Not every public sector is a field: evidence from the recent overhaul of the English NHS

ABSTRACT A structural interpretation of institutionalism has become dominant in public management research. Yet, studies tend to assume an institutional-level phenomenon without specifying how an organizational field was identified or whether structural characteristics can indeed be found in the organizational population studied. This lacuna is illustrated by exploring the structural interpretation of the field construct in the case of the recent overhaul of English primary care. Findings demonstrate the need for a more robust application of institutionalism in empirical research. Possible research problems for public management and a future research agenda based on a more relational approach to fields are discussed.


Introduction
Institutional analysis of public policy and management reform has gained momentum in recent years with an increasing number of studies demonstrating Ferris and Tang (1993) assertion that 'Institutions do matter' (9) for public administration. Drawing on an established sociology of organizations and political science, public administration commentators have applied institutionalism, mainly in its structural interpretation (Scott 2001), to the study of historical public sector developments. Management reforms in England, Australia, and the United States, for example, have been addressed (entirely or partially) as happening within or across organizational fields and the framework has been productively applied to understanding the dynamics and processes of public sector organizational change (Botterill 2011;Kickert and van der Meer 2011;Ongaro 2013;Mizrahi and Tevet 2014;Ho, Alfred, Tat-Kei, and Tobin 2015).
While gaining hold in public management research, the specific application of a structural interpretation of institutional theories has been problematized in broader management and organization literature. Scholars suggest that institutional theories tend to concentrate on institutional dynamics while failing to consider the heterogeneity inherent in organizational populations (Greenwood et al. 2011) and that institutional approaches might fail to appreciate critical perspectives (Suddaby 2015;Willmott 2015). More specifically, a recent review by Zietsma et al. (2017) highlighted how studies tend to conceptualize the existence of an organizational field without specifying how a field was identified empirically nor explicitly addressing the extent to which field characteristics can be found in the domain under study.
Considering these critiques of institutional approaches to the study of organizational change, this paper observes how the same problem is prevalent in public management studies and aims to discuss possible consequences of a loose application of the field construct to heterogeneous organizational populations in public management research.
Drawing on the review by Zietsma et al. (2017) and their four criteria for the empirical identification of an institutionally defined organizational field (i.e. status hierarchies, bounded network structure, shared meaning and practices, and shared identity), we explore the recent, radical overhaul of the English National Health Service (NHS) by legislation enacted in 2012 and the fragmentation of healthcare commissioning, provision, and planning it initiated. The current state of English primary care planning and commissioning provides an excellent temporal point to examine the application of institutional analysis. First, the NHS has been predominantly theorized in the literature as an institutional-level phenomena and is frequently approached as a highly regarded institution of the British state, both in academic as in political discourses. Second, as will be discussed, the current complex governance structures found in England pose significant challenges to the successful empirical identification of an institutionally defined phenomenon and thus provide an illuminating case for a call to reorient research concerns. This paper therefore contributes to understandings of current changes in healthcare organization in England and to understandings of public policy reforms more generally. By showing how fragmentation introduced in 2012 calls into question claims that the healthcare system can be analysed as a coherent organizational field, our analysis suggests that a process of de-institutionalization (Oliver 1992) might be taking shape in England while also pointing to the possible future obstacles to this process. Our focus upon the more general problem of under-specification of the relevant empirical population in public management research means that our analysis is relevant to a range of public sector contexts in different countries.
Furthermore, by taking an empirical case study and examining the pitfalls of unwittingly applying the theoretical construct of an organizational field, this paper contributes to public management and administration scholarship by demonstrating the need for a more careful, nuanced use of the framework in future research of management reform. In particular, we discuss the research problems which may arise when fields are under-specified including problems of generalization, under-utilization of the institutional framework, and the possibility of naturalizing political discourses. These problems, together with the detailed survey of current organizational populations in the English NHS also lead us to suggest a possible research reorientation towards a more relational understanding of fields.
The argument comprises three sections and a discussion of findings. First, we highlight the centrality of the organizational field concept for institutional studies and present Zietsma et al.'s (2017) four criteria for the empirical identification of a field. Second, we survey the previous use of institutional approaches in studies of the NHS and suggest that previous studies fail to specify what they mean by organizational field. Third, we examine the current state of the English NHS and question the empirical identification of an organizational field using the explicit identification criteria. A brief historical outline of the structure of the NHS precedes an examination of recent changes following the 2012 Health and Social Care Act (HSCA12). Analysis focuses on newly established Clinical Commissioning Groups (CCGs), their organizational structures and practices. This section demonstrates problems inherent in misidentifying an organizational field when studying a particular public sector reform. The paper concludes with a discussion of the findings, their implications for future policy and research into public organization and management.

