Understanding implementation in complex public organizations – implication for practice

The effective implementation of politically initiated public service innovations to the front-lines of the public service organization, where the innovation is to be applied, is a challenge that both practitioners and researchers struggle to solve. We highlight the importance of analysing contextual factors at several levels of the implementation system, as well as the importance of considering how the practical everyday work situations of the front-line workers influence their application of the innovation in question. We illustrate this by exploring the implementation process of a specific work inclusion measure, looking at its wider context and some of its implementation outcomes at a specific public agency. The intention is to illustrate the significance of considering the contextual complexity influencing implementation work as a reminder for practitioners to take this into account in their planning and practices.


Introduction
Implementing centrally initiated innovations into the daily operational practices of frontline workers in public agencies is a challenge that practitioners and researchers have struggled with for more than half a century (Hupe, 2014;Moulton & Sandfort, 2016).
Assumptions that initiatives of change from top levels of political authority and leadership will automatically be executed at the front-lines of the organization are a thing of the past (Hill & Hupe, 2014).Implementation and innovation researchers describe the process as complex, multi-levelled and dynamic, but in many cases 'underemphasize[s] the interactions between [the] different levels' in their research agenda (Wong, 2005, p. 2).Approaches that integrate this multi-level complexity into innovation and implementation research have been called for (Hill & Hupe, 2003;Hupe, 2014;Wong, 2005).Current scholarship aims to understand the complexity of innovation and implementation processes by applying various sociological perspectives.This includes a critical realist holistic perspective to help explain social change and phenomena, drawing on a rich understanding of change as happening in the interactions between the structures or contexts of the implementation processes and the people acting within them (Wong, 2005;Mihailescu, Mihailescu & Carlsson, 2013).In addition, a framework of multi-level nested implementation systems of strategic action fields have been applied to analyse the complexity involved (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015).Moreover, Wegener has recently stressed the need to consider the drivers of innovation in the everyday practices of service providers (Wegener, 2015).In the context of this article, we use the concept of an implementation system to emphasize the contextual understanding of interacting levels, from the policy level through the organizational level and the operational frontlines.We specifically emphasize the importance of seeing the wider context of the implementation system in relation to the specific work situations and everyday practices of the front-line workers.In the end, the front-line workers are the ones who bring the intended innovations to use or not, and their everyday decisions have important implications for the policy outcome.
The term innovation has numerous definitions.Hartley (2005) emphasizes that innovation is not just a new idea, but a new practice.Innovations may originate from policy and can be technical, administrative or organizational in character ( Van de Ven, 1986).The purpose of innovation in public services is to meet certain societal challenges, or to achieve more with the resources available (Hartley, 2005).In order to reach political ambitions through public agency, innovations are often planned at central levels of government, though with the intention to be executed at the front-line of service organizations in which service providers, such as social workers, meet service users.One example can be seen in Norway's major emphasis on employment as the foundation of welfare for individuals and society, in addition to its political goal of improving the quality of work inclusion services for citizens with disabilities or other needs for publicly assisted facilitation to join the work force (Norwegian Ministry of Labour, 2006Labour, , 2013;;Norwegian Ministry of Labour and Social Affairs, 2016).The organization responsible for providing work inclusion services to the citizens is NAV.
NAV grew from a merger in 2006 between the central government administrations of public employment and national insurance, as well as the local governments' social service administrations (Christensen, Fimreite, & Laegreid, 2014).It consists of the Directorate of Labour and Welfare at the top level, overseeing various NAV units, such as County and Municipal offices at increasingly lower levels of the organization. 1  The political emphasis on work inclusion has resulted in a myriad of public service innovations, strategies, agreements and measures being implemented through NAV (Norwegian Ministry of Labour, 2012).One such measure, which will be the illustrative focus of this article, is the Facilitation Guarantee (FG). 2 The FG was derived from the Tripartite Memorandum of Understanding on a more Inclusive Working Life, better known as the Inclusive Working Agreement (Spjelkavik, 2014), between the Norwegian government and their social partners, the employer and employee organizations, all of which are central in the Nordic Work Model (Moene, 2010).The Inclusive Working Agreement was a response to high disability benefit uptake rates and sickness absence found throughout the OECD countries in the 1990s (OECD, 2005).The White Paper, 'On an action plan for people with disabilities 1998-2001' (Norwegian Ministry of Labour and Social Affairs, 2006, p. 5) stated that, 'It is the government's goal that as many people with disabilities as possible shall be permitted to participate in the labour market.People with disabilities constitute an important labour resource.'Parallel to the White Paper, the Norwegian government initiated a working group to investigate the causes of increased sick leave and the increased uptake of disability benefits (Rambøll management, 2008).As a result of the working group's report, rather than proposing a tightening of benefits as in some other countries, the Norwegian government set out to give more responsibility to solving these issues to employer and employee organizations (OECD, 2005).Aiming to reduce 'the outflow from the labour market into health-related benefits and early retirement schemes', the government and social partners signed the Inclusive Working Agreement for the period from [2001][2002][2003][2004][2005] 'to cooperate on strengthening active measures at the workplace' (OECD, 2005, p. 5).
The idea was that the workplace is the primary arena where progress could and should be made (OECD, 2005).The initial Inclusive Working Agreement was evaluated in 2005, with the conclusion that sick leave had in fact decreased in the period from [2001][2002][2003][2004][2005], but that the rate of employment among people with disabilities had not changed (Norwegian Ministry of Labour and Social Inclusion, 2009).
