Everything Changes, But It All Stays the Same. Labour Market Parties, Corporatism and Norwegian Sick Pay Policies 1978– 2014

Since the adoption of a generous and universal sick pay scheme in 1978, the key elements of Norwegian sick pay policies have remained the same. The present study focuses on the gradual developments in welfare corporatism and policymaking during this period, arguing that these changes paved the way for new and surprising strategies and behaviour among the labour market parties. Tracking several retrenchment attempts across decades, the analyses show how policymaking in corporatist committees was gradually replaced by less predictable processes. Successive governments of different colors have tried to bypass the social partners and legislate hierarchically, thus signaling a break with traditional corporatist norms and decision rules. Labour and business groups adapted by negotiating a pact that kept the existing distribution of economic risks in the sick pay scheme off the political agenda, and by backing each other and creating negative attention to government in the media to protect the pact. In sum, although sick pay policies have remained largely unchanged, this is a status quo upheld by processes of welfare policymaking that have changed substantially. The pact between the social partners and the state is currently a new vetopoint for welfare policymaking. But the piecemeal institutional transformation witnessed in this period, together with the need for conflictual media strategies and new alliances to protect the pact, suggest that it could be a fragile veto- point.


Introduction
This paper follows the development of Norwegian sick pay policies from the introduction of a universal and generous sickness insurance scheme in 1978, to 2014 when the last in a series of retrenchment attempts failed. The focus is not on the policies themselves, which have been remarkably stable for more than three decades. Rather, the purpose is to illuminate and reinterpret a case of welfare state resilience by theorizing and empirically tracking how changes in the political-institutional context of welfare politics paved the way for new strategies and behaviour among interest groups representing employers and employees.
Prior to the late 1980s, welfare policymaking processes and actor configurations had been relatively stable and predictable in Norway. This was in part due to a long-standing and extensive system of corporatist committees through which interest groups were involved in the development of policies, in welfare as well as in other policy areas (Nordby 1994;Christiansen et al. 2013). Additionally, in the labour market, tripartite relations between employers, employees and the state were characterized by stable cooperation (Engelstad 2015, 283). This relationship was arguably even strengthened in the 1990s, following successful negotiations over wage restraint (Dølvik & Stokke 1998). The assumption that institutions are stable or inert, a key feature of much institutional theory focusing on explaining continuity (Mahoney & Thelen 2010, 5), is naturally close at hand when addressing the question of resilience in such a context. Both path dependency and the continued strength of labour and corporatism are thus part of the story underlying the lack of sick pay policy reform in Norway (Hagelund & Pedersen 2015).
This paper reinterprets the case of Norwegian sick pay policies with the aim of supplementing existing knowledge. It is a study of politicalinstitutional change and group strategies in the context of welfare state resilience. Although results inform our understanding of Norwegian sick pay policies, this study will not in itself serve as a basis for causal inferences about resilience. Instead, the reinterpretation concentrates on providing new insight about the relationship between gradual changes in corporatism and interest group strategies. The contribution rests on challenging the assumption about institutional stability. Studying the politics of sick pay policies over this long timeframe, a number of changes in corporatist structures and processes surface. The gradual shift from traditional, Scandinavian 'committee corporatism' (e.g., Christiansen et al. 2013) to 'social pacting' (e.g., Avdagic et al. 2011) represents a key development in this period. I argue that the changes sum up to a piecemeal transformation of welfare corporatism in Norway, which substantially altered conditions of policymaking and of tripartite relations.
The first part of the theoretical section discusses elements of corporatism that could be linked to the influence, strategies and behaviour of interest groups in policymaking processes. In accordance with the empirical focus of the study, I do not concentrate on the outcome (reform or resilience). Instead, the purpose is to sketch how a corporative interest group system is characterized by specific decision rules and norms that shape the expectations and actions of the actors involved. The second part of the theoretical discussion subsequently connects institutional change to both norms and 3 decision rules. This section is informed by my reading of the empirical case and represents an attempt to generalize the findings by building a theoretical framework applicable to the study of welfare state reform processes and interest group strategies in a corporatist setting. More concretely, the goal is to spell out how developments in corporatism -but also related changes in the political context -change established decision rules and norms, which in turn affect the strategic choices of interest groups.
Next, the empirical section tracks developments in the institutions and processes related to sick pay policy reform attempts in Norway over more than three decades, paying attention to the role and choices of employer and employee groups. The analyses show how the traditional system of policymaking in corporatist committees was gradually replaced by less predictable processes. Successive governments of different colours tried to bypass the social partners and legislate hierarchically, thus signalling a break with traditional corporatist norms and decision rules. Labour and business groups adapted to this development and the reform attempts that accompanied it. First, by negotiating a pact with the state that kept the existing distribution of economic risks in the sick pay regime off the political agenda. Second, by backing each other and creating negative attention to government in the media whenever status quo was threatened by reform proposals.
