Reorganizing the welfare state administration

Abstract

This article addresses the big welfare administration reform in Norway. The reform is a merger of the employment and national insurance administrations, combined with more formal collaboration with the local government social services administration. The reform introduced a mandatory partnership model between central and local government. This model is a hybrid of hierarchy and network. A substantial dilemma in the particular partnership model chosen is how to enhance vertical control at the same time as sustaining the autonomy of local government. The partnership model created to solve this dilemma represents a delicate and ambiguous balance between accountability to the central government and to the local council.

Key words:

• Partnerships
• network
• accountability
• reform
• welfare administration
• Norway

Introduction

This article addresses one of the largest public-sector reforms in recent Norwegian administrative history. The reform – named the NAV-reform – is a merger of the employment and national insurance administrations into a new employment and welfare administration, combined with more formal collaboration between this new administration and the local government social services administration. The aims of the NAV-reform are to increase work participation and to make the administration more user-friendly, more holistic and more efficient. The merger of the employment and insurance administrations in Norway is rather unique. A merger of two large welfare administrations, which combined count for around one-third of the state budget, has never taken place before. The NAV-reform is further complicated by the fact that the social services administration – a local government responsibility – is also included. A central component in the reform is thus the introduction of a mandatory partnership between central and local government. This is an innovation in the Norwegian political administrative system. Never before have central and local government co-operated as a formal partnership in the form of a one-stop-shop system (the NAV-service offices) at one geographical level. The reform can be seen as a ‘whole-of-government’ or ‘joined-up’ government initiative intended to increase the capacity of the political administrative system to address ‘wicked problems’ cutting across existing policy areas and to improve vertical and horizontal co-ordination in the fields of policy and implementation (Christensen and Lægreid 2007).

Our main focus here is the mandatory partnership model as a way of fulfilling the aims and intentions of the reform. We will particularly focus on the effort to provide some of the most important services of the welfare state through an organizational structure that in many ways resembles a network. Three questions are central: first, what are the main features of the partnership? Second, why has the partnership been established? Third, what are the consequences of the partnership, with a special focus on how the partnership challenges accountability relations in the political-administrative structure in Norway.

Our empirical data are a combination of personal interviews with forty-three central political, administrative actors and stakeholders at central government level. The interviews are conducted from November 2007 to January 2008. We have also used reports from public commissions, white papers and propositions from the Government and Parliamentary documents. Findings from other studies of the central reform process and of the implementation of the partnership locally (Alm Andreassen et al. 2007; Nyhuus and Thoresen 2007; Christensen 2008; Dahle 2008) have also been used.

We will start by presenting our theoretical approach, focusing on the concepts of partnerships, networks and accountability. Second, we will present the partnership empirically by focusing on the central features of the partnership model chosen in NAV and how it has been implemented. The driving forces and central actors behind the partnership, and potential implications of the partnership are our main focus here. Third, we will end the article with a discussion of this particular partnership model.

Theoretical Approach: Partnerships, Networks and Accountability

Partnerships have become a popular tool in the governance of welfare. They are designed to enhance collaboration and co-operation across boundaries in public services (Glendinning et al. 2002; Sullivan and Skelcher 2002). Repeated efforts to achieve co-ordination are a main argument behind using partnership models in the public sector. Partnership is an umbrella concept that covers many types of co-operative arrangements (Linder 2002; Grimsley and Lewis 2004). Analytically this makes it an elusive concept because it is difficult to define precisely (Powell and Glendinning 2002). A minimum definition of partnership is that it involves at least two organizations with some common interests or interdependencies. Partnership is thus a joint working arrangement that requires a certain level of autonomy to determine or implement a programme. The arrangement involves a degree of trust, equality, reciprocity and also commitment. Central features of partnerships are lack of hierarchy and lack of cohesiveness. A common assumption is also that the involved actors have clear goals regarding the aims of the partnership. Normally the structural arrangements around the partnership are rather loose.

