Why improvements fail in the public sector – management models as barrier or enabler for improvements

Over the years, the public administration and management model of public sector has moved from a “Weberian” and bureaucratic governance model to the marketized new public management (NPM) model based on private sector ideals (Hood, 1991; Lapuente and Van De Walle, 2020; Osborne, 2006). These management models are said to be a response to insufficiencies in previous models but are not free from flaws themselves and are often blamed for contemporary failures within the public sector and public service organizations (PSO) (Bryson et al., 2014; Hood, 1991; Osborne, 2006). Over the past three decades other concepts have been proposed to be the “next governance model” (Alford and Hughes, 2008; Bryson et al., 2014; Denhardt and Denhardt, 2015; Denhardt and Denhardt, 2000; Kinder and Stenvall, 2023; O'Flynn, 2007; Osborne, 2006; Osborne, 2020; Osborne et al., 2013; Stoker, 2006; Sønderskov and Rønning, 2021). However, to further complicate the context, an organization does not consist of just one management model; the models most often exist in parallel, in a so-called hybridization (Fossestøl et al., 2015).

Tied to these management models is the creation of public service logic (PSL), where a fundamental part is the service-oriented approach, shifting the linear goods-manufacturing logic of NPM into a dynamic co-creation amongst involved actors (Osborne, 2006, 2018; Osborne et al., 2013). With this change, the intra-organizational focus of NPM is switched to that of the user, as “PSOs do not create value for citizens – they can only make a public service offering. It is how the citizen uses this offering and how it interacts with his/her own life experiences that creates value” (Osborne, 2018, p. 228). Hence, it is the PSO that is added into the system surrounding the user, not the other way around, which implies that value is co-created by the PSO(s) and the user (Osborne, 2018).

To meet the increasing demands of society and users, public services must be constantly improved (Elg et al., 2017). However, it is not always straightforward to realize the benefits of improvement initiatives (Fryer et al., 2007). While we do have some knowledge about what drives results (Raharjo and Eriksson, 2017) and how to facilitate improvements (Gyllenhammar and Hammersberg, 2022), it is not clear how to succeed in improving in the context of the public sector (Al-Haddad and Kotnour, 2015; Gravesteijn and Wilderom, 2018; Prebble, 2021), especially when multiple actors are involved in delivering a service (Bryson et al., 2017).

In general, the area of improvements has received significant attention (Bhuiyan and Baghel, 2005) and a multitude of different approaches exist, including lean, six sigma and TQM (Andersson et al., 2006). However, when it comes to the context of public sector, and especially when multiple actors are required, there have been calls for further research (Gyllenhammar et al., 2022). The multi-actor context of the service ecosystem is becoming increasingly common in contemporary society (Bryson et al., 2017; Eriksson and Hellström, 2021). Combined with increased complexity, the problems of this context are sometimes even called “wicked” (Geuijen et al., 2017; Prebble, 2021; Rittel and Webber, 1973).

It is at this intersection – containing management models and improvements in a public service ecosystem – that the present article aims to contribute, a context that is increasingly multifaceted, containing complex and wicked problems (Geuijen et al., 2017; Prebble, 2021), where organizations use a hybridization of management models (Fossestøl et al., 2015). This has created a need for more research that focuses on this research gap, in understanding how management can help or hinder development and improvements in the public sector.

Hence, the purpose of this article is to explore the relationship between current public administration and management models and improvements, and especially how management models can be developed to support improvements.

Notably, the article takes a quality management approach to the sphere of the public sector and combines the fields of public management and quality management. The purpose is achieved by analysing a four-year improvement initiative between researchers and public service providers, combined with current knowledge of management models in the public sector and improvements.

In this section, we explain the notion of improvement and present three current co-existing management models in the public sector. We also formulate three guiding research questions based on the presented theory.

Improvements are often based on the plan-do-study-act (PDSA) learning cycle, in which small changes are tested and evaluated systematically for continuous development. The foundation is to identify problems and use scientific methods to solve the identified problems (Bhuiyan and Baghel, 2005). Improvements are a common element in many popular management concepts, such as TQM, lean and six sigma (Andersson et al., 2006).

William Edwards Deming, a pioneer in the field of quality management, identified four domains of improvement knowledge that must be grasped to gain profound knowledge in the area: (1) appreciation of a system, (2) knowledge of variation, (3) theory of knowledge and (4) knowledge of psychology (Deming, 1994). The introduction of quality thinking and kaizen (the Japanese term for continuous improvement) gained leverage to a large degree in Japan thanks to the teaching of Deming and Joseph Juran. The Juran trilogy and quality management emphasize not only the small incremental improvements but also the larger and radical breakthrough improvements (Juran, 1986; Deming, 1994). Today, shared core principles in quality management, besides continuous improvement, are customer focus and teamwork (Dean and Bowen, 1994; Gremyr et al., 2020). Other principles that are often included in quality management frameworks and success factors for improvements include process understanding and aspects of management and employee motivation (Eriksson, 2020). Much of the early knowledge from the application of quality management and improvements is derived from the private sector, but the ideas and knowledge are now widespread in the public sector (Eriksson, 2016; Raharjo and Eriksson, 2017).