Research on employment and disabilities shows that work inclusion prospects are enhanced when a person is trained on the job as part of the ordinary work force, a socalled Work First strategy, rather than being trained for the job in sheltered activities (Spjelkavik, 2012).An effective Work First strategy requires collaboration between public service agencies and sectors, and importantly includes collaboration between the public employment agency and the ordinary labour market (Spjelkavik, 2014).However, barriers have been found to exist, deterring employers from hiring people who are in need of special adaptations in their job situation (Dyrstad, Mandal, & Ose, 2014).Research on employers' views of barriers against employing people with disabilities finds that elements such as security and relationships of trust with the public employment agency are essential ingredients for promoting work inclusion into the ordinary work force (Gustafsson, 2013;Gustafsson, Peralta, & Danermark, 2014;Schafft & Spjelkavik, 2014b).Responding to employers' need for a secure and trusting relationship with NAV has been found to be of major importance for increasing work inclusion for people with disabilities (Schafft & Spjelkavik, 2014a).
During the first period of the Inclusive Work Agreement, a test project called 'Flexible jobs', with a Work First focus, tried out interventions such as salary substitutions and a close follow-up of employers by public employment agencies in six counties (Rambøll-management, 2008).The evaluation of this project resulted in the introduction of the FG in the renegotiation of a new Inclusive Working Agreement for the period from 2006-2009.The parties agreed on a need to focus on recruiting people with disabilities into ordinary working life.The FG was one of the solutions proposed to help with employers' need for close contact and collaboration with public agencies during the process of work inclusion for people with disabilities.By ensuring that potential employees and employers would receive guaranteed facilitation and followup from public agencies, the risk for employers would be decreased (Norwegian Ministry of Labour and Social Affairs, 2006).The FG was explained in detail in an attachment to the new Inclusive Working Agreement from 2006-2009.In 2007, the central government decided that the FG would be subject to a one-year test project in five of the original 'flexible jobs' counties.By the end of the period, the FG was merged with an existing, similar work inclusion measure called the 'Guarantee of facilitation' available at the Centre for Assistive Technology and Adaptions that became a part of NAV during the reform.In 2008, the FG was made into a permanent work inclusion measure at the national level of NAV (Riksrevisjonen, 2013).The FG entails a written document for the employer, the employee and NAV that lists contact information and the commitments of the three parties.A central element of the FG is that the employer and employee receive a specific contact person, a front-line worker at the NAV office who is responsible for coordinating the granted measures from NAV, as well as following up and supporting the employer and the employee, both before and during the work inclusion process (Rambøll management, 2008;Riksrevisjonen, 2013).
In 2012, the FG was specifically included as part of a new work inclusion strategy for young people, called the Job Strategy, resulting in a sharp increase in its use (Riksrevisjonen, 2013).An audit by the National Audit Commission on the implementation of the FG, in the period from its introduction through 2012, concluded that it had not been implemented as intended the previous years, and pointed to several shortcomings that needed to be addressed (Riksrevisjonen, 2013).This led to a call for more goal-oriented work by NAV to secure the proper implementation of the FG, and for the local NAV offices to ensure that work on implementing the FG was prioritized. 3Starting in 2013, NAV increased the efforts to make the FG an integral part of a work inclusion methodology at the local offices.This contributed to the increase in the use of FG started in 2012 due to the Job Strategy (Riksrevisjonen, 2013).However, at the onset of this case study in 2015, seven years after its initial introduction and despite increased implementation efforts, the use of the FG by front-line workers in NAV was still of varying character at many local offices.Specifically in the local NAV office of the case study, after being counted as one of the local offices using the FG most frequently in one of the most effective counties in terms of the number of FGs used during the period from 2013-2014, the FG usage dropped during 2015 after dedicated implementation efforts had ceased.This may indicate a lack of internalization into the practices at the front-lines of this specific office.
To understand such a relative lack of attainment in incorporating the FG into existing practices in general and particularly at this local office despite considerable implementation efforts, we point to the importance of considering the wider context of an implementation process in addition to the micro-processes involved at the front-line.
In this article, we use a critical realist case study approach and draw from a theoretical framework of complex multi-level implementation systems set forth by Sandfort and Moulton (2015) and Moulton and Sandfort (2016).This article is part of a case study exploring innovation processes in a complex public organization using the implementation of the FG in NAV as the empirical illustration.More specifically, the study attempts to find explanations for the extent and the way that the FG was implemented at this selected local NAV office.The study emphasizes both the wider contexts at the macro-and mezzo-levels of the implementation system, as well as specific micro-processes at the level of the everyday practices of the front-line workers who, in the end, are the ones who bring the innovation into their daily work or not.Our aim in this article is to explore selected aspects of the wider context and work situation of the front-line workers in relation to the FG implementation outcome at the local NAV office studied.In line with a critical realist case study approach, we identify one of several possible mechanisms that can help explain the outcome.Our intention is to illustrate the importance of recognizing the contextual complexity in implementation planning and practices.
We start by exploring the FG as a public service innovation, and present a conceptualization of implementation and innovation processes.We then give a brief account of the structural elements of Sandfort and Moulton's (2015) and Moulton and Sandfort's (2016) framework for implementation research.We move on to the methodological section, introducing the project as a critical realist case study and describing the critical realist implications for the study, as well as for the case selection, sampling, data collection and analysis.On the basis of an analysis of the FG implementation system, inspired by Sandfort and Moulton's (2015) and Moulton and Sandfort's (2016) model, we present some selected parts of its wider context in the policy field and the organizational field, as well as relevant parts of its specific context at the front-line.We trace an example of contextual implications that we found had an influence on the final FG outcomes in the case, namely the outsourcing of work inclusion services to external providers influencing the implementation of the FG at the micro-levels of front-line workers.We discuss the impact that the complexity of the context has on this process, and we conclude with a brief summary, as well as a discussion of the limitations of the study and some implications for practice and further research.