There are different kinds of partnerships. Mörth and Sahlin-Andersson (2006) classify partnerships along two dimensions – degree of formality and degree of permanence. We will add two further dimensions to these: the degree to which private actors are involved and the degree of voluntariness. Partnerships can thus be very informal, time-limited, voluntary and with a strong private component, but also highly formalized, mandatory, permanent and with a weak private component. Even though it normally is a common understanding that partnerships are based on collaboration between equal partners, it is easy to find examples where this is not the case, where one partner is stronger than the others. In partnerships between levels of government central government will in many cases be the senior partner and local government the junior partner. The basic point in the federal US system of ‘shared governance’ is different forms of authority in these relationships (Elazar 1987).

Partnership is closely related to networks. Networks – particularly between public and private actors – have recently been introduced in most western democracies as a way to increase the capacity of the public sector to deliver services. The use of networks in political-administrative structures has, however, also been highly contested both politically and academically. Klijn and Skelcher (2007) distinguish between four conjectures on democracy and networks: an incompatible, a complementary, a transitional and an instrumental conjecture. The first argues that representative democracy and networks conflict; the second that networks support democracy by engaging a wider network of actors in the policy process; the third argues that governance networks are part of a transitional process from state-centric government to a network form consisting of decentralized nodes of authority; while the last conjecture is that powerful governmental actors increase their capacity to shape and deliver public policy in a complex world through the instrumental use of networks (Klijn and Skelcher 2007: 598). This last conjecture challenges the ‘governance without government’ thesis of Rhodes (1996) that networks are self-organizing, that the Government is only one of many players and that there is a strong horizontal component in the networks (Bache 2000; Davies 2002). In contrast this instrumental conjecture implies that the central government is a powerful actor that creates networks and partnerships in order to realize its projects, and that the networks can have an agency form, created in response to a national mandate to be a delivery arm for a national policy initiative that requires inter-organizational co-operation at the local level (Skelcher et al. 2005). Networks and partnerships might then be seen as an effective response to the predictive contingencies for co-ordinating different tasks and levels in the public sector (Hudson 2004).

As a rule, modern polities are organized according to the principle of purpose, which makes them vertical in nature and characterized by strong sectors and weak co-ordinating mechanisms (Kettl 2003). This implies that vertical co-ordination within specific sectors may be good. The tasks of the welfare state, however, represent a complex and fragmented area of government, and a growing number of cases and problems do not fit into this traditionally sector structure. In political-administrative structures solving such problems is a challenge, mainly because it implies horizontal co-ordination between different sectors.

In the NAV-reform purpose (sector) and territory as principles of specialization are active at the same time. Three main tasks are involved – employment, insurance and social services – as are three levels of government: the NAV-service offices (one-stop-shops) are situated at the municipal level, and the NAV-agency (the merged central government organization) has regional branches at the county level. In addition to the partnership model we thus have a multi-level governance system in which tasks are carried out at different levels of government, implying increased interdependence of public agencies operating at different territorial levels, often in a complex system of overlapping jurisdictions (Bache and Flinders 2004; Marks and Hooghe 2004). Tasks can rarely be treated independently of each other, the different levels have to collaborate and co-ordination between levels is as important as co-ordination between sectors.

There are several approaches to co-ordination in such a multi-level system with overlapping principles of specialization (Lægreid and Serigstad 2006; Morris et al. 2007). One is the classical hierarchical view that assumes that work is divided by sector, often used synonymously with purpose or tasks (Gulick 1937). Co-ordination is achieved by the use of hierarchy, legal authority and specialization of tasks and is a vertical internal process with clear lines of authority. Its strengths are related to vertical accountability, role specifications and handling of routine tasks. This is a top–down approach based on the hierarchical model. Such a model presumes that the organization of the welfare administration must begin at the top and be directed downwards, implying strong political control from central government. The idea of top–down co-ordination is derived from the notion that the organizations to be co-ordinated have already been identified by headquarters co-ordinators, that the relationship of these organizations to each other is well understood, that agreement has been reached about what objectives will be accomplished by altering certain of these inter-organizational relationships and that the authority and means to alter these relationships exist. In other words, it assumes that having a hierarchy will facilitate implementation. The problem in the context of welfare administration, however, is that several of these assumptions are only partly fulfilled, and the problems of co-ordination do not lend themselves well to hierarchical direction (Wise 2002: 141).

