The ‘service turn’ in a new public management context: a street-level bureaucrat perspective

ABSTRACT It is increasingly argued that public management should build on a service logic instead of the prevailing manufacturing logic of New Public Management (NPM). Drawing from three cases in Swedish public healthcare, key features of a service logic such as value creation, co-production, and collaboration are prominent in formal documents and everyday talk. However, the 67 interviews in this study reveal that the service logic ideal is practically unreachable in a context impregnated by NPM. Instead, we suggest that street-level bureaucrats often need to address service logic expectations (public values, relationship-building, etc.) using an NPM logic (measurements, control, etc.).

participating in designing services, interacting with frontline staff, etc.An ecosystems view entails that value is created not in a linear fashion but in complex networks including actors across sectors.Value is more complex in PSOs than for private companies by not only addressing organizational or individual values for citizens, but also public values that benefit the broader citizenry (e.g.Osborne 2020).
Theoretically, PSL has been described as offering something other than NPM does (e.g.Grönroos 2019; Osborne 2018).However, there have been few empirical interrogations of PSL (Dudau, Glennon, and Verschuere 2019), with studies predominantly focusing on service ecosystems (Petrescu 2019), expanding the co-production concept (Eriksson 2019), and the possibility of value destruction (Engen et al. 2021).PSL is by no means the first concept to declare that NPM is unfit for contemporary challenges (Osborne, Radnor, and Nasi 2013) or even 'dead' (e.g.Dunleavy et al. 2006).However, there is a gap between theory and practice in that practitioners in public administration have seen little of the claimed death of NPM (Pollitt 2016).The empirical research rarely considers the fact that NPM is institutionalized in public organizations of today (Andersson and Liff 2012).PSL as practice does not appear in a vacuum and does not exchange NPM, but does appear in an NPM context.The consequences of this are far from obvious and require empirical research.An important question is who should 'do' PSL in practice?If NPM had managers and management systems in focus (Hood 1991), PSL focuses on 'frontline staff' (Döring 2022;Hardyman et al. 2021;Strokosch and Osborne 2020).However, frontline staff as a term in PSL has its origin in service research from the private sector (e.g.Normann 2001) and has not considered the specific characteristics of frontline staff in public organizations.
The present paper addresses this issue by turning to research on street-level bureaucracy.Starting with Lipsky (1980Lipsky ( /2010), street-level bureaucracy is a theoretical framework within public administration and management that analyses the specific characteristics of the street-level bureaucrats (SLB) that constitute the frontline staff of public organizations.When Lipsky (1980Lipsky ( /2010) ) introduced the concepts of SLB, he also directed more attention to the people who met citizens and were therefore the implementers of policies.Instead of viewing these people as powerless executors of policies that government policy-makers had decided, Lipsky (1980Lipsky ( /2010) ) argued that even though SLBs are at the bottom of the organizational hierarchy, they have considerable agency based on exercising wide discretion in decisions about citizens with whom they interact.This direct interaction with citizens was seen as a major difference from policy-makers, policy analysts, senior officials and elected officials, who conducted their work largely without daily interaction with citizens.Therefore, SLBs constituted the 'human face' of policy.SLBs interpreted policy in direct interaction with citizens, which meant that their situated agency (Bevir and Richards 2009) was wider than their organizational positions would indicate.Consequently, such situated agency would be crucial for PSL in practice, but then the institutionalized NPM would be one of the major influences of such situated agency and NPM may privilege system efficiency above the derived value to users (Kinder et al. 2022), although there is a lack of empirical work in this regard.Lipsky (1980Lipsky ( /2010) ) highlighted the street-level bureaucrat discretion as an important topic for research, not least to reduce the gap between the policy-as-designed and its actual implementation.To understand the implementation process, it is essential to understand the structures within which the SLBs operate.For instance, SLBs do not merely respond to performance incentives typical of NPM, but use their discretion to produce informal practices that are different from what policymakers and managers often recognize (Brodkin 2011).Davidovitz et al. (2022) concluded that meso and macro levels are important when studying SLBs and that managerial influences may enable SLBs to use their discretion to move towards clients.Zhang et al. (2022) showed how leadership may nurture such professional discretion by enhancing learning.This supports previous results that low-level managers could be included in what is conceptualized as 'street-level', since they can either have coordination functions that make them interact with citizens directly (Arnold 2015;Evans 2011) or they can influence the discretion of street-level workers and thereby indirectly influence the interaction (Borins 2002).
While PSL research has made some important contributions to public management, it is still largely de-contextualized and conceptual.The present article seeks to contribute to the theoretical development of PSL by recognizing that PSL practices, like other post-NPM concepts, take place in contexts impregnated by an NPM logic.This has consequences for the SLBs.Specifically, by focusing on the service delivery processes as perceived by frontline staff, the paper also contributes knowledge of how SLBs respond and act in terms of PSL practices impregnated by NPM.It has been argued that such a hybrid approach is important when studying SLB (Klenk and Cohen 2019).An additional theoretical contribution to PSL is to provide the called-for empirical examples (Dudau, Glennon, and Verschuere 2019).
NPM has had a strong influence on Swedish public healthcare since at least the 1980s (Eriksson et al. 2020).Recently, various post-NPM features have been introduced that are aligned with PSL: the active citizen, interorganizational collaboration, and emphasis on more collective forms of values rather than individualized ones.Thus, the three cases used in the current study were selected because they represent typical PSL features and therefore have potential to help address the abovementioned theoretical contribution of the paper.In particular, we focus on how the actors who should 'do' PSL -street-level bureaucrats -respond to co-existing and competing PSL demands and demands based on NPM-influenced control.Thus, our main contribution is the contextualization of PSL practice alongside different forms of hybridity between PSL and NPM demands.The three empirical cases were chosen because they are examples of different aspects of PSL.Data were collected through 67 semistructured interviews (Bell, Harley, and Bryman 2022).
To fulfil our aim, the remainder of the paper is structured as follows.First, we describe hybridized forms of public management.We then describe NPM to enable an understanding of how institutionalized NPM influences public sector organizations, followed by the recent development of PSL, which we intend to contribute to.This is followed by a description of the method and then the result section.The paper finishes with a discussion section and conclusion.