Innovation, implementation and the innovation process
The FG can be explored as different types of innovation, such as policy innovation leading to process innovation and eventually leading to service innovation (Fagerberg, 2005;Hartley, 2005), all surfacing at different times throughout the innovation cycle (Van de Ven, 1999).First, initiated by the Inclusive Working Agreement and given political authority and funding from the central level of government, the FG starts as a public service innovation, a tool for collaborative planning between the NAV representative, employers and jobseekers, as well as between different units of NAV.
Second, the implementation of the FG calls for process innovation, a new way of organizing the work process for front-line NAV staff in their collaborations with employers and jobseekers.Third, this has potential to lead to service improvements as a result of a strengthened cooperation between the parties involved.The result is dependent upon the innovation process taking place, especially the extent to which the 'new', in this case the FG, is actually implemented into existing work inclusion practices.
The term to 'implement' has originally been defined as 'to carry out, accomplish, fulfil, produce, complete' as suggested by Pressman and Wildavsky (1973 pp. xiii-xv), and is commonly considered a crucial part of the innovation process.The innovation process may be defined as 'the development and implementation of new ideas by people who over time engage in transactions with others within an institutional context' ( Van de Ven, 1986, p. 591).This definition includes the wider context as an important element in that process.Scholars often divide the innovation process into various phases (Fagerberg, 2005;Wegener, 2015).These broadly tend to include: 1) the development of an idea for a new solution to an identified problem or need, 2) the implementation of this new idea into existing practices, and 3) the diffusion of the new practice into different organizational settings (Hartley, 2013).Wegener contributes to the model of innovation phases by adding 'drivers for innovation' that make innovation possible or not at the practice level, i.e. the everyday work situation of the service providers (Wegener, 2015).In addition, cyclical phases such as evaluation, adaption to change and the implementation and diffusion of the results of these evaluations are important parts of an innovation process (Van de Ven, 1999).As such, the actual sequence of innovation processes is rather cyclical and messy (Van de Ven, 1999).In this article, we focus on the implementation part of the innovation process and the work situation of the front-line staff, but we also illustrate that it must be seen in relation to the innovation process as a whole.

The implementation system
To analyse the context of the implementation process as a system consisting of interacting levels of the macro, mezzo and micro (Wong, 2005), we draw on a multilevel conceptualization of implementation systems suggested by Moulton and Sandfort (2016) and Sandfort and Moulton (2015) .This framework takes into account the influence of the innovation itself on its implementation process.It also provides a rich understanding of implementation systems by conceptualizing the multiple interacting levels into three nested fields (Moulton & Sandfort, 2016).Because our aim in this article is to explore the FG's context, we mainly focus on the structural elements of the framework.Sandfort and Moulton (2015) use the metaphor of water running through a natural three-layered water filtration system in a pond to illustrate how a new policy or public service innovation flows through a multi-level system during its implementation process.The structural elements of the implementation system, as illustrated by Sandfort and Moulton (2015) and Moulton and Sandfort (2016), contain: 1) a policy field at the macro-level, consisting of a bounded network of organizations carrying out the particular policy, 2) an organizational field at the mezzo-level, where the policy is authorized and operationalized, and 3) a front-level field consisting of the micro-level where the implementation system interacts with the target group to carry out the innovation, specifically through the front-line staff in their everyday practices and interactions.Each layer has unique social structures that act as filters and shape the innovation as it passes through, at the same time the social structures are embedded in the rest of the context (Sandfort & Moulton, 2015).Both the strength and flexibility of the social structures, and the innovation itself, influence the implementation processes (Sandfort & Moulton, 2015).By emphasizing the role of the innovation itself in affecting the process, as well as analytically differentiating the three levels in the system, Sandfort and Moulton (2016) provide a tool for conceptualizing the complex contexts of implementation at interacting levels.This allows us to more systematically analyse the implementation of the FG.