For complex, unstructured and rapidly changing problems, which, we have argued, are characteristic of the welfare sector, a network approach may be more suitable (Wise 2002; Kettl 2003). Hudson (2004) claims that networks are more likely than the main alternatives – markets and hierarchies – to address the key factors that bring potential partners together. A network approach understands co-ordination as the interaction of interdependent actors from different traditional hierarchical structures and from outside such structures. They pay less heed to formal top–down authority and rely more on negotiations and mutual adjustments and on bringing together organizations to pool resources and knowledge. This network model scores high on adaptability and flexibility, but accountability may be reduced and ambiguous, and steering may be more difficult.

A third model, which can be labelled the hybrid model, combines the hierarchical and the network models. In this model there is a stronger supervisory and regulatory role for semi-independent central agencies, which is supplemented by a partnership model in each local employment and welfare office. Along the vertical dimension the central agency has a semi-autonomous status, which represents more than ‘delegated hierarchy’ from the parent ministry. The agency operates on the principle of professional knowledge and should be free from instruction by the cabinet or individual ministers (Christensen and Lægreid 2005). It carries out regulation using its own delegated regulatory power, resources and responsibilities. On the horizontal dimension this model tries to enhance overall responsibility for welfare state policy through local partnerships with municipalities and also central framework agreements with the municipalities' interest organization. Thus this hybrid model gives statutory power to the central agency and some co-ordinating responsibility to a superior ministry and shares responsibility for welfare state tasks with the local authorities through a partnership model.

A main concern that arises when partnership is used within the public sector is the problem of accountability. Accountability is a multi-dimensional concept. In a hierarchical model the concept of accountability is here primarily related to upward accountability to political sovereigns (Christensen and Lægreid 2002). The network or partnership models will make a model of strictly hierarchical responsibility from the top less applicable. Partnerships need some level of independence but at the same time they shall be accountable upwards to politicians, horizontally to other agencies and local government and downwards to citizens. They thus have to face the challenges of political as well as administrative and bureaucratic, legal and professional accountability (Pollitt 2003).

In the complicated organizational structure chosen in the NAV-reform, mandatory partnership between central and local government is an important measure for co-ordinating processes as well as results between the involved agencies and actors. A central accountability issue that rises from the new arrangement is how the relationship between government and networks impinges on ministerial responsibility. It is generally accepted that ministries should be allowed to give some interpretative guidance in the partnership on how to carry out tasks. This is expected to be general guidance and not directed at specific cases but in practice they allow rather strong governmental control. The partnership model might also weaken accountability to political bodies at the local level. A joint front-line unit is supposed to balance local accountability to the municipal council with vertical ministerial accountability. Weaker upwards accountability to the parliament and accountability to the local council may, however, be supplemented by stronger downwards accountability to users, clients and citizens. A partnership model scoring high on formality and permanence is constrained by procedural and substantive rules that define its discretion. Autonomy from direct political control does not automatically mean immunity from public accountability. Thus, responsiveness towards users and clients might become a substitute for accountability in the control of autonomous central agencies.

The governance literature in general and the related literature on accountability in governance is in large part concerned with governance and partnership as phenomena in which private actors are a central feature (Rhodes 1996; Skelcher et al. 2005). But there is also a more state-centric approach to governance (Peters and Pierre 2003) in which a public–public partnership is a main component. In our case the partnership is explicitly public–public. Outward accountability relations to civil society organizations and to private sector stakeholders become less relevant, but also the public–public partnership has to handle situations in which hierarchical accountability is challenged with accountability to local actors as well as to different professional actors. And the problems of many hands (Thompson 1980) are as important in public–public partnerships as in public–private partnerships. Challenges connected to accountability to whom and for what are thus also crucial for public–public partnerships (Boven 2007). We will claim that networks as a co-ordinating mechanism in the chosen partnership model in NAV supplement rather than substitute for the traditional welfare state hierarchy (Verhoest et al. 2007). One consequence of this is that accountability relations are challenged. In the following paragraphs we will scrutinize this argument further.

The Partnership Model In Nav – Some Central Features

In 2001 the Norwegian Parliament (the Storting) requested that the Government (a Centre-Right minority coalition government) should consider a merger of the three main welfare administrations – the insurance administration, the employment administration and the local social services administration. The Government's proposal, which was based on the work of an inter-ministerial working group, was that the welfare administrations should continue to be divided into three parts, but a one-stop-shop at the local level should be created encompassing the latter two services (St.meld.no 24 (2002–3)). The Storting rejected this proposal, and the Government set up an expert commission to elucidate anew the organization of the welfare administration. Their proposal (NOU 2004: 13) was very similar to the one that had already been rejected. It stated that there should be one administration for income and employment, one for pensions and social rights and one for social services, and that the latter should remain a local government responsibility. The Government was, however, reluctant to present the Storting with a proposal that was more or less the same as the rejected one.

When the expert report was delivered, a new ministry had been established owing to inter-cabinet rotations. For the first time responsibility for both the insurance administration and for the employment administration was integrated into one ministry, and the laws regulating local governments' social services administration were also part of this new super-ministry's portfolio. Merging political responsibility paved the way for a co-ordinated effort to implement the reform. The minister in the new Ministry for Labour and Social Affairs was a former director of the National Insurance Administration, but he had also had experience as an executive administrative manager at local government level and been a director of the Norwegian Association of Local and Regional Authorities (KS). He signalled early on that he would not follow up the proposal from the expert commission and that his priority was to build a user-friendly organization. Therefore he strived to create a welfare administration in Norway that was oriented towards user-participation. To do this he advocated merging the administrations of insurance and employment at all levels but leaving intact the autonomy and tasks of local government and creating a joint front-line service in each municipality bringing together the merged employment and insurance administrations and the social services administration. The partnership model was introduced as a way of fulfilling these three goals and was promoted as the missing link between local self-government and ministerial responsibility. This model differs from the ‘collaboration’ and ‘collaborative models’ that were so important in the Norwegian public sector when welfare politics was implemented in the post-World War II period (Fimreite et al. 2007) by having less partners and also more specific partners but first and foremost it differs by being mandated.

The partnership model was intended to create joint operative solutions with two owners, the municipality and the government agency. The solutions imply dual accountability relationships – upwards within the sector to central government and horizontally to the local government authorities. The model was based on the following principles:

•	The partnership is compulsory by law and mandatory for all municipalities.
•	There should be one welfare office in every municipality (there was also an opening for solutions where two or more municipalities shared an office).
•	The welfare office should be a joint front-line service implying co-location of the social services administration and the new employment and welfare administration (the amalgamated employment administration and insurance service administration).
•	The welfare office could either have a joint management or a dual management arrangement – with one manager from the municipality and one from the employment and welfare administration (government).
•	From the municipal side the welfare office should as a minimum include financial social assistance, financial advices, housing of homeless people and the right for each individual to have a plan worked out connected to social and welfare services.