In effect, it appears that the pact between the social partners on the subject of sick pay has been established as a new veto point for welfare policymaking. As an example of neo-corporatist tripartite relations (e.g., Ebbinghaus & Hassel 2000), such a pact could, on the one hand, signal the flexibility and continued strength of a corporatist system of welfare policymaking. However, the piecemeal institutional transformation witnessed in this period, in addition to the apparent need for conflictual media strategies and unorthodox alliances to protect the pact, points to a paradox in existing interpretations of this case that emphasize the continued strength of corporatism: Although status quo has been upheld in terms of policies, this has happened alongside a process of institutional displacement. The closing discussion therefore looks at how we can understand this, raising questions about the future prospects of the new veto point and 'the most generous sickness insurance scheme in the world' (Hagelund & Pedersen 2015, 223).

Welfare State Reform Processes, Corporatist Decision Rules and Norms
The literature on welfare state reforms has described how economic challenges related to globalization, as well as demographic changes, put pressure on European welfare states and pushed retrenchment to the top of government agendas (cf. Pierson 1994;Jensen et al. 2019). The role of welfare corporatism -broadly defined as the different institutionalized processes through which interest groups cooperate with authorities about the formation of welfare policies -in subsequent reforms has been a debated element in this literature. One the one hand, corporatism could be regarded as an institutional veto point (e.g., Crepaz & Moser 2004, 272), a perspective that has received empirical support in a comparative study showing that corporatism decreases the likelihood of reform (Allan & Scruggs 2004). On the other hand, others have argued that corporatist institutions do not necessarily obstruct reform (cf. Hemerijck & Van Kersbergen 1999;Ebbinghaus & Hassel 2000;Anderson 2001). For instance, Anderson's (2001) analysis of Swedish pension and unemployment policies documents how class-based interest groups in a corporatist system of policymaking have worked constructively to achieve compromise on retrenchment. The argument echoes that which Swenson (1991) put forward in his reinterpretation of the dominance of labour and left in Scandinavia: it is less about a 'balance of class power' and more about a 'cross-class alliance' that has allowed both labour and capital to achieve favourable outcomes.
From the perspective of this case, where the puzzle relates to gradual institutional change and new actor strategies and coalitions, policy outcomes do not occupy a key role in the theoretical discussions. Instead, I focus on the elements of corporatism that are likely to affect the expectations and actions of the labour market parties. I do this with the help of two theoretical tools, the first one relating to corporatism as a decision rule and the second highlighting the role of norms in corporatist arrangements. Scharpf (1989) offers a theoretical argument about decision rules that can be used to analyse how institutions shape actor perceptions about, and preferences for, different strategic choices in policymaking processes. Corporatist systems are de facto characterized by consensual settlement and unanimity, even though formal rules subscribe hierarchical government authority and majority decisions. From the perspective of tripartite relations, majority decisions and hierarchical government authority are theoretically linked to less predictable and more volatile policy processes and outcomes because they create an opportunity for involuntary redistribution. Policies can be changed, if a government or majority so wishes, without the consent of all partners/parties. In an established corporatist decision-making system, the mechanisms are very different as partners cannot easily withdraw without high costs. Thus, when they do not agree, this likely implies continuation of earlier policy choices to which all partners at one point have agreed rather than a breakdown of cooperation. Any signal that governments, which through formal rules ultimately possess the authority to initiate and implement legislation, want to change decision rules will therefore be crucial input to the social partners about their opportunities to influence welfare policies.
Another important ingredient in the complex relationship between corporatism and actor strategies relates to norms. Visser and Hemerijck (1997), for instance, compare corporatist arrangements to social capital, 1. because they enhance norms of reciprocity, trust and sense of duty and responsibility to other social partners. Distrust and myopic, self-interested behaviour are key challenges to stable cooperation and compromise. In a game theoretic approach, these problems are reduced by iterative interactions (i.e., Axelrod 1984). Empirical case studies of tripartite relations seem to support this, pointing to how repeated interactions induce social partners to converge normatively and ideologically and recognize the benefits of cooperation (i.e., Falkum 2015). Corporatist social capital, or shared norms, should be helpful when social partners try to compromise on retrenchment, ensuring that actors commit to each other and to established processes of cooperation about policymaking.
Both corporatist decision rules and corporatist norms provide predictability and increase the likelihood of a specific type of interest group behaviour. They work together to create a strong but partly informal blueprint for how negotiations and policy processes should work. However, this naturally means that developments which threaten rules and/or norms have the potential to shift actor strategies considerably. I will return to this in the next section, where the assumption of stable institutions is left behind and replaced by a discussion of how a number of political-institutional changes could be linked to these two theoretical tools.