Tied to the system perspective in Deming's (1994) first domain, it is important to adhere to the different perspectives of actors in the public sector. In a larger system, different actors can have diverging goals as well as different solutions for how they seek to solve identified problems (Gyllenhammar and Hammersberg, 2022). Furthermore, in some cases, even though actors are jointly delivering a service, they can have different names for the beneficiary (Gyllenhammar et al., 2023).

Moreover, there are major obstacles to undertaking systematic improvements, and there are many reasons and factors that cause improvement initiatives to fail (McLean and Antony, 2014; Reed and Card, 2016). Apart from general factors of change management, such as lack of management commitment, lack of vision/goal/motivation, lack of staff knowledge and expertise, inadequate communication, overly top-down management style and lack of employee commitment, there are some specific issues when it comes to implementations in the public sector (Beer and Eisenstat, 2000; Kotter, 2008; McLean and Antony, 2014; Radnor and Osborne, 2013). Firstly, public services, are services not goods, which means that the linear value-chain logic (Porter, 1985) is ill-suited (Osborne, 2018) – more about this in the next section regarding management models. Secondly, success indicators in the public sector are different from those in the private sector, which implies a risk that indicators will be ill-suited if they are transferred from private contexts. Thirdly, the operating system of PSOs is often internally focused, missing the multi-actor reality of contemporary society (Radnor and Osborne, 2013; Bryson et al., 2017). Furthermore, there are also differences in how public organizations and private organizations apply quality management and improvements (Raharjo and Eriksson, 2017). Therefore, the present article explores the success and failure of improvement initiatives in the public sector to better understand influencing factors, leading to the formulation of the first research question:

The answer to this research question will also be a stepping stone towards accomplishing the purpose of the article and understanding the relationship between improvements and management models.

The theoretical foundation draws on seminal works (e.g. Hood, 1991; Osborne, 2006) that remain widely cited and foundational in the field (Kinder and Stenvall, 2023; Lapuente and Van De Walle, 2020; Sønderskov and Rønning, 2021), as these works established the conceptual distinctions between the models that continue to structure contemporary debate. In this section, we present the three management models studied: (1) public administration, (2) NPM and (3) post-NPM, describing the current movement of the next governance model in the public sector.

Public administration, sometimes referred to as traditional or old public administration (cf. Denhardt and Denhardt, 2000; Stoker, 2006), has its roots in the Weberian bureaucratic organization and political science (Osborne, 2006). Some of the most common attributes are a clear division of labour and a strong hierarchical order. Here, knowledge resources are seen as scarce and should be concentrated higher up in the hierarchy (Adler, 2001). In this bureaucratic organization, the state is seen as a monolithic and unitarian entity where power is concentrated at the top, which follows set democratic objectives (Bryson et al., 2014; Osborne, 2006). There is a clear divide between administrators and the executing workforce, all of which follow a certain set of rules for how to perform their tasks.

Research on PA and improvement suggests a dual role: the bureaucratic structure provides stability, procedural legitimacy and professional autonomy that can support systematic improvement work (Dahlström and Lapuente, 2017; Eriksson et al., 2025). However, studies also show that rigid rule-following, hierarchical control and compliance-oriented audit systems can impede local learning and improvement initiatives, particularly when they crowd out professional discretion (Fossestøl et al., 2015; Radnor and Osborne, 2013).

While there is no unified picture of what NPM is, there are some commonalities between the varieties (Alford and Hughes, 2008). Christopher Hood's (1991, pp. 4–5) seminal article presented seven (slightly overlapping) doctrines: “‘hands-on professional management’ in the public sector”, “explicit standards and measurements of performance”, “great emphasis on output controls”, “shift to disaggregation of units in public sector”, “shift to greater competition in public sector”, “stress on private sector styles of management practice” and “stress on greater discipline and parsimony in resource use” (italics in original). These doctrines were used later in publications by other authors but using other terms (e.g. Alford and Hughes, 2008; O'Flynn, 2007; Osborne, 2006), such as managerialism, contractualism and multi-divisional structures (Alford and Hughes, 2008). A central part of the NPM is the intra-organizational perspective (Osborne, 2006), which is often blamed for sub-optimization of the multi-actor networks required in public service (Bryson et al., 2017; Osborne et al., 2013).

Furthermore, NPM is argued to take a production-oriented focus, usually based on Porter's (1985) value chain, which is unsuitable for public services due to the separation of production and consumption, placing the customer/user in a passive role (Osborne et al., 2013). However, other authors have argued that the value chain approach is just one of several value configurations; two others are the value shop and the value network (Eriksson et al., 2021; Stabell and Fjeldstad, 1998).

Empirically, NPM's market logic and performance regimes have shown mixed results in relation to improvement. While some studies find efficiency gains in narrow output measures (Lapuente and Van De Walle, 2020; Pollitt and Dan, 2011), others conclude that NPM conditions reduce continuous improvement to tool-based, cost-cutting exercises disconnected from user value (Radnor and Osborne, 2013). NPM has also been shown to act as a persistent barrier to co-production and collaborative improvement by framing citizens as passive customers and non-profits as business-like contractors (McMullin, 2023).