The centrality of fields for institutional analysis
New Institutionalism is a diverse, 'broad tent' conceptual framework widely utilized in both the public administration and management literatures (Lowndes 1996). In public administration scholarship, while the 1990s and early 2000s might be characterized by a theoretically oriented debate about the future shape and direction of institutional research (see Blom-Hansen 1997;Ferris and Tang 1993;Lotte, Asbjorn, and Sorensen 2004;Merino 1996) the last decade has seen a growing number of empirical studies using the framework to facilitate understanding of historical public sector developments globally (e.g. Botterill 2011; Kickert and van der Meer 2011;Ongaro 2013;Mizrahi and Tevet 2014;Ho, Alfred, Tat-Kei, and Tobin 2015;Pillay, Reddy, and Morgan 2017).
Although varied, the dominant strand of institutionalism applied in studies of public administration reform is a political-science influenced interpretation (Guy 2011) based on institutional theories of isomorphism and legitimacy (DiMaggio and Powell 1983;Meyer and Rowan 1977) whereby formal structures of bureaucracy are understood as 'thickly' institutionalized, themselves adaptive products, responsive to environmental and relational influences (Selznick 1992(Selznick , 1996. In line with this thinking, changes to public organizations or the behaviour of individual actors within them thus cannot be explained by efficiency, self-interest, or formal governance arrangements alone. An institutional analysis: 'comprises [. . .] an interest in institutions as independent variables, a turn toward cognitive and cultural explanations, and an interest in properties of supraindividual units of analysis that cannot be reduced to aggregations or direct consequences of individual's attributes or motives' (Friedland and Alford 1991, 8). This leads to a focus on organizational context and associated pressures towards conformity and uniformity, or the 'exogenous nature of change, which emanates from the realm of ideas and legitimacy' (Greenwood and Hinings 1996) while also paying attention to contextual structures and dynamics within and across organizations (e.g. Fareed et al. 2015;Pillay, Reddy, and Morgan 2017).
This shift in approach, amounting to accepting that organizational phenomena cannot be adequately understood in terms of efficiency and instrumental action and that economic and technological variables do not solely explain organizational developments, is dependent on the theoretical construct of an organizational field. Termed 'institutional field,' 'societal sector,' or 'institutional environment,' the construct of organizational field has become the accepted term for 'the constellation of actors that comprise this central organizing unit' (Wooten and Hoffman 2016, 131). That is to say, for us to conceptualize our empirical terrain as having to do with a field-level phenomena, something 'above' or 'outside' the mere technological and economic variables, there is a need to speak of, identify, and analyse an 'organising unit' which is at a different levela field level. Its centrality is evident in the theory's origins with, e.g. Warren (1967) indicating 'the need for research focusing deliberately on the "field" within which organisations interact' (397) and Lowndes (1996) presenting the field level as the feature distinguishing new from 'old' institutionalism. It continues to feature as a principal concern in more contemporary works with, e.g. Tina, Goodstein, and Richard Scott (2002) seeing it as 'the most useful level of analysis to emerge in recent years' (13) and thus 'the level at which most of [their] analysis takes place' (17). Recently, upon reviewing over 100 studies drawing on institutional approaches, Zietsma et al. (2017) concluded that 'institutional fields are presumed to be the predominant source of pressures for institutional conformity and the site of institutional embeddedness, [and] is the core idea of institutional theory' (5).

A problem of specification in field studies
The application of institutional theory to empirical studies has been a welcome and important development for public management research. Nonetheless, this framework has undergone sustained critique in the broader management literature, mainly due to a failure of studies to address the specificities of the organizational population being theorized and thus to convincingly identify the existence of a field-level phenomenon. That is to say, the main explanatory construct of the theory tends to be taken for granted in empirical research, with Zietsma et al. (2017) noting that: 'in the empirical literature on fields [there] is very little specification of the kind of field being studied, with the assumption that a field is a field is a field' (12).
This can be seen in public management studies, for example, in Frumkin and Galaskiewicz (2004), where an empirical comparison uses an institutional framework to argue that 'public sector organisations' are different than 'business and non-profit organisations' in how they respond to mimetic, normative, and coercive institutional pressures. Such institutional dynamics are invoked while empirical description remains highly abstracted, refraining from addressing the particularity of the organizational population in question, its boundaries or any other empirical specifications that could demonstrate the existence of a field-level phenomenon. As another example, in a study of New Labour's reform of local government, Lowndes and Wilson (2003) apply 'new institutionalism tools' without specifying the empirical scope of the organizational field under study. While demonstrating shifts in New Labour's approach to public management reform, the analysis does not address the question of the organizational population on which or in which institutional dynamics are said to play out. Adler and Kwon (2013) explore the implementation and impact of clinical guidelines on the medical profession, explicitly considering individual, organizational, and field-level influences. However, what constitutes the relevant field is unspecified. Professional associations, accountability to external bodies, and market forces are all referenced as 'field-level forces,' but the population and boundaries of the field remain unspecified.
With DiMaggio and Powell (1983, 148) arguing that: 'the structure of an organisational field cannot be determined a priori but must be defined on the basis of empirical investigation,' the lack of systematic consideration of characteristics found in a set of activities subsequently analysed as a field potentially impinges on our understanding of the phenomena under analysis, here public management and the public sector more broadly. In particular, lack of elaboration of exactly what is the 'field' under consideration might mean that the field construct is misapplied to empirical material and some of our theorizing is not as robust as it should or could be.
Addressing this issue in the broader management literature, and building upon DiMaggio and Powell (1983) assertion that fields 'only exist to the extent that they are institutionally defined, ' Zietsma et al. (2017) propose four elements to be identified empirically in studies aiming to utilize the field construct in an analytically meaningful way. Those are as follows: (1) The existence of status hierarchies among actors: identifiable empirically by the emergence of inter-organizational patterns which are definite; (2) The existence of a bounded network structure: identifiable empirically by the extent of increased interaction between organizations in the domain studied; (3) The existence of shared meanings and practices: identifiable empirically by an increase in the information load organizations are confronted with; and (4) The existence of shared identity: identifiable empirically by a furthering of mutual awareness between organizations.
In the following, we examine the appropriateness of applying the field construct according to this framework in the case of a particular public management reform. We use the latest overhaul of the English NHS and discuss challenges to claiming that this public sector environment can be meaningfully analysed as a field-level phenomenon. Before we do that, we turn to discuss this lacuna of underspecification as it manifests in past studies.