The one-stop-shops are based on fixed, regulated and binding co-operation between the central and the local authorities, which will be laid down in local agreements, negotiated between the regional NAV-office and the individual municipality. These local partnerships are the manifestation of the partnership model and they are not voluntary. There is however a trade-off in the partnership arrangements between central government's need for standardization and local government's need for local adaptation and flexibility. The central agency worked hard for a mandatory arrangement, for the principle of a unitary management model and for a standardized task portfolio for the local front-line office. This was not acceptable to the actors defending local self-government, such as the KS – which negotiated at the central level on behalf of the 431 municipalities and the Ministry of Local and Regional Government. In line with the Norwegian consensus-oriented policy-making style the result was a compromise: it was agreed that the arrangement should be mandatory, that the unitary management model should be recommended but a dual management model could also be accepted, that there should be some defined minimum municipal tasks included in the front-line unit, but that other municipal tasks could also be included in a specific local partnership. Thus the partnership arrangements avoided too much detailed top–down steering and allowed for local flexibility. The flexibility regarding which municipal tasks could be included and the flexible management model were both concessions to local self-government and corresponded with the existing Local Government Act, which grants local authorities a high level of discretion when it comes to organizational matters.

The basis for the local partnership agreements was a framework agreement at the central level between the Government and the KS. This was the first time such a central partnership agreement had been settled between the Government and the KS and was supposed to send an important political signal to the government agency and to the individual municipalities on the establishment of a local partnership agreement for the front-line service. The central partnership differs from the local because it is not mandatory and because the KS cannot take decisions that commit the municipalities. The central partnership constitutes a consulting arrangement between the Government and the KS, but nevertheless plays an important symbolic role. The presence of the central partnership is thus more important than the actual substance of it.

So far about 160 local partnerships have been established based on local agreements. Most of these offices have unitary management, but there are also some with dual management models. Most of the managers come from the former employment administration or from the insurance service, but there are also some managers who are local government employees. There are substantial variations in what tasks the municipalities include in the partnership. Many municipalities have added tasks such as alcohol and drug prevention, immigration, psychiatric health care and child welfare. Some partnerships have also evolved their own specific aims for the local one-stop-shops in the agreements while others have not. It is obligatory for services included in the partnership to be co-located. Some one-stop-shops are also co-located with other local government services, but most of them are not.

Summing up, the partnership model in the NAV-reform is a public–public partnership. The model is based on partnerships at the local level. Joined-up government at the central level is addressed to some extent by involving the municipalities' interest organization (KS) in formulating a general framework through a central partnership agreement. The NAV-partnership is mandatory, permanent, formalized through contracts and agreements and includes only public partners at the central and local levels. The NAV-partnership model thus differs from a ‘classic’ public–private partnership (PPP) in two essential ways: private actors play a minor role in the partnership, and it is mandatory. It is also based on rather strict formal arrangements. The similarities between the NAV-partnership and a PPP are that both represent a close collaborational and contractual venture with some durability and have been concluded between organizations that rely upon agreement between the partners in return for some positive outcomes for each of them and provide citizens with public goods and services (Carroll and Steane 2000; Mörth 2007). Other overlapping features between the NAV-partnership and PPPs are that they are about power sharing, and there is a common understanding that the partnership involves equal partners. We will claim that these features make the model chosen in NAV a partnership – although a special type of partnership.

Why Was a Partnership Model Chosen in Nav?

The minister of the newly merged ministry for Labour and Social Affairs proved to be a clever political entrepreneur, who succeeded in getting active support from central civil servants in constructing this particular partnership model, by changing the government position through negotiations within the cabinet and co-opting the KS. The minister made great efforts to engage in a dialogue with the Storting. By gaining support from the Labour Party, which was the main opposition party, he managed to make the Storting more receptive to a solution that was not its first priority. He made a great effort to integrate the central stakeholders in the decision-making process and to involve the political parties at the Storting in informal consultations.

The partnership idea came up in a creative co-operation process between this minister and a few central top civil servants in his own ministry. A main strategy was to achieve joint understanding and agreement with the KS and also with the opposition in the Storting. The KS was involved in creating reference groups to counterbalance the dominance of central agencies favouring a standardized model that would limit the autonomy of the local authorities. The KS also played an important role in legitimizing the process in the municipalities.

The government proposal received the support of the Storting, even though it did not include a full merger. Several of the political parties said it was imperative that the new model at the local level should produce better cooperation among the services, and they also stressed the user-friendliness and cost-efficiency aspects of the proposal.