Regarding the post-NPM concept, there is no unified term; different authors have tried to give this post-NPM concept different names, such as new public governance (Osborne, 2006), public value management (Stoker, 2006) and new public service (Denhardt and Denhardt, 2000, 2015), as well as broader terms such as “the Emerging approach of Public Administration” (Bryson et al., 2014). However, a central notion is that of trust (Denhardt and Denhardt, 2000, 2015). Either it is explicitly stated or as part of other terminology, using terms such as relationships or collaboration (O'Flynn, 2007; Osborne, 2006).

Although the concept of trust has existed in the academic debate since the middle of the last century, the discourse has some problems, including defining trust itself, its antecedents and separating trust and risk (Mayer et al., 1995; Schoorman et al., 2007). The widely cited article by Mayer et al. (1995) (cf. Schoorman et al., 2007) addressed the above-mentioned issues and defined trust as “the willingness of a party to be vulnerable to the actions of another party based on the expectations that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control the other party” (p. 712). Central to the concept of trust is the trustor (the one that trusts) and the trustee (the one that is trusted). The antecedents are defined as: (1) benevolence, the trustor's perception of the positive orientation that the trustee has towards the trustor; (2) integrity, based on the perception the trustee has acceptable principles according to the trustor; and (3) ability, which could be seen as the competence, skills and characteristics needed to perform the domain-specific task that the trustor seeks.

Post-NPM approaches are empirically associated with more user-centred and collaborative improvement, emphasizing co-production, trust and inter-organizational learning (Kinder and Stenvall, 2023; Sønderskov and Rønning, 2021). However, research shows that post-NPM rarely replaces NPM but layers on top of it, creating hybrid organizations where market logics and bureaucratic structures persist alongside collaborative ideals (Funck and Karlsson, 2024; McMullin, 2021). This hybridization shapes the conditions for improvement in complex and often contradictory ways.

Empirical research over the past decade indicates that the dominant management model in a public organization systematically conditions which improvement practices are adopted and how successfully they perform. Studies of NPM-driven contexts find that performance regimes, market logics and managerial control tend to reduce continuous improvement to a tool-based, efficiency-oriented exercise that fails to create user value (Radnor and Osborne, 2013; Lapuente and Van De Walle, 2020; Pollitt and Dan, 2011). Bureaucratic-PA structures both enable improvement – by providing professional autonomy, capacity and rule-of-law legitimacy and impede it where rigidity and red tape dominate (Fossestøl et al., 2015; Eriksson et al., 2025). Post-NPM and NPG approaches, emphasizing coordination, co-production and learning, are empirically associated with more user-centred improvement, but operate in hybrid settings where NPM and bureaucratic logics persist and produce both enabling and constraining conditions for improvement work (Sønderskov et al., 2022; Kinder and Stenvall, 2023; Gyllenhammar and Hammersberg, 2022). The current empirical conclusion is that misalignment between governance logic and improvement logic rather than any single management model is the principal explanation for why public sector improvement initiatives stall or succeed (Bryson et al., 2014; Eriksson and Hellström, 2021; Sønderskov and Rønning, 2021).

While the theoretical properties of these management models are well-established, empirical research on how they relate to improvement initiatives in practice remains limited. Prior studies have largely examined management models from an organizational or governance perspective (Bryson et al., 2014; Lapuente and Van De Walle, 2020; Osborne, 2006), with less attention paid to how the coexistence and hybridization of models shape the conditions for systematic improvement work (Fossestøl et al., 2015). This represents a notable gap, particularly given the increasing complexity of public service delivery and the growing importance of multi-actor improvement efforts (Bryson et al., 2017; Gyllenhammar and Hammersberg, 2022). The present study addresses this gap by empirically examining how management models can hinder or support improvement initiatives in a public sector context.

Table 1 provides an overview of the three management models. In short, public service organizations operate through a hybridization (Fossestøl et al., 2015) of these models to varying degrees. This article explores the link between these management models and improvements. While each model has its merits, they can both impede and facilitate improvement efforts. To focus the research and achieve its purpose, the following research questions are also posed:

Even though existing management models in public organizations have their merits regarding some aspects (such as legal and political aspects), it is argued that the models can do more when it comes to supporting improvements. Therefore, based on the findings relating to RQ2, our intention is to explore what a management model that supports improvements can look like.

The studied improvement initiative was a programme for improving the sick leave system in a Swedish healthcare region. In Sweden, healthcare is managed by different geographical regions and within each region, several different healthcare organizations provide healthcare services. Furthermore, to receive sickness benefit in Sweden, there are three separate main actors in the sick leave system: healthcare, The Swedish Social Insurance Agency and the patient. The improvement programme was jointly set up by the Region and the researcher's university. Here, the Region financed the researchers from the University who were involved. The studied healthcare region is one of the largest in Sweden and consists of many different hospitals and healthcare providers.