Methodology
The empirical data used in this article is derived from the case study described above.
The case study was designed to explore and explain the implementation process of a public service innovation in a complex public sector setting.Requiring an in-depth understanding of the contexts and processes involved, a qualitative intensive case study methodology is appropriate (Flyvbjerg, 2006;Tsang, 2013).A critical realist organizational case study approach, which takes into consideration the broader context in a temporal manner to explain the phenomenon in question (Vincent & Wapshott, 2014;Wynn & Williams, 2012) is especially fitting.Critical realist case studies are equipped with tools to 'investigate complex organizational phenomena in a holistic manner', allowing researchers to 'develop in-depth causal explanations for the outcomes' of these phenomena, taking into account the wider contextual factors that, over time, may have had an influence on its occurrences (Wynn & Williams, 2012).This article mainly reports on the exploratory phase of the case study, where the focus has been to develop a case description in order to identify causal links to be analysed in the explanatory part of the study (Yin, 2013)

Case selection and sampling
Using a critical realist informed grounded theory approach during data collection and analysis (Belfrage & Hauf, 2016;Kempster & Parry, 2014;Oliver, 2012), the sampling procedure followed a purposeful and unfolding process over several stages (Patton, 2014), aiming to find a somewhat 'exceptional' (Vincent & Wapshott, 2014) or 'extreme or pathological case' (Danermark, Ekstrom, Jakobsen, & Karlsson, 2001) that could shed light on causal configurations and mechanisms, which help us understand the implementation process.The sampling process started with a few interviews of key informants in NAV.The FG stood out early in this process as a relevant public service innovation considered important by key informants, but which at the same time showed some intriguing implementation results.The initial key informant interviews led to a natural and purposeful expansion of the sample to informants at several levels of NAV and to relevant documents in a snowballing manner (Patton, 2014).These included leaders and implementation coordinators at different hierarchical levels of the system, as well as front-line staff.Because of the need to limit the study, we chose not to include informants from the target groups of the FG.However, the chosen informants all had important roles in the FG implementation process and provided thick descriptions (Patton, 2014) of their personal interactions and experiences with these targets groups.
The case can be defined as the implementation process of the FG during the time period prior to its inception in 2008 until a new version of the FG was put into effect on 1 January 2016.The case context (Harrison & Easton, 2004) is the field of work inclusion and the hierarchical line of the implementation system, focusing on a specific local NAV office and its specific group of front-line workers responsible for work inclusion with the target group.It also includes the corresponding county NAV office, the top level of the state-run part of NAV (called the 'Directorate of Labour and Welfare') and the central government's Ministry of Labour and Social Affairs for a wider contextual understanding of the case.An insight into the role of the Ministry in the implementation process is important because of its responsibility for policy formation, and for its collaboration with the Directorate in planning the implementation of the FG at the operational level.The county NAV office had an important influence on the implementation processes and strategies in the local offices under its control, through which the FG is put into practice.The specific local NAV office in the case study was chosen in collaboration with the county office as part of the snowballing method, identified as an 'exceptional' case (Vincent & Wapshott, 2014).The local office chosen was significant because it belongs to a county NAV office that was highly prosperous in terms of FGs used overall, as well as for the local office's unusually high usage of FGs in the period of 2013-2014 and the drop shortly after.

Methods of data collection and analysis
For the purpose of triangulation, we used a mixture of different qualitative methods (Patton, 1999).These included: 1) examining documents (Bowen, 2009), e.g.relevant policy documents, audits, internal organizational reports and secondary data sources in order to acquire an insight into the context and history of the work inclusion field, 2) observations of case meetings of the two teams of front-line workers at the selected local office (Bøllingtoft, 2007) to gain insight into the practices and work situations of the front-line staff, and 3) individual interviews (Smith & Elger, 2012) with key informants in leadership and implementation coordinator positions, plus front-line workers, to attain insight into the implementation context, process and drivers at the different levels of the work inclusion field.The documents were examined and relevant sections extracted.The interviews were carried out using a semi-structural interview approach, with thematic interview guides being constantly revised as the insights in the field deepened (Oliver, 2012).By continuous updating of the interview guides to fit the informants' focus as the research progressed, the findings emerged in a constant dialogue and continuous analysis of the empirical data (Oliver, 2012).After each interview and case meeting observations, memos were jotted down and added during the analysis process.Each interview was transcribed verbatim and reviewed together with the observation notes and relevant documents.The data analysis of the exploratory phase of the case study (Yin, 2013) relevant to this article is primarily focused on the initial memos, analytical notes and document readings continuously prepared in the initial phase of the study to gain insight into the context and history of the FG.The empirical illustrations used in the article are findings that emerged during this continuous process in the early phase of the case study.Key informants have validated central parts of the case summary and findings during the process of analysis (Healy & Perry, 2000;Yin, 2013).

The FG and its wider context
To help understand the implementation outcome of the FG at this local NAV office, we emphasize the importance of analysing the FG itself, and how it fits into the context at the different levels.In this section, we consider important elements of the FG and selected aspects of the structural contexts at the macro-, mezzo-and micro-levels of the implementation system drawn from the case study.In particular, we draw on Sandfort and Moulton's (2015;Moulton & Sandfort, 2016) framework of multi-level implementation systems.First, we present some central traits of the macro-level work inclusion policy field to which the FG belongs, including important public and private actors in the field, and the policy landscape of the FG.We then present some selected points from the organizational level, focusing on the structure and selected contextual influences, such as a prioritization of resources at the mezzo-level.Lastly, we discuss some selected contextual findings at the micro-level and aspects of the work situation of the front-line staff.More specifically, we illustrate how the macro-level condition of outsourcing work inclusion services, the mezzo-level resource allocation and microlevel front-line coping strategies all combine to influence the outcome of the implementation effort at the local NAV office in the case study.