Hybridized forms of public management
Public management has increasingly been understood based on co-existing and competing demands, which has led to hybridized forms of public management as important explanations.Such hybridized forms may involve hybrids of professionalism and managerialism (Andersson and Liff 2018;Noordegraaf 2015) or hybrids of bureaucracy and network (Laegreid and Rykkja 2015).The latter is relevant for the present paper because it involves different forms of coordination that resemble PSL (more horizontal coordination also involving actors outside the organization) and NPM (more vertical coordination and intra-organizational focus).Laegreid and Rykkja (2015) established that although hybrid coordination exists between hierarchy and network, the vertical hierarchical coordination dominates, which in relation to our focus would mean that PSL would have difficulty having a real impact in NPMdominated contexts.In a similar approach, Fossestøl et al. (2015) related hybridity in public management to different post-NPM and NPM demands.Since PSL has been described as one of the major post-NPM trends (Osborne et al. 2022), we use their identified responses as a guide to hybridized forms of responses to NPM/PSL demands.Fossestøl et al. (2015) identified four main types of responses that street-level bureaucrats may use in meeting the demands of two reforms: non-hybridity, in which the new post-NPM/PSL demands are ignored; ad hoc hybridity, which is indecisive adherence to both NPM and post-NPM/PSL demands; negative hybridity, in which NPM and post-NPM/PSL demands are separated; and positive hybridity, in which both NPM and post-NPM/PSL demands are integrated.Klenk and Cohen (2019) claimed that the effect on SLB of NPM and post-NPM had created an ambiguous working environment because neither NPM nor post-NPM features, such as co-production, have replaced bureaucratic structures; rather, they are layered 'on top' of traditional public administration.The sometimes conflicting values and rationalities of the different models of reforms have been recognized as complicating SLB practice (Klenk and Cohen 2019).For example, the increased accountability of SLBs in many post-NPM reforms has been difficult to combine with managerial and consumerist features of NPM, not least in healthcare, in which physicians' and nurses' values for helping people may collide with managerial demands of delivery of measurable efficiency (Eriksson et al. 2020;Marston and McDonald 2006).However, performance indicators may be compatible with professional values if they are developed by and for professionals (Benish 2020).Researchers have also considered that it can be difficult to combine demands of efficiency with post-NPM aspects emphasizing democratic, often slow, processes, such as co-production (Eriksson 2022).Sager et al. (2014) emphasized another aspect of hybridity; that not only SLBs as public servants, but also private actors play an important role as implementation agents in new modes of governance.This is not unproblematic; for instance, these for-profit SLBs may find it difficult to address both the rules of the state and market incentives and/or client needs (Thomann, Hupe, and Sager 2018).
An often-mentioned aspect of post-NPM is digital information technology.In contrast to Dunleavy et al'.s (2006) claim that digital information technology is replacing NPM, Jorna and Wagenaar (2007) regarded this technology as one of many contextual factors that influences frontline staff -citizen communication rather than entirely changing their relations.Studying an unemployment fund in Switzerland, Buffat (2015) found that a new information technology system provided managers with more data on co-workers' decisions and compliance with the law, at the same time as the system created a distance that weakened the quality of supervision.Thus, control of NPM was enabled by the ease of collecting data, and the quality of the control deteriorated at the same time.
Other important post-NPM aspects are trust and values, not least in a Swedish public management context; trusting the competence of SLBs rather than the control of NPM is highlighted in so-called trust-and value-based governance.Paradoxically, as these models try to move away from managerial and hierarchical detailed steering, they have often been implemented in a top-down manner (Jacobsson, Wallinder, and Seing 2020).Also, recent French healthcare reform has emphasized a value-based approach and trust in SLBs, for instance by letting physicians participate in the governance of hospitals.However, rather than something new, this is often a step back to how healthcare was organized before NPM (Simonet 2015).
This section has elaborated on SLBs and hybrid responses.To understand such responses, we first need to better understand the underlying logic of the two reforms.Therefore, the next two sections will describe NPM and PSL.

