The chapter has also pointed to two further aspects of the work of infrastructuring that we need to take into account: that is, their temporal and spatial elements. First, starting with the concept of time, any infrastructural investment has a temporal element that is not only evident in the passage of chronological time, but is also palpable in the transformational intent and the promise of a utopian perfectibility; the latter is a key epistemological premise of the discipline of economics, with growth being seen as the goal of achieving everlasting improvement. This promise of an anticipatory better future is central in the work of the SDGs: when it comes to SDG4, it has almost taken a moral dimension and sense of urgency (Grek, 2020), capitalised to either speed up or slow down the process depending on context. The SDG4 discursive analysis of reports and declarations (see also Chapter 1) shows infrastructural meaning to be produced through gathering past failures and future ideals into an unfolding anticipation in the present, or in other words economics’ pursuit of unhindered future growth. The case showed that apart from the anticipatory talk, a certain slowness of time was important in laying down the foundations of the new agenda, avoiding shocks and too sudden changes. Once the groundwork was done, after 2015, we see the process speeding up, coupled with an emphasis on expanding the infrastructure spatially and including a great variety of actors, both geographically, in terms of sectors as well as the ideas and interests that contributed to its production. Again, as already discussed, such expansion of the SDGs in regard to the inclusion of diverse policy actors, including state and non-state ones, was another reason for the prevalence of economics as the quick and accepted disciplinary perspective that would be able to put in line such a wide range of interests and ideas.

Finally, to return to the chapter’s earlier discussion, recent years have seen the rise of the sociology of ignorance, a new field of studies that examines the other, less visible side of the politics of constructing knowledge: that is, the politics of ignorance, or as this chapter prefers to call, the politics of ‘non-knowledge’. Linsey McGoey has been one of the key advocates of the need for social science to examine ‘the mobilisation of ambiguity, the denial of unsettling facts, the realisation that knowing the least amount possible is often the most indispensable tool for managing risks’ (McGoey, 2012a, p. 3).

The consideration of the symmetry of knowledge/non-knowledge is of course not new. Socrates insisted that his ‘wisdom’ was derived by his knowledge of what he didn’t know. Philosophically and historically the realisation of the limits of the human knowledge has always been present; nevertheless, our over-emphasis on examining the political uses of knowledge in governing societies has resulted in not engaging nearly enough with non-knowledge. Non-knowledge (or, for others, ignorance) here is not seen as an impediment and obstacle to knowing, but as a productive force, that strengthens the role of knowledge and of the knowing subject. For scholars in the field of ignorance studies, we need to investigate non-knowledge as ‘regular’ rather than ‘deviant’ (Gross & McGoey, 2015, p. 4). Yet, to date these discussions lack a coherent, agreed-upon nomenclature (Smithson, 2008). Although some scholars use ignorance and non-knowledge interchangeably (e.g. Kleinman & Suryanarayanan, 2013, p. 495), others distinguish between the two (e.g. Gross, 2012), emphasising the need to avoid the negative connotations that the word ‘ignorance’ implies. Further, there are also scholars who develop taxonomies of different types of ignorance and non-knowledge (e.g. Aradau, 2017; Beck & Wehling, 2012; Gross, 2016).

A review of the literature in the growing field of ignorance studies would be beyond the scope of this chapter. However, the key message that most of this literature appears to agree upon, despite the differences in terminology, is that non-knowledge is productive and not just the negative side of knowledge. Actors may actively try to nurture and preserve ignorance to use it as a resource to advance their interests be it in claiming more funding, denial of responsibility, or assertion of expertise (McGoey, 2012b, p. 555). Importantly, McGoey emphasises that such production and use of non-knowledge may be strategic and deliberate, but not necessarily conscious. Mallard and McGoey go further to propose an epistemological position ‘which asserts as a general maxim that ignorance can be an equally powerful political resource as knowledge’ (2018, p. 3). They suggest that

A second exploration by social scientists of how policymakers, experts and bureaucrats contribute to the production of soft forms of ignorance in international affairs… is the literature on the production of indicators, ratings, benchmarks which now circulate everywhere in the world of IOs and global media (Davis, Fisher, Kingsbury, and Merry 2012; Espeland and Sauder 2007; Espeland and Vannebo 2007). As scholars of transparency and auditing practices have long pointed out (cf. Strathern 2000; Power 1997), such indicators help to make policy decisions appear as if they belong to the realm of the certain and unquestionable even when policy options are based on the flimsiest set of factual observations. Most ‘global governance’ apologists who applaud the increasing use of benchmarking in policy research rarely acknowledge that the production of most indicators (like ‘rule of law’ indexes) is based upon fragile methodological foundations, and that the process of turning measurements into policy recommendations most often turns uncertainties and approximations into certainties… (Davis et al. 2012).

Indeed, it is precisely the construction of the doxa of a governable, manageable world that paradoxically the mono-disciplinarity of economics has resulted in: in such a world, actors that participate in its making have to be selective and actively ignore inconvenient data, or, as the empirical example above illustrated, systematically disregard the development of some measurement tools versus others. As recently one of the METRO interviewees emphatically suggested, ‘it is art, not science’. This art of assembling knowledge, while actively and strategically constructing non-knowledge, is necessary in order to leave the epistemic authority of the solutions uncompromised (no matter how ‘clumsy’ these may be), as well as having the door always open to the construction of new problems and solutions once the previous ones fail.

Although the field of ignorance studies has put a lot of emphasis on classifying kinds of non-knowledge, it has so far not achieved a coherent set of ideas about how to investigate the process of producing non-knowledge. The most notable exception to this is Scheel and Ustek-Spilda’s (2019) work; the latter use the notion of enactment from STS, while also making references to the concept of controversies, and in particular the examination of cases of non-transfer of knowledge—the moments of distortion, reinterpretation and loss that may occur when ‘data move between people, substates, organizations, or machines’ (Edwards et al., 2011, p. 669). The attention to the particularities, representations and often visualisations (through graphs, maps and other visuals) that the enactment agenda allows could be seen as a helpful way of investigating the tools and effects of the production of ignorance.

To conclude, this chapter mobilised relevant literature and used empirical examples in order to offer two propositions: first, that instead of disciplinarity, global education governance is primarily dependent on a monodisciplinary knowledge production orientation; and second, that an investigation of metrological realism needs to focus on the social construction of non-knowledge as a vital component of studying the epistemic authority of transnational institutions. Perhaps a sceptical turn in the study of transnational regulation, evaluation and monitoring must lead to an ‘un-settling’ of the classic studies of the political use of statistical knowledge, and offer the promise of a more creative, at times even inconvenient, analysis of the unaccounted and thus invisible processes of the construction of non-knowledge that the making of quantification requires.