The work inclusion policy field of Norway
In any implementation setting there are various actors who, because of their interest and expertise, engage in-or have an influence on the implementation of new policy or public service interventions (Sandfort & Moulton, 2015).Norway is a strong welfare state with a wealthy economy, and with important regulations between employer and employee organizations, as well as an increasing focus on a Work First strategy.
Though divergent, political parties in Norway are focused on preserving the welfare state through prioritizing high levels of employment.Publicly funded work inclusion services are allocated from within NAV itself, not to mention being outsourced to either publicly owned labour market rehabilitation services, non-profit service providers or forprofit entrepreneurs.In addition to these various service providers, the work inclusion field in Norway consists of governmental and municipal bodies, both political and administrative, employer and employee organizations, and as user organizations (Duell & Tergeist, 2009) that protect the rights of citizens with various disabilities.These actors all have different expertise, legitimate authority, influence and interests (Sandfort & Moulton, 2015) that they bring to the work inclusion issue.
The different actors in the work inclusion policy field in Norway therefore all have an interest in how public service innovations are shaped, how resources are allocated, which user groups are prioritized and the nature of the innovations in focus (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015).The different roles, authorities and interests of these stakeholders may be political, administrative, legislative, operational or financial, and their activities and collaborations all influence different implementation processes ( (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015).As a central example for our illustration in this article, the recent NAV reform led to a proliferation of work inclusion service providers in the market (Andreassen & Aars, 2015).These providers have an interest in the different work inclusion alternatives available to frontline staff in NAV because their existence as service providers relies on getting work from NAV.
A substantial amount of money is allocated every year through the National Budget for the procurement of these work inclusion services (Fjeldstad, 2016).If the services bought are not used, the legitimacy of NAV and its ability to maintain its mandate will be brought into question.

The FG in the context of the work inclusion policy field of Norway
***An analysis of the policy field of an implementation system should include an overview of the legal grounding and funding streams for the innovation in question, as well as existing and potentially competing or complementary measures to that specific intervention (Sandfort & Moulton, 2015).The FG is not statured by law, but instead is formalized through guidelines and propositions from the Ministry of Labour and Social Affairs, and acted upon through guidelines drawn by the Directorate.As we have seen, it was originally initiated from the political collaboration in the Inclusive Working Agreement between central government, employer organizations and employee organizations.The FG is funded through the annual National Budget approved by the Norwegian Parliament.In response to demands from the elected governments and its ministry, the Directorate continuously monitors the use of the FG through specific performance measures.
The context of competing and complementary measures of the FG must be understood in relation to the fact that the local NAV offices responsible for putting the FG into practice also function as partnerships between the state and the municipalities.Some Being a state measure and intended to promote collaboration between employer, employee and NAV during work inclusion efforts, the FG can be used in conjunction with work inclusion programmes from the municipality or state, measures that are either internally provided or outsourced to other service providers.The formal intention is that the NAV front-line worker writes an FG contract together with the employer or the outsourced service provider and the employer, plus the employee early in the process, listing all the work inclusion measures that NAV will be providing for that specific case, as well as contact information, a follow-up plan and an overview of the rights and responsibilities of the different parties involved.As such, the intention is that the FG is complementary to the other measures available, functioning as a tool to support the collaboration between the different parties, in addition to giving the employer the crucial security (Schafft & Spjelkavik, 2014a) that NAV will coordinate and follow-up when necessary.As we will see, however, the use of externally provided services was not seen as compatible with using the FG in the local NAV office in the case study.Some of the reasons for this may be understood by analysing the organizational context at the mezzo-level of the implementation system.