New public management as institutionalized in public organizations
In order to improve the alleged inefficiency, inflexibility, over-centralization, hierarchy and costliness caused by the bureaucracy of traditional public administration (Lindberg, Czarniawska, and Solli 2015;Osborne et al. 2015), PSOs across sectors and countries increasingly adopted approaches from the private sector in the late 1970s and early 1980s.This became known as the umbrella term NPM (Hood 1991).While academic interest in NPM may be declining, several decades of reform have made it an inevitable part of the public sector today that continues to influence different actors (Karlsson 2019).
Despite some divergencies and lack of clarity, there are commonalities among NPM reforms.For instance, Ferlie and Geraghty (2005) noted that there were typical hard (control, measurements, etc.) and soft (customer orientation, quality focus, etc.) versions of NPM.Another distinction is that between market focus and managerial/ production focus (Andersson and Liff 2012).Market focus is achieved by splitting PSOs into purchasers and providers, creating business-like relationships and, at the same time, aiming to offer a diversity of providers, also by increasing the numbers of contracted private actors due to the increased use of public procurements and privatizations (Dunleavy and Hood 1994;Green-Pedersen 2002).Free choice of citizens among a diversity of competing providers across sectors is expected to improve public service delivery (Nordgren 2009).
The managerial/production focus of NPM addresses the efficiency of internal production processes by measuring and controlling output and outcomes within delimited units that have also decentralized accountability of their performance (Christensen 2012;Verbeeten and Speklé 2015), not least by using practices from industrial quality management to organize many public services in a sequential fashion, leading to results orientation (Alford and Hughes 2008;Hood 1995;Osborne, Radnor, and Nasi 2013), controlled through standardization and performance indicators (Almqvist et al. 2011;Dunleavy and Hood 1994;Hood 1995).
It is commonly argued that NPM has contributed to improved customer focus and efficiency (e.g.Olsson et al. 2014).However, the side effects of these (often local and sector-specific) results are often not mentioned (Pollitt 2002).Overall, the reported output (for example, lessons delivered by a university) -and, even more rarely, the outcomes (such as students' learning), that are particularly important in a public sector context (Alford and O'Flynn 2009) -of various NPM initiatives are often lacking (Pollitt and Bouckaert 2017;Pollitt and Dan 2011).
It has been argued that the efficiency aspect of NPM is better suited to economically and technically oriented policy areas such as transportation and garbage collection than to areas in which goal fulfilment is harder to quantify and measure (such as education and the context of this paper: healthcare) (Christensen and Laegreid 2003;Gregory 2003).Thus, the focus on output-related performance measures and pay-forperformance may not be relevant in these contexts (Verbeeten and Speklé 2015).

Public service logic and value creation
Public service logic (PSL) has developed by drawing from central ideas from the governance (Osborne 2006) and public value (Moore 1995) literature.It has been almost a decade since the initial development of a PSL (Osborne and Strokosch 2013;Osborne, Radnor, and Nasi 2013).At the heart of this 'service turn' is the alleged weakness of NPM that has often been identified by service researchers to build on an inherent manufacturing logic (Grönroos 2019;Quist and Fransson 2014), in which PSOs have been organized as if they produce and deliver tangible goods rather than intangible services (Osborne 2020).
The consequence of a manufacturing logic is that value is conceptualized to be produced in a linear assembly line-like fashion and delivered to a passively waiting 'customer' (Osborne, Radnor, and Nasi 2013;Quist and Fransson 2014).In this sense, value is considered transactional when goods (or services) are exchanged for money (value-in-exchange) (Vargo and Lusch 2004).In a service logic, however, value production and consumption cannot be separated due to the inherent intangible characteristics of services (Parasuraman, Zeithaml, and Berry 1985), which take place when frontline staff and users interact -the so-called moments of truth (Normann 2001)which emphasizes the importance of relational aspects in the roots of PSL: service management and marketing (e.g.Grönroos 2015).
According to Grönroos (e.g.Grönroos and Voima 2013), it is only through direct interactions that value co-creation between provider and user occurs.In the provider's and user's separate spheres, the co-is dropped and the actors may independently create value for themselves.However, the provider may facilitate the user's value creation process by developing and providing resources (potential value) that may support the user's creation of value (Eriksson et al. 2020(Eriksson et al. , 2022;;Grönroos and Gummerus 2014;Skålén et al. 2018).Thus, potential value from the provider can only be realized to real value by the user in their sphere during usage, so called value-in-use (Grönroos and Voima 2013;Heinonen, Strandvik, and Voima 2013).
Users may contribute with their knowledge and skills to develop and design services with the provider, as a co-producer or co-designer (Dudau, Glennon, and Verschuere 2019;Eriksson 2019).The more active role of users than in the manufacturing logic also means that they may take over some of the provider's routine tasks, such as selfcare and recycling garbage (McColl-Kennedy et al. 2012;Nordgren 2009), delivering services to other users, such as educating other users (Osborne, Radnor, and Strokosch 2016).Engen et al. (2021) noted that PSL initially focused on user input on services that were then executed by PSOs, but has lately shifted focus towards how public organizations can support, facilitate, and enable value creation for the user.
The focus on the user's value creation also opens up for actors other than the main PSO Here, the user may not only realize value from the provider, but also combine resources from a multiplicity of actors, including friends and family, private enterprises, and third-sector organizations (Eriksson and Hellström 2021;Osborne et al. 2015), suggesting an actor-to-actor approach in public service logic (Skålén et al. 2018).
In this ecosystem view on value creation, the integration of resources is considered mutually beneficial for all the involved actors in resource integration (Kinder et al. 2022;Petrescu 2019).In this view, public services should be understood as systems rather than the concern of single organizations (Osborne et al. 2015;Radnor et al. 2014).Thus, managing the relationships among the actors in the system is essential, which particularly concerns relationships with citizens/users (Osborne et al. 2015).
Like NPM, PSL originated in the private sector.With few exceptions (Grönroos 2019), it has been argued that there are essential differences between private and public services that need to be addressed.This concerns the relationship between provider and user that may be fundamentally different: a returning 'customer' is good news in the private sector, but a returning client to a social service office may be understood as a service failure (Osborne 2020).Public service users are more likely than private service users to approach a service with fear (surgery, for example) given the level of risk or discomfort (Berry and Bendapudi 2007).Similarly, patients in drug rehabilitation or inmates in prisons may best be understood as reluctant customers, given that they are likely to not want the service (Moore 1995;Nordgren 2009).
In particular, PSL has recognized a need to balance an overly narrow focus of value perceptions of the individual user to focus on public values (Alford 2016), such as public interest and common good (Beck Jørgensen and Bozeman 2007), that concern the broader citizenry or specific groups -and the potential conflict within and between these levels of value (Eriksson and Nordgren 2018).Petrescu (2019) argued that public values on the society level, which are essential for public services, can be analysed at the individual level -because it is not individual value perceptions accumulated (Stoker 2006) -but also at the collective level; and therefore, the ecosystems' view may be particularly relevant for understanding value co-creation for public services occurring at a multiplicity of levels.Recently, it has been recognized that PSL needs to recognize that value may not only be created in such a system but also destroyed, and that value may be created and destroyed unequally among the involved actors (e.g.Cluley, Parker, and Radnor 2021;Dudau, Glennon, and Verschuere 2019).
The present paper addresses the following research question: (1) How do SLBs respond to co-existing and competing demands of PSL practices and demands related to the NPM-influenced context?(2) What kind of hybrid approaches appear in response to the two different demands?