Institutional analysis of the English NHS
Institutional analysis of healthcare reform in England has developed considerably in a short time (Lockett et al. 2012;McDonald et al. 2013). Such explanations are now widespread, drawing on diverse data from ethnographies of healthcare practices (Staniland 2009) to large-sample factor analysis (Harris et al. 2014). Institutional explanations shed light on pressing issues facing healthcare in England, specifically relating to organizational change and managerial reform as they provide accounts of observed developments which sidestep individualistic conceptualizations of action (Currie and Suhomlinova 2006;Macfarlane et al. 2011;Checkland et al. 2012). Currie and Suhomlinova (2006), for example, eschew atomistic explanations to show how knowledge sharing across professional boundaries is hampered not by individuals' traits, capacities, or motivations but by different occupational groupings' reactions to institutional pressures. Macfarlane et al. (2011) argue that throughout extensive structural changes, there endured an 'NHS brand' encompassing values, norms, and ways of working (see Osipovič et al. (2016) for an alternative view drawing on an institutional framework). Checkland et al. (2012) documented how the healthcare commissioning and provision split, which relies on commissioners' discretion, is hindered not by incompetent management, but by misalignment between culturalcognitive assumptions prevalent in the NHS and their interaction with new regulatory structures. The analysis favours an institutional explanation seeing managers as constrained by cognitive frames of 'what is commissioning/commissioned,' highlighting the decisive influence of shared, institutionalized meanings and approaches to practice. This has also been shown in the assessment of dental services (Harris et al. 2014) and the adoption of health information technology (Currie and Guah 2007;Fareed et al. 2015). We enthusiastically support these treatments of public sector reform which broaden our understanding and take us away from narrower, economistic and instrumentalist explanations. Problems arise, however, when claims are made on the basis of an organizational field, the existence of which remains implicit. For one, it allows for considerable variations. In the case of health services in England, scholars have favoured widely varying scopes for the relevant field, ranging from a country-wide healthcare system, through particular professional groupings, to parts of a healthcare system, without always elaborating on the characteristics or dynamics of their chosen domain. Currie and Suhomlinova (2006) speak of a broad 'healthcare organisational field' comprising a whole country, including higher education alongside NHS organizations. Instead of examining individual action or the action of individual organizations, they focus on this broad field in order to consider 'specialised segments of the field' and explore both 'field-wide similarities' and 'intra-field differences' relating to institutional pressures and knowledge sharing. Focusing on the narrower case of dental care, Harris et al. (2014) make a similar move; drawing on Meyer (2010) and Scott (2001), they depict dentists as 'acting from within the dental practice, itself structured as an organisation set in a wider organisational field of dental health care provision, across which are woven multiple [. . .] institutional logics' (82). McDonald et al. (2013) refer to a 'general medical practice' and 'as opposed to medicine more generally, [conceptualise it] as a field, with its own set of practices' (48) to maintain the focus of analysis on changes within the field.
Alongside population variability, this lack of specification means the field construct remains underutilized in healthcare studies. Considering the specificities of the domain can further improve our understanding of the nature of interactions between a group of organizations, including their level of homogeneity or heterogeneity, the purpose of their interaction, and the structure of relations among them. In studies of the NHS, much emphasis has been given to concerns relating to criteria 3 and 4, perhaps at the expense of the others. There has been much work on the existence of shared meanings and practice between organizations operating in the broadly defined English healthcare field, while studies examining the status hierarchies (criterion 1) in the NHS or those mapping the boundedness of organizational networks from an institutional perspective (criterion 2) are less prevalent. Studies of the NHS considering the latter do exist (see, e.g. Ferlie and Pettigrew 1996; Allen 2013), but not using an institutional approach. This leaves part of the institutional theory framework underutilized.
Case study context: the healthcare system in England Our analytical starting point is that an implicit, taken-for-granted assumption of the existence of an organizational field in healthcare poses a research problem. With the NHS undergoing a period of turbulent change following the UK Coalition Government's (2010-2015) HSCA12, the features of the current system throw the problem of field into sharp relief. As it was empirical conditions we encountered that led us to this conceptual argument, this paper considers to what extent these criteria for the existence of an institutionally defined organizational field apply in the current English NHS. Examining the existence of a field in the past remains outside the scope of this paper. We move to provide a brief historical overview of the English NHS (n.b. responsibility for the NHS was devolved to other United Kingdom countries in 1999) and explore key elements of the HSCA12 managerial reform.