Summing up, there was a strong political wish in Norway to establish a front-line service with an employment and welfare office in every municipality. A partnership between central and local government was designed to provide co-ordinated services better adapted to users' needs and to replace the present system of three different offices in each municipality. A network of local offices would constitute a co-ordinated front-line service with responsibility for employment, sick leave, medical and occupational rehabilitation, disability pensions, financial social assistance, pensions and family benefits. Central government responsibility would be concentrated in one agency: the employment and welfare agency (NAV), which was to co-operate closely with the individual local authorities. The partnership was invented by a new minister in a new ‘super’ ministry and his civil servants who were searching for a ‘Columbian egg solution’ that would simultaneously establish a one-stop-shop in every municipality in which all three services were included and accept the present division of tasks and responsibilities between central and local government.

NAV is a radical departure from the traditional employment and welfare administration, but it also represents a complicated arrangement of central–local government co-operation and division of responsibility. This arrangement differs from former collaboration between the two government levels. Even so the formal division of responsibility between the central and local authorities has not changed in a fundamental way. The hierarchy is still there: political responsibility for the national insurance service as well as for labour-market policy remains with central government, while financial social assistance remains a discretion-based, means-tested benefit under local politicians' control. Both co-ordination between the three administrations at local level and the one-stop-shop idea, which is akin to a network, represent challenges for administrations/services and government levels accustomed to territorial as well as cultural distance.

As a consequence of the Norwegian consensus-oriented style of decision making there is a strong wish to come up with compromises that everyone can accept. In the NAV case this means that ambiguity is built into the chosen solution that provides flexibility and room for different interpretations. The solution – the partnership model – represented something new, which has the potential to change one part of the Norwegian ‘living’ constitution; the principle of local government autonomy as strong and hard to challenge. Therefore the partnership model between local and central government as an intermediate organizational form combining the principle of local self-government and the principle of ministerial responsibility was quite an innovation. What made it possible was the ambiguity of the new principle; how it will be specified and practised is, however, an empirical question. The partnership is thus the product of a strong belief in the present political-administrative structure in Norway combined with a simultaneously felt need to modernize this structure. We now turn to focus on possible consequences and implications of the partnership model.

Implications of the Partnership Model

The plan is that the front-line service units (the one-stop-shops) based on local partnerships will be implemented in all municipalities by 2010. It is therefore too early to measure most effects of the reform directly. Concerns about implications and effects have so far been connected to whether it is possible to create holistic, integrated and seamless services locally (Alm Andreassen et al. 2007; Nyhuus and Thoresen 2007), to whether actors representing central government services will hold the upper hand in the local partnerships and by that out-maneuver the local government services (Dahle 2008) and to whether actors at central level can communicate with their local counterparts in a less standardized way and grant enough discretion for local offices (Askim et al. 2008). In this article we will not discuss consequences of such implications any further. Our main concern is potential effects of what can be interpreted as the rather blurred lines of accountability in the partnership model.

In a political-administrative structure based on a representative democracy clear accountability to the political executives is a central concern. An important question therefore is whether the NAV-reform will lead to more unambiguous and transparent accountability or whether it will instead produce an ‘accountability deficit’ (Baldwin et al. 1998; Christensen and Lægreid 2006). In a representative democracy elected elites are answerable to the public. The desire to get re-elected acts as a disciplining force on elected representatives. Publicity and transparency are thus important for accountability, because the electorate needs information in the re-election process (Aars and Fimreite 2005: 244).

Partnerships and networks operate according to different standards. They are often self-elected, their decision making is normally not as public as decision making in representative political institutions and they are not as formally organized as those (Aars and Fimreite 2005: 244). To claim accountability in network structures is more challenging than in politically formalized institutions. These arguments apply not just to decision making but also to service delivery in a political-administrative structure.