Setting and participants
The present paper draws on three cases in Swedish public healthcare in the secondlargest of 20 regions in the decentralized Swedish healthcare system, which houses approximately 2 million inhabitants in 49 municipalities.Sweden's regions are responsible for providing healthcare in hospitals or primary care centres and the municipalities are mainly responsible for providing care and support for people with disabilities and the elderly at special accommodations or at people's homes.Both regional and municipality services can be provided through private or third-sectors actors.
In total, 67 respondents were interviewed.The sampling of both cases and respondents was purposive (Bell, Harley, and Bryman 2022); see Table 1.The cases were chosen because they were believed to reflect typical post-NPM cases: citizen coproduction (Case A), inter-organizational collaboration (Case B), and inter-sectorial innovation (Case C).Case A focuses on public administrators' and healthcare professionals' perceptions on co-production with patients and relatives.Case B focuses on the perceptions of healthcare professionals from different levels in the Swedish decentralized healthcare system in which they work together with patients (and relatives) and at patients' homes.Case C draws on the perceptions of public employees, patients and others in designing the first centre for cancer-affected people in Sweden.
The main respondents for each case were working as either public administrators with direct contact with citizens or as first-line public managers (both referred to as 'street-level bureaucrats', in accordance with the definition of Arnold (2015) and Evans (2011)).Other actors were also interviewed (public managers on higher levels, patients, business-owners, association representatives, politicians) to make sense of interviews with the main respondents and to get an understanding of the SLBs place in the service ecosystem (Eriksson and Hellström 2021).
Before data collection began, the research project was approved by the Swedish Ethical Review Authority (registration number 2019-02280).All interviews took place approximately one year before the COVID-19 pandemic (September 2020-January 2021) and were predominantly conducted face-to-face.The interviews took between 30 and 90 minutes to complete.

Data collection and analysis
All interviews were semi-structured in order to allow the respondents to answer the questions freely and to adjust questions as the interviews went on (Kvale 1994).The goal of the interviews was to get the perspectives of the respondents regarding their work related to patients and citizens.This included a focus on the process of service delivery, as well as collaborations in developing and designing services, rather than quality of the care and outcome of care.Prior to the interviews, the respondents were provided with information and consent was collected.All respondents agreed for the interviews to be audio-recorded and the interviews were then transcribed verbatim.Secondary data were collected through documents, field notes from workshops and meetings with managers, and were used to look up issues highlighted in the interviews, prepare the interview guide, and so on (Nowell and Albrecht 2018).The questions asked related to matters such as the service meeting with patients/citizen (Could you please describe a typical care meeting at a patient's home?), the relationship with management (Are the measures expected to be reported back to management relevant from your professional perspective?)and recent Swedish reforms (In what way are you working in a trust-based manner?).Analysis of the transcribed interviews was used adopting a thematic analysis approach similar to template analysis (Brooks et al. 2015;King 2012).First, tentative themes were clustered based on literature on NPM and PSL; these were based on codes such as output control (NPM), competition (NPM), decentralization (NPM), collaboration (PSL), citizen involvement in interactions (PSL), collective values (PSL), and management's trust in frontline staff (PSL).Second, an initial template was developed based on a random selection of the transcribed interviews, which led to tentative themes being omitted, modified or expanded.Next, the themes of the template were further adjusted while the remaining transcriptions were analysed (Brooks et al. 2015;King 2012).Saturation for certain themes was reached halfway through the analysis and the last quarter of the analysis involved making minor adjustments to the developed themes (Nowell and Albrecht 2018).To avoid misunderstandings, themes were presented both verbally and in reports to both some of the respondents, as well as other stakeholders at various occasions (Lincoln and Guba 1985).A few minor adjustments were made based on the respondents' feedback.Figure 1 presents the final themes, concepts (expressions close to the respondents' answers), and overarching dimensions.
In the final step of analysis, we turned to theory by Fossestøl et al. (2015) to better understand the connections between the PSL and NPM categories in Figure 1.Here, the categories and their connections were applied to Fossestøl et al. (2015) four main types of responses that street-level bureaucrats may use in meeting the demands of two reforms (NPM and a Norwegian welfare post-NPM reform, similar to PSL): nonhybridity, in which the new post-NPM demands are ignored; ad hoc hybridity, which is indecisive adherence to both NPM and post-NPM demands; negative hybridity, in which NPM and post-NPM demands are separated; and positive hybridity, in which both NPM and post-NPM demands are integrated.