Before the 2012 reorganization
The NHS was established in 1948 offering healthcare free at the point of delivery, funded through general taxation, to the UK populace. A tripartite structurecomprising General Practice (primary care/family practice), Hospitals, and Community Health (provided by local authorities)was established, with each component enjoying considerable autonomy within its domain. A significant structural reorganization in 1974 introduced an additional tier to the system in the form of Area Health Authorities (later, District and Regional). These 90 organizations, coterminous with local government boundaries, were intended to unify the system by taking on planning and provision responsibilities for most community health services alongside oversight of most NHS hospitals.
In 1990, a split between purchasers and providers of services was introduced (Flynn and Williams 1997). Existing healthcare providers became NHS Trusts, each with an independent board, which sold its services to NHS purchasers in a quasimarket. The initial purchasers included District Health Authorities, and GP Fundholders (groups of GPs that volunteered to take on virtual budgets to pay for their patients' services). In the early 2000s, further reorganization replaced these purchasers with Primary Care Trusts (PCTs), responsible for commissioning most services. The policy aspiration was for PCTs to commission care from a mixed economy of competitive providers, including quasi-independent NHS Foundation Trusts (Allen et al. 2012). Strategic Health Authorities were created to oversee the work of PCTs and hold them to account.

Post Health and Social Care Act 2012
The HCSA12 was a significant reorganization including: changes to procurement and competition regulations; revised governance and statutory arrangements; and a shift towards a more local, clinically led commissioning function in the form of CCGs (Greer, Jarman, and Azorsky 2014;Timmins 2012).
The HSCA12 introduced new organizations and altered the form and remit of existing ones. The role of the Department of Health was changed to one which focused on strategy, with responsibility for NHS operation and oversight given to a new executive, non-departmental 'arm's-length' public body: NHS England (NHSE). NHSE also took on responsibility for commissioning specialized and primary care services. The Care Quality Commission (CQC) was given responsibility for care quality assessment and provider licensing. Monitor (since subsumed into NHS Improvement), an existing organization responsible for regulating Foundation Trust Hospitals, had its role extended to market regulator.
At the centre of this new system are CCGs. Replacing Primary Care Trusts (PCTs), CCGs are membership organizations (composed of GP practices) which took statutory responsibility for approximately two thirds of the NHS budget (NHS England 2017) in April 2013. There are currently 207 CCGs, and they commission the majority of care for their populations. NHSE was created with 27 Local Area Teams across four regions. A re-organization in 2015 saw Area Teams abolished to reduce management costs. NHSE retained its four regions but now operates subregional teams, with an overall reduction in staff. This new structure exhibits a 'scooped out' middle that stands in contrast to the more visible hierarchy of PCTs and the Strategic Health Authorities that previously sat 'above' them.
Initially, CCGs' role was the commissioning of secondary and community care. Responsibility for commissioning low-volume, high-cost specialized services was NHSE's, along with commissioning primary care services from GPs. More recently, NHSE has sought to delegate responsibility for commissioning of primary care to CCGs. There is an inherent conflict of interest in involving GP members in the commissioning of their own services, which can only be partially mitigated. Each CCG has adopted a form that reflects local context and the legacy of responses to previous policy driven reorganizations in each area. Even the roles clinicians play in CCGs varies, together with how their 'added value' is understood and perceived (McDermott et al. 2015). CCGs vary in their population size, affiliation with local authorities, governance structures, operational strategy, and the managerial language used to describe these. As membership organizations, CCGs also differ in the relationships they have with their members, how they monitor their performance, and their healthcare commissioning priorities (Checkland et al. 2014). This raises questions about the commonality of meaning, expectations, and ways of doing which are at the core of an organizational field assumption.
Thus, the HSCA12 represents another watershed moment in NHS history. The resultant significant organizational and structural change suggests now is an appropriate time to closely evaluate the use of the organizational field concept in relation to the English NHS and by extension other healthcare and public administrative systems more broadly, with CCGs an appropriate focus given their statutory centrality to the Actif a field-level phenomenon is to be found in the English NHS, CCGs would be a pivotal element thereof.

An evolving organizational landscape
This organizational complexity continues to increase. NHSE has adopted a role of de facto policy maker and proposed the creation of New Care Models (NHS England 2014a), defined as locally specific inter-organizational arrangements to facilitate care planning and provision across broader geographical areas and integration across primary, secondary, and community care boundaries. Launched in January 2015, there are currently fifty New Care Model 'Vanguards,' each identified as one of five loosely specified types (e.g. Primary and Acute Care System). A key inspiration for the development of some Vanguards is US Accountable Care Organizations where multiple care providers work together, with a fixed budget, to provide services for a specific patient population.
In 2015, NHSE announced the creation of Sustainability and Transformation Partnerships (STPs). Presented as a response to perceived failures of the HSCA12, especially its core quasi-market dynamics (Ham et al. 2015), NHSE now prefers organizations across the healthcare system 'come together' and develop a joint approach to care planning and delivery in order to access a Sustainability and Transformation Fund (England 2016c). STPs are not formal organizations. Although the plans have implications for de facto governance structures and accountability relations, these are not explicated in legal and regulatory arrangements. STPs vary considerably; some include as many as twelve different CCGs while others include one, and some have a foot-print crossing local authority boundaries while others are coterminous with one. Another inter-organizational arrangement that has proliferated post-HSCA12 is the GP Federation or super-partnership (England 2016b), where groups of GP practices establish a joint working arrangement as service providers (Royal College of General Practitioners 2008). Some Federations have become so large that they cross CCG boundaries.
Findings: Challenges to the identification of an organizational field The above description enables us to consider how the NHS might fit the theoretical framework demanded by the field construct. As explored below, the introduction of member-led CCGs, non-statutory STPs, and the accompanying complexification of organizational geographies, accountabilities, and regulatory responsibilities make it challenging to characterize the current state of healthcare organization in England as a field-level phenomenon according to the four criteria (the following is summarized in Table 1).