The organizational field of NAV
The organizational hierarchy of the public work inclusion field in Norway starts at the central level of the Government.The Ministry of Labour and Social Affairs has a responsibility to ensure that labour and welfare policies correspond to the aims set out by the Norwegian government.The Ministry's responsibility also includes defining the outcomes they are aiming to reach, as well as reporting on the results from the previous year through a continuous dialogue with the Directorate.The Ministry is responsible for the policy formation, and has the formal responsibility for laying down the priorities and implementation strategies to be put into practice by NAV.
NAV was established in 2006 as a result of the largest public reform of recent times in Norway (Christensen, Fimreite, & Laegreid, 2014) (Riksrevisjonen, 2013).
The Directorate of Labour and Welfare, the county and local NAV offices with municipal partnerships, as well as these specialized units, are all part the NAV organizational field at the mezzo-level of the work inclusion sector.The NAV local offices are where the innovations are put into practice with the intended target groups.We found that the implementation strategies of various work inclusion measures and innovations are planned and operationalized at the organizational level through leadership and communication channels, the integration of information systems, resource allocation, and through prioritizations at the specific NAV county or local office level.Since the FG was made permanent in 2008, a coordinator at the Directorate has been organizing the implementation effort through allocated FG coordinators at each of the 19 county NAV offices.Their responsibility has been to support the implementation of the FG at the local NAV offices in their respective counties (Riksrevisjonen, 2013).Together with other relevant coordinators and leadership at the different levels of NAV, these FG coordinators communicate the intentions of the FG and how it is to be used and prioritized in work inclusion practices at the local level.At the specific NAV county office in the case study, especially from 2012, the FG was communicated from the highest level of leadership as very important, and to be used in all 'work training' cases with a Work First approach, both those that front-line workers followed up internally, and those outsourced to external service providers.The use of FGs at the local NAV offices was monitored by the county FG coordinator, and evaluations of quantity and quality were reported back to the local offices annually in order to provide examples of good FG work, and to make necessary adjustments to the progress.
Contextual influences at the organizational level that we found through the empirical enquiries at the specific NAV municipal office include the way the resources were distributed and prioritized by local leadership, such as the number of service recipients in each front-line worker's portfolio, the organization of work between the front-line workers and the organization of the workers' responsibilities for each service recipient.
In addition these include work inclusion practices and collaborations among the front-line workers, work inclusion measures and service providers usually chosen at this front-line division and the local leader's communication of which performance measures to focus on at any given time.As an example, this specific office had a strong focus on following up people who had been out of work for a long time through the municipal 'Qualification programme'.By trying out different ways of organizing work directed at this target group, communicating a strong focus on this work at a leadership level and providing flexibility in the use of resources in this municipal programme, the office had some very good results for this specific target group.At the time of the case study, the selected division at this local office was organized into two mixed teams, consisting of front-line staff working with the municipal programme and staff working with a state-run 'Work Assessment Support Programme'.The state-run programme was based on a diagnosis-centred and medical assessment of work ability, while the municipal programme was more flexible and targeted people with socio-economic or health related needs for support.The staff in the state-run programme had tight assessment schedules, while the municipal programme was organized in a more flexible way.At this specific NAV office, we found that the front-line staff in the municipal programme had a substantially lower number of service recipients in their portfolio compared to staff in the state-run programme.This meant that staff working on the municipal programme had more time and resources for direct follow-up.The 'Work Assessment Support Programme' workers had tighter deadlines, less time for each service user and tended to more extensively outsource the work inclusion followup to external service providers.

The front-line of NAV
Communication at the various levels of the FG implementation, as well as the evaluation efforts by the FG coordinators and the resources allocated to the different work inclusion programmes, are all examples of contextual influences at the organizational level of the implementation system, which may have a direct or indirect impact on whether front-line staff implement the intended change in their everyday practices of micro-level interactions with colleagues and with the target groups.The specific work situation of the front-line workers had an important influence on the implementation outcome at the end of the line.As the running example shows, the intent at the levels of policymakers, leadership and FG coordinators was that the FG should be used in all types of work inclusion efforts where ordinary employers were involved in order to formalize the relationship between NAV, the employer and the employee.This also included using the FG when Work First follow-up services were outsourced to external public, non-profit or for-profit work inclusion service providers in order to secure continuity for the employee and security from NAV for the employer.
In the preliminary analysis of case-meeting observations and interviews, we found that the practical understanding at the micro-level of when to use the FG was different at the front-line in the specific local office in the case study than centrally intended.The front-line staff described challenging work situations and a strong desire to provide the best services to their service recipients in compliance with the resources available to them.They did not see the practical point of using the FG when outsourcing the followup to external service providers.In addition to making sure that the recipients were given a service provider who had the time and resources to follow up, and who was specialized in the field, as well as having access to an extended network of employers, a crucial point of the outsourcing, at least as many front-line workers saw it, was to cut down on their own work load to cope with their demanding work situation.The external service providers had the role of following up the employer and employee, so using the FG for outsourced cases seemed purposeless to the loaded NAV front-line worker.We found this to be the case, especially for front-line staff responsible for the state-run programme.

Implications of the contextual complexity for implementation
Factors linked to the innovation to be implemented, its original intentions and use, in combination with contextual factors at all levels of the implementation system, will eventually influence the implementation outcomes at the front-line level of the organization (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015).When we analyse the FG as both a public service innovation being shaped by actors at the different levels of the implementation system, and as a new process for delivering a service that needs to be blended with existing practices in the front-line, we see that the innovation itself and the technicalities surrounding it will affect the implementation process and outcome.Moreover, the possibility of incorporating the innovation into everyday practices is dependent upon contextual factors at all levels, especially those directly constituting the specific work situation of the front-line staff.Using the metaphor of the water filtration system (Sandfort & Moulton, 2015), the FG as a public service innovation has been filtered through a work inclusion policy, where several actors with various interests and authorities moulded its development.Illustratively, market conditions at a macro socio-economic level have created a work inclusion industry consisting of both private and non-profit sectors, adding strong actors in the policy field with diverse authority and interests.The policy field's budgeting structures and political priorities placed on NAV at the mezzo-level of the organizational field cause NAV to continue the outsourcing of work inclusion services to these private and non-profit actors.In addition, the policy intent has been for the FG to be applied, both when using internal follow-up and with external work inclusion providers.When the FG further filtered through the organizational field of NAV consisting of a strong, hierarchical structure and several levels of leadership, focused on in the case study, the intention for the FG to be used in both internal and external work inclusion cases remained.
However, at the desks of the front-line staff in the local NAV office, the FG appeared as only applicable with cases of internal follow-up.We found that the original intention at the policy level, and the organizational level for the FG to also be used in outsourced work inclusion cases, did not match the practical application at the front-line level in the study.
The mismatch between the policy intent of the FG and its practical use implies that implementing the FG into the everyday practices of the front-line workers requires an understanding of how the innovation fits into their specific work situations.
Understanding the context of the current work situation, and the practices of these front-line workers, may provide insight into factors that influence their decision of which work inclusion methods to use.Compared to the ones working with the municipal 'Qualification Programme', the front-line workers responsible for the state-run 'Work Assessment Support Programme' had a larger portfolio of service recipients, as well as stricter deadlines, thus leaving fewer resources and less time for direct follow-up work with employers and service recipients.The outsourcing of follow-up services to external providers was a widespread practice among these staff members, both because they did not see themselves as having enough time to give good quality follow-up, and because outsourcing services freed up time for them to cope with their own work situation, e.g. in meeting necessary deadlines in the state-run programme.
As a result, staff responsible for the state-run programme did not see the point of using the FG for outsourced services.Outsourcing the work inclusion services also implied outsourcing the responsibility of following up and working with the employer, leaving the FG redundant in the eyes of many front-line workers.The mismatch between the intention of the FG at higher levels of the implementation system, and the way it was practised at the front-line, shows how interests and policy intentions, prioritization and resource allocations at the macro-and mezzo-levels of the system affect the implementation of public service innovations at the operational level where the actual application of the innovation is to take place.
We have provided some insight into the FG implementation system (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015) in the study, including parts of: 1) the wider context at the macro-level of the work inclusion policy field and the policy context of the FG, 2) the mezzo-level of the organizational field of NAV, and 3) the micro-level of the work situations of the front-line staff at a specific local NAV office.This understanding of the context in which the implementation occurs is at the heart of using a critical realist case study approach to explore and explain a social phenomenon.
Insights into the history and context of the innovation in focus, and links between these and the context of macro-, mezzo-and micro-levels of implementation, allow us to find feasible causal tendencies or mechanisms that can explain some of the outcomes of the FG in the case study at the local office.By exploring the context at the three levels, the work situations of the front-line staff and the practice of outsourcing, we point to one plausible explanation for why the FG is not internalized at this specific front-line office, namely that the use of outsourcing has an influence on its implementation.We maintain that this analysis demonstrates some of the contextual complexity that needs to be considered when planning and practising implementation work.The front-line exclusion of FGs when outsourcing services is not even close to a complete explanation of the FG implementation outcome at the NAV office in the case study.It does, however, illustrate how contextual factors at the macro-and mezzo-levels of an implementation system impact on its application at the micro-level of case-to-case decisions, which eventually accumulate into its implementation (or otherwise) at the organizational level.

Conclusion
We have illustrated how the FG as a specific public service innovation has moved through several levels of leadership, including the political, legislative and then operational levels of NAV, finally ending up in the hands of front-line staff at a local office.The action of 'carrying out, accomplishing, fulfilling, producing, completing' (Pressman & Wildavsky, 1973 pp. xiii-xv) has to occur at the micro-levels in the local NAV office, where front-line staff to a greater or lesser degree put the intended change into practice, case by case, in their interaction with the target groups, and in response to the work situation in which they find themselves.To understand implementation outcomes in these complex processes, it is necessary to analyse them in relation to the multi-level, nested contexts, which influence the micro-level processes in the work situations of the front-line staff.In particular, we explored how the outsourcing of work inclusion services at the local office of the case study seemed to compete directly with the application of the FG.This can be understood as a result of the way the FG has been applied by front-line workers to fit their practical everyday work situation, despite the original intentions at central levels of the implementation system.We pointed out that the practical everyday work situation accountable for the mismatch is shaped by decisions about resource allocation, priorities and performance measures at the levels of leadership in the organizational field.As such, we use this as an illustration of the complexity involved in implementation work, and hold that analysing the implementation process in relation to its wider context using a critical realist lens, allows us to understand how the character of the intervention itself, its context and individual decisions by front-line workers may act as causal mechanisms to influence the final outcome.Such an exploratory phase of a case study is of substantial importance in finding explanations for 'why things are as they are' (Easton, 2010, p. 119) in the implementation process, as well as when planning implementation work.

Limitations of the study and implications for practice
In this article, we have given a narrow account of the inception of the FG and some insights into important features of its implementation system.Several important elements of the innovation process and the implementation system are left out.Also, individual decisions and implementation strategies at the various leadership and coordinator levels of the implementation system are not emphasized.This is primarily to simplify our example used for this article.Our intention is to give a simplified illustration of the importance of analysing the wider context, in addition to the specific work situation into which a public service innovation is to be applied, in order to find explanations for implementation outcomes.To demonstrate this, we discussed: 1) the demands for using outsourced services from the policy and organizational fields, 2) the demanding workloads and limited resources originating at the organization field, and 3) how the practical experience of the work situation at the micro-level at the front-lines became amalgamated to give one (of many) plausible explanation(s) for the limited implementation of the FG at the NAV office in question.We are not proposing that this is the explanation of the FG implementation result at the NAV office in the case study.
But it does illustrate the complexity involved in implementation work, and may serve as a reminder to implementation practitioners at all levels to consider the wider context, as well as the specific work situations of front-line staff in their implementation work.
As such, we have also provided some tools for analysing the contextual setting by exemplifying our point in the application of the three-level implementation system (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015) in our illustration.

Implications for further research
In this article, we have focused on a limited illustration of one possible mechanism explaining the implementation outcomes of the FG at the specific NAV office in the case study.This mechanism, in line with a critical realist approach, cannot be seen in isolation from other mechanisms that combined will provide a more complete explanation of the implementation outcome (Wynn & Williams, 2012).It will be fruitful to tease out other plausible explanations for the implementation outcome in this specific NAV office, in addition to analysing the roles of the FG coordinators and leaders at various levels of the organizational field, and the roles of the front-line staff in this specific implementation process.We propose that using an in-depth explorative single case study of a specific public service innovation, particularly explored through the lens of a critical realist case study, is of substantial value for researching innovation and implementation processes as complex, multi-levelled and dynamic without 'underemphasiz[ing] the interactions between the different levels' (Wong, 2005, p. 2) of the implementation system (Moulton & Sandfort, 2016;Sandfort & Moulton, 2015).
We maintain that similar studies may be repeated for new insights into various policy fields; or for the purpose of comparison could be conducted with different public service innovations in the same contextual setting, or with the same public service innovation in different contexts.
2. The translated name, 'Facilitation Guarantee' from the Norwegian 'Tilretteleggingsgaranti', is somewhat limited.The Norwegian name implies both making any necessary adaptations in the work place, as well as facilitating the relationship between the employer and the employee.In the following, we call the Facilitation Guarantee FG for simplification purposes.
3. The report also resulted in a call for an evaluation of the FG by the Control Committee of the Norwegian Parliament, eventually leading to a revision process of the FG from 2014, and ending in a revised version put into effect on 1 January 2016.This case study is focusing on the implementation process of the FG up until the revised version was put into effect.
4. Among the reasons for the revised FG of 1.1.2016was a tidying up of the jungle of work-inclusion measures that have grown forth in the last few decades.
of the alternative work inclusion programmes available to front-line workers are financed by the municipality, such as the 'Qualification Programme', which delivers work inclusion measures to the target groups internally in NAV, or outsources the services to external service providers.Other work inclusion measures are financed directly by the state, such as the 'Work Assessment Support programme', with work inclusion efforts followed up internally in NAV through measures such as 'salarysubstitution' and 'work training places', or often outsourced to external service providers.The emerging focus on a Work First strategy and attention to the needs of the employers over recent decades, as well as the NAV reform of 2006, have resulted in an increasing number of available work inclusion measures from both the state and municipality to front-line workers in NAV.According to our document review, these included approximately 40 initiatives for work inclusion purposes in 2015. 4 . The purpose was to design a new public organization that could offer a 'strong structure for early intervention and coordinated support' by 'integration of the public employment service, the social insurance and parts of the municipal social assistance into [one] Labour and Welfare Administration' (OECD, 2013, p. 5) An important consequence of the NAV reform was the establishment of local NAV offices working as one-stop shops with the purpose of providing citizens with all their labour and welfare needs through one office and one front-line worker (Andreassen & Aars, 2015).Structurally, the Directorate is at the top hierarchical level of the state-run part of NAV, and is divided into different lines of command, including a service department, a benefits department and an accounting department, as illustrated in the organizational chart in the Appendix.The service department, relevant in this study, includes 19 NAV county offices.These offices are responsible for the local one-stop shop NAV offices at the municipal level in partnership with 428 municipalities (NAV, 2013).The specific NAV local office in this study is organized into different divisions at the front-line level, serving distinctive target groups according to their levels of need for work inclusion facilitation.Especially for this case study, one division of front-line workers is responsible for the work inclusion efforts for people who have a specific need for facilitation and follow-up because of socioeconomic or health-related reasons.In addition to the local NAV offices that work directly with the target groups, NAV consists of several units at the county level with different expertise, which support the local offices on particular issues.These include units with expertise on the local labour market (NAV Market), expertise on working with companies that are voluntarily part of the Inclusive Working Agreement (NAV Inclusive Work Place Support Centre) and expertise on health-related challenges and work inclusion adjustments (NAV Employment Counselling) (NAV, 2013).In addition, a separate division in the service department with units in each of the 19 counties has a special expertise on assistive technology and work facilitation for people with disabilities (NAV Assistive Technology and Adaptions).Together, these units play various roles in the implementation of work inclusion measures depending on the implementation strategies set forth at the Directorate level and the county levels of NAV.The NAV Inclusive Work Place Support Centre and the NAV Assistive Technology and Adaptions are particularly involved in the role of implementing the FG