Trustworthiness
To ensure trustworthiness (Lincoln and Guba 1985), the researchers carefully took field notes and recorded interviews.Audio recordings were important to ensure that quotes were verbatim, which would have been difficult if only taking notes.The recordings also allowed us to listen to how things were expressed by the respondents rather than only what they talked about.We addressed the transferability of findings by providing as much information as possible of settings and contexts.Both authors had a long engagement in and knowledge of the local context and could identify key actors to discuss early result findings in order to establish credibility and ensure that the authors' pre-understanding did not overly impact the research findings (Alvesson 2003).The presentation of preliminary results -similar to member checking (Lincoln and Guba 1985) and utilization-based evaluation (Greenwood and Levin 2007) -was also important to avoid misunderstandings.

Findings
The main findings of the study include the role of citizens, which reflects both typical NPM logic (users receiving services) and PSL logic (users involved in co-designing services); top-down control of management requiring frontline staff to report outputs (NPM), as well as governance in informal intersectorial networks (PSL); and creating value mainly for the organization (NPM) or addressing collective values such as equity (PSL).

Citizen's roles
Case A focused on street-level bureaucrats' and managers' ideas about citizen coproduction across the policy cycle.This ranges from citizens' active engagement in coproducing their own or relatives' care plans to their participation in the development of existing or new services.The former was the most common type of co-production that healthcare professionals mentioned: If we do not include the patient, we cannot expect compliance.(Physician) The role of the citizen as a receiver of services was evident in the interviews and the street-level bureaucrats often restricted the citizen's role in co-production to express satisfaction with received services.As a typical NPM feature, the measurability of the citizens' perceptions is essential and aggregate levels of opinions are taken as quality indicators, even though some staff felt this was insufficient.The quantitative notion of the patients' voices was also evident in that an 'average patient' who was 'representative' of the whole population was sought, as opposed to recognizing the different needs of unique patients, or patient groups.However, in some cases, citizens were involved in more active ways, as developing new services or when (experienced) patients and staff jointly educate and inform other patients.
It is a totally different thing when those with the same disease inform other patients about it, together with the nurse.(Manager) In these examples, the public organization's relationship with citizens were more in line with the PSL ideal.Case B focuses on healthcare professionals from different levels collaboration to 'deliver care' at elderly patients' homes.In this case too, the notion of the citizen is seen as a receiver of services.Much has been done to improve services for the elderly.Despite the emphasis on person-centeredness in the project (as in the Swedish healthcare discourse overall), this proved to be lip service, according to some of the participants in the project.A reason was that the patient group (older patients with multiple chronic illnesses) were deemed too ill to be involved.
Instead, people from a senior citizen association were involved, but they are often healthy and may have little in common with our patient group.(Physician) Nevertheless, the perceived patient needs were used for the organization to jointly map a typical process from the patient's perspective, which was new to some: For the first time, we could see the whole patient journey, not just isolated to our own organization.(Manager) This process orientation was a typical heritage from the industrial NPM quality improvement movement that had gained increased attention in healthcare systems.As in Case A, the success of the model was often motivated by increased patient and relative satisfaction on the aggregate level.
The mantra for Case C was that it had been 'created with, for and by canceraffected' People.Here, the starting point was not to develop an existing public service but rather to co-innovate a new welfare service that addressed the citizens' 'life events' of receiving a cancer diagnosis.However, much of the focus was on issues that occurred for the cancer-affected person when treatment was finished and when they were expected to get back to 'normal lives', which is when many difficulties arose: Public healthcare deals with the tumour; we take care of the rest.(Patient representative at the centre) An important aspect was to address the loneliness that many cancer-affected people experienced, not least after treatment.The intention was to create an arena for people to meet: Here we can have good conversations, here is the energy that drives you forward, here is the healing power.No one should have to be alone with their cancer.(Patient visiting the centre) At the centre, citizens engaged with other actors (public, private) to identify needs and develop a non-profit organization through charity and funds to fill 'a gap in the welfare system'.The physical centre was run jointly by public staff and citizens.Both the codesign of the project, as well as how the physical centre operates in a public service ecosystem, are typical PSL.