Identifying status and hierarchy in cases of complex governance
The dissolution of a more pronounced, hierarchical governance with the abolishment of Strategic Health Authorities has yielded a complex situation in which organizations find themselves in conflicting positions with regard to their organizational functions (Checkland, Dam, et al. 2018). This generates fuzziness in clearly situating oneself in field positions and thus ambiguity in status hierarchies among actors, challenging the identification of criterion 1. First, the relations between the Department of Health, NHSE, CCGs, GPs, provider Trusts (encompassing hospital care, ambulance services, mental health services, community care), and regulatory organizations (i.e. CQC, NHS Improvement) are contingent on organizational function and task. The HSCA12 introduced NHSE as the statutory body responsible for overseeing service delivery. Although it might be conceived of as a regulatory, state organization and thus 'above' other organizations in the hierarchy, in practice NHSE, while public, is not part of Government. NHSE is not bound by the Civil Service Code and sits outside Government's formal structure (Greer, Jarman, and Azorsky 2014) and thus poses difficulties for a straightforward placing within a multi-level analysis. As an example, although NHSE holds CCGs to account, CCGs also have some responsibility for holding NHSE to account for its conduct of the commissioning of specialized services.
Second, the problem of clearly identifying actors' positionsin this case, a clear demarcation between regulatory, state actors, and those outside formal state structuresbecomes almost insurmountable when considering the commissioning of primary care. NHSE has delegated the operational side of primary care commissioning to CCGs, with CCGs assuming varying levels of responsibility (England 2014b). Consequently, there are no clear boundaries between what would be considered a regulatory agency and an 'extra-state' entity as commissioning responsibility for primary healthcare becomes intermingled between NHSE and CCGs to varying degrees (Checkland, McDermott, et al. 2018).
Third, as membership organizations CCGs are governed by their GP members while also acting as their commissioners. Thus, the regulatory function, so central to institutional explanations, is fragmented across the supposed organizational field making it problematic to designate clear chains of above and below in terms of accountability and command. For example, CCGs are encouraged to establish their own Local Incentive Schemes, replacing centrally administered frameworks aimed at ensuring quality and standardization across primary care such as the Quality and Outcomes Framework (QOF) or Directed Enhanced Services (DES). These changes are meant to allow CCGs to manage services in a locally tailored manner (thus placing CCGs 'above' GP practices), but actual decisions regarding practice development such as mergers or property development remain under the discretion of individual GPs via their contractual commitments (thus placing GPs 'above' CCGs). This is also the case with primary care performance monitoring, which is fragmented between NHSE, CCGs, and CQC. Although CCGs are statutorily obligated to assure NHSE on the continuous improvement of primary care services, CCGs only performance monitor practices, while quality monitoring of individual GPs is reserved for NHSE. At the same time, CQC monitors other practice performance aspects, such as the quality of management and premises.