Starting with the philosophical quote attributed to Kurt Lewin that “the best way to understand something is to try to change it” (cf. Greenwood and Levin, 2007, p. 18), the present article takes an action research approach. Our approach is similar to what Svensson et al. (2015) called interactive research. This approach allows for two linked cycles, one with the researchers who can distance themselves from the context and make critical evaluations and the other with the practitioners who can understand their problem formulation from their own reality. These cycles are then connected and learn from each other, making use of the other's gained knowledge to reinterpret their own. Hence, the researchers have combined the distance of research to contribute to critical thinking, while the occasional closeness has enabled an in-depth understanding of the context (Herr and Anderson, 2015). Next, we utilized the PDSA quality management tool (Bhuiyan and Baghel, 2005; Bergman and Klefsjö, 2010), which corresponds to the action research cycle of planning, taking action, evaluating the action, (re)planning, etc (Coughlan and Coghlan, 2002). The intention of this article is to examine the systemic conditions for improvements and management models in the public sector; the specific empirical details are not the primary focus of this article.

Interactive and action research were chosen as they are particularly well-suited for studying systemic aspects of complex, multi-actor problems in public sector settings, where understanding cannot be separated from active engagement with the phenomenon (Herr and Anderson, 2015). The oscillation between closeness and distance enabled the researchers to systematically move between system levels and actor perspectives, ensuring a comprehensive analysis rather than a narrowly bounded case study.

The main actors in the programme included two programme leaders from the healthcare organization, approximately 40 process leaders from the region, three main researchers from the university, and three other researchers who were involved occasionally. The process leaders were responsible for improvements in their organization. At the start of the programme, initial plans were made with two of the main researchers and the two programme leaders. The first contract that was formed stipulated that the two parties (the university and the healthcare organization) would work together, and that the university would receive funding for education, research and utilization in connection with improving the sick leave process and system. The programme commenced on December 17th, 2018. Table 2 summarizes some of the key happenings during the programme. At the same time as the improvement programme took place, following major change and improvement initiatives in the region, two events occurred: (1) a major strategic initiative called “The Transition”, which affected healthcare organization and healthcare delivery, was launched; and (2) a new comprehensive digital care system in the region, including a knowledge management system, was introduced. Additionally, (3) due to the COVID-19 pandemic, many workshops and meetings were held digitally, and some participants may have been clinically engaged elsewhere.

To contextualize the empirical setting, the sick leave system in Sweden involves three main actors: healthcare providers, the Swedish Social Insurance Agency and the patient (service user). The purpose of the system is to facilitate the return of the service user to sustainable working capacity while providing financial support during rehabilitation. Failure to improve the system can be shown at multiple levels: at the micro level, prolonged sick leave can erode professional skills and self-confidence; at the meso level, organizations lose experienced employees and continuity; and at the macro level, the cumulative effects include a shrinking workforce, escalating public costs and increased pressure on welfare systems (Löfgren et al., 2024).

Problems in the public sector are becoming increasingly difficult to manage and associated with those of wicked and complex problems (Geuijen et al., 2017). Within the definition of complex and wicked problems, it is impossible to grasp every problem aspect, improvement initiative, and all possible diffusion effects of different initiatives beforehand (Geuijen et al., 2017; Snowden and Boone, 2007). Hence, this article has adopted an approach in which all solutions are not claimed to be known, but increasing knowledge makes it possible to take identified problems and bring them towards the domain of complicated problems, which can be solved beforehand (Camillus, 2008; Snowden and Boone, 2007); and, for those problems that still exist in the complex and wicked area, create guidance for how to approach those problems.

During the project, various sources were used to gather data and enabled triangulation (Bryman and Bell, 2015). Triangulation was employed by combining three distinct data sources: (1) practitioner improvement reports, (2) observational data from programme sessions and (3) reflective journal notes. This made it possible to cross-validate findings and reduce the risk of single-source bias (Bryman and Bell, 2015). The set-up included the researchers teaching the practitioners improvement methodologies and simultaneously guiding and supervising improvement projects conducted by the practitioners in their organizations. Furthermore, from the fall of 2019 to the end of the improvement programme 2022, journaling was performed, enabling the integration of experiences and information, which allowed a reflective practice (Coughlan and Coghlan, 2002). The journaling allowed for recording the experience while being close to the data (for example, when helping the practitioners in their improvement projects), as well as zooming out and creating a deeper understanding and judging whether the insights have been correctly interpreted, enabling further decisions and actions to be taken in the educational programme as well as for the practitioners' improvement projects (Svensson et al., 2015). Tied to this research-internal learning loop, feedback was also sought from the participating process leaders at the end of each programme session. These two learning loops, one researcher-based and one practitioner-based, allowed for continuous improvements throughout each programme course. As such, the improvement programme as a whole could be seen as three layers of reflexive cycles containing the steps of planning, acting, evaluating re-planning (Coughlan and Coghlan, 2002). One large cycle, containing the whole improvement programme, and within it two cycles (one for each course round), contained double cycles (practitioner/researcher) for each session. Moreover, reports were written by the practitioners describing their improvement projects, which were also used for this study.

In total, the empirical material comprised approximately 220 conference observations, 40 supervision sessions related to improvement projects, and around 150 journal entries from workshops and working groups conducted during the programme. It should be noted that the service user was not explicitly included in the data collection, as the research focus centred on the organizational and systemic conditions for improvement rather than the user's individual experience. This aspect has been addressed elsewhere (Järvi et al., 2018).