Management or governance
In Case A, the NPM logic of controlling is evident.Grass-roots bureaucrats mentioned that they report various measures as a common practice.While some aspects of care, such as equality and patient involvement, were perceived as difficult to measure, it was more important to discuss these issues.However, top management expected streetlevel managers to report these aspects in a similar manner as they were expected to report economic measures: We have to treat soft issues [such as patient involvement] as hard issues [such as economy] for management to take them seriously.(Manager) Moreover, people at this level are often told by top management to do things, such as involve citizens.However, some physicians and other staff do not understand the relevance of including patients and there is an apparent risk that they 'do as they are told', albeit with a risk that they are just 'ticking the box', with little impact in practice.Involving citizens becomes a 'paper product' to please managers.Co-production was mainly done with individual patients (often a physician's own patients) rather than patient associations.Thus, the governance ideal of the involvement of multiplicity of actors in society was often absent.The focus is clearly based on the main provider's organization.
Measuring and reporting numbers was also commonly mentioned in Case B. For instance, physicians visiting patients in their homes state that they have to report all kinds of non-medical things.This is also seen from other stakeholders: The poor physicians are constantly haunted to report this and that.(Manager) Similarly, standardization guides how long certain activities -such as showering or eating breakfast -should take at patients' homes for nurse assistants.This is calculated by someone who is 'up there' and has no experience of how things work in practice, according to the staff at the floor level: According to a standard, the shower should take 20 minutes, but for Lisa it takes maybe 30.(Nurse assistant) Because of the inter-organizational nature of Case B, rather than the responsible organization's management, it is mainly the newly created coordinated function (holding the network together vertically and horizontally) that requires the data to be reported to them.It was also clear that there was a gap between top management and street-level bureaucrats in which the former's perceptions of the model seemed to stem from reports that were several years old (and described as a success story), whereas the latter reported that the model largely did not function as intended.However, there were aspects of collaborating in a non-hierarchal and non-formal manner.
All respondents in Case C mentioned the importance of including a plurality of actors across sectors.The point of departure -focusing on the 'life event' of getting a cancer diagnosis -was important: Cancer challenges people in many ways, different ways, and affects the person affected by cancer, but the family is also affected, physically, mentally, social, existentially and economically.(Patient visiting the centre) The 'life event' perspective also meant that the collaboration addressed matters of societal nature rather than healthcare matters, which made it necessary to include actors from wider society.Many respondents stated that these collaborations between actors were non-hierarchical.Moreover, informality prevailed and typical NPM features such as measurability were emphasized less than in the other two cases.This could be a consequence of the absence of a management in charge and, instead, a coordination function.It was mentioned that the voluntariness proved a fragility and that some public organizations were less present as time passed.Moreover, it was also clear in the interviews that the reasons for participating may also be self-interest of private businesses to profit from participating and supporting the centre, which may be problematic in the long run.However, the fact that the centre was run as a non-profit organization rather than a public organization was a reason why many private companies had been interested in contributing: If you want to get someone to donate to something, it shouldn't be something that looks like it's paid for through taxes; it must be something that is non-profit.(Patient representative at the centre)

Value
In Case A, citizen involvement was argued to have a positive impact on value for the individual patient.This often concerned patients' involvement in their own treatment plan.The influences from the industrial value chain were evident, in which it was argued that standardization of processes, reduction of PUBLIC MANAGEMENT REVIEW 'waste' and other expressions, suggested a linear logic in which value is refined in a sequential logic and produced in order to be eventually delivered to 'customers'.Case A also included citizen involvement in co-designing services, which was mainly done in order to improve services or prevent services from being used at all for the greater population.Here, value was created for the patient group, rather than the individual patient participating in developing services: One wants to use one's experience to improve things for others.(Patient) Many administrators noted the relevance of reaching vulnerable groups in citizen involvement, addressing public values of reaching those 'in greatest need': If we don't actively work for equality, we work against equality.(Manager) Value in terms of both individual and public was created jointly with citizens.However, these public values were articulated by regulations and management.Thus, values related both to NPM (individual) and PSL (public).
Value in Case B was created by the collaborative public organizations and the locus of value creation was 'person-centred' because it occurred at patients' homes.Here, the ideal was that professionals from different organizations would work together at patients' homes.However, it was mentioned that this network logic was no longer working, and the different professionals did not meet.The nurse assistants were not surprised that they did not meet physicians very often since . . .we are at the bottom of the ladder (nurse assistant) The above quote clearly shows the presence of hierarchy.Despite the personcentredness, staff said that they did not have time to talk and do what the patients wanted; they could only provide the basics related to care before having to rush to the next patient's home.This was unfortunate, since patients often wanted staff to stay for a chat, not just to give them medicine: Now I have 'produced' a medication here, but I barely had time to say hello [to the patient].There is as much medicine in that -talking -as in a pill.(Nurse assistant) Thus, value was mainly understood in a NPM rationale, creating value for the organization rather than the patient.
In Case C, the respondents commonly understood that the cancer centre was supposed to 'complement the welfare system' and the responsibilities of healthcare and public agencies.Value was created in an interconnected way in which actors from not only public sector but also the private and third sectors participated, as well as patients.
It is very easy to understand this as something that the healthcare system is responsible for, but there is something that we can actually do to help each other in society at large [. ..] that is why it is so important that civil society and associations are included in the whole thing, that there is something that can make life easier, that you have a need in some way that cannot be met by healthcare.But there must be other actors involved.(Business owner) By integrating resources, such as different forms of knowledge and skills, the different actors created value for themselves and also for all actors in the service ecosystem.However, the main purpose of the organization was to fill a gap in the welfare ecosystem and, in so doing, address public value on a societal level.In this case, value was co-created among the participating actors in the ecosystem.