The partnership model in NAV tends, as we have pointed out, not to clarify lines of accountability. A key question in this model is how one can have joint action, common standards and shared systems on the one hand, and vertical accountability for individual agency performance on the other. The challenge is to balance better accountability to central government, accountability to local council and responsiveness downwards (Christensen and Lægreid 2007). There is an inbuilt inconsistency in the NAV-reform. It claims to empower users and clients, free managers, enhance administrative accountability and strengthen political control both from central and local political bodies. But in reality it is difficult to achieve these things simultaneously.

The NAV-reform is in many ways built on post-NPM views of co-ordination. Vertical co-ordination, meaning how the central government should secure control over the new services and standardize them, is important and can be based on (a) political control, indicating a traditional, centralized structure, (b) production, suggesting more devolution-oriented solutions and (c) rights, for example making greater use of independent appeal bodies. When it comes to vertical accountability the relationship between the ministry and the central NAV-agency is of great importance. The organizational solution apparently envisaged here is a more traditional ministry–agency relationship, where there is a balance between control and autonomy. Added to this is the internal vertical co-ordination inside the NAV-agency and the inter-governmental co-ordination between the central agency and the regional and local parts of the apparatus.

Horizontal co-ordination on different levels also has an impact on administrative accountability. The horizontal intra-ministerial co-ordination between the employment and insurance administrations and other areas in the ministry is no exception to this; neither is the inter-ministerial co-ordination between the employment and welfare administrations on the one hand and other related policy areas such as education and health on the other. Added to this there are also concerns about vertical co-ordination involving the mechanisms the ministry has to influence social services locally in the tension between sector-based and territorial specialization. The advantage here is that all the relevant areas now come under the ministry, potentially furthering co-ordination. The disadvantage is that the Ministry of Labour and Social Inclusion1 is huge, and the political leadership may have capacity problems. Whether the different levels of the NAV-agency will manage to co-ordinate employment and insurance services as planned is also an important question, as is local co-ordination with the social services inside the one-stop-shops. At the very local level co-ordination with locally elected political and administrative executives is also a challenge (Christensen et al. 2007a). Another major concern is whether the reorganization is overly geared to coping with multi-service users. While these users are indeed likely to be better off with a co-ordinated structure, the outcome of the merger for the remaining eighty-five percent who normally use only one of the merged services is more uncertain.

Given that these decentralized nodes are responsible for the most important services in the welfare state, it is important for accountability that the partnerships are publicly visible. The NAV-partnerships have so far been rather transparent. The actors are known, decision making is public and there are also procedures for appeal. In addition the attention of the media, researchers and also organized users is focused on the partnerships and their ability to deliver. To accept such public scrutiny is part of being accountable in a transparent democratic society, and it is important that politicians are involved in the partnership (Aars and Fimreite 2005: 245). Mayors take part in the NAV-partnership locally, and at the central level the minister is also directly involved. Decentralized nodes as networks may be held accountable through the internalization of values and norms (March and Olsen 1995: 154). To create democratic attitudes in the partnerships is of great significance if they are to operate with the kind of publicity and transparency needed to enhance democratic accountability. To create such attitudes is mainly the task of the political authorities – at both the local and central levels (Aars and Fimreite 2005: 245–6). The partnership is, however, a rather hybrid model given the structural constraints that constitute it and it is too early to know how the balance between different accountability relations will play out in practice.

The partnership arrangements differ from traditional political and administrative decentralization or delegation (Pollitt 2005) by the hybrid character of combining political decentralization to local elected bodies and administrative decentralization to subordinate administrative units. The public–public partnership represents an innovation in the Norwegian political-administrative system. In contrast to the traditional arrangements of either delegation to subordinate administrative bodies focusing on ministerial responsibility or political decentralization to local government in line with the doctrine of local self-government, the novelty of the partnership model tries to combine these doctrines. The organization of the welfare administration constitutes a double balancing act because it requires co-ordination between the responsible ministry and the central agency and its subordinate bodies on the one hand and between the government administration and the municipalities on the other hand. The partnership model is a hybrid of hierarchy and networks. It is an inbuilt inconsistency in the reform trying to handle objectives that are difficult to achieve simultaneously. A substantial dilemma is how to enhance vertical accountability and control within the welfare state administration at the same time as sustaining the autonomy of local government in this policy area. To solve this dilemma the partnership model was introduced, but it represents a delicate and ambiguous balance between accountability to the central government and to the local council and between specialization by sector and territory that has yet to be tried out in practice.

Conclusion

If we relate the NAV-partnership to Klijn and Skelcher's (2007) four conjectures on democracy and networks discussed in our theoretical approach, it is obvious that the model and its local manifestations bear most resemblance to what the authors call the instrumental conjecture. The central government has been a powerful actor in creating the partnership in order to realize its own project. The partnership is an agency form created in response to a national mandate to be a delivery arm for a national policy initiative that requires inter-organizational co-operation at the local level (Skelcher et al. 2005). Powerful governmental actors increase their capacity to shape and deliver public policy in a complex world through the instrumental use of partnerships. It might also be understood as a step in a transitional process from state-centric government to a network form consisting of decentralized nodes of authority. The power relations and tensions between central and local government within the partnership is resolved through a combination of hierarchy and negotiations, but in practice the central government seems to have the upper hand and the outcome tend to favour the interests of central government.

The main intention of this public–public partnership is to make it possible to operate a co-ordinated welfare administration through a one-stop-shop at local level and at the same time preserve the present division of tasks and authority between central and local government. As an organizational phenomenon such a partnership is ambiguous and not well understood, and how well it works in practice is an open question (Pollitt 2003). This kind of collaboration requires a high level of trust and healthy and active relationships between the involved partners to foster improved outcome (May and Winter 2007). There is no evidence to conclude that the partnership model represents a more advanced form of democracy than traditional representative democracy. One reason for this is that accountability relations are more ambiguous in partnership/network structures than in structures based on representative political institutions. There are less clear lines of accountability for decision making and service delivery in partnership – and hence also in the new Norwegian welfare administration.

The idea of a partnership that can cross territorial and sector-based silos in the welfare state system appears very attractive. Yet in the NAV-model the legal responsibilities and local government responsibilities of a traditional vertical organization are combined with an ambiguous partnership model with competing lines of authorities. The managers in the partnerships are supposed to be multiply accountable to their superiors in the regional and central NAV-agency, to the local council and political and administrative executives in the municipality as well as to users and clients. Different actors are responsible for different aspects of joint activity and the managers have several lines of accountability – towards the central government political executives but also political accountability to the local government political executives. Added to this they have legal, professional and bureaucratic accountability (Pollitt 2003).

Another problem is linking the sectoral objectives issued from the top down with the local formulated targets in the municipalities. Unless cross-cutting local targets are given equal status in the partnership with top–down organization-specific targets, the partnership will have difficulty becoming a major tool (Pollitt 2003). Partnership models have a strong positive flavour and are generally seen as a good thing. When partnership is on the agenda, it is, however, also important to pay attention to the fact that the ‘silo mentalities’ these structures and reform initiatives are supposed to attack often exist for very good reasons (Page 2005). Well-defined vertical and horizontal organizational boundaries are not only a symptom of obsolescent thinking (Pollitt 2003). Very often they are justified by principles underlying our political structure, such as division of power, predictability, impartiality, rule-of-law, professional confidentiality and protection of privacy. Breaking these boundaries may pose new challenges to the political administrative system, but a more serious issue is perhaps that it may also change the way our democracy operates.

Acknowledgement

This article is part of the ongoing evaluation of the reform in the Norwegian employment and welfare administration, funded by the Norwegian Research Council. An earlier version of this article has been presented at the workshop ‘The Shadow Line of Accountability, Performance and Control across Public Private’, Venice, 17–18 April 2008. We want to thank participants at this workshop for comments. We also want to thank Jostein Askim, Tom Christensen and Paul G. Roness for valuable comments and help.

Notes

1 The name of the Ministry of Labour and Social Affairs after the central-left coalition took office in 2005.