The article follows a methodology akin to Dubois and Gadde's (2002, 2014) systematic combining, where empirical data and theory are continuously re-matched to advance both practical knowledge and theory. An overview of the analysis steps is provided in Table 3. To consolidate data from the various projects, the first step involved merging the data for clarity (see Table 4). Main activities were identified and then, through consensus, discussed the outcomes, improvements and influencing factors. A draft of Table 4 was reviewed by process leaders for validation in September 2022. Subsequently, a theoretical framework (Table 1) was developed in Step 2, facilitating a deductive coding approach like Hsieh and Shannon's (2005) direct content analysis. This approach mapped factors from Table 4 to management models in Table 1, with authors discussing and agreeing on relationship constructions. The aim was to ascertain if factors could be linked to specific aspects of the management models.

This chapter is based on the three research questions and starts by presenting the results answering RQ1, followed by an analysis that addresses RQ2 and RQ3.

An overview of major improvement initiatives is further presented in Table 4. The table is divided into (1) improvement initiative, (2) main outcome, (3) improvement and (4) main explaining factors in relation to management models in the public sector.

Deming (1994) noted that every system yields the results it is designed for, necessitating changes to the system for different outcomes. Hence, improving the way of working is crucial for achieving better results; mere discussion or increased knowledge will not change the system. However, altering underlying structures, like the management model, to favour development could lead to improvements. The analysis identifies three main outcome categories of the improvement program – firstly, a few real improvements for an actor in the system, mainly derived from a setting in which the right prerequisites among the individuals and the individual's home organization existed. Secondly, increased knowledge within improvement science for participating actors, depending on the motivation among the individuals, the competence development differed. Lastly, a common language and view for improvements, as well as a willingness to solve challenges jointly among process leaders. To conclude, the program has increased system comprehension and linked its various components. Due to the heterogeneity of participants, roles, organizations and professions, discussions gave the participants the opportunity to grasp challenges from a broader perspective. Yet, it remains uncertain how the second and third categories improved the system.

Like Radnor and Osborne (2013), we argue that the operating processes and systems are suboptimal for improving PSOs; in their case, Lean in particular. Based on the current study, we claim that many improvement initiatives have no effect, since the management models effectively “block” most improvement initiatives. This line of thinking is developed below, along with an elaboration on the factors' connection to the management model. The section is divided into two sub-sections: (1) Public administration and management and (2) Improvements.

The analytical process involved mapping the identified factors (A–L) against the defining characteristics of each management model as described in Table 1. Through iterative discussion among the research team, each factor was assessed in terms of which model's underlying logic best explained its occurrence: (1) hierarchical rule-following (PA), (2) performance measurement and market orientation (NPM) or (3) trust-based collaboration (Post-NPM). In several cases, factors were found to reflect tensions between two or more models operating simultaneously, consistent with the hybridization described in the theoretical framework (Fossestøl et al., 2015). The empirical basis for this analysis is illustrated below, with examples of quotes from project reports and practitioner conference entries collected during the programme.

Working with improvements is not part of the main process, nor is it the main part of the purpose of the PSOs; hence, there is a lack of political and management interest. The communication was mainly horizontal and the vertical communication was limited during the programme, and there was Insufficient interest and backing from senior management (C), a known factor for failure to improve (Beer and Eisenstat, 2000). This was consistently reflected in the empirical data, as when practitioners noted that they needed “a senior management team that is receptive to improvement suggestions” (practitioner conference entry), and when programme journals recorded that “many attempts to make the management team support improvement projects” were unsuccessful, as “something else was always a higher priority” (journal entry). As seen from the programme's improvement initiatives, none of the PSOs' management felt they had accountability for the (wicked) problem (Geuijen et al., 2017; Rittel and Webber, 1973). Instead, management has other urgent matters to deal with in their linear, “production” organization to function as a bureaucratic machinery (Osborne, 2018; Porter, 1985). Insufficient support and interest from adjacent management (H) resulted in the absence of an improvement culture at the micro-level of the service ecosystem (Radnor and Osborne, 2013; Petrescu, 2019). Furthermore, as seen in the traditional role of a public administration manager, it is to ensure that procedures, laws and rules are followed. Meanwhile, the NPM-esque manager focuses on performance targets, which are market-oriented and not necessarily focused on increasing value (Bryson et al., 2014; Stoker, 2006); hence, neither the NPM nor the traditional public administration managers seek improvements.

In connection to this, the fact that there is no straightforward way to measure effects (D) improvements of the sick leave system makes it ill-suited for the NPM-oriented organization in general, as its main performance focus is the monitoring of measurable input/output from a financial perspective (Hood, 1991; O'Flynn, 2007; Osborne, 2006; Stoker, 2006). Meanwhile, as it is difficult to find performance targets, politically provided goals are either non-existent or irrelevant for practice, and it can even be a risk destroying value (Altermark, 2020), paving the way to failure according to traditional PA (Bryson et al., 2014; Stoker, 2006). The effect of this difficulty to measure is manifested in unclear goals and purpose (B), a troublesome system flaw given that the hybridization of the three existing models are dependent on some type of goal or purpose, if it is politically provided (as in PA), market-oriented (as in NPM) or outcome-focused (as in Post-NPM) (Hood, 1991; O'Flynn, 2007; Osborne, 2006; Stoker, 2006). This leaves the micro-level without a direction, and low motivation (K) to organizationally improve.