Discussion
In this section, drawing from the analysis of the empirical cases in the previous section, we elaborate on how street-level bureaucrats respond to practices typical of PSL in a context that has not abandoned NPM.As mentioned in the introductory section, Fossestøl et al. (2015) four main types of SLB's responses in meeting the demands of both NPM and post-NPM are: non-hybridity, ignoring new post-NPM demands; ad hoc hybridity, adhering indecisively to NPM as well as post-NPM demands; negative hybridity, separating NPM and post-NPM demands; and positive hybridity, integrating NPM and post-NPM demands.
As mentioned, the three cases were selected because, at first glance, they appeared to be typical PSL cases.All cases do show typical PSL features and could also be categorized as collaborative governance in a broad sense (Emerson and Nabatchi 2015).Below we account for the three cases in relation to three of the types of responses of Fossestøl et al. (2015).

Citizen co-production and ad hoc hybridity
Case A is a typical PSL case in that it emphasizes co-production between public employees and citizens, which is important when both the bureaucracy of a traditional public administration and NPM's market orientation are said to have failed (e.g.Osborne 2020).
There was a clear desire from many SLBs in Case A to address typical PSL features, such as co-production.However, some respondents seemed to be more interested in checking the 'co-production box' than addressing genuine patient involvement, which illustrates the stronger influence of hierarchical coordination before network coordination (cf.Laegreid and Rykkja 2015).Similar to what Brodkin (2011) found, NPM features limited the ability of SLBs to use their discretion to move closer to citizens.However, in Case A PSL features were used symbolically to create room for action based on their professional background; for instance, in medicine.Thus, patient stories were collected, but sometimes not used at all to develop existing or new services.In this sense, the case could be seen as an example of non-hybridity (Fossestøl et al. 2015).
Moreover, despite demands from management to work with citizens to co-produce, NPM features were well-rooted and restricted 'how far' PSL could go in practice (cf.Laegreid and Rykkja 2015).Citizens were still seen as 'guests' who visited the organization and helped improve the internal processes.Thus, the reasons for involving patients (and relatives) were discussed more in terms of improving efficiency than with a focus on the benefits for the service users or citizens (Engen et al. 2021;Quist and Fransson 2014).However, even if PSL was restricted, involvement of patients did enable what could be labelled ad hoc hybridity.
PSL was restricted because the value concept was mainly associated with the industrial quality management movement that had become popular with NPM, particularly in a healthcare context (Batalden and Stoltz 1993).Consequently, value was understood at the individual level such as perceived service quality (Grönroos 2019) or customer satisfaction (Bergman et al. 2015) -and the collective level of value was simply understood as an aggregation of the individuals' perceptions of value.It is widely accepted that aggregation of individual value perceptions should not be confused with public value, which instead addresses the common good, the public interest, democratic values, etc (Bryson, Crosby, and Bloomberg 2014;Meynhardt 2009;O'Flynn 2007;Stoker 2006).
The respondents also suggested that public values, such as equity, were addressed using NPM logic focusing on measurements.In the interviews, the respondents emphasized that post-NPM aspects such as collaborative governance, including citizen co-production, as well as how these function -including so-called person-centredness (Ekman et al. 2011), equality, trust -were important, but these aspects were devalued as 'soft issues' in presentations.These soft issues were dealt with in a 'hard issues' logic and quantified in the same way (number of men and women going through a specific surgery, for example), as commonly noted in research on diversity and gender (Due Billing and Alvesson 2000).Many of the respondents argued that this logic did not function for equality and gender issues, since democratic ideals, participation, etc. could not be reduced to numbers.
Despite the presence of non-hybridity, in which PSL demands were ignored or only addressed symbolically, there was a genuine overall desire among SBLs to work more in line with PSL, although it was considered a struggle in an NPM context.Thus, it was clear that they had to respond to both PSL and NPM demands for the same service in what could be labelled ad hoc hybridity (cf.Fossestøl et al. 2015).

Interorganizational collaboration and negative hybridity
Case B is a typical PSL case as it focuses on interorganizational collaboration between PSOs to solve the wicked challenge of ageing society (Eriksson et al. 2020).
It was clear from the empirical material that the ideal of self-governed network was difficult to achieve and that, therefore, the collaboration had created special management groups and coordinators -similar to network administrative organization (Provan and Kenis 2008) -that did what managers typically do in NPM: collect quantitative data of in-patients at hospitals, number of home visits, etc.
The demands of NPM and PSL were clearly separated, which formed a negative hybridity (Fossestøl et al. 2015).The respondents mentioned that collaboration among the actors in the healthcare system was encouraged and articulated to benefit patients, but that the system also encouraged competition with other primary care centres, not least through the reimbursement system of getting paid per patient.The fact that the relatively small and competing primary care centres did not collaborate between them also created difficulties in terms of 'talking with one voice' in collaborations with hospitals and municipality care.Davidovitz et al. (2022) showed how important the institutional environment and the organizational level is to understand SLB actions.In this study, it was manifested by that quantitative measures of output also decided whether the collaboration was successful or not.Despite the importance of the knowledge and skills of SLB in Swedish contemporary public sector reforms (SOU 2019), the SLBs' top managementand especially frontline managers -dismissed reports of malpractice as 'narrative evidence'.Instead, top management presented statistics of patient satisfaction and visits to emergency departments as proof of the collaboration's success, at the same time neglecting stressful work environment for healthcare staff and so forth.It has been argued that trust, relationships and other typical network organization and governance features are difficult to pinpoint because they are relatively abstract (Provan and Milward 2001).Thus, the holy grail in PSL -the frontline staff -citizen encounter as the 'moment of truth' (Hardyman, Daunt, and Kitchener 2015;Normann 2001) -was hard to live up to as relationship-building at micro-level between healthcare staff and elderly patients that took place at the patients' homes were constrained by the shortage of time they had for each visit.Thus, efficiency may increase with shorter visits, but value for both patients and staff may decrease due to having insufficient time to establish a relationship at the micro-level (Järvi, Kähkönen, and Torvinen 2018).
The collaboration did not involve the active involvement of patients in developing services.The ideal of PSL would be to separate it from other forms of collaborative public management (e.g.Agranoff and McGuire 2003).However, the collaboration also clarified how difficult it may be for SLBs as they are dependent on the interaction of citizens to enact their situated agency, which means that certain citizens who cannot fully interact in a professional way, such as children and the elderly, need special efforts that make the SLBs situation more difficult (Sommer Harrits and Østergaard Møller 2014).The elderly in Case B were elderly patients, who often had multiple chronic illnesses, which made involvement of collaboration difficult.