Identifying increased interaction when networks are unstructured
This organizational complexity also affects the ability to empirically identify criterion 2, namely increased interaction as signalling a bounded network structure. The construct of an organizational field includes organizations bound together not only by geography or goals, but also sharing of similar services, products, suppliers, users, regulatory agencies, and others (DiMaggio and Powell 1983). Such 'similarity' or assumed homogeneity is precisely at issue in the current state of the English NHS. Regardless of how statutory responsibilities, accountabilities, control, or funding are fragmented across hierarchical levels (as discussed above), changes in NHS structures also make it challenging to identify a community of organizations in an analytically meaningful way.
First, under current arrangements, it is challenging to see CCGs as part of a bounded network structure comprising a countrywide primary care field. Rather, healthcare planning and delivery appears structured more as silo fields across the country, putting criterion 2 in question. As Wooten and Hoffman (2016) note, when considering different types of work performed by organizations, an analytical language of 'buyer,' 'supplier,' or 'regulatory agency' does not provide sufficient institutional explanation. Healthcare in England post-2012 exemplifies an empirical case in which defining these roles is particularly problematic. CCGs might bear the same designation or appear to administer the same services, but in fact vary considerably. For example, under NHSE guidance, CCGs' governance structures are not predetermined and each organization addresses healthcare commissioning differently. With minimal unifying governance structures, and dependent on regional partnerships, CCGs have developed piecemeal, answering local and contingent needs. This makes it particularly difficult to see all CCGs as the 'same' organizations in a cross-country bounded network.
Second, although all CCGs serve 'patients,' these vary on clinical, demographic, and socio-economic profiles and thus no unified 'demand side' population exists. Similarly, although all CCGs have some degree of involvement in commissioning 'primary care,' in practice this has led to the establishment of various local incentive schemes with different ways of assuring delivery and through the commissioning of different providers, thus rendering a 'supply side' categorization also problematic. Consequently, a field perspective aimed at securing the ability to analyse how organizations enact their environment and are simultaneously acted upon by the same environment (Wooten and Hoffman 2016) becomes questionable when dealing with the English NHS, as defining what Fligstein (1997) referred to as a 'local social order' becomes almost CCG specific. If indeed field-level processes focus on strategic action through framing, in the hope of finding resonance with other field members Fligstein and Doug 2015, CCGs' role as broker-like organizations, investing resources in engagement activities to translate, explain, and communicate healthcare policies and schemes developed by NHSE to their members, testifies to the existence of a local, CCG-specific institutional environment.
These issues raise questions over the existence of a single, unified, and bounded country-wide network structure. As CCGs exhibit intensified interaction with local partners in a more partial, place-based capacity, their interaction with other actors in the broad healthcare system is likely to be limited.
Identifying shared meaning and practices when there is no 'best practice' As discussed, the bulk of healthcare research in England using institutional approaches has focused upon criterion 3 and the existence of shared meaning and practices across organizations. Nonetheless, the HSCA12 and subsequent developments call for a careful re-evaluation of the empirical terrain. The use of the organizational field construct is meant to signal the existence of a more or less defined space from which organizations draw institutionalized 'social facts,' the recognition of which grants them legitimacy when determining appropriate action (Meyer and Rowan 1977). These organizations, it is posited, share a common meaning system and interact more frequently with each other than other organizations (Scott 2001). Authors have already commented on the difficulty of identifying shared meaning empirically, especially when reform is perpetual (Ferlie, Fitzgerald, and Pettigrew 1996). We argue that CCGs position at the heart of the healthcare system and the current shift towards STPs, GP Federations, and Vanguards further exacerbate this possibility.
As illustrated in Figure 1, although a tree-like perspective might give the impression that CCGs are operating on the same organizational layer, CCGs' target patient populations are, in fact, constructed from a complex, and highly locally contingent, layered structure. CCGs' patient catchments might span multiple municipal geographies, they might include a GP Federation, could be negotiating with a Vanguard that itself might include practices outside the CCGs' original catchment, and they will need to consider local STPs' decisions.
Even where a unified approach seems to exist, such as with the seemingly standardized accountability mechanisms governing CCGs (articulated in the Improvement and Assurance Framework (IAF) (England 2016a), the accommodation of local contingencies by NHSE make it difficult to identify shared meanings and practices. In the IAF, for example, formalized accountability is defined as dependent  Figure 1. A tree-like perspective of the 'system' vs layered complexity of each CCGs spatial situation.
on the 'quality of relationships' (8) and left partly unspecified. Thus, even frameworks that apparently apply 'across the board,' are not providing straightforward, common definitions or operational instructions, further enhancing the development of locally sensible 'silo' arrangements. Although CCGs and STPs might be interested in what other CCGs and STPs are doing, adopting a fellow organization's plans does not necessarily serve as a legitimizer. As there is no clear 'best practice' and although CCGs and STPs develop their strategies together with local partners, in relation to specific healthcare providers, and in terms of their own resources and demands, scope for institutionalization is weak.

Identifying shared identity under conditions of competition
The issues above all highlight problems in the identification of shared identity among CCGs. The assertion of the empirical existence of shared identity among healthcare organizations becomes more problematic when considering the regulation of competition specified in the HSCA12. The Act's legislative changes have seen the NHS moving from a sector-specific regulation model to one based on competition law (Sanderson, Allen, and Dorota 2017). NHS providers and commissioners are now subject to regulation by NHS Improvement (economic regulator of the whole NHS), and by The Competition and Markets Authority. This has critical implications for the identification of criterion 4: shared identity in the organizational field. First, the shift towards a competition based regulation model has entrusted NHS Improvement with issuing and regulating provider licences subject to regulations on anti-competitive behaviour. Consequently, any attempt at mergers or joint ventures by NHS organizations (e.g. CCGs, hospital Trusts) can be scrutinized for their impact on competition. There are now a number of cases in which mergers or integration attempts that bring organizations in the NHS closer have been reviewed and rejected by NHS Improvement (Sanderson, Allen, and Dorota 2017). This suggests the existence of a regulatory structure acting to subvert institutionalized sharing of identity.
Second, in the post-HSCA12 healthcare system, NHS Trusts find themselves in direct competition with each other as they are asked to assume the behaviour of 'market actors' (Davies 2013). Competition laws are based on the advancement of self-interest and the exhibition of strategic behaviour known to fragment identities and work counter to cooperation. This has been observed to affect accountability arrangements and to shift the system away from the NHS' constitutional ethos (Davies 2013).
Third, the HSCA12 intensified the possibility of private involvement in healthcare. The mandating of open and competitive tendering for services commissioned by CCGs makes it easier for private companies to challenge decisions to award public bodies with contracts (Krachler and Greer 2015). Private providers can often claim lower costs because they are not required to provide the full range of services, such as emergency or urgent care, and are thus outbid traditional public providers. This further problematizes any straightforward identification of shared identities in the health system. Zietsma et al. (2017) requirement for empirical identification of criterion 4 is greater awareness of the organizations in a field. Current circumstances might indeed lead to further awareness, but it is not one based on relations of cooperation or reciprocity but of competition and self-interest. Although an organizational field can signify both common purpose and an arena of strategy and conflict (DiMaggio and Powell 1983), the further opening up of the NHS to private providers means that the identity of providers participating in an area is not pre-known before tendering happens, thus making it hard to claim an established arena exists. Under such conditions, the emergence of shared identity is questionable.