As one participant articulated: “We are only measuring the people that are sick [not the ones getting well]” (journal entry), capturing how the current measurement logic, rooted in NPM's financial input/output focus, fundamentally misaligns with what practitioners understood as meaningful improvement.

Connected to the above reasoning regarding the lack of clarity regarding goal and purpose, there was no sustained or coherent plan for using the training material (E), leading to valuable work having no dissemination or uptake. An issue that could be derived from laissez-faire management (Beer and Eisenstat, 2000), in combination with the (internal) contractualism of NPM (Denhardt and Denhardt, 2000; Osborne, 2006), leading to the fulfilment of the contract – that is, creating educational material – but no more, as this was not prioritized by management (C). The PSO strives to do something tangible that makes a difference but relies on the voluntary effort of the employees, in a management system that is steered by bureaucracy and top-down authority (Denhardt and Denhard, 2000). For example, the produced videos were intended to be watched voluntarily by the employee (in line with the trusting approach of post-NPM models) and were treated as a compliance exercise in line with traditional public administration thinking. However, some sort of expected improvement was absent, especially since the funding came from the government and just needed to be spent on an activity regarding collaboration. Within this issue, it has been hinted that the current system residing NPM and PA effectively hinders any post-NPM structure based on trust and relationships to work proficiently since any attempt by a trustor to believe in the ability of a trustee is likely to be hijacked by one of two problems; The first is the top-down hierarchical system, requiring the mangers approval, hence making the trustees ability in some sense obsolete (Bryson et al., 2014; Denhardt and Denhardt, 2000; Mayer et al., 1995); The second is that the desired action will not rhyme with the NPM-esque performance target of the trustee, which can potentially create an internal competition of resources (Bryson et al., 2014; Hood, 1991; Stoker, 2006).

Furthermore, the no clear accountability for the sick leave system (J) and no mandate to act (I) in the sick leave system shows tendencies of classic failures regarding change programs (Kotter, 2008). As is typical when it comes to wicked problems, it was unclear who “owned” the problems regarding the sick leave system. This lack of clarity meant that no one was accountable, and no one could be blamed for the results, as a manifestation of the disaggregated units with their individual performance targets of the NPM model (Osborne, 2006; Stoker, 2006). This was visible empirically when attempts to act on improvement ideas stalled over questions of who had the authority to implement: journal entries from the data visualization group recorded discussions about “who to turn to for a ‘decision to implement’”, and practitioners noted that “regulations from government agencies are a hindrance” to cross-organizational collaboration (journal entry).

In terms of time spent by the individuals on each improvement initiative, analysis and discussions were favoured over testing and experimenting. The absence of focus on executing improvements (A), merely just understanding the context, is often the result when problems are seen as solvable beforehand; in reality, the solution cannot be solved without testing (Snowden and Boone, 2007). This “solvable approach”, where employees are not taking action, mirrors the hierarchical system of bureaucracy, where managers are those who approve and initiate actions for “solvable problems” (Bryson et al., 2014; Denhardt and Denhardt, 2000; Snowden and Boone, 2007). Furthermore, a result of the constant endeavour to find “the solution”, initiatives was also overshadowed by the introduction of a new care information system (F) and other relevant aspects to consider in the changes. As in other sectors, iterative or cyclic testing and experimentation is vital for solving complex and wicked problems (Camillus, 2008; Geuijen et al., 2017; Snowden and Boone, 2007); this point is not supported in the linear thinking of either PA or NPM (Bryson et al., 2014; O'Flynn, 2007; Stoker, 2006).

As a result of the bureaucratic system, both through the separation of management and operations and through hierarchy, the process leaders are distanced from the core business and its staff, which means they cannot influence the daily operations of the sick leave system (Bryson et al., 2014; Denhardt and Denhardt, 2000; Osborne, 2006; Stoker, 2006), especially as the core operations are largely performed by physicians with a strong professional basis, who are used to being autonomous (Von Nordenflycht, 2010). Hence, the work is perceived as too remote from direct patient care (L), and the organization consists of too many administrative layers for improvement initiatives to be realized. Even though there is a customer focus, in terms of the end customer/patient/insured citizen, as emphasized in NPM (Bryson et al., 2014; Stoker, 2006), the user perspective is lacking. This, combined with the administrative layers of bureaucracy in the traditional public administration (Bryson et al., 2014; Denhardt and Denhardt, 2000), leaves the improvement initiatives ineffective because they do not reach the end customer. Since there are several layers of “users”, such as other employees in the public service system, improvements are required before reaching the end user to enable any real benefit in the multi-actor system (Gyllenhammar and Hammersberg, 2022).

Also, there were varying conditions across individuals and home organizations (G) regarding driving improvements; this, combined with the work being perceived as too remote from direct patient care (L), creates another hindrance as improvements require both professional knowledge of the specific context, as well as improvement knowledge (Batalden and Stoltz, 1995). Therefore, it is unsurprising that employees working with the sick leave system in general had low motivation (K) for driving improvements, as barriers regarding the aspects of competence, organizational structures and leadership (Beer and Eisenstat, 2000) stood in their way.