Theorizing the Link between Changes in Corporatism and Interest Group Strategies
Inspired by institutional theory focusing on gradual transformations (e.g., Mahoney & Thelen 2010), this section adds to the preceding by exploring the relationship between changes in corporatism and interest group strategies. The main argument is that decision rules and norms vary according to the type of corporatist arrangements that characterizes a specific (sub) system, at a specific point in time. Thus, I will discuss the two main manifestations of corporatist relations relevant to this case: the policy preparation (or implementation) committees typical of Scandinavian corporatism and the policy agreements or contracts of social pacts. The purpose is to theorize how changes in corporatist institutions and processes, from the former to the latter, could affect interest group strategies and actor coalitions in welfare policymaking. Given that this case study concentrates on Norway, existing research on state-interest group relations in the Scandinavian context is used as a starting point. Comparative studies measuring corporatism from a political economy perspective have repeatedly ranked Norway among the most corporatist countries (cf. Lijphart & Crepaz 1991;Jahn 2014). At the same time, a parallel literature approaching corporatism more broadly as a variety of democracy has traced the representation of organized interests in public boards and committees in Scandinavia (e.g., Christiansen et al. 2013). This structure is a key element in what is termed the corporatist channel of interest representation in Norway (Nordby 1994;Rommetvedt 2017). In these boards and committees, with the mandate to analyse problems, prepare and sometimes implement policies, employer and employee groups have acted as important agents in the development of the welfare state. Corporatism has thus been a suitable label for Norwegian policymaking since WW2, both when looking at macro-economic management relevant to welfare regimes and when looking at the making and implementation of labour market and social insurance policies (Nordby 1994;Engelstad & Hagelund 2015).
However, from the beginning of the 1980s, the corporatist system was significantly down-scaled in Scandinavia (i.e., Lewin 1994;Christiansen et al. 2013). Measuring corporatism through the number of public boards and committees with interest group representation, the case of Norway is characterized by a peak around 1980 (Thesen & Rommetvedt 2009). Thereafter, the numbers drop significantly. Although interest groups still enjoy access to approximately the same share of committees, the dramatic decline in the absolute number of committees in 2005 amounted to a 66 percent reduction (from 1980). From the perspective of this case study, a key question would be whether the downsizing of the traditional corporatist arena is also reflected within welfare and social policies, and more specifically with regard to sick pay policies? And if so, what kind of processes replace the traditional one, and to what extent do these changes affect the strategies of labour market parties in negotiations over welfare state reform?
The interest group literature points to an increased use of lobbying towards parliament and civil servants as a key strategy for groups that seek to maintain their influence in the context of this reduction in corporatist committees (cf. Rommetvedt et al. 2013). The lobbying trend suggests a more pluralist political system, but still does not provide an exhaustive answer to questions about the role of interest groups in welfare policymaking. Interactions between the state, employee and employer groups are still much closer in Scandinavia compared to the rest of Europe (Avdagic et al. 2011). Consequently, a decline in corporatist structures do not necessarily imply a disappearance of corporatism, but rather a change in the processes of bargaining (Molina & Rhodes 2002). Applying Schmitter's (1982) categorization, 'our focus should shift from "neo-corporatism 1" (the structure of interest representation) to "neo-corporatism 2" (the system of policy making)' (Molina & Rhodes 2002, 319). In other words, if we want to explore group influence on welfare policies when traditional structures change or disappear, we should direct attention to their role and strategies in the processes where these policies are developed.
In this Scandinavian case, tracking corporatist relations over many decades, such a task inevitably involves studying groups in two very different settings: the traditional corporatist committee explained above, and the (revived) corporatist pact to which I turn below. Although the latter links our case to the theoretical and empirical debate about the 'variable and constantly evolving' nature of corporatism (Molina & Rhodes 2002, 322), it is important to bear in mind the Scandinavian setting. Much of the theoretical interest in pacts was initially sparked by the inability of previous theories, focusing on the structural prerequisites of corporatism, to explain or predict the revival or appearance of pacts in unexpected places. In contrast, the establishment of a pact in my case would come as no big surprise. Still, the question is how we could interpret a change from committee to pact and how it affects the groups involved.
Ebbinghaus and Hassel operationalize the evolving nature of corporatism when discussing the 'renaissance of concertation through social pacts during the 1990s' (2000,45). They supplement an initial focus on competitive economic performance in an era of globalization and budget constraints, emphasizing that social policy reforms and social insurance 'are as important elements of social pacts as wage restraints' (Ebbinghaus & Hassel 2000, 45). Social pacts could be defined as 'publicly announced formal policy contracts between the government and social partners over income, labour market, or welfare policies that identify policy issues and targets, the means to achieve them, and the tasks and responsibilities of the signatories' (Avdagic 2010, 637).
Theoretically, pacts are tools that could serve social partners in several contexts. From the perspective of this study, pacts constitute likely strategic responses for interest groups that have lost stable, institutionalized participation in welfare policymaking. Particularly in light of potential alternatives such as exclusion and undesirable decisions by majority rule. While a pact, on the one hand, still reflects a key role for interest groups in policymaking, its format is very different from that of a traditional Scandinavian corporatist committee. For instance, the tools offered by pacts are the tools that govern contractual relationships: partners have formal obligations or deliverables, they are monitored and breaches are sanctioned etc. The exchange between social partners and the state is usually explicit and concrete, in contrast to traditional corporatist committees where the exchange is a rather abstract or implied element of the relationship.