Discussion
Our argument calls for 'putting back organization' (Hirsch and Lounsbury 1997) into public management research as we emphasize the importance of demonstrating the integrity of the organizational field concept prior to analysis based upon it. This problematization rests on the degree of organizational diversity (Kondra and Hinings 1998) that exists within and between NHS organizations and the implication of this heterogeneity for institutional explanations and predictions. Our findings highlight how, when the criteria provided by Zietsma et al. (2017) are considered, the current landscape of healthcare planning, commissioning, and provision in England resists straightforward designation as a field-level phenomenon. We turn to discuss the implications of these findings for our theorizing about the English NHS and then more broadly for public management research, both in terms of the problems our analysis suggests as well as possible future research trajectories.

Deinstitutionalization of the English NHS?
Our 'failure' to empirically identify an organizational field speaks to current and future changes in the system. Although our analysis does not focus on the continuity or discontinuity of particular organizational practices, but rather on the possibility of theorizing primary care in England as a field-level phenomenon, there is evidence to suggest that a process of 'deinstitutionalisation' (Oliver 1992) has been taking place in the England NHS. Among the empirical indicators for antecedents of deinstitutionalization proposed by Oliver (1992), our findings point to the prevalence of both organizationallevel pressures (e.g. social pressures in the form of a decreasing of historical continuity and changing institutional rules and values, functional pressures in the form of increasing competition for resources or increasing technical specificity, and political pressures in the form of conflicting internal interests) as well as environment-level pressures (mainly, changing external dependencies and increasing structural disaggregation) for the erosion or displacement of institutionalized practices.
Supporting this deinstitutionalization view is the current direction of travel in healthcare policy in England towards consolidation of organizational structures. In this vain, the head of NHSE has suggested that STPs might garner more 'decision rights' and the authority to 'recommend changes to the configuration or governance of constituent statutory organisations,' including CCGs (Wickwave 2017). Examples of this are already emerging with, for example, CCGs in the Midlands working within the STP process to create an 'STP-wide' single acute and mental health contract (Thomas 2017) and eleven CCGs in the North of England working together under a 'joint commissioning committee' aligned with the STP (Gammie 2017). Both developments suggest a recognition of a need to move away from current fragmented, multi-layer arrangementsor to 're-institutionalising' the NHS. However, we have demonstrated how elements of the HSCA12, namely the expansion of competition law, is likely to run counter to current attempts to 're-institutionalise' the NHS. STPs are predicated on the notion of groups of organizations identifying with a specific geographical place and working collaboratively, seemingly with scant attention to competition regulations. Without changes to the latter, current attempts to reinstitutionalize lack a coherent legal underpinning structure and may be unable to accrue legitimacy.
Thus, rigorously applying the criteria of Zietsma et al. (2017) highlights a tension between institutionalization efforts and the post-HSCA12 legislative landscape, ones that would be lost without careful examination of the organizational population. With some of the forces described in the English NHS, such as increased agencification and fragmented organizational landscapes resonating with issues facing public administration in a variety of contexts and countries, future research might usefully apply this approach of organizational field interrogation in order to further detail the contours of the challenge in identifying fields, the implications associated with ascribing the label, and its dynamic.