Inter-sectorial innovation: positive hybridity
The last case, Case C, is characterized by positive hybridity (Fossestøl et al. 2015) in that NPM and PSL demands are integrated by SLBs.
Of the three cases, Case C shows the most 'unproblematic' hybridization of NPM and PSL practices.This could be explained by the fact that this is the only case to have emerged outside existing structures of the Swedish welfare system, which means it also evolved outside an existing NPM structure.Therefore, NPM and PSL features could develop simultaneously, enabling a better fit than in the other cases where PSL features were introduced into an already established NPM structure.In terms of the positive hybridity of NPM and PSL, there seem to have been advantages in that there was no established collaboration for the created network to build on, there were no borders to cross, and no single PSOs' responsibilities to guard.Klijn (2012) argued that collaborations as continuations of NPM partnerships and the emergence of governance collaborations are often intertwined and difficult to separate in real-life public administration.In this sense, Case C can be understood as a continuation of NPM in which PSOs seek private and non-profit actors to participate in service delivery (Klijn 2012).However, an important aspect is that the initiative of the present case was taken from patients rather than by a PSO.
As mentioned, Case C is the only one of the three cases that has an ecosystem (Petrescu 2019) approach, with actors across sectors as well as citizens.The different actors contributed with different resources in their effort to combine resources to create value for one another (Eriksson and Hellström 2021).The PSOs contributed with knowledge of legislation and professional expertise, private companies contributed with computers, furniture and design and business models, and patients and their relatives with their experience of getting and living with a cancer diagnosis.Thus, PSL features of relationship-building and focusing on the lived experience of the service user in designing service (Trischler and Scott 2016) were combined with influences of organizing work from the private sector, as in NPM (Hood 1991).Although the charity model has become increasingly popular, it remains relatively uncommon in the Sweden in which citizens largely rely on the welfare state (Vamstad and von Essen 2013).This may be the reason why many PSOs disappeared from the collaborative partnerships at an early stage.Case C is an example of value creation that is in line with PSL and later (and less managerial) developments of public value literature that focus on how a multiplicity of actors, including private and third-sector organizations, in society may contribute to create public value (Beck Jørgensen and Bozeman 2007).In particular, public value is what is determined together by citizens, through participatory democracy such as user involvement in which public value may be identified and emerge through deliberation among multiple actors (Stoker 2006), even though actors may have different and conflicting perceptions of value (Alford 2016).In the series of workshops of Case C, value was created in a similar fashion.

Conclusion
This paper contributes to the development of PSL by drawing from three cases from Swedish public healthcare that are still heavily impregnated by NPM.NPM and post-NPM/PSL can both be understood as being 'on top' of traditional structures (Klenk and Cohen 2019).The present paper takes a micro-level perspective, as emphasized in PSL, particularly SLB, rather than the managerial focus of NPM.Building on Fossestøl et al. (2015) research on SLB responses in NPM and post-NPM responses, none of the three Swedish cases entirely ignores new PSL (as a post-NPM example) demands.One case is an example of how SLBs indecisively adhere to both PSL and NPM demands (ad hoc hybridity).In another case, PSL and NPM demands are separated (negative hybridity).Finally, in the last case, PSL and NPM demands are integrated by SLBs (positive hybridity).
Confronted with empirical evidence, PSL proves to be unreachable in practice, not least because of the institutionalized and ubiquitous NPM.Thus, control, measurements, efficiency of intra-organizational processes and other NPM features dominate (Laegreid and Rykkja 2015), whereas PSL and other post-NPM concepts need to work in conjunction with these features.The present paper has shown that SLBs often need to address PSL expectations of both value (for example, a diversity of public values) and value creation (for example, dynamic and interactional), using an NPM logic of focusing on value for the PSO or as perceived by single 'customers' and value creation as linear and transactional.In everyday practice in public administration and management, these are often intertwined.In sum, the implicit notion that PSL should or could replace NPM is simply not possible, and the PSL premise of the SLB and citizen perspectives at the heart of service delivery is unrealistic.
We suggest that PSL, and other 'post-NPM' varieties, need to take existing managerial reforms into consideration to make them more realistic and likely to be implemented in practice.Pure PSL or similar normative ideas are doomed to fail.We believe that our three cases in this paper are cases of more balanced PSL cases, mirroring practice.
We believe the findings are largely transferable to similar contexts, despite the notion that the Scandinavian countries have retained democratic ideals in public