These two varieties of corporatism in other words differ substantially, which in turn holds implications for both decision rules and norms. Looking at the former first, a system of corporatist committees does not guarantee a particular outcome but grants selected labour market groups stable access to the processes where welfare policy is developed. In effect, this is corporatism as an institutionalization of the policymaking process. A pact, on the other hand, guarantees the outcome, the actual policies, for a specific period of time. But the process by which the agreement is reached is unclear or up for debate: 'Not only the content, but also the process can be the subject of negotiation' (Molina & Rhodes 2002, 318). The process will always involve negotiations somehow, but there is no guarantee that interest groups will be able to take part in future developments of policies. Theoretically then, pacts involve increased uncertainty for labour market groups about long-term influence and decision-making in comparison to corporatist committees.
Secondly, these two versions of corporatism also vary in terms of the accumulation of corporatist norms. In the traditional Scandinavian corporatist practice, organized interests have been institutionally integrated in the policymaking process through representation in public boards and committees (cf. Christiansen et al. 2013). Repeated exchange within such institutions, characterized by restriction of access and a principle of codetermination (Thesen & Rommetvedt 2009), increases trust and commitment. This is important because it depresses incentives to apply alternative and more conflictual strategies whenever government and social partners find it hard to reach agreement on welfare reforms. Although pacts might involve renegotiations and some level of iterative interaction, the extent is lower. Furthermore, the design and context of the bargaining situation mean that actors' perception of their relative power is more consequential to their strategies than norms and trust (Avdagic et al. 2011, 10).
The different links between varieties of corporatism and interest groups strategies are likely to be enhanced by other processes of societal and political change. A particularly relevant case in point is the so-called mediatization of politics, a process whereby the media has become more and more important for political actors and institutions (Esser & Strömbäck 2014). In theory, interest groups that are included in stable corporatist decision-making systems should be less prone to 'go public', using media to voice criticism and draw negative attention to the government. However, Häusermann et al. (2004, 33), for instance, argue that increased media coverage opens 'up the traditionally confidential and selective sphere of corporatist negotiation and weakens the social partners' ability to reach agreements' . Given the continuing strong support for the welfare state and many of its services (Haugsgjerd & Kumlin 2019), this trend would seem very challenging to reform-seeking but risk-averse political actors. Mediatization changes the game and might make social partners feel less obligated to act in ways that characterized previous arrangements. Media strategies after all beckon with a potential to blame government and stir reform opposition. 9 Still, the extent to which mediatization affects interest group strategies depends on how the groups perceive their position and opportunities in the policymaking process. This again makes the differences between pacts and committees of potential interest. The trust and close connection built through iterative interactions in corporatist committees should provide a stronger barrier towards using media in short-sighted attempts at gaining an advantage. While the higher levels of uncertainty and stronger focus on power associated with pacts, lower the threshold of using the media.
This section has discussed how gradual developments in corporatism affect decision rules and norms. Note in addition that the strategies of governments matter for how social partners perceive their bargaining situation. Based on the case of Denmark, Mailand (2006) points to how the state has become more self-assured, showing a 'great will to act unilaterally' (ibid: 382). If governments try to bypass corporatist arrangements and legislate based on hierarchical government authority, this will further strengthen the signals about changes in decision rules and norms arising from gradual, institutional change.
Summing up before turning to Norwegian sick pay policies, I have put together a theoretical framework that will be used to (re-)interpret a case of welfare state resilience supported by new actor strategies and surprising actor coalitions. The idea in short is that changes in the political-institutional context influence actor perceptions of how to sustain their influence in welfare policymaking.

Research Design
Sick pay schemes are crucial to one of the pillars of the welfare state: income security. While sick pay policies thus are, undoubtedly, important, this particular case of Norwegian sick pay policies was not selected because it is considered influential (or critical) in the literature. At the outset, and from a distance, it is a 'closed case' with an implied causal interpretation: Norway is still considered one of the clearest examples of neo-corporatism (Engelstad & Hagelund 2015, 2-3), and welfare state resilience is therefore not necessarily surprising. However, as explained in the introduction, taking a closer look while at the same time questioning the institutional stability of the corporatist setting reveals interesting and paradoxical characteristics. The goal has therefore been to supplement existing interpretations of this case. The broader relevance is to be found in the theoretical framework built while studying the case. There is no clearly defined population to which one can generalize the findings from this single case. But the theoretical framework highlights mechanisms underlying (lack of) welfare state reform in a corporatist setting, paying particular attention to how gradual institutional change could affect interest group strategies. It should therefore be relevant for a range of welfare policymaking processes in systems with corporatist features.