Answering the third research question and addressing the shortcomings highlighted in Table 4, a Management model for improvements is proposed that takes a quality management perspective and is suitable for public administrations. The proposed management model is described with the help of the same elements as shown in Table 1 in the theory chapter; each element of the management model is described in greater detail below and summarized in Table 5. However, it is emphasized that this is not a “new” model but should rather be seen as an evolution of previous model(s), as we are acknowledging the existence of the current system, and that this system cannot start from a “tabula rasa”. Hence, it is not a replacement of everything but an advancement, especially when it comes to the aspects of the post-NPM model(s). However, at the core of this advancement is the issue of negligence of both complex and wicked problems seen from both traditional public administration and NPM.

The debate of public value is said to have started with the seminal article by Moore (1994), and even though we do not claim to solve the still ongoing debate of what public value is and how to measure it (e.g. Brunetto et al., 2022; Prebble, 2021), we do argue that it cannot be described simply with a price mechanism, nor a hierarchical and rule-based system (Adler, 2001). One of the core principles of quality management is to have value-adding activities (Bergman and Klefsjö, 2010; Eriksson, 2020), and the consequence of this is that the governance mechanism(s) and accountability system within the public sector must be value-centric and focus on value creation, even though it might be difficult to define value.

The monolithic structure of bureaucracies and NPM has been shown to be too slow and rigid for contemporary problems (Bryson et al., 2014, 2021). However, the call for more distributed leadership and increased trust (Adler, 2001; Denhardt and Denhardt, 2000; Osborne, 2006) should not come at the cost of a system perspective, as this is an essential factor for success (Deming, 1994), especially when it comes to system improvements in the public sector (Gyllenhammar and Hammersberg, 2022). Hence, the inherent conflict between locally made decisions and a system perspective must be acknowledged.

There is no single universal solution to all problems, and in a society where problems are even labelled “wicked” (Geuijen et al., 2017), and as suggested by the post-NPM models, a pragmatic approach should be taken to address these problems (Bryson et al., 2014; Denhardt and Denhardt, 2000). For example, using a range of value configurations – such as shop, chain, and network (Eriksson et al., 2021; Stabell and Fjeldstad, 1998) – could serve as a basis for designing the system of delivery where each problem should be evaluated to decide which type of value configuration is appropriate.

As seen from the empirical material, lack of management interest and support, lack of motivation and a non-existent focus on improvements are factors that contribute to the failures of improvement initiatives. To address these shortcomings, where management models do not prioritize improvements, management must help employees by enabling the integration of system components and directing the organization(s) towards a common purpose. This is a reality that consists of multiple goals and multiple types of goals, not just policy/rule or price-based targets (Bryson et al., 2014; Stoker, 2006). These goals should also be directed towards improvements, as this is a vital aspect for success (Bergman and Klefsjö, 2010; Eriksson, 2020).

In contrast to NPM and PA, the emerging management post-NPM model of trust and relationship is based on an inter-organizational focus, which enables a system perspective. However, the co-existence of all three management models creates barriers for the system perspective. NPM focuses on measurable aspects within the organization to increase competitiveness, while PA prefers a stable, non-changing environment, for (the monitoring of) the policy system (Adler, 2001; Osborne, 2006). However, even though the system perspective is vital for improvements (Gyllenhammar and Hammersberg, 2022), customer focus is at the core of all organizations (Dean and Bowen, 1994; Gremyr et al., 2020) or, in the more unclear case of the public sector, the beneficiary (e.g. Gyllenhammar et al., 2023). That is, the focus, or foci, should be on both the system and the beneficiaries.

Even though decisions should be made locally – as argued regarding organizational structure above – this should not be at the cost of the system perspective (Deming, 1994). Hence, the performance focus should be set at a system level to foster locally made decisions to support the system instead of only the immediate and local context. Furthermore, the previous management models are based on the capability to measure performance (Denhardt and Denhardt, 2000). However, given that value in public services is difficult to measure and has a complex nature (Gyllenhammar et al., 2023; Prebble, 2021), there is a risk when measuring what can be measured that the performance focus is drawn from what actually enables value. Hence, the performance focus must also acknowledge the complexity and the immeasurability of the system. Therefore, the performance focus should be on fulfilling a purpose and value-based rather than on achieving a measurable target.

As value can be both created and destroyed in the public sector, it is important to distinguish who the service is made for, and with this, who it benefits (Gyllenhammar et al., 2023). In many cases, this will most likely be the local citizen, but as some public services are more ambiguous regarding who the beneficiary is – for example, a prison or school (Osborne, 2018) or the possibility of multiple (unintended) beneficiaries/users (Gyllenhammar et al., 2023) – context-specific definitions are needed.