Implications for research into public management reform
In terms of public management research, our findings point to a lacuna which is hindering our ability to (a) generalize knowledge, (b) might be a missed opportunity to further differentiate our knowledge of public sector changes, and (c) a risk to naturalizing contested political discourses. We discuss these problems now before turning to point to a possible future research agenda, based on a more relational understanding of the field concept.
First and foremost, implicitly assuming field-level phenomena without specification means that comparisons between and generalization based on such studies of management reform are highly questionable. Although a growing subset of studies might address the 'same' public organizational field, e.g. as in the case of past examinations of NHS reform, generalization becomes difficult unless these studies clearly designate the empirical population considered. With institutional theory already suffering from a case of too many interpretations and a lack of criticality which means that it became 'too difficult to understand what it is' (Hirsch and Lounsbury 2015, 96), if the idea of empirical commensurability commonly understood is removed from our studies, then institutionally focused public management research loses not only explanatory power, but ontological integrity, as it invokes a conceptualization which cannot tolerate close empirical examination. In the case of the English NHS, our findings show how a study taking into consideration STPs and Vanguards as part of the field in question will be completely different from one which does not. Similarly, a reliance on the appearance of similarity of identity among CCGs as exhibited in tree-like presentations of the system without proper scrutiny of the underlying layered complexity means that general rules might be inferred from cases which are highly particular.
Second, assuming that a particular public sector domain exists as a field-level phenomenon without delving into an analysis of field characteristics might mean that institutional frameworks are underutilized in public management research. Our close examination of the organizational landscape in primary care in England points to the possibility of organizations working in what we called 'silos' or fields which are almost organization-specific, and whereby collective action is haphazard and contingent. Alongside identification criteria, Zietsma et al.'s (2017) review also provides field classifications, separating between exchange fields and issue fields. By laying out each of the four criteria, our analysis makes a first step towards considering the English NHS as an issue fieldan organizational domain in which fields are formed around common issues and not around structured exchange relationships (Wooten and Hoffman 2016). With CCGs developing idiosyncratically and organizational populations varying greatly between locales, there are indications that healthcare is now more akin to how social movement organizations exchange with one another rather than how a statutory state-led arena of practice might usually be envisioned. This demonstrates a need to further engage with the specificities of fields in public sector studies so to better take into account differentiation in the public sector.
Third, and closely related, usage of an organizational field as a theoretical lens through which to talk and think about public management reform might end up uncritically supporting and replicating a political discourse aimed at masking the fragmentation and dissolution of state-run welfare systems. In the case of the NHS, a highly politicized social institution, the way that government reforms mobilize 'single NHS' or 'One NHS' rhetoric might be replicated by studies taking the unity of the health system as an analytical starting point. In the recent UK election, the Conservative party manifesto addressed the health system as 'Our NHS,' discursively reasserting that healthcare is a single entity even though the bulk of fragmentation and reorganization detailed in this paper accrued under Conservative-led governments. Thus, research using the organizational field construct to study welfare system reform should ensure proper empirical identification of objects of study, or risk naturalizing political rhetoric.
Towards a more relational understanding of fields in the public sector The analysis presented has focused on a structural definition of an organizational field, one that first and foremost examines the organizational forms existing in a particular public administration domain. There are other definitions in use, mainly those found in research programmes such as 'institutional logics' (van den Broek, Boselie, and Paauwe 2014). Research done under these labels acknowledges the heterogeneity found in organizational fields and thus might be a better framework for cases such as the one discussed. Nonetheless, the very construct of a field as a bounded, recognizable institutional context is still fundamental to these programmes and it is thus expected that studies using these frameworks will also need to devote more time to the explicit identification of a field before analysis of either 'logics' or 'work' is done. We would welcome further research able to comment particularly on these programmes, but it is worth mentioning that the bulk of research in public administration is based on the structural variant present here.
Arguably, a more promising candidate would be to change our understanding of a field towards a more relational one. In their review of investments into fieldoriented institutional approaches, Wooten and Hoffman (2016) discuss contemporary developments which we believe hold promise for field-oriented organizational analysis in public management. Namely, they point to the possibility of relinquishing prior understandings of a field as a 'thing' or 'mechanism,' and suggest the need to adopt frameworks that conceptualize fields as relational spaces. This suggests the need for studies which change analytical focus from an alreadyestablished, always-there field towards questions of field emergence, the contingency of field 'activation,' and the specificities of how field actors relate to one another.
This focus of analysis seems to be more fitting to the analysis of the English NHS as it currently stands, suggesting that a qualified return to the original usage of the construct as put forward by, for example, Warren (1967), might be particularly helpful in analysing contemporary public management developments. This closer alignment between Warren's interpretation and welfare reform might be expected as his theorization came about when studying community organizations, rather than industry bodies and their commercial dynamics as in the case of more frequently used 'founding founders' of institutionalism. Understanding the field as closely bound by geographical proximity, Warren (1967) posited an institutional framework based less on similarity of organizations, values, sharing similar ways of doing things, or being part of the same group of organizationsthose structural features of an institutional-level phenomenon. Rather, his focus remained on organizational actions which can only be carried on with relation to those of another. This framework allows for an institutional analysis that proceeds regardless of an organization's relation to a larger set of organizations, and thus allows consideration of relations between organizations before assuming an institutional-level phenomenon that might or might not act as context. Theoretically, Warren's relational drive changes research orientation towards questions of multiple contexts, and empirically the focus on action brings back the centrality of localities, key to the current state of the NHS and whose influence over public sector structure and practice is growing on the back of decentralization and devolution. Future research on the possibilities of using a more relational approach to both fields and public sector contexts is thus supported, and welcomed, by our findings.

Conclusion
Institutionalism is frequently employed in empirical studies of public management and administration. Such research commonly posits the existence of an organizational field, a key component of the theory, without its existence in the context in question being subject to interrogation. By applying Zietsma et al.'s (2017) four criteria for the identification of a field to the case of the English NHS, this study highlighted how field identification in a rapidly evolving, complex public sector context is currently difficult due to ambiguity in clearly articulating inter-organizational hierarchies, and the existence of forces acting to counter entrenchment of shared identity between organizations. Some of the forces we have described in the English NHS, such as increased agencification and a fragmented organizational landscape, are resonant with issues facing public administration in a variety of contexts and countries. We believe that this suggests that sufficient attention and time must be devoted to at least considering the applicability of the organizational field construct to research studying public administration domains before scholars embark on a full blown analysis of institutional dynamics.
In particular, future research might usefully apply this approach of organizational field interrogation in order to further detail the contours of the challenge in identifying fields and the implications associated with ascribing the label. The organizational field concept is integral to institutionalism, and it is important that research adopts a reflective attitude towards the concept in a given context to offer practitioners a route towards meaningful generalization and research recommendations. NHS England. 2016c. Re: Developing Sustainability and Transformation Plans to 2020/21 (Gateway