A key design choice stemming from the theoretical interest in gradual, institutional change was to cover a long timeframe. It is otherwise easy to overlook the fact that while outcomes (status quo) and broad institutional labels (corporatism) stay the same, a number of slow moving but transformative changes in corporatist institutions and welfare policymaking processes have taken place. Following a single issue over a long time is useful because it allows us to study how actor reconfigurations are affected by institutional and political changes, while many other factors are kept constant. However, recent scholarship has highlighted the importance of comparing issues in order to capture the multidimensionality of modern welfare politics and the reconfigurations of actor coalitions (Häusermann 2012, 112). Future studies could hopefully test and develop the framework from the present paper through a comparative approach. Hagelund and Pedersen's (2015) comparison of sick pay (no reform) and pension policies (reform) is an already existing example, which I rely on for both theoretical inspiration and empirical descriptions. It clearly indicates that there are scope conditions not captured in the present single case study, relating to, for instance, the complexity of different policy schemes and their relative importance to the social partners. However, this only makes it more timely to continue the exploration of the relationship between gradual institutional change, group strategies and coalitions for a larger set of cases where reforms have proven difficult.
The chosen case represents a topic which has received attention not only in research papers but also in a number of evaluation reports and Green papers. This allowed me to emphasize the construction of a theoretical framework for reinterpretation, instead of building the empirical case from scratch. This construction involves recontextualizing the familiar facts of an empirical case, available in a combination of sources: policy documents, secondary literature and news articles (see Appendix A.1). Such a design means that the empirical part does not occupy the centre stage of the study. Still, although the review is short, I have as far as possible identified and used the relevant official sources to map positions and decisions instead of relying solely on secondary literature. Green papers have thus provided information on which groups were included in the policy processes as well as their primary and secondary positions. White papers and Propositions to the Storting give insight into the government's policy wishes, while Recommendations from the Storting spell out the outcome of reform attempts. The same prioritization of sources apply to questions relating to media strategies and content, where I have used the news archive Retriever to search for relevant articles in the dominant broadsheet (Aftenposten), tabloids (VG, Dagbladet) and news agency (NTB) in Norway.
Finally, when touching upon aspects of the argument that has not figured prominently in previous interpretations, I have tried to supply additional information. That is, in order to explore the fate of traditional corporatist structures within welfare and labour market policies, and specifically sick pay policies, I have collected data on the number and composition of corporatist committees (see Appendix A.2). Put together, these sources allow me to investigate what kind of policymaking processes have replaced the traditional ones, and how the role and influence of interest groups have transitioned alongside institutional developments. Consequently, the empirical review of the case will focus, among other things, on which groups are included in the policy processes, at what point in time they are included and through what types of arrangements or institutions.
Four different policy processes will be described and analysed. One of the main criteria for choosing these four subcases were that they should be spread out over the period of study, allowing me to address the development over time. Additionally, it was important that the selection did not neglect important policy proposals in this period, in order to avoid a bias towards less consequential policy processes. Determining importance 'objectively' is not unproblematic, but the selected cases are regularly addressed in both existing literature and public debates about sick pay policies. Consequently, they should provide a relevant empirical basis from which to discuss welfare corporatism in this particular case.

Analysing Welfare Corporatism and Sick Pay Policy Processes in Norway
As already mentioned, Norway experienced a 2/3 reduction in corporatist committees from 1980 to 2005. Table 1 illustrates how this general trend also affected welfare policymaking, showing the development of interest group representation in public boards and committees belonging to the jurisdiction of ministries handling labour and social affairs. As we are interested in the tripartite (or multipartite) relations between the state and the social partners, only committees that have a minimum of two interest group representatives are part of the sample.
The clear trend in this overview raises questions to be addressed in the review of the sick pay case. The decreasing number of committees seems to indicate a decline in welfare corporatism, but at the same time poses a puzzle: Where is welfare policy developed when the traditional corporatist channel shrinks? And what is the role of interest groups in these processes?