The empirical findings provide a direct basis for each element of the proposed Management Model for Improvements. The recurring unclear goals and purpose (B), combined with the inability to measure meaningful outcomes (D), demonstrate that existing governance mechanisms, whether rule-based (PA) or market-oriented (NPM), fail to orient organizations towards value creation, motivating the model's emphasis on value-centric governance and accountability. The insufficient interest and backing from senior management (C) and adjacent management's insufficient support and interest (factor H), rooted in PA's hierarchical separation of administration and operations and NPM's performance target logic, underscores the need for a redefined managerial role focused on enabling and integrating rather than controlling. The absence of mandate to act (factor I) and unclear accountability (factor J) across organizational silos, a structural consequence of NPM's disaggregation, point directly to the need for a network-based organizational structure with distributed leadership. Finally, the distance from the patient (factor L) and the varying conditions across individuals and home organizations (factor G) highlight that the system of delivery must be flexible and pragmatically adapted to the problem at hand, rather than imposed uniformly from above.

To fulfil the purpose of this article, which is to explore the relationship between current public administration and management models and improvements, and especially how management models can be developed to support improvements, data from a four-year improvement initiative between researchers and public service providers were utilized. A framework describing management models was used to illustrate the management models' hybridization and shortcomings in the Swedish sick leave system.

Today, agile thinking and the shaping of autonomous individuals and teams in organizations are seen as a core component of a successful management model (Laloux, 2017; Eriksson, 2020; Hamel and Zanini, 2020). It is these engaged and autonomous teams and individuals that enable improvements (Eriksson, 2020). It has been argued that many of the identified factors that hindered the improvements, and the overall development of the organizations, would probably not have occurred if a more suitable management model had been in place. As several authors have pointed out, bureaucracy (Hamel and Zanini, 2020) and NPM (Dunleavy et al., 2005) are, or should be, replaced by other more favourable governance models (Appelo, 2011; Hamel and Zanini, 2020; Laloux, 2017). The empirical data demonstrate that these governance logics remain firmly entrenched and continue to constrain improvement initiatives and undermine employee motivation.

Each management model causes its own problems. However, as the empirical material depicts a hybridization of management models that are used in parallel (Fossestøl et al., 2015), the different systems for governing the organization cause even more trouble. For example, there is a lack of motivation due to the distance to the beneficiary caused by a hierarchical organization (PA). At the same time, the motivation of the employee is essential to drive the improvement initiatives forward (Post-NPM) and the system of sick leave insurance is difficult to measure from a qualitative point of view; even though it is quantifiable, it does not capture the soft aspects of health and welfare, leading to less managerial focus as they seek good quantifiable measures (NPM). Furthermore, because the silos created by NPM have left the sick leave system without any “owner”, it becomes even more difficult for management to give employees the mandate and trust to act, since the managers do not “own” the issue and therefore cannot delegate or give the mandate for employees to act.

This article contributes to practice in terms of highlighting the dysfunctionalities of management models in the public sector, some of which are due to single management models and some due to hybridization between several models. A development of the current system is proposed, acknowledging that not all aspects of preceding management models are without merit; rather, each model exhibits certain structural limitations. The core of the proposal is therefore to advance the current system by addressing its structural shortcomings, including the contradictions arising from hybridized management logics. The suggested Management Model for Improvements is value-centric, acknowledging that public value is multifaceted and sometimes difficult to define (Brunetto et al., 2022; Moore, 1994; Prebble, 2021) but is centred around the beneficiary (Gyllenhammar et al., 2023). Furthermore, it takes a pragmatic approach to problem-solving (Bryson et al., 2014), where leadership is distributed but at the same time adheres to the importance of a system perspective of improvements (Gyllenhammar and Hammersberg, 2022).

From a theoretical perspective, the present article contributes to the field of public management by infusing quality management and improvements within its realm. The article also contributes with empirical findings that link empirical issues with the more theorized management models of the public sector. Furthermore, it also answers the call for more coherent management models within the public sector that acknowledge the particularities of this context (Elg et al., 2017). More specifically, the research develops the knowledge of why improvements fail and what to do about it, enabling future research to further develop management models and management thinking within the public sector. Moreover, the framework created in the theory chapter can be utilized to evaluate other public management contexts than that observed in this article. More specifically, this article makes several contributions to the intersection of quality management and public sector governance. Firstly, it empirically demonstrates how dominant management models – PA and NPM – and their hybridization, systematically shape the conditions under which quality improvement initiatives either succeed or fail, thereby extending the quality management literature into the governance domain (Elg et al., 2017; Radnor and Osborne, 2013). Secondly, by applying a management model framework to analyse improvement failures in a multi-actor public service system, the article bridges two fields that have largely developed in parallel: public administration theory and quality management practice. Thirdly, the proposed Management Model for Improvements offers a theoretically grounded and empirically derived framework that is specifically designed for the complexity of public sector contexts, addressing limitations identified in both NPM-oriented quality management approaches (Lapuente and Van De Walle, 2020; Pollitt and Dan, 2011) and purely bureaucratic structures (Fossestøl et al., 2015). For practitioners, the framework provides actionable guidance on how to restructure governance mechanisms, managerial roles and accountability systems to enable rather than obstruct improvement work. The management model framework developed in this article can furthermore be applied to analyse improvement conditions in other public sector contexts beyond the Swedish sick leave system studied here.

There are no ethical issues with the publication of this manuscript.