Facing "a tipping point"? The role of the OECD as a boundary organisation in governing education in Sweden

This paper focuses on the role of the OECD as a boundary orga- nisation in Swedish education policy during the last 10 years. This has been a critical period of the transformation of the Swedish education system. Post-PISA, Swedish education moved from a position of con ﬁ dence and relative strength to one of crisis and slow recovery. So far, most educational research has either tried to identify a top-down linear in ﬂ uence from the international to the national “ levels ” , or has disputed such in ﬂ uence by suggesting that PISA has not causally led to substantial national reforms. On the contrary, this article suggests that education policy research has to move away from viewing the relationship of the national with the global as hierarchical, linear and hegemonic. Through the operationalisation of the concept of boundary orga- nisation, this paper is asking pertinent questions about the conditions, processes and practices that led to the entanglement of a large international organisation, the OECD, in the day-to-day policy-making of a country such as Sweden. This requires a nuanced theoretical and empirical approach, attentive to the political work of material artefacts, actors and organisations in the negotiations and ultimately the making of educational consensus.


Introduction
Numerous scholars have written in depth about the effects of quantification in the field of education governance in Europe (Kamens, 2013;Lawn, 2013;Lindblad, Pettersson, & Popkewitz, 2018;Ozga, Dahler-Larsen, Segerholm, & Simola, 2010). The key actor in these analyses, and as it seems, the epicentre of all this scholarly attention, is almost always the Organisation for Economic Cooperation and Development (OECD) and its successful Programme for International Student Assessment (PISA) (OECD, 2019). Academic research has sought to examine the ingredients of this success and the impact it has had not only on national education systems in Europe and globally, but also on the shifting policy and public perceptions on the value of education and how it should be governed (Grek, 2009).
In developing this argument, the paper will examine the period between the 1990s until today, paying particular attention to the period from 2011 until 2015. In this latter period the OECD has been particularly active in participating in discussions about the reform agenda in Sweden, as a result of the performance downturn indicated in PISA 2012, but also through the commissioning of two country national reviews of education and other related research. As suggested, this article offers a novel analysis of the role of the OECD in shaping the policy agenda, by focusing on its "boundary work" (Gieryn, 1983). Bringing together hard statistical evidence of the Swedish education downturn (PISA) (often in the form of emotive graphs of a spiral decline), alongside creating a discourse of the need for change, the OECD acted as the negotiating arena for both those believing in data for improvement, as well as those actors identifying the need for reform as more of a political and ideological matter, rather than a scientific one.
However, what is particularly interesting in the relationship of Sweden with the OECD? The Swedish case is the focal point of this analysis because, during the last 20 years, it has served as the great OECD experiment. Having followed the OECD 1980s' core narrative around the need for more deregulation, school choice and accountability, Sweden is now characterised by OECD experts as having gone too far and being at the "tipping point" (Blanchenay, Burns, & Köster, 2014) of a major educational calamity. Amidst this crisis the OECD yet again becomes the expert actor of choice that has the data and know-how to offer a way out. This know-how does not only entail PISA data and the two country expert reviews of 2011 and 2015, events that sparked the establishment of the Swedish School Commission and supplied it with most of the data and the narrative for change. Crucially, this article will show how the OECD has been acting at the boundary of different policy actors' interests, values and perceptions. In other words, the OECD, the dominant technocracy in the European education policy space, became the mediator and "buffer" of different forces and positions, in an effort to find a governing answer to the market vs. equivalence conundrum that has plighted the Swedish education system in the twenty-first century.
Analytically, the paper will show how, after almost a decade of sustained involvement within the Swedish education policy arena, the OECD now occupies a position that speaks persuasively to all relevant actors: policymakers, researchers, education professionals, media and the Swedish public. As a result, it has become the policy actor and influencer par excellence in Sweden, with the vast majority of education actors ascribing it with a legitimacy that is unprecedented for an external organisation entering a national education policy arena. Through using its technical competencies to legitimise pushing for reform agendas that aim to exercise a tighter grip on accountability in a system that lacks, according to OECDspeak, "systemic strategic vision", the OECD performs boundary work (Gieryn, 1983) that is transforming the shape and mode of education policymaking in Sweden.
Empirically, the analysis is based on an examination of interview data with eight prominent Swedish policy and research actors, as well as discourse analysis of an extensive range of OECD and ministerial documents in the period 2011-2015. This paper analyses three major OECD reports in more detail and also makes use of the relevant literature that discusses education reforms in Sweden since the 1990s. The interviews informing the analysis were conducted in the period October 2016-March 2017 and were taken by ministry representatives, academics, national agency officials and headteachers' associations.
Before we move on to the analysis, it would be worthwhile to first outline some of the key conceptual underpinnings of the notions of boundary work and boundary organisation, as well as explain the methodology of the empirical investigation. The paper will then move on to a historical contextualisation of the developments that led to the current situation. Although perhaps not at the core of this paper's analysis, yet those historical developments, as I will show, created the necessary breeding ground for some of the contemporary events. Finally, the paper will continue with the presentation of the analysis of the interviews and documents and finish off with a discussion of the research findings.

Boundary work and boundary organisations: theoretical and methodological considerations
The concept of "boundary work" was devised to delineate the relationship of "science" with politics (Evans, 2005;Gieryn, 1983). Despite linear accounts which see scientific evidence as constructed separately from the domain of politics in order to neutrally inform it, boundary work suggests that science is imbued with political considerations (Guston, 2001;Jasanoff, 1990); in other words, the boundary between science and politics is fluid. Boundary work then reflects the activities of organisations that seek to mediate between knowledge and action (Cash et al., 2003).
More particularly, boundary work is practised when tensions arise between actors who inhabit spaces at the interface of the production of "useful" knowledge and policymaking. Too little permeability of the boundaries separating science from politics would suggest too little learning from or contribution to the practice; in an era of evidence-based policy, such a proposition would be unthinkable. However, if the boundary was to be entirely dissolved, not only science would risk being politicised (an anathema to the objectivity and reliability of evidence-making), but politics and politicians could also risk being viewed as mere mouthpieces of technocracy (Guston, 2001;Jones, Jones, & Walsh, 2008). Boundary work is therefore required to manage the tensions and contradictions of the process of harnessing knowledge to promote action (Jasanoff, 1990).
Research on boundary work is plenty, as evidenced by the literature (Evans, 2005;Guston, 2001;Hellström & Jacob, 2003;Raman, 2005). Despite differences in emphasis and language, most authors suggest that boundary work is more likely to be effective when specific "boundary objects" are produced as a result of the process. Boundary objects are scientific products such as maps, models or reports that "are both adaptable to different viewpoints and robust enough to maintain some separation across them" (Star & Griesemer, 1989, p. 387). They simultaneously inhabit different social worlds (both of science and politics, in our case education science and education politics) and they satisfy the informational requirements of each of them. According to Star and Griesemer, 'their boundary nature is reflected by the fact that they are simultaneously concrete and abstract, specific and general, conventionalised and customised (1989, p. 393). Boundary objects have to have two specific characteristics to be successful: they need to provide some kind of architecture, i.e. offer a specific scale and structure. Second, they need to truly operate at the boundary, offering flexibility of interpretation by a wide range of competing actors (Star, 2010).
Finally, the third related concept and the one more pertinent for this paper's analysis is that of "boundary organisation". Boundary organisations do not just perform boundary work and produce boundary objects: more crucially, they provide forums and arenas where actors from both the science and the policymaking world can interact, exchange and learn. The primary function of boundary organisations is to communicate and translate their knowledge production in order to subsequently build joint knowledge that can be perceived as credible, legitimate and salient (Cash et al., 2003;Guston, 2001). In what way is this any different from boundary work? Boundary organisations formalise the work they organise by distributing accountability across the boundaries; they are the space where the science-policy interplay is accounted for and thus legitimised. According to Guston, First, they [the organisations] provide the opportunity and sometimes the incentives for the creation and use of boundary objects and standardised packages; second, they involve the participation of actors from both sides of the boundary, as well as professional who serve a mediating role; third, they exist at the frontier of the two relatively different social worlds of politics and science, but they have distinct lines of accountability to each. (2001, pp. 400-401) Nevertheless, Guston has been criticised for having a too linear conceptualisation with his complete separation and dichotomy of the two communities, in addition to describing the process as static. By introducing the concept of hybrid management, Miller (2001) expands on Guston's ideas to suggest the hybrid nature of the actors that are put together and managed within boundary organisations. In other words, Miller suggests that boundary organisations do not deal with the duality of the research vs the policy communities; instead, "what is managed within boundary organisations are multiple relations among institutions, people and artifacts, hybrids constructed by a mix of social worlds, norms, facts and values" (Latour, 1993). With the use of the notion of management, Miller emphasises the performative aspects that are part and parcel of the making of the organisation and its internal structures and processes, as these reflect the external environment in which they operate. Therefore, for Miller, boundary organisations Put scientific and political elements together, take them apart, establish and maintain boundaries between different forms of life and coordinate activities taking place in multiple domains. (2001, p. 487) Moving on, in terms of research design, the paper operationalises the concept of boundary organisation through moving beyond top-down accounts of the mere and one-directional transfer of policy from the international to the national, towards more attention to the interaction and mediation across "levels" and actors. The empirical research for this paper focused mainly on the critical analysis of discourse (Chouliaraki & Fairclough, 1999) in the examination of three key texts, through a focus on their "texturing" work; by "texturing", the paper refers to the "work" that is done textually, i.e. the textual "working up" of the role of the OECD as a key knowledge broker and problem solver across (and increasingly as time progresses) these documents.
In addition, a set of eight actors from both the Swedish education policy scene and the OECD, as well as other relevant research agencies, was identified and interviewed. The interviews focused on the actors' role in processes of coordination of the OECD's research and policy advice role in education in Sweden. The policy actors interviewed and quoted in this paper have had positions of power and significant decision-making leverage, and therefore in all cases first-hand experience and participation in meetings and debate between the Swedish government and the OECD in regard to its influence in Swedish education policymaking. The interviews complement the critical discourse analysis of the texts; following CDA's realist epistemology, they are significant accounts of the experiences of these actors working in the field during the years of the OECD's increasing involvement. In this way, actors' stories and the discursive construction of the enmeshing of science and politics are not separate but go hand in hand.

Swedish education policy after the 1990s
From the arrival of a coalition government of conservatives, liberal, centre party members and Christian Democrats in 1991, the links between the economy and Swedish education were strengthened (Lundahl, 2007). Lundahl argues that this was accomplished in three ways: "increased importance is attached to education as infrastructural investment, supporting economic growth; the market and enterprise serve as models of education and school and education is privatized or marketed" (Lundahl, 2007, p. 123). Thus, the early 1990s was a time of radical transformation of Swedish education policy, a transformation so deep and fundamental that its ripple effects can still be felt.
Prior to this time, Sweden had a very uniform and centralised education system. It built exclusively around the public school and run by the central government. Nonetheless, already by the second half of the 1980s, the seeds of change were sown; a debate around the efficiency of the public school system began and it led to demands for more school autonomy and greater parental school choice. As a result, the then Minister of Education, Bengt Göransson, pushed for a new "steering proposition" in 1988, which laid the foundations for a more outputoriented education policy (Ringarp, 2012).
The first major change was the decentralisation of decision-making from the centre to municipalities in 1991. From then onwards, municipalities were given extensive decisionmaking power over many aspects of schooling, such as hiring teachers and even (after 1996) setting teachers' wages. The funding for schools changed radically too; although before the 1990s it was the central government that controlled the resources and their allocation, with this new reform agenda, municipalities were given a lump-sum from the central government which they would distribute themselves, dependent on perceived needs in regard to teaching hours, textbooks, school premises and other matters (Waldow, 2008).
In parallel to these transformative changes, Sweden also pushed for parents to have much greater school choice (Forsberg & Román, 2014). Liberal regulations to establish and run independent (private) schools were introduced. Parents were given choice over which school their children to attend, irrespective of catchment areas. The government set up a voucher system in which both public and private schools would receive resources on the basis of the number of enrolled students, while private schools were allowed to operate on the basis of tax funds, therefore ending the state monopoly on public education, and introducing marketisation in Swedish education (Rönnberg, 2012). Curriculum setting also became much more free and ad hoc, dependent on the school's decision-making, whilst the government actively pushed schools to give parents information on school performance levels so that they could make informed choices (Lundahl & Waldow, 2009). According to the OECD, "the reforms therefore marked a profound redirection from a centrally run education system that was micromanaging inputs, towards a system oriented goals and objectives". (Blanchenay et al., 2014) Independent schools were closely monitored by the National Agency of Education, which was later on (in 2003) coupled with the "National Agency for School Improvement". Nonetheless, the need for tighter control of the system led to the establishment of the Swedish Schools Inspectorate in 2008, which took over the task of ensuring that independent schools and municipal schools complied with laws and regulation, while the National Education Agency focused on supporting and evaluating the work of municipalities and schools. The Agency has a crucial role in relation to generating and managing education performance data in Swedish schooling: it publishes education statistics through two publicly available databases -SIRIS (Information System on Results and Quality) and SALSA (Local Relationship Analysis Tool)which present information on the characteristics and results of municipalities and schools (Baxter & Hult, 2013).
In addition, the Swedish Association of Local Authorities and Regions (SALAR) publishes its own analysis of NAE data through the development of 15 success indicators and rankings of individual schools: this is the "Open Comparisons" database, and its primary aim is to inform the public debate and increase the efficiency of the system.
Finally, the Swedish School Inspectorate is in charge of site visits and the production of qualitative data that accompanies NAE hard data. Much of its work is driven by the notion of "equivalence": 1 the need to safeguard an equivalent standard of education throughout the whole of the country (Rönnberg, 2012, p. 74). The inspection model emphasised "control, results and formal or judicial aspects of education" (Baxter & Hult, 2013, p. 4) and therefore the intensity of inspections increased, as the following quotation illustrates: In 2011 the Inspectorate assessed 2400 comprehensive schools, 550 secondary schools and 660 publicly funded educational enterprises. In their annual report to the government they stress the increase in productivity of around 1000 visits compared to the previous year (or a 41% increase in productivity, our calculation). (Skollinspektionen, 2011in Hult & Segerholm, 2012 The formation of the new inspection agency marked a clear discursive shift towards a more regulative and control-oriented direction, articulated in a new and harder language, including the frequent use of terms such as audit, rules, prescriptions, supervision, take sufficient measures, penalty, injunction and other (Clarke & Lindgren, 2015, p. 143).
The reforms that took place in Sweden in the 1990s radically transformed the education landscape: from an entirely centralised system, municipalities now had the task of carrying out a large share of decision-making in schooling, data production and management and resource allocation. The reforms were motivated by three major principles: (1) moving decisions closer to the school and to parents; (2) giving more freedom of choice to citizens regarding education; and (3) introducing marketisation and competition between education providers, both public and private.
Although it is not in the scope of this paper to examine the extent to which some (or all) of these radical education policy reforms were inspired by global education policy shifts, yet they do suggest the construction of a changing policy context, that was trying to follow closely the global trends of evidence-based policy, accountability, decentralisation, and the use of comparison as a soft regulatory instrument (through performance naming and shaming). All the policy reforms discussed above happened at a very short timeframe and shook the foundations of the Swedish education system at its heart: was education still contributing to the construction of the Swedish welfare state, guaranteeing egalitarianism for all? Or had the waves of marketisation transformed it to its core? (Alexiadou et al., 2018;Blomqvist, 2004;Larsson, Letell, & Thörn, 2012;Lundahl, 2002Lundahl, , 2011 These questions led to a number of contestations, between local and governmental actors, but also between national academic research and "emerging" statistical evidence from international performance measurement tools; indeed, the launch of PISA in 2000 was about to turn these internal antagonisms, questions and conflicts into a full-blown education crisis. With this important historical context in mind, we will now turn to the core of the analysis, which is the examination of key OECD reports on the state of Swedish education.

An educational crisis? The OECD narrative of the Swedish education downturn
The discourse analysis will focus primarily on three texts: these are the Country Note for Sweden which formed part of the "OECD Review on Evaluation and Assessment Frameworks for Improving School Outcomes" (2011); the 2014 Governing Complex Education Systems Swedish case study (Blanchenay et al., 2014); and, finally, the Swedish OECD country review of 2015 ("Improving Schools in Sweden: An OECD perspective").
2011: To start with the earliest of the three texts, the 2011 OECD country review was invited by Sweden, and coordinated by Ms Kerstin Hultgren, a Senior Advisor in the Swedish Ministry of Education and Research. As with all of these reports, there was an extensive background country report, already prepared by Swedish actors and submitted to the OECD in 2010. This background report was giving the OECD representatives information that would assist with their visit in 2011. After some general information about the education system and specifically the reforms of the early 1990s, the report suggests that there are "concerns about a decline in student learning outcomes", with results in reading literacy "having significantly decreased", "results … in mathematics have also decreased", "a significant decline", and 'science results of Swedish 15 year olds were for the first time below the OECD average in 2009ʹ (Nusche et al., 2011, p. 25Nusche et al., 2011. The stark language of decline continues with a hit to the Swedish soul: Sweden used to be among the countries with a below-average impact of socio-economic background on performance. However, in PISA 2009, the impact of socio-economic background on reading performance has markedly increased and is now above the OECD average (Nusche et al., 2011, p. 25).
The report continues with statistics that show "a widening gap between high-and lowachievers", "a large gap in education outcomes between immigrant students and their native peers" (ibid, p. 26) suggesting that "segregation has increased in recent years" (ibid, p. 27).
Although the language of crisis and failure continues throughout the report, the more interesting discursive construction is elsewhere: that of a system that although has evidence and transparency, it lacks coherence and coordination, especially in regard to sharing responsibilities amongst all relevant actors: It is a system of shared responsibilities where the quality of evaluative feedback to the education system and its use is determined by the behaviour of several actors, and effective evaluation can be operated only though the cooperation of all of them (ibid, p. 33).
The report identifies the focus on governing through a focus on outcomes as a strength of the system; nonetheless, it suggests that "it carries risk in terms of equivalence of education across the country"of course, the language of equivalence is uniquely Swedish and its choice here over equity, which would be the commonly used one by the OECD, is deliberate. Further, the report highlights that "while a lot of quality assurance work happens locally within classrooms, schools and municipalities, there tends to be little documentation of such practices … the roles and responsibilities for implementing different aspects of evaluation and assessment are not clearly laid out in writing." In other words, although the report commends Sweden for its decentralised governance of education, and its focus on outcomes, data and transparency, it also points to certain deficiencies of the system which relate to what the OECD calls "a major challenge", being "the lack of an external assessment framework that would allow monitoring whether national learning goals are being achieved". It recommends that "Sweden should consider ways to increase the reliability of the existing tests, as well as possibilities of introducing additional types of student assessments such as a sample-based survey" (ibid, p. 39).
This discursive move away from discussing the declining results towards offering an analysis of a system that appears to have the evidence but does not know what to do with it continues throughout the report. On page 40, the authors of the report suggest that the most major drawback of the system is its "lack of knowledge management", thereby presenting the OECD as having the know-how and managerial competencies to identify how knowledge produced in the system needs to be analysed and disseminated. Lack of information of tracking students over time is discussed as problematic, but also "no information is available on how many schools have computerised systems", "no national plan to deal with this or standardise the approached to facilitate comparison" (p.41).
Further, the report makes a strong claim that although data exists in the system, "the priority now is to ensure that the existing data and information are actually used for improvement", thus implying that this has not been happening so far. Unsurprisingly, the recommendations for new policy reforms are the following: (1) Develop a strategic plan for evaluation and assessment and sustain efforts to improve capacity; (2) Develop an externally moderated student assessment system; (3) Increase clarity and support from the national level; (4) Improve knowledge management and strengthen links to classroom practice.
(2011, p. 41) The report examines the issues of student assessment, teacher appraisal and school evaluation; in all three areas, the focus continues to be on the need of more coordination and knowledge management in the system, since there are 'concerns about the reliability of national tests as a measure of system performance (2011, p. 103), in addition to "emerging evidence that collection, presentation and analysis of data could be further improved" (ibid, p. 104).
To conclude, Considering the existing strengths and proposed reforms of the Education Act, the OECD Review Team suggests the following potential policy pointers to increase efficiency in evaluating the Swedish education system: (1) Further build and develop evaluation capacity at central and municipal levels.
(2) Improve the mobilisation of existing information from evaluation and assessment.
(3) Explore the ways to more reliably track educational outcomes at the system level (ibid, p. 105).

2014:
The 2014 "Governing Complex Education Systems" Swedish case study has a telling title: "Shifting responsibilities -20 years of education devolution in Sweden" (Blanchenay et al., 2014). According to the report, This case study examines the consequences of important education decentralisation reforms that took place in Sweden in the early 1990s. The sudden shift away from a traditionally centralised education system towards a decentralised one meant that municipalities had to quickly accommodate new responsibilities. Difficulties related to this shift were noticed early on and then confirmed by international surveys, in particular PISA, which revealed that student performance was deteriorating while the gap increased between top-and bottom-performers … The central government, steering education at arm's length, has few tools to incentivise compliance with national goals. At the municipal level, financial resources are often allocated based on tradition and local politics rather than actual needs. This is in part due to misuse of available data and of expert knowledge by decision-makers (Blanchenay et al., 2014, p. 5).
This quotation, although lengthy, is a precise summary of the main gist of the report, which, similar to the previous report owes the majority of its data to Swedish analysis and findings (here in particular the report "Municipality responsibility in practice", submitted to the OECD by the Swedish National Agency in Education). The report discusses the "lack of systemic vision" which was the result of the fast decentralisation that did not allow for enough thinking and planning in regard to the distribution of responsibilities. The report recommends "clearer lines of responsibility and better use of existing data" and for the system to follow "a holistic and strategic approach to accountability" (ibid, p. 9). Elsewhere, the report recommends "harnessing relevant expert knowledge, prioritising the creation of a broad and holistic culture of evaluation, and creating networks and mentoring relationships". (ibid, p. 10). Although this is a very long and detailed report, due to the need for brevity, we will not go into great depth here. What is important is to show that, similar to the 2011 report, the OECD does not recommend any significant change of direction in terms of collecting more or better data. On the basis of the PISA data, but also other international, national and local evidence, what is suggested is the need to produce reliable data and a knowledge management mechanism where best practice will be shared across the different municipalities. The OECD yet again appears as the main actor with the know-how to coordinate minds and datasets towards a system of complete alignment and accountability. In other words, the "municipalisation" of the education system when "deliberately not accompanied by support from the central government" (emphasis in the original, ibid, p. 27) is condemned; the report suggests that the reforms were 'designed and guided by the philosophy that the "local authorities knew best", having "adopted a deliberate policy of not intervening" (ibid, p. 27). Interestingly in terms of the analysis here, the report goes further to claim that This led to spontaneous organisational rules, which in turn resulted in a variety of governance arrangements. Even today the perception of responsibility seems to vary greatly, both across people in similar positions in different municipalities, and between different bodies within the same municipality. (ibid, p. 27) In its concluding discussion, the report continues with the same narrative of lack of cohesive vision, lack of clarity in relation to responsibilities in the system and lack of capacity or competence to govern education effectively at the local level. It suggests that "crucially, there is no real enforcement mechanism for central authorities to ensure compliance with the goals".
A very interesting discussion in the paper regarding the use and misuse of data closes off with this comment, which is surprising when it comes from the OECD, but less perhaps so, since the report is written by CERI: "it is apparent that there is a hierarchy of knowledge which prioritises media-friendly rankings and is useful more for political purposes than a long-term strategic development of assessment in schooling" (ibid, p. 33). Last but not least, It is now clear that Sweden is facing a tipping point, and the timing is right to harness the momentum for change. What remains unknown, however, is whether the change can be channelled in the desired direction. (ibid, p. 37)

2015:
The Swedish OECD country review of 2015 was not the first one in the country; as already discussed, another one had preceded it in 2011 (Nusche et al., 2011). However, in light of the negative PISA 2012 results, as well as the general downward spiral of Swedish education performance, it quickly led the Ministry of Education and Research (MoER) to commission the OECD for yet another report of the country's education system. The objectives of the review were to, (1) identify the main reasons for the decreasing trends in Swedish students' performance; (2) draw on lessons from PISA and other benchmarking countries/regions with an expert analysis of key aspects of education policy in Sweden; and 3) highlight areas of policy and its implementation which might add further value to Sweden's efforts to improve student performance (OECD, 2015, p. 13).
The process followed the usual pattern: a background report prepared by the Swedish government, an OECD pre-visit which defined the key areas for review, an OECD team review visit to Sweden in October 2014, as well as a series of other exchanges with experts and stakeholders in Sweden and internationally (OECD, 2015). The OECD report acknowledges the contribution of a number of actors within Sweden and in particularly the Ministry, such as all members of the OECD -Sweden Education Policy Review Steering Group (namely Annica Dahl, Anna Westerholm, Johan Lindell, Kerstin Hultgren, Merja Strömberg, Anna Barklund, Peter Johansson and Annica Hellewell) (OECD, 2015). The two external experts in the team were Richard Elmore, Gregory R. Anrig, Research Professor of Educational Leadership, Harvard Graduate School of Education, and Professor Graham Donaldson, the former Scottish HMI Chief Inspector and current president of the Standing International Conference of Inspectors (SICI) -Donaldson was one of the chief architects of the self-evaluation model in Scotland. The expertise of both external evaluators (on leadership and accountability) is quite evident of where the focus of the review lay.
The OECD visit took place between 13 and 22 October 2014 and involved a number of meetings with key actors such as the Ministry; the Swedish National Agency for Education (Skolverket); the Swedish Schools Inspectorate (Skolinspektionen); the two teacher unions (Lärarförbundet and Lärarnas Riksförbund); academics in education research and teacher education (Stockholms universitet and others); the Swedish Association of Local Authorities and Regions (Sveriges Kommuner ochLandsting); and visits in different municipalities and local schools. (OECD, 2015) The report uses quite damning language to describe the state of Swedish education: "no other country … saw a steeper decline" (ibid, p. 7); "a school system in need of urgent change" (ibid, p. 11); "a position significantly below the average" (ibid, p. 27). A discursive analysis of the text reveals a language that describes a system in crisis. However, similar to the previous texts analysed, the educational crisis serves as the context in a system presented by the OECD as in need of radical change. According to the OECD, Sweden's disappointing performance on PISA has sparked the national debate on the quality and future of education in Sweden which seems to have resulted in a broad consensus on the need for change and support for the various school reforms and policies that the Swedish government has embarked on in recent years (2015, p. 8).
Indeed, the report points out repeatedly that the gap between the performance of high and low achievers cannot only be explained on the basis of increased immigration; this, as in other reports, is again pointing towards the lack of equivalence in the Swedish system, an issue that many Swedes would find more problematic and unacceptable than the country lagging behind other nations. The report suggests that there is a lack of capacity and clarity in roles and responsibilities at various levels of the education administration and recommends that, To respond to these challenges, Sweden should implement a comprehensive education reform to bring about system-wide change and strengthen the performance of all Swedish schools and students. It needs to define priorities, establish clear education responsibilities across the system and consistently provide appropriate support and challenge to schools, municipalities and private organisers in their improvement efforts (ibid, p. 8).
A lack of reliability of student assessment is a critical challenge as identified in the report, and this is emphasised (ibid, p. 54) at several points in the text. More critically though, the report maintains that the reaction of the Swedish Government to the negative performance trend has been unambitious: "the reforms of recent years are important, but evidence suggests they are also somewhat piecemeal, and simply too few, considering the serious situation of the Swedish school system". Instead, what is suggested is "a comprehensive reform agenda". This agenda is outlined in the report through a series of recommendations that require tighter accountability measures and closer steering of municipal decision-making; an agenda reflecting many of the arguments used by the research participants as outlined in the next section.
To conclude this section, it would be important to briefly re-cap how the OECD positions itself at the boundary of science and policy throughout these reports. Although one would have expected the OECD to put much more emphasis on their technocratic, data-producing power, the arguments put forward consistently play more of a role of knowledge broker and advocate, rather than a knowledge producer. By interpreting evidence and proposing new areas of improvement in the policy agenda, the OECD uses the country reports to create a forum where actors from all sides of the science and politics divide can interact. Through inviting the writing of background country reporting, it translates its respective knowledge and, more importantly, builds joint knowledge that is perceived as credible, legitimate and salient (Cash et al., 2003;Guston, 2001).
Receiving "the to-do list": an internal perspective of OECD's influence in Sweden Eight key actors were interviewed for this analysis. Their reflections on the OECD's influence were enlightening. They described the influence of the OECD in shaping the public and policy debate in Sweden as having started as soon as the first negative results were published. In the interviewees' analysis of how the OECD PISA became dominant discourse, there was unanimity in suggesting that the OECD became quickly the golden standard of education research in Sweden, at the expense largely of national education researchers who were seen as of lesser quality and relevance.
Secondly, although they do not often use the term "crisis", they all agree that PISA has been a legitimate source of evidence of the declining quality of Swedish education. However, more crucially than that, the OECD was seen as indispensable in relation to having the know-how to promote change. This is significant, as most interviewees did not only comment on the PISA data as the source of OECD's authority; they often referred to the organisation itself acting as a knowledge and policy broker, negotiating and translating political values, ideas and visions for the future. Interestingly, some of them went as far as to describe the OECD as the single force for halting the wave of marketisation in the country. Here is an extract discussing the OECD as being at "the left of almost all political parties": It was the OECD which was the promoter of efficiency, market-oriented solutions etc. But over the years, as PISA has developed at least in the Swedish or the Nordic context, the OECD has more and more become the voice of some kind of reason … for example, when it comes to the market orientation of the system, the OECD I would say is one of the actors who would claim that that the market orientation is too strong, which is quite contrary to what you would have expected. (policy maker 4) This is extraordinary given OECD's declared ideological standpoint, but evident also of the extent of marketisation in Sweden, as well as OECD's strategy to always appear as the a-political voice of reason. This is another instance, where the OECD does not simply operate as a technocracy, i.e. the voice of scientific reason. On the contrary, it offers political guidance and acts at the boundary between evidence and policy. Indeed, in the 2015 OECD report, it is stated that Sweden should, Revise school choice arrangements to ensure quality with equity, arguing that Sweden could benefit from managing school choice to prevent segregation and increased inequities. Providing full parental school choice can result in segregating students by ability and/ or socio-economic background and generate greater inequities (OECD, 2015, p. 101).
Finally, interviewees refer to the brokering role of the OECD in processes of consensusmaking through the involvement and active participation of a diversity of new expert and policy actors, such as economists or legal experts. These people were brought together to debate the evidence but also the political and policy implications. Indeed, some of the actors described a very polarised debate on the way forward for Swedish education. In this kind of sharp political climate, it is only the OECD that appears to have the brokering capacity to act as a boundary organisation and unite actors under its techno-managerialist wings.
Although they do not themselves use the term socialisation, all interviewees in their interpretation of the influence of PISA in Sweden, offered a similar story of staggered events that followed one another; of the involvement of an ever wider set of actors; of the importance of the OECD experts in guiding actors and advocating suggestions; and of the central role of the establishment of the Swedish School Commission as a forum of meeting, debate and learning for all the actors involved.
Indeed, the title of the report of the Commission, "Samling för Skolan" (Gustafsson et al., 2017), denotes precisely the notion of "congregation" or "gathering"the meeting and consensus of different actors around the core of the commission's study, which were the OECD numbers themselves. Numbers and data are central in the interviewees' narratives, but so are the meetings, the debates, and the continuous political work of the OECD in bringing actors together in socialising and learning events.
Interestingly, perhaps simultaneously with the rise of the OECD as the ultimate go-to education experts, we observe the slow decline of Swedish education research as valid and trustworthy enough to even take part in the PISA data collection process. Instead, Andreas Schleicher acquires an almost divine quality that matches closely the religious adherence to PISA in Sweden: What has happened, you can go back to 2003, TIMMS and PISA were at MidSweden university, now they are all run by the educational board (Skolverket). And they contract fewer and fewer education researchers for very little time to do some coding, to offer some comments. We were really independent from the government and at the time we were in a lot of the OECD meetings, we were involved. But now it is the educational board which does all thatand they don't have any researchers, they have project managers but they do not have researchers, they have government bureaucrats….But when Andreas Schleicher is in Sweden it is like we have a visit from God, it is very strange. I think this is problematic. (Academic 2) As a result, education researchers do not have an alternative voice in Sweden anymorewhen they take part, the majority of them is to validate rather than dispute the PISA results: Today no one can [criticise PISA] really. PISA has in some sense got so much status that I don't meet many who can say we can contrast PISAbut a lot of people say we need to discuss the implications of PISA. (Academic 2) The Swedish debate is much less active than for example the Norwegian debate. I think the criticism is louder in other countries. (Policy maker 3) However, Some of these recommendations are problematica lot of these things they are writing, a lot of secondary analysis is made by economists, statisticians not so much from the scientific community in education. (Academic 2) Although the academic community appears to have lost its central position in informing policy, there appears to be a much more diverse and horizontal participation of different actors in policymaking, even if it involves a lot of "cherrypicking". Here, speaking about how the OECD report was commissioned, an interviewee, suggests: They have asked for it. Policy makers, Björklund (the Minister of Education and Research) was the one who contacted the OECD and asked for the analysis of the reason for the decline … I think that it may be seen as well as giving up asking our national organisations but I don't think it needs to be interpreted in that way. He asked also many Swedish organisations and persons, researchers, people in the professions for suggestions and listened to them very selectively of course and did a lot of cherry picking of what he liked to hear and what he didn't like to hearthis is all what politicians do. But I think the OECD report was quite well received in Sweden; it is written by people in the OECD and parts of it reflect a certain Ideology and policy … but a big part of the report has been written by many researchers who are very proficient and very apt in doing this kind of research. I think that the suggestions they are making are seen by and large as making a good sense as providing a useful perspective (Commission member 4).
Two developments that appear to have dominated the Swedish education policy scene since 2000 are of essence here. The first one was the unequivocal rise of the OECD as the golden standard of education research in Europe generally and in the country particularly (with the simultaneous downgrading of national education researchers). Second, it was the role that the OECD played, through its country reviews, in facilitating a debate about a system that was portrayed as in crisis. This declining performance, given the history of Sweden as a model European education system throughout the twentieth century, in addition to the success of close neighbours (Finland, for example) became instrumental in allowing the OECD to enter the country in order to socialise and "educate" all relevant actors about the critical need for change. That process began slowly since the mid-2000s but became cataclysmic after the damning PISA 2012 report. Of course, national actors had a role to play in this, as they themselves instigated a series of initiatives, and particularly the establishment of the Swedish School Commission: It was a response to the OECD report. If I can give you a bit of the timeline: in December 2013 we have the PISA report, week after that there was a big debate at the parliament about the school crisis. There after Björklund invites the OECD to write the report, even before the report is released and they organised this school commission with Anna Ekströmnow the chair is Jan -Eric Gustafsson. Their task was to study the report of the OECD in order to make a Swedish analysis, do we agree what is the to do list, but this commission has been criticised as being only in favour of this particular view that the PISA results are the only ones that show the truth about Swedish schools today (Academic 1, my emphasis).
Indeed, the Swedish School Commission was launched by the Government in April 2015 and was initially headed by Anna Ekström, the director general of Skolverket. When she was appointed as Minister of Upper Secondary School and Adult Education and Training, Jan-Eric Gustafsson, a professor of Education, replaced her as the head of the Commission. The task of the Commission was set out as follows: 'partly based on the OECD's recommendations, the schools' commission will submit proposals aimed at improving learning outcomes, teaching and equity in Swedish schools 2 . Interestingly, the debate was very closely associated with the OECD data. The OECD and its recommendations were central to the debate and in many ways, framed it. For example, the Commission was purposefully staffed by a broad range of actors and that met regularly over two years in a process of learning, socialisation and translation of the OECD recommendations to national policy.
Yes, it is really an answer to the bad results of PISA 2012and I think it was a good responsenot follow the OECD in their 'do that, do this' but take some time and think it over…. Because there is so much political tension about schooling in Sweden just now. You really, everyone identifies you pro or against the free schools etc etc (Policy maker 5).
Indeed, it was precisely the slow up-take and translation of OECD's recommendations in the national context that created the consensus necessary around its central role.
According to a key ex-actor/member of the Commission: It is an advisory board but on the other hand if it is unanimous I think it would have an impact because of its membersthe commission is very broadly set up by persons with huge responsibilities in the education field and few people who are not directly connected with education but with a very high degree of public trust. So I would say that if they are unanimous and they come up with solutions that are possible, it will have an impact, it would be very strange if it didn't (Policy maker 3).
The Swedish School Commission met regularly for two years (2015-2017). Its members were asked to look at the evidence and draw conclusions about the direction of travel for Swedish education. Interviewees described these meetings as learning opportunities for all participants involved. They described the Commission as broadly reflecting the wider public and policy debate in Sweden, and suggested that its priority is to take the time necessary to offer a "Swedish solution", nonetheless following closely the OECD research and recommendations. Again, in their narratives, they never claim that the OECD data are not central; in fact, they are the spine that holds them all together. But they do also suggest that the national filtering process that is happening through their meetings is necessary for the interpretation, adaptation, persuasion and at the end adoption of the OECD perspective. Given the polarisation of the public debate, however, as well as the urgent nature of the need for change as suggested by the OECD, this work is not easy and often involves the participation of the OECD in pushing for specific agendas forward through the commissioning of more OECD research in the country. According to a Swedish academic, It was convenient to call external expertise because there was a disagreement about the status of education in Sweden because in the reforms in the 1990s the state and the national were weakened in relation to the local: the local was becoming stronger, this was debated and the teacher organisations weren't (becoming stronger), so they disagreed. Nor did the parties (agree), the alliance of right and liberal parties. So the OECD PISA report of 2006 was an opportunity to start to talk about things as a mistake, as something that needed to be organised differently (Academic 3, my emphasis).
The OECD as a boundary organisation: a hybrid role?
The analysis of the OECD reports and the interviews above suggest an education system which, after a series of radical reforms which transformed Swedish education governance in the early 1990s, was to be exposed by the OECD PISA and other international comparative assessments as facing a sharp performance downturn. Even before the news of the apparent Swedish education crisis, there were already voices in the system that suggested that the reforms had gone too far and that the level of decentralisation and marketisation was creating problems of steering and equivalence. In an effort to counteract some of these challenges, reforms such as the establishment of the Swedish Schools Inspectorate were meant to create bodies that had an overview and that would provide some direction from the centre. Nonetheless, the results of PISA 2009 and 2012 were to show that these reforms were fruitless; even worse, the results provided evidence that the problems facing Swedish education were much larger than it was previously thought. When the OECD spoke of a crisis and "the lost soul" of the Swedish education system, their negative assessment arrived at a point when the debate about marketisation had reached a level of polarisation that needed urgently a way out. The central government had already made some failed attempts to take back some control and steer more effectively. Teachers were disgruntled with loss of status, low wages and increased requirements for quality assurance. Education researchers felt powerless in front of the big OECD data machine. Last but not least, quite a few municipalities did not have a firm handle on how to administer a system that varied a lot, with new private education providers and parents eager to exercise their consumer rights to choose the best performing school for their children. In what previously was the model European education system throughout the twentieth century, a country that was proud of its welfare state and an education system that offered equivalence for all, irrespective of class and ethnic background, the recognition of these challenges became a shock and an opportunity for a new actor to enter the scene and offer an alternative, positive, future-proof vision: this actor was the OECD.
One could characterise the influence of the OECD in Sweden as having undergone two phases: the first one between 2010-2013, from the publication of the PISA 2009 results until the announcement of PISA 2012; and the second one, from the end of 2013 until today.
During the first phase, the OECD established itself as the unequivocal source of knowledge on the performance of the Swedish education system. Not only did it offer results that compared the country with other systems abroad but the OECD also offered substantialand "shocking"evidence in regard to the degree of equivalence, i.e. the levels of equity of the system across municipalities and schools. Although the Swedish system was awash with data, the OECD offered a new comparative parameter that was not there before: performance trends. Although there was a lot of data being produced by the system it was impossible to compare these data over time; crucially, it was the longitudinal nature of the OECD statistics that strengthened the force of the argument against the decentralisation reforms. Now for the first time, those sceptical of the 1990s reforms could use those "killer graphs" to show the downward spiral of Swedish education and demand change.
In addition, the teaching profession, due to the complexities of decentralisation and the local nature of the evaluation systems, was accused of being too "esoteric" in their assessments of classroom work, and therefore unreliable. Education research in universities was also seen as of lesser quality, as it could have never competed with the level of statistical expertise and the international comparability dimension offered by the OECD data juggernaut. As a result, amongst the ashes of the 2009 PISA results, a new phoenix for Swedish education had arisen: that was the OECD and its golden standard of education research. Hence, this first phase was one of consolidating the knowledge work of the OECD as robust, reliable and relevant; these are the characteristics of all powerful technocracies.
Nevertheless, the story does not seem to end there. As I have shown in the analysis of the reports and interviews in this paper, the OECD acquired a certain new rolea boundary rolein policymaking in the country, one that went beyond the technocratic power of its statistical expertise. The OECD operatedand still operates, despite the improved PISA 2015 resultsas a form of a boundary organisation. As we have seen earlier, boundary organisations are those which work to facilitate collaboration and communication across the research and policy communities (Guston, 1999(Guston, , 2001. Why and how did the OECD assume this role? One of the main reasons is that technocratic expertise was not enough to render the OECD as a key player in the field. Following their well-developed capacity to translate PISA data into policy recommendations, the OECD entered the Swedish education policymaking world with some key ideas for the improvements needing to be made to reverse the downward tide. Through a slow process of country visits, detailed reporting and numerous conversations with local actors at all levels of government and the education professionals and academic world, the OECD became a boundary organisation constructing a very carefully maintained equilibrium amongst the different powers and interests of the actors in the field. On the one hand, the central government which needed tighter controls of the decentralised, market-oriented system; and on the other, municipalities, teachers and academics who appeared helpless in regard to how to make sense, assess and in the end find governing solutions for a system at the brink of a major crisis. With a careful orchestration of socialisation and learning events around the conduct of the two country reviews, but also other research (as discussed above), the OECD created a new hybrid environment in which the provision of knowledge and policy suggestions would often intertwine so closely that it would be impossible to discern between the two. Therefore, the OECD managed to construct a new but very effective data/policy mix, one that could be seen as the reason for its success not only in Sweden but elsewhere, too. This is not purely knowledge. It is not policy per se either. Instead, it is a hybrid of both knowledge and policy closely intertwined. It is situated at the boundaries of scientific knowledge production and national education policymaking; this is the work that a boundary organisation does. As this paper has shown, OECD's influence is the outcome of sustained and careful boundary work, necessary at a political context with national and historical sensitivities that cannot be merely guided by cold, rational data. The OECD produces a hybrid construct mixing hard numbers, administrative advice, managerial know-how and best practice recommendations in a big, versatile, complex and ever-changing mixture of facts and values. The work of a boundary organisation, such as the OECD, is necessary to broker and translate knowledge taking actors' diverse interests and values into account.