In order to investigate these questions, the rest of this section focuses on tracing sick pay policies in Norway over almost four decades. The universal sick pay scheme introduced in 1978 replaced a mix of private and public arrangements that in practice provided different groups of employees with substantially different degrees of coverage. The new policy ensured full compensation from the first day of sickness for all categories of employees. In all of the major questions regarding the design of the sick pay scheme, LO (The Norwegian Confederation of Trade Unions) and the employee side enjoyed more influence than NAF (The Norwegian Employers' Confederation, from 1989 NHO, The Confederation of Norwegian Enterprise) and the employer side. The Labour government thus ended up proposing 100 percent income compensation (as opposed to 90 percent), no waiting period and the two first weeks covered by employers (as opposed to 1 week) before national insurance takes over financial responsibility. Although the original sick pay scheme thus had a labour stamp, it was passed by a unanimous vote in the Storting (Hagelund & Pedersen 2015, 223). Table 2 provides an overview of the selected policy processes, the main steps in these processes, including information about the existence and composition of corporatist committees and agreements (pacts). During the 1980s, the issue of sick pay was hardly politicized (Pedersen 1997), but a number of changes in eligibility criteria and maximum income level for calculation of benefits meant that varying governments adjusted somewhat. Later on, new rules and procedures with regard to sickness absence have been introduced, demanding more of both employers and employees in order to reduce the number and length of such spells. Hagelund (2014) examines this in detail, arguing that although economic incentives have remained the same, the governance of sickness absence in the workplace has changed. This has taken place through an intensification of dialogues and activities for which the labour market parties are responsible (see overview ibid: Table 1, page 80). We should not ignore the sum of stepwise reforms and adjustments, thereby underestimating the level of policy change (Van Kersbergen & Vis 2014, 25). The term 'resilience' thus need qualification, in the sense that it applies to the core elements that are crucial to the distribution of economic risk between employers, employees and the state: mainly compensation levels Official Norwegian Report (NOU). 2 Proposition to the Odelsting. 3 Committee recommendation to the Storting, followed by first and second reading. 4 The Confederation of Academic and Professional Unions (AF). 5 Confederation of Vocational Unions (YS). 6 The Norwegian Medical Association. 7 The Norwegian Association of Local and Regional Authorities (KS). 8 The Enterprise Federation of Norway (HSH). 9 The Federation of Norwegian Professional Associations (Akademikerne). 10 The Norwegian Federation of Organisations of Disabled People (FFO).

11
The Norwegian Forum of Disabled Peoples' Organizations (SAFO). and employer funding. Apart from an increase in the employer period from 14 to 16 days in 1998, these elements have remained nearly untouched since 1978. The first major attempt at 'politics of austerity' started at the end of 1980s when budget deficits were rising. A budget proposal from the Conservativeled government in December 1989 suggested reduced compensation levels for employees and increased economic responsibility for employers. In terms of media debate, this first subcase displays some fuzz in the early stage of the reform attempt when the government had attempted to bypass the social partners. 2. It is worth noting that employer and employee strategies seem to reflect traditional lines of conflict. While LO attacked the Conservative-led government and explicitly suggested increasing employer financing of the scheme, 3. NHO remained relatively silent offering few comments. 4. The proposal did not make it through parliament (Budsjett-innst. S. II. (1989)(1990)). The parliamentary committee encouraged the government to discuss the matter with the social partners, explicitly referring to positive results in recent wage settlements. The government subsequently initiated multipartite negotiations, where both employers (NHO, KS) and employees (LO, YS, AF) agreed to enter into a binding cooperation to reduce sick leave -given that there would be no changes in the current sick pay (page 16 in St.prp. Nr. 95 (1989-1990). From the government's point of view, the main condition of the agreement was that it should have a substantial effect in cutting costs for sick pay over the following 3 years.
The resulting agreement in practice restrained political intervention in the fundamentals of the sick pay scheme. The public committee that at that time already had been appointed to review and suggest changes in sick pay policies counted members from six different interest groups. In their final Green paper from 1990, the committee explicitly noted this intervention ('the agreement has intervened directly in the work of this committee' (page 9, NOU 1990: 23)). Of the remaining policy questions that were up for debate in this committee, many received broad and general support from both employers and employees. The so-called 'arbeidslinja' (workfare) where the idea is to reform policies in order to stimulate people to work instead of receiving welfare benefits was -at least at the most general level -accepted by both unions and businesses.
The 'sick pay project' was extended throughout the 1990s and eventually included in the basic agreement between the social partners in the labour market. After an initial positive trend, a new rise in sick leave figures from the mid-1990s prompted the second major retrenchment attempt towards the end of the decade (Ose et al. 2009, 34). A new public committee was appointed in 1999, signalling a commitment to traditional corporatist arrangements. Still, both employer and employee costs and risks were targeted in this process. NHO was willing to accept an increase in their financial contribution to the scheme (guaranteeing 20 percentages of the costs after the first 16 days of illness) on the condition that income compensation was cut by 20 percentage points in the first 16 days of illness. Five of the six groups representing employees strongly opposed, both in the Green paper produced by the committee (NOU 2000: 27) and in subsequent negotiations with the government (Ose et al. 2009). The media debate was not intense, and the few comments given by the respective groups mostly repeated the stalemate situation in the committee. 5. The result was in the end a continuation of the pacts of the 1990s -now called the IA-agreement about an inclusive labour market (Inkluderende Arbeidsliv). For a second time, the explicit condition of the social pact was that government would stay away from the sick pay scheme (St.prp nr. 1 Tillegg nr. 1 (2001-2002)). And for a second time, the pact was conditional on the development in sick leave. If the operative goal of a 20 percent reduction was deemed unattainable by an evaluation in 2003, the agreement would terminate.
The final subcase is in fact a collection of various reform attempts since 2001. The three coalitions in office since the adoption of the IA-agreement have proposed changes to the sick pay scheme, starting with the centreright government of Bondevik II (2001-2005, Conservative Party, Christian Democratic Party, Liberal Party). In its struggle to balance the budget, the government proposed that employers should bear 10 percent of the costs throughout the whole period of sick leave (page 14 in St. meld. Nr 1 (2004Nr 1 ( -2005). The proposition was dropped, as neither employers, employees nor the opposition could accept it. It is not particularly surprising that employer groups voiced their resistance and criticism in the news. But also LO and other employee groups reacted strongly and were quick to publicly label it a breach of contract, despite the fact that the proposal targeted employers only. 6. Only 2 years later, in 2006, the Stoltenberg II government (2005-2013, Labour, Center Party and Socialist Left Party) put forward a strikingly similar proposal. The government argued that the conditions of the IA-agreement had not been met. Pointing to the development in sick leave figures, they concluded that employers might not have the right economic incentives to improve the work to combat sick leave (press release 6.10.2006). The reactions were strong, and both employer and employee groups rushed to criticize the government, yet again calling the proposal a breach of promises and the IA-agreement. 7. The fact that the proposal came from a Labour-led government did nothing to change LO's commitment to the IA-agreement and support of their traditional counterpart, NHO. Prime Minister Stoltenberg invited the parties to the IA-agreement to negotiate and suggest alternative ways of reaching the same reductions in sick leave and costs. The resulting measures (focusing on early activation and stronger control/sanction within the existing sick pay regime) were subsequently adopted, while the proposal to increase employer responsibility was dropped. Only 4 years later, negative developments in sick leave numbers spurred the establishment of an expert committee (without interest groups). One of its solutions included an increased economic burden for employers. But with the failed attempt of 2006 fresh in mind, the Stoltenberg cabinet signed a continuation of the IA-agreement instead of proposing a controversial reform of the sick pay scheme.
Shortly after the Solberg coalition (2013->, Conservative Party, Progress Party) entered office, a new proposal threatened to double the minimum required income for sick payment (page 24 in Prop. 1 S (2014-2015)). According to the government, this was not an intervention in the IAagreement but more of a 'technical adjustment' (press release 20.10.2014). Neither employers nor employees agreed and the proposal was quickly dropped. Although the media debate was less salient than with earlier reform attempts, NHO did not leave any doubts as to where its loyalty could be found. 8. If considered as an attempt to bypass established practices of welfare policymaking processes in order to push retrenchment, the first reform process in the late 1980s/early 1990s clearly failed. It did nevertheless initiate a sequence of events that gradually transformed welfare corporatism in the area of sick pay. At that time, however, traditional lines of conflict between the social partners were salient and decisive for strategic choices visible, for instance, in media debates (see footnotes 2, 3 and 4). Even around the turn of the century, during the second major reform attempt, there were still traces of the traditional corporatist decision-making system left in the shape of the committee appointed for the Green paper (see Table 2). Media debates did not seem to serve strategic purposes for the involved parties, in the sense that they mostly mirrored the stalemate situation played out in the corporatist committee and bargaining process where employers and the government wanted to reform while employees refused (see footnote 5).
Gradual institutional transformation was nevertheless already underway. At first this could be directly experienced by the social partners through a reduction of corporatist committees in general (see Table 1). Furthermore, to the extent that committees actually were appointed, a higher number of groups was included which in turn complicated negotiations and diluted the corporatist channel (see Table 2). The reform attempts from 2004 onwards sped up and intensified what had been, up until that point, a rather slowmoving change of welfare policymaking processes. First of all by repeatedly trying to bypass both the newly established pact as well as traditional corporatist committees (see Table 2; no Green papers or committees appointed).
Instead of appointing corporatist committees that produce Green papers in the early stages of the process, the post-2001 period was characterized by governments skipping straight to budget proposals -in some cases, only involving the social partners later on when media criticism had been strong. Instead of influence through the traditional corporatist structures, the IAagreement was institutionalized -through repeated retrenchment attempts, multipartite negotiations and media debates -as a veto point that limited political intervention (see also Hagelund & Pedersen 2015).
Interestingly, this has worked regardless of whether the attempts at retrenching have targeted the employer or the employee side. When one has been threatened, the other has provided support. It would seem that the employers and employees have been brought closer together, while their distance to the state in this tripartite relationship has increased. The commitment to each other and to the pact thus means that the social partners have acted to block welfare retrenchment -regardless of the colour of government and traditional ties between specific interest groups and specific parties. In relation to this, Hagelund and Pedersen (2015, 227-29) point to how both the employer and employee sides have enjoyed the opportunity to get what they wanted. When nothing happened, part of the plausible interpretation that is offered emphasizes strategic considerations: Each side was worried that getting what they want could open up the field to unwanted reforms in the future.
This interpretation is compatible with the perspective of the present study, although important underlying factors are partly undercommunicated. Experiencing gradual institutional change since the late 1980s, employer and employee groups have had to cope with increasing uncertainty about their role and influence in the processes that determine future sick pay policies. I have theorized how different varieties of corporatism could be linked to different group strategies. In this case, the reduced influence and gradual disappearance of traditional corporatist committees were particularly important. A pact that ensured continued group influence was established early on, which could be interpreted as the continuation of an integrated, corporatist system. However, the policymaking process was substantially changed, meaning that groups lost what used to be a predictable and highly valued strategy for political influence (Thesen & Rommetvedt 2009). Furthermore, by skipping straight to proposals, various governments signalled an intention to use formal, hierarchical decision rules to achieve reform. A more self-assertive state made for a less trustworthy partner in tripartite relations. In other words, groups were no longer guaranteed that decisions could not be made unless through a compromise that they themselves were parties to.
In sum, this period has witnessed a long-term transformation of welfare corporatism where the threshold for governments to initiate reform without consent from groups has been lowered. This comes at the cost of higher uncertainty for the social partners. Within the traditional corporative arrangement, groups could be relatively sure about two things: They would be included in the next policy process, and if they could not accept a proposal, it would most likely not be passed. When these institutions are bypassed, this is no longer the case, as government threatens to form majorities with whomever backs their reform proposals.
Characterized by increased uncertainty -about the present policy process but also with regard to future processes -employee and employer groups seek to maintain influence by means of three approaches. First, they want to keep issues -in this case, the existing sick pay scheme's distribution of economic risks -off the political agenda, since they are no longer guaranteed a seat at the table when decisions are made. The establishment of a pact, such as the IA-agreement (see Table 2; the agreement/pact column), thus easily makes sense from a group perspective, responding to the reduced predictability of political processes.
Second, if the core aspects of the sick pay scheme nevertheless are threatened by reform proposals, groups can reduce uncertainty by entering into new coalitions. This is perhaps the most surprising element of this case. Labour and business take turn to back each other even when they could have sided with the government and ensured a majority for their primary position without any immediate costs to themselves (see footnotes 6, 7 and 8). But given the changes that they have experienced, and the signals that governments are continuously sending through reform attempts, it is also a strategic move to protect oneself from ending up in a minority position in the future. In many ways, this could be interpreted as a continuation of an historical cross-class coalition that both employers and employees deem beneficial (Swenson 1991). But the context is very different, as Swenson's cross-class analysis depicts the coalition as a sign of strong and proactive strategizing that shaped corporatism and social democracy in Scandinavia. The present case, several decades later, finds that a gradual transformation of welfare corporatism in combination with a more self-assured state has left the social partners more vulnerable. Arguably then, the coalition could, to some extent, be interpreted as a strategy of last resort, where the social partners try to make up for lost opportunities in an increasingly unpredictable environment.
Third, they create negative attention to government in the media when cornered. This is made possible because the institutions that fostered trust and reciprocity between government and social partners, and thus also confidentiality, no longer occupy the same role in the policy process. Interestingly, this could be interpreted as a functional response to government strategies. By downsizing the traditional corporatist channel and signalling an interest in hierarchical decision rules, governments are indicating that the electoral channel should matter more in policymaking, at the cost of corporative codetermination. In response, groups' media strategies force governments to confront public opinion. In practice, they are putting the basis of hierarchical decisions in democratic systems -representation -to the test. In cases where public opinion is favourable to existing welfare schemes, governments will have to withdraw or accept electoral costs. In sum, the outcome is that governments have found it hard to push radical reforms, even though their flexibility to do so should have increased.

Conclusion
The long traditions of corporatism in Norwegian welfare policymaking and the continued strength of labour groups are not the only keys to understanding this case. By emphasizing gradual institutional change and theorizing how this affects interest group strategies, I have examined how welfare state policies have been upheld through new alliances and strategies in the context of a transformation of welfare corporatism. Thus, we can conclude on three accounts. First, zooming in on the actors and their choices, reading them in the context of how gradual institutional changes affect their perception of decision rules and norms, the alliances between the social partners and their use of conflictual media strategies makes sense. The concluding part of the preceding analysis section, in addition to the theoretical framework, already summarize the inferences to be drawn from the present study on this account.
Second, this is also a story about a broader transformation of welfare corporatism. In fact, the starting point for the gradual displacement of traditional welfare corporatism witnessed in this case was arguably a political goal of abandoning corporatist decision rules. In the perspective of Mailand (2006), the state has tried to claim a stronger ownership of sick pay policies. Through policy proposals and various attempts at bypassing groups and traditional corporatist committees, consecutive governments have indicated that reform authority should rest much more firmly in the electoral channel of representation at the cost of the corporative channel. Although they have not succeeded in terms of policy content, they have effectively shook up the policymaking process, initiating a development whereby welfare corporatism, group strategies and influence have changed substantially over the last 30 years. From the perspective of employer and employee groups, we have moved from a situation where the state was a reasonably reliable partner within a system that secured future influence, to a situation where shifting governments appear as an unreliable partner, while a pact serves as temporary protection of policy status quo. This institutional displacