Quality in School Trusts: A Comparative Study in Four Countries

Abstract

Recent decades have seen the emergence of new forms of governance in education. Market-oriented education policies have seen schools removed from local government oversight, and instead sees them run by independent school trusts. This begs the question of what effectiveness and quality in a trust framework mean. In this paper, we review the literature from four countries with relatively comparable trust structures: England, the Netherlands, Sweden, and the US, to look at the evidence base on trust quality, leadership, and effectiveness. We find that the evidence base itself is relatively limited in scope. It is strongest in relation to impact on the organisation, especially with regards to the importance of HR, relationships within the trust, and quality assurance mechanisms. There is evidence at the network level on the importance of knowledge and skills of governors, effective forms of collaboration, and a shared vision across the trust. Trusts can improve the quality of education through professional development of teachers, developing consistent approaches, and sharing knowledge. Evidence of how trusts can enhance their impact in the community is limited.

1. Introduction

Recent decades have seen the emergence of new forms of governance in education. Market-oriented education policies promoting school choice and accountability have led to reforms to the governance structure of schools in a number of countries, typically through removing them from local government control. These moves have led to the development of charter schools in the US, academies in England, and free schools in Sweden, for example. In these cases, schools are no longer funded through and run by local governance structures, such as school boards, local government, or local education authorities, but are funded directly, either through a contract with the ministry of education or through voucher schemes. One motivation for such changes has been to provide more autonomy for schools. The extent to which that is the case is disputed, as the new structures are seen in some countries (e.g., England) to have de facto transferred more control to national government [1,2]. Another key driver has been to improve the system through competition between schools, which would allow market mechanisms to drive up quality [3].

While in the US, in particular, these new forms of governance were initially envisaged as giving autonomy to individual schools, it became clear relatively early on in the development of these governance arrangements that there are risks to the sustainability of standalone schools, this both in terms of financial stability and in their ability to sustain school improvement. This has led to the formation of joint overarching structures encompassing groups of schools, such as Charter Management Organisations (CMOs) in the US and Multi-Academy Trusts in England [4,5]. In some countries, such as England and the Netherlands, such structures have been explicitly promoted by government, leading to their rapid growth and even dominance within the education system [6,7,8]. In most cases, these structures take the form of charitable trusts, though in the US, and to a greater extent in Sweden, for-profit organisations also exist. In this paper, we define a trust as a group of schools working together in a single entity, outside of the local authority or government. This broad definition allows us to look at structures and systems in a range of countries.

The growth of such trusts within the school system raises a number of questions. Issues such as democratic control, accountability, and the role of the state in education have rightly attracted significant discussion (e.g., Wennstrom [9] for the case of Sweden and Kulz [10] for England). Therefore, in countries where trusts are a significant or dominant factor in the education system, the question of how to optimise their quality should be addressed. A number of studies have looked at how trusts themselves are best led and governed [8]. There is also some research on attainment and added value of CMOs and trusts (see, for example, Tuttle et al. [11] on the impact of the KIPP CMO in the US, or Andrews [12] on the performance of Multi-Academy Trusts in England). These studies suggest that there are differences in performance between trusts, which from a leadership and governance perspective begs the question as to what factors affect their quality. Some researchers, such as Greany [8] in England, have looked at this question. However, to date, there have been few attempts to summarise the state of the knowledge on what quality in leadership and governance of a school trust looks like. As such forms of educational governance grow in importance, it is worth looking more closely at this question, to ascertain to what extent there are common lessons that can be drawn for both policy and practice with regard to the effective management and governance of such entities. The fact that we are seeing similar though by no means identical developments in different countries makes it interesting and timely to explore differences and similarities across contexts to explore if we can come to a common quality model and to see whether policy differences lead to different models or emphases.

In this paper, we will, therefore, look at the research base on quality in school trusts, using studies from four different countries that have reasonably comparable systems in which school trusts play a prominent role: England, the Netherlands, Sweden, and the US. The research questions we will attempt to answer in this literature review are as follows:

• What evidence exists on the factors that appear to determine trust quality?
• What is the quality of this evidence base?
• To what extent does the evidence on trusts align with existing models of collaboration and governance in education?

2. New Forms of Governance in Education

The development of trust structures can be located within the new forms of governance that came to increasingly characterise public sector governance in many countries since the late 1990s [13]. This has been described as ‘networked governance’, whereby rather than direct bureaucratic control, the state works through a range of private or public sector networks to provide greater flexibility in the delivery of public services [14]. This form of governance has been promoted as being more suited to the complexities of the contemporary state, and better able to grapple with hard and interconnected problems, such as, in education, the issue of inequalities in educational outcomes. However, the extent to which network governance has indeed been better able to solve such problems is disputed, and network structures are also seen as weakening democratic control over the public sector [15,16,17].

Trusts can be seen as a form of networked governance, in that governance rests with an overlapping set of entities, in the first place the trust itself, but also a range of funding and oversight bodies, including national governments, trust sponsors, such as charities or individuals, religious bodies, national inspectorates, local government, and quasi-governmental organisations, which may all play a role in the provision of education and related services (e.g., in support for students with special educational needs). While network governance can describe the relationship between trusts and external actors, a question that can be asked is the extent to which trusts themselves work as networked governance structures internally. As Wang and Ran [14] stated, networks are structures in which actors are interdependent and need to work together to solve problems or require resources, while collaboration in networks entails shared decision-making and consensus creation to reach shared goals. Delegated structures, in contrast, involve a clear hierarchical structure, in which tasks and competencies are handed over to actors within the structure, who then may or may not collaborate to achieve goals. A key issue in this relationship between centre and periphery is the trade-off between motivating the local agents (schools) by giving them autonomy and losing central control over processes and outcomes [18]. As we will outline below, trusts respond to this trade-off in different ways.

Theoretically, collaboration can have a number of benefits. Muijs [19] pointed to a number of potential reasons, such as closing structural gaps in knowledge between organisations; complementary understandings and meaning-making leading to better adaptation to a complex environment; building collective capital; and combatting organisational anomie. There is empirical evidence of the benefits of collaboration for school improvement [20]. Collaboration between schools has been found to build capacity for improvement, allow sharing of practice, strengthen opportunities for CPD, strengthen opportunities to develop leadership across the organisation, foster better cross-community relations in divided societies, and in a limited number of studies, improve student attainment [20,21]. This evidence has attracted some criticism, with Croft [22] claiming that the evidence for collaboration is a lot weaker than often stated. Generally, however, the consensus in research on the impact of collaboration between schools is largely positive, and in theory trust structures should provide strong opportunities for collaboration, though the extent to which they actually do so is less clear [23]. As well as collaboration, in theory, trusts could support school quality through the opportunity to rapidly scale up good practice, provide a strong leadership structure for supporting schools, standardise back-office functions for greater efficiency, and provide stability and support to individual schools [4]. The extent to which this is the case is again debated, however.

When looking at the impact of trusts on educational quality, the question of course arises of what we mean by the latter. A range of definitions of quality in education exist [24]. These can be ranged on a continuum from narrow definitions, which stress measurable outcomes, and in particular attainment (as in much school effectiveness research), to views that take a holistic approach focussed, among other things, on the personal growth of the child. One can also place quality in education on a continuum from a primary focus on student outcomes to a broader focus on the characteristics of the school or trust as an organisation. In this paper, we take a broad and organisation-based definition, seen as most appropriate in discussing characteristics at the trust level (as opposed to school or classroom levels). We therefore define quality as the extent to which an organisation excels in meeting its objectives, and we are thus looking for those factors that allow trusts to do so. Objectives can of course be varied. Pupil attainment is one, but by no means the only objective of a trust, which may include quality of curriculum, equity, pedagogy, sustainability, etc. In their discussion paper on trusts in England the Confederation of School Trusts [4] defined five quality domains: expert governance, quality of education, workforce resilience and wellbeing, efficiency and effectiveness, and public benefit and civic duty. In the Netherlands, the PO-raad (advisory board for primary schools) distinguishes impacts in three main areas: quality of education, HR, and business processes [25], while in their quality framework for charter school support organisations, the US body ‘Building Charter School Quality [26]’ put forward financial performance and sustainability, board performance and stewardship, and parent and community engagement. These lists have obvious similarities, but also some differences, which may reflect the national context. It is therefore possibly most useful to look at trust quality in terms of the areas they can impact.

Looking at accountability in networks, Ehren and Perryman [27], drawing on Popp et al. [28], distinguished four key areas in which networks of schools can have an impact:

-

On the individual, that is both the ‘clients’ of the organisation, such as students, but also the teachers and leaders in the schools that are a part of it (e.g., impact on teaching and learning through support from the centre). In accountability terms this is related to the contribution of the teacher to trust outcomes.

-

On the school as an organisation (e.g., support from the centre for school leadership and common approaches to behaviour). In accountability terms this in turn relates to the contribution of the school to the performance of the trust.

-

On the trust as a network (e.g., back office, relationship strength, and collaboration within the trust), which includes the synergies brought about by schools being part of a network.

-

On the community more broadly (e.g., improved employment prospects or social cohesion), which speaks to the social role of trusts as an integral part of the communities they serve.

This framework provides a holistic but structured view, that looks at trust impact across levels and allows us to incorporate the range of outcomes from our broad definition of quality. In this paper we will draw on this framework to look at the literature on trust quality in each of these four main areas.

3. Methodology

In order to map the evidence, this study followed a narrative literature review methodology. While alternative approaches, such as systematic review, were considered, it was determined that these would not fit the questions or the field, in which much of the evidence base is qualitative. A search was conducted using a number of databases, such as Web of Science, EBSCO, ERIC, and Google Scholar. These databases mainly produced articles from peer-reviewed journals.

In addition, we interrogated the so-called ‘grey literature’ by searching websites of government agencies (e.g., ministries, inspectorates, and representative bodies in each country) to uncover non-peer-reviewed sources and conducted a general Google search, which uncovered further sources from non-governmental bodies, such as charities and advocacy groups. We also drew on the references to other studies in those sources. We used search terms including trust, MAT, CMO, chain, friskolar, Stichtingen, governance, effectiveness, quality, governance, leadership, and impact.

The volume of papers recovered in this search was relatively limited, which allowed us to individually filter each source according to the aims of the study. The filtering criteria were as follows:

-

Relevance: the paper needed to discuss issues related to quality or quality management in school trusts.

-

Clarity: the paper needed to clearly describe methodology and scope.

-

Quality: the paper needed to be based on an empirical study or evidence base.

A total of 64 articles were extracted from this initial search, with a further 13 added following peer review. The majority were in English. Some papers in Dutch and Swedish were also recovered. The latter were translated into English by a student of the authors.

In this study we looked at evidence from four countries with reasonably comparable trust structures: England (‘Multi-Academy Trusts’), the Netherlands (‘Stichtingen’), the US (‘Charter Management Organisations’), and Sweden (chains of free schools). These systems are comparable in that they each have seen the emergence of government-sanctioned new forms of school governance, whereby schools are governed by entities other than national or regional government or state-sanctioned religious denominations which have traditionally run schools in most Western countries. In each of these four countries, independently governed but government-funded schools now form a significant part of the system. What these systems also have in common is that in each case such schools have increasingly formed groupings or chains of multiple schools working under a common governance structure, which in this paper we have called trusts. Reasonably comparable still requires the caveat that significant differences exist in the structures of these organisations and the education systems they work in.

Multi-Academy Trusts in England are directly funded by national government and can be set up by any party with approval from the ministry of education. Most are regionally based, though there are some national trusts. Trust schools form most of the secondary sector, and a significant proportion of the primary sector. School Trusts are independent legal entities (charitable companies) governed by a Trust Board. It is the Trust itself that is the legal entity, and not the schools that are its constituents. The Trust Board is appointed by the Members, and Trustees have legal duties under Charity Law and Company Law. The nature of the academies policy has changed over time—while initially the primary aim was to encourage novel providers, in particular in disadvantaged areas, the policy has moved towards academisation of the majority of the education system through the conversion of previously local authority-run schools into academies. Academies have gradually come to form the majority of schools in the hybrid education system [4].

Stichtingen have considerable autonomy within the Dutch education system and are financed through a block grant. As in England, the government has promoted the formation of boards covering multiple schools, which currently cover over 60% of schools in the system. Of these, the majority run between 2 and 10 schools. Stichtingen in the Netherlands tend to be geographically clustered. They usually have a dual structure, with a supervisory board overseeing an executive board. The latter is typically small, often made up of just one or two governors. In turn, these governors tend to delegate a lot of their legal responsibilities to individual schools and their leaders, meaning that there is often a significant amount of variance in practices within the trust. Accountability is concentrated at the trust level, for example, in inspection [29]. In the Netherlands, the impulse towards the foundation of Stichtingen has come primarily from government, with full-scale conversion of all state-funded schools to trusts making this the organising principle across the education system. The current structure, which separates governance from management of schools, dates back to legislation enacted in 2009 [30].

Charter Management Organisations in the US arose out of the charter school movement, and now run a significant proportion of charter schools in that country. Similar to trusts and Stichtingen, they are non-profit (there are also a small number of for-profit EMOs, but no relevant studies were found on these for this review). They predominantly operate in low-income neighbourhoods, and often serve non-white ethnic communities [31]. The impetus for charter schools and CMOs came largely from philanthropists, grassroots organisations, and school choice advocates, rather than being set up by government. They typically receive less per-pupil local government funding than district-run schools, though in many cases they receive philanthropic top-up funding which makes up the difference or can in total exceed public school funding [31]. The US, of course, is not a unified education system, and the form and number of charter schools differs significantly between states. In most cases, the aim of charter schools was to encourage innovation and/or competition within local education systems, in particular in areas characterised by disadvantage and social deprivation [32]. Though there are over 11,000 school districts that contain charter schools, they are still very much a minority nationally (approximately 6% of enrolled students are enrolled in charter schools, and of those around one-quarter are part of a CMO), though CMOs do predominate in certain local systems such as New Orleans [33].

In Sweden, a law change in 1992 instituted public funding for private schools, so-called friskolar (Free Schools). These schools are frequently run by chains, such as Kunskapsskolan. A key difference with the other countries in this review is that the Swedish system allows and is dominated by for-profit chains. This significantly changes the incentives within the system and has been the subject of criticism and controversy in Sweden. Free schools, the majority of which are part of a chain, account for around 25% of schools (and more in the non-compulsory upper-secondary sector) [34]. All free schools are required to follow the national curriculum, but are free to pursue their own pedagogical approaches, which is intended to spur innovation and competition.

4. What Does the Evidence Base Look Like?

There is a relative dearth of research on trust quality in the international evidence base. While in the field of educational effectiveness there is a long tradition of comparative studies of schools, the same is not true of trusts. The majority of quantitative studies focus on comparisons between trusts and local authorities or school districts, and as such are not very illuminating with regards to what makes a trust work well [29]. Complexity is generally seen as a key reason for the lack of research in this area. It is of course the case that the impact of trusts is likely to be largely indirect, work through a number of different mechanisms, and be contingent on a range of influences within and outside of the schools and the trust itself [29,35].

Most of the studies in this field are qualitative, with the exception of a number of survey studies in the US and England. Studies come predominantly from England, the Netherlands, and the US. There is a notable lack of research on quality factors in trusts in Sweden, with almost all published research being about systemic impacts on attainment and the distribution and selection of students.

In the US, the charter school movement has received a lot of research interest, especially in the area of competitive effects on attainment, a portion of which is dedicated specifically to CMOs. Much of the US-based research is framed in terms of debates around school choice. There are a number of useful large-scale quantitative studies, though these tend to be somewhat older.

In the Netherlands, there is relatively little evidence on the quality of trusts. A lot of the literature is descriptive or prescriptive, with limited empirical research, with the exception of the valuable work of Honingh and Ehren and reviews by the school inspectorate. There are a number of quality frameworks, but the evidence base supporting them is often unclear.

The English evidence base has a substantial number of qualitative studies on trust characteristics as well as quantitative survey research and comparisons of attainment between trusts and between trusts and local authority-run schools.

Sweden has a smaller evidence base in this area. Almost all papers found were either focussed on the overall impact on attainment and equity of free schools rather than trusts, or largely polemical in nature, written primarily by opponents of free schools.

5. Findings

5.1. Impact on the Individual

Studies looking at the impact of quality factors on student outcomes most typically take the form of correlational analysis, with attainment measures as the outcome variable. There are relatively few such studies internationally, with most originating in the US. Some US researchers have compared charter schools that are part of a CMO to those that are not and found that, on average, those in a CMO outperformed standalone charter schools [36,37]. One interesting study in England compared schools that converted to academies either as standalone academies or by becoming part of a trust. They found that the latter showed higher levels of attainment and more rapid improvement [38]. Trusts tended to focus more strongly on changes to leadership and management, while standalone convertors were more likely to focus on increasing pupil numbers or on curriculum change.

In another such study in the US, Furgeson et al. [37] matched their sample to comparator schools and found a positive but small impact of being a secondary school in a CMO (there were no test data for primaries). They found two main factors that were related to a positive impact of the CMO on test scores:

-

Comprehensive behaviour policies, including clear behaviour standards, zero tolerance for dangerous behaviour, codes with rewards and sanctions, and signed agreements with parents.

-

Professional development of teachers through provision of tailored CPD and feedback, and instructional coaching.

Many CMOs in the US operate so-called ‘no excuses’ policies. While these have been associated with better attainment outcomes in some studies, analyses from New York City and Massachusetts, in which different elements of the approaches of successful charter schools were separated, suggested that the very strict behaviour policies associated with this approach are not a necessary correlate of improved attainment [39,40].

Kelly et al. [41] studied a range of CMOs and local districts in the US. They found three main drivers of improvements in student attainment:

-

Improving the quality of teaching through investment in high-quality CPD and an aligned curriculum.

-

Allowing school leaders to exercise instructional leadership, not least by matching resources to goals. The authors also recommend some linking of reward to performance.

-

Driving culture through shared vision and goals.

A number of authors have stressed the importance of an emphasis on teaching and learning, or relatedly on instructional leadership. What is meant by this differs between studies, however. Hopkins [42], reviewing successful trusts in England, suggested a focus on five areas: curriculum, assessment/accountability, teaching, learning, and leadership. This does, however, leave a lot of questions as to how a trust should approach this in practice.

Furgeson et al. [37] identified three main approaches to improving learning and teaching in the CMOs they studied in the US:

-

Data-driven, with a strong emphasis on the use of formative data and performance-related pay.

-

Time-on-task driven, with an emphasis on school discipline and increasing instructional time.

-

Incremental, where a more gradual approach to innovation was taken.

Their data suggested that the data-driven and time-on-task approaches were more impactful than the incremental approach. Factors relating to a time-on-task-driven approach, such as behaviour policies, were more likely to be significantly related to attainment than factors relating to the data-driven approach. In addition, Furgeson et al. [37] found that use of formative assessment across the CMO was also positively related to outcomes. Performance-related pay and more assessment of pupils were not significant.

A follow-up study showed that CMOs that scored particularly strongly on behaviour policies focussed on developing social norms and behaviours, encouraged consistency across classrooms, ensured adults modelled desired behaviours, and encouraged parental support for behaviour norms, for example through home-school contracts. In terms of teacher coaching, the strongest CMOs employed coaching that strategically targeted teachers’ needs but aligned these to school and CMO goals, frequently monitored and observed new teachers, provided rapid feedback on teacher observations, selected expert coaches, and ensured that school policies supported coaching [43].

Fryer [44] used five practices seen as characterising high-performing charter schools: increased instructional time, high-dosage tutoring, data-driven instruction, high-quality teachers and leaders (though their definition of quality was rather unclear), and a culture of high expectations, and tried to implement them in non-charter schools in one school district. The programme showed modest positive impacts on math attainment and no impact on reading.

Crawfurd and Hares [45] found some evidence that the variance in student attainment is reduced in chains compared to standalone schools, though this evidence was limited to relatively smaller chains in England and Pakistan. Similarly, Baude et al. [46] found that variance in quality between charter schools in Texas reduced over time, as the better CMOs came to dominate the sector. There is also evidence from the US of lower levels of variance in teacher effectiveness in CMOs, which appears to be related to the level of support for new teachers [47]. In England this was not found to be the case, however, with attainment gaps between trusts staying pretty much similar between 2014 and 2018 in a series of comparative studies [48,49,50,51,52]. Less variance is also not universally seen as an indicator of quality. Peeters et al. [53] argued that we do not sufficiently understand what variance means in the context of trusts, and that valuing choice means valuing variance. However, valuing variance in the system does not necessarily imply that variance within trusts themselves is always desirable. Choice can involve variance between trusts, but homogeneity within trusts.

5.2. Impact on the School as an Organisation

Studies in this area tend to focus on three main aspects:

-

Relations between centre and school leaders,

-

CPD and Human Resources Management,

-

Quality assurance.

Relations between governors/trustees, CEOs, and school leaders are highlighted in a number of studies. Social capital across the trust and among board members was found to be a key feature of effective Dutch trusts in Honingh et al.’s [35] review. US research shows similar patterns, with good relationships between CEOs and school leaders characterising effective CMOs. Internal human capital is also a major driver of the capacity of CMOs to grow and expand [54,55]. The extent to which trustees engage with schools in the trust, and the extent to which these bottom-up perspectives actually influence what happens at the board level, can be problematic, however. Baxter and Cornforth [56], for example, reported a lack of connection between the board and school levels in their case studies in England. In Constantinides’ [57] study of successful trusts in England, he found a strong emphasis on developing relationships between CEOs and school leaders, to “enable principals and their school leadership teams to make decisions that would respond to the shifting and changing context of their school”. There is some evidence from Furgeson et al.’s [37] study of CMOs that principals spent less time on administration and more on instructional leadership than those in comparison schools, while Greany [8] reports school leaders saying that high-quality back-office support freed up time for headteachers to dedicate to educational leadership. Also in England, Stokes et al. [58] found that academies employ more middle and senior leaders, but that while this is associated with better performance in standalone academies, this is not the case for trusts, possibly due to lower levels of autonomy for leaders there.

An important element of workforce resilience and well-being is opportunity for professional development. The role of the trust in facilitating this was seen as a major advantage by leaders interviewed by the English inspectorate Ofsted [59] in their study on Multi-Academy Trusts. Greany [8] reported that most school leaders in his study claimed that professional development was strong in their Multi-Academy Trusts, and that most MATs offered staff significant initial teacher education (ITE) and continuous professional development and learning (CPDL) opportunities. The same emphasis on professional development was found in Constantinides’ [57] study of five successful trusts in England. Similarly, the Dutch inspectorate reported that the most effective trusts supported their schools to become learning organisations and empowered leaders in those schools [60]. In the US, Furgeson et al. [37] found staff turnover to be lower where the CMO provided central CPD. While there is limited evidence on the impact of school boards, the impact of school leaders has been demonstrated, so creating the conditions under which school leaders can be optimally effective would appear to be a key task of the trust (just as creating the conditions under which teachers can be effective is a key role of school leaders) [29]. High-quality CPD is seen as central to this task.

An advantage of the trust structure is the flexibility it can allow in managing the workforce across trust schools. In this regard, Worth [61] reported on data demonstrating that movement between schools within a trust is significantly larger than that between non-trust schools at similar distances, but also that within trusts teachers are more likely to move from a school with a more advantaged intake to a school with a more disadvantaged intake, which is the opposite of the typical pattern of teacher mobility. This suggests that some trusts are proactively managing staff to support schools in greatest need. There are also some studies suggesting faster career advancement for teachers in CMOs in the US, compared to both standalone charter schools and local district-run schools [47,62]. This flexibility can also have negative consequences, however, with some studies reporting higher levels of workload and greater staff turnover in CMOs [63].

One area of focus in the literature is the role of the trust in quality assurance (QA). Ofsted [59] reported that trusts in their study played an important role in QA, though the extent of their specific involvement varied significantly between trusts. Furgeson et al. [37] found that frequent observation of teachers was linked to greater coherence in approaches in CMOs. The extent to which these different approaches were related to differences in quality of education, or indeed outcomes, are not clear, however. In the Netherlands, the inspectorate stated that when QA is delegated too much to schools in the trust, they perform less well, in particular on pupil support [60]. However, if QA is centralised at the trust level a key challenge for trusts is avoiding an approach that becomes overly bureaucratic [31].

Monitoring the performance of schools through use of data is seen as central to accountability within CMOs in the US [41] and is also an element mentioned by Hill [64] in his commentary on school improvement in Multi-Academy Trusts, by Ofsted [52], and by the Dutch PO-Raad [25] in their guidance. Conversely, lack of effective use of data was a factor in the failure of MATs in Evans’ [65] study in England. According to Kettlewell et al. [66], there is a risk that boards can become too inward-looking and insular if they are not themselves subject to effective monitoring and evaluation. External input may be helpful to this monitoring and evaluation process. All these findings suggest that internal accountability, the extent to which the trust holds itself accountable, is a potentially important factor in ensuring trust quality [6].

How much trusts need to systematise and standardise processes is a disputed question, with few clear answers from the literature. There is little clear evidence of a differential impact of more and less centralised trusts on student attainment [18]. Nonetheless, it has been a major point of discussion in policy and research [59]. In England, Menzies et al. [67] distinguished between a cluster of trusts that provided broad autonomy to school leaders and one where trusts aimed to standardise approaches to pedagogy and curriculum. In the latter, standardisation was achieved in two distinct ways. In a first group, a central team imposed specific processes and practices across the trust. In a second, convergence was achieved through collaboration and mutual learning. Hopkins [42] recommended that trusts systematise approaches to teaching and learning across the trust, while Hill [64] stated that trusts need to allow adaptation to context in their schools while applying a number of non-negotiables in back-office functions such as finance and HR systems. In Menzies et al.’s [67] study, greater standardisation appeared helpful for ensuring that low-performing schools in the trust improved but created a ‘glass ceiling’ for the highest-performing schools, which were less able to innovate. There is a tendency for trusts in England to show greater levels of standardisation over time, as they have been encouraged to do so as a matter of government and inspection policy [68]. Nevertheless, looking at 23 case studies it was still the case that only 3 fully standardised across curriculum and pedagogy, with 7 allowing full autonomy to all their schools. There tended to be greater alignment and standardisation in pedagogy than curriculum [69]. Neri et al. [18] found that trusts where there is greater alignment in educational/professional background between trustees/governors and school managers are more likely to decentralise. They relate these differences to principal–agent theory, wherein the key difference between whether trusts centralise or decentralise lies in information asymmetry, i.e., trusts will decentralise when information is most likely to reside in the local schools rather than at the centre.

Comparing the four countries, in Sweden there is a tendency towards standardisation of approaches to support the brand of the chain within the educational marketplace, also often seen when these chains expand abroad [70]. In the Netherlands, in contrast, there tends to be little standardisation and a lot of autonomy for schools within a trust. In England, Ofsted [71] reported significant variance in the extent to which trusts directed policies and practices in areas such as teaching and learning and curriculum. The same appears to be the case in the US, though the extent to which CMOs prescribe curriculum and teaching appears, on average, greater than in England [31]. This does not necessarily imply that curriculum in charter schools is more innovative, however. Wohlstetter et al. [32], in an early study, found that charter schools innovated primarily in instructional delivery, while curriculum stayed quite traditional.

5.3. Impact on the Trust as a Network

At the network level, the emphasis has been on:

-

the role of collaboration,

-

the role of the governing board or trustees, and

-

a shared vision across the trust.

Collaboration is frequently seen as a key lever for school improvement, and one that trusts are well placed to use. Being able to collaborate as part of a network of schools is seen by many school leaders as one of the main benefits of being part of a trust. Getting schools to work well together is seen as beneficial, though conditions need to be in place to ensure that this is carried out effectively [64,71,72]. This means ensuring that there needs to be sufficient capacity within the trust to engage in collaboration, that collaboration must based on clear goals and desired outcomes, and that a culture of trust should prevail [19]. The extent to which such network effects actually materialise is disputed [72]. While some studies suggest positive network effects in trusts [57], and an earlier study of two CMOs and school districts suggested that information and good practice were shared more easily in the smaller and less bureaucratic trusts than in the larger school districts [73], a more recent review of evidence on charter schools in the US saw only very limited evidence of network effects, albeit in a field of which there isn’t much research [23].

Effective governance of the trust as a whole is seen as key to ensuring quality, and thus to reducing poor performance, and has been the focus of several studies. As part of quality of governance, there has been research into the experience and skills of governors and trustees, whose expertise in a range of areas, including educational quality, finance, legal, and HR is essential to the effective functioning of the trust [25,66,74,75]. In an interesting MBA dissertation on why Multi-Academy Trusts fail based on interviews with CEOs, Evans [65] found that the behaviour, skills, and motivation of trustees and board members were the key factors determining the success or failure of trusts. In some less successful trusts, governors and trustees did not fully understand their role in holding the CEO to account. The skills and motivation of the CEO were also central to success, with not all CEOs having made the transition from previous roles as heads successfully. The importance of challenge and support from the trust was also emphasised by school leaders interviewed by Ofsted [71] who, similar to Evans, found that not all governors fully understood the role and goals of the trust. The extent to which conflict within the board was helpful or not was studied by Heemskerk [76] in the Netherlands. The results suggest that while discussion and conflict that were task-related enhanced the effectiveness of the board, the opposite was true of relational conflict.

Kettlewell et al. [66] reported a mismatch between the skills CEOs thought trustees had and those trustees reported themselves, especially with respect to knowledge of education. Other characteristics of boards and trusts have also been studied. A study in the Netherlands looked at a range of factors in 131 boards which run 658 schools [6]. They measured self-perceived efficacy of the board (i.e., do board members feel qualified to contribute to educational quality), the extent to which the supervisory board was seen as an effective counterweight to the executive board, number of board members, and number of schools in the trust. Results showed that the supervisory board acting as a counterweight to the executive board was the only significant factor influencing pupil outcomes. There is further Dutch research showing a link between better inspection results and a better understanding of what is happening in schools (monitoring) by the supervisory board, and the Dutch inspectorate has also reported that boards better able to manage the quality of school leadership obtain improved outcomes [35]. The Dutch inspectorate has also stressed the importance of boards being open to feedback and disagreement [60].

In their study of CMOs and local districts that had successfully improved their schools, Kelly et al. [41] pointed to unified governance and a clear vision that aligns school practices as characterising successful CMOs. Similarly, in their study of Stichtingen in the Netherlands, Hooge et al. [77] found a shared vision and coherence of governance and policy to be among the main factors identified by trust and school leaders in effective trusts. These, in turn, lead to collective responsibility and collaboration across the trust. In Greany’s [8] case studies of MATs, he found that the most effective trusts had a clear vision underpinned by shared values and culture. In those trusts, staff knew and were able to articulate the vision of the trust. This, in turn, helped to develop shared values. Extensive communication helped staff to know and understand the vision. The importance of shared vision and good governance may also be related to the finding from England that small- and medium-sized trusts outperform larger ones in terms of student attainment [78], though research in the US shows a more positive impact of larger CMOs [33].

Research in England has looked at the strategic role of trustees. Baxter [79], using case study research, pointed to the key role of strategy, but also to some confusion about what is actually meant by this at the trust level. Trustees tend to leave key aspects of strategy, such as horizon scanning, primarily to the trust CEO [80]. Constantinides [57] likewise found that strategy at the trust level is largely set out by the CEO rather than by trustees.

5.4. Impact on the Community

While involvement with the community is often said to be a key part of the civic role of trusts, there is very little research evidence on their impact on communities. The role of the trust as a community anchor is particularly stressed in the Dutch literature, for example in Honingh et al.’s [35] review of effective practice in board governance. A code of practice drawn up for Dutch primary boards specifically mentions the need to look beyond the trust and to the national interest [81]. Strong links with the local community and a proactive orientation to working with stakeholders were reported by the Dutch inspectorate as characteristics of effective trust boards [60].

A moral imperative to provide pupils with the best possible education was a major driver for trusts according to trust and school leaders interviewed for Ofsted’s [71] study. Locally based trusts saw their role as providing more continuous joined up provision in their locality. By contrast, a loss of focus on moral purpose was sometimes evident in failing trusts in Evans’ [65] study, especially where a focus on growth became all-encompassing. In the English context, while some trusts deliberately engage with their communities, there are others in which relationships have been conflictual, with parents at times protesting trust policies [82].

Learning from other schools outside the trust was mentioned as a characteristic of high-performing trusts by Hill [64] and Kelly et al. [41]. The latter also stressed that the successful CMOs they studied created strong links with external stakeholders. This is also the case for networks more generally, where there is clear evidence of the need to remain open to external stakeholders and influences to avoid becoming overly internally focussed [19].

There is some evidence from Sweden that this public commitment may be lessened due to the structure of the system; in this case, the emphasis on for-profit operations leads to a focus on extraneous factors that do not necessarily improve education [9]. This is clearly also a risk in countries such as England and the US, where a system of competition between schools may limit the extent to which schools belonging to different trusts in a particular area collaborate with one another. Greany [83], studying trusts in five different locations, found widely differing responses, with the new trust structures providing both opportunities for new and emerging actors to work together, as well as incentives for fragmentation and division. There is further evidence from England that trusts may lead to a less cohesive local education system [68], especially where trusts are in competition with one another and with other non-trust schools [84]. Interestingly, when forming trusts, schools often do not do so with schools with which they have previously collaborated. This again appears related to competition between schools in local areas, and may exacerbate local fragmentation, as over time schools in a trust may tend to work more with each other and less with local schools outside the trust [85]. As well as these negative impacts of competition, there is, however, also some evidence from Sweden in particular of positive effects on attainment in localities in which school chains have opened provision and compete with local authority-led schools [86].

6. Discussion

One key finding of this review is the relative paucity of research and evidence on which factors improve trust quality. This limits the extent of the conclusions we can draw. Nevertheless, there are some findings that tentatively point us in specific directions in relation to the field.

If we look at Ehren and Perryman’s [27] framework, we see that evidence is strongest on impact on the organisation, especially in relation to the role of HR, relationships within the trust, and the importance of quality assurance. There is evidence at the network level on the qualities required from trustees and governors, effective forms of collaboration, and a shared vision across the trust. Trusts can also impact on quality of education through professional development of teachers, developing consistent approaches, and sharing knowledge. Evidence of how trusts can enhance their impact in the community is limited, however. Together, these findings suggest the start of a theory of change for how trusts can improve quality, which could in turn lead to enhanced educational outcomes, and better outcomes for the communities they serve. This is illustrated in Figure 1 below.

The model depicted above represents the best guess at a theory of change that we can currently take from the evidence reviewed. The model suggests that quality of governance and governors, along with high-quality central leadership and management, will help develop strategy and a shared vision across the trust, as well as effective processes and support. This refers to back-office processes, such as HR and finance, as well as more direct support through, for example, the development of shared approaches to behaviour management or school culture. The literature reviewed here suggests that these central elements have four main outcomes in terms of trust quality: better recruitment and retention of staff, effective quality assurance mechanisms, high-quality CPD, and collaboration and knowledge sharing between schools, leaders, and teachers. These, in turn, lead to high-quality leadership and teaching and learning. There are, of course, contextual factors that influence this model, for example the fit between centre and schools and the influence of information asymmetry on the extent of centralisation [18].

An important caveat is the relatively limited evidence base we are drawing on. This means that we may well be missing important elements. The area of community relations and impact is largely missing from the literature. The same is true for factors such as financial management or multi-agency work. Similarly, it may well be that there are relationships that should be added to the model, but which are not present in the literature as it currently exists, or that some of the relationships posited may, upon further research, turn out to be overstated or absent.

While the above model points to the similarities between systems found in the literature, there are also some key differences. One of these relates to the differing emphasis on collaboration or centralisation as a core mechanism for school improvement. In the Dutch and English literature, there is a strong emphasis on the former, while in the US and, to a lesser extent, Sweden, centralised approaches are highlighted more frequently. This difference may result from the context in which school trusts were set up. In the US, the charter school movement was often related to particular approaches to pedagogy, such as the so-called ‘no excuses charters’. In many cases CMOs consist of entirely new schools that were explicitly set up to develop a particular approach [87,88]. In the Netherlands and England, the governments’ emphasis on conversion and incorporation of existing schools into these groupings has led to more diversity and reliance on the sharing of existing practice rather than the development of new top-down models [60,69]. In Sweden, the for-profit nature of many trusts has resulted in an emphasis on a coherent brand parents can buy into [70,89].

Quality of governance is another area in which emphasis differs, with a particular concern around this evident in the Dutch literature, which is strong on recommendations for governance. This is likely the result of the relatively limited development of the governance structure in the Netherlands which consists of a small number of people [35], as compared to that in the other countries reviewed. The Dutch literature also emphasises community work more, which may result from the fact that trusts in the Netherlands are all local, whereas in the other countries both local and national trusts exist. These differences in emphasis point to the influence of national policies in the way they encourage specific structures or forms of governance, which may in turn significantly influence the structure and functioning of trusts as they grow within a particular context. They also limit the extent to which any model is generalisable across contexts. In addition, as mentioned above, the evidence base itself differs between countries. One could, of course, draw more broadly on the literature on school-to-school collaboration [20,90] to draw lessons for trust structures. However, the specific nature of trusts, which are both hierarchical delegation structures and (in the best cases) networks, makes this a less clear fit than might be the case.

It is a bit of a cliché that a paper reviewing research evidence will always end with a call for more research. But that is what we are going to do here. Trusts and other networked forms of governance have grown strongly in recent years in a range of countries. This inevitably leads us to two key questions: one (which we have not covered in this paper) is whether this development is a good thing in the first place. The second, which we have, is how such groupings of schools can be made to work as effectively as possible. There is a (albeit limited) body of research on the first question, which is somewhat inconclusive. As we have seen, there is only limited evidence on the second. This is clearly an issue for education systems that have, or are, going down this route. We hope this review will give leaders in trusts and the schools that are part of them some pointers, but there are clear lacuna in the evidence base, which we hope further studies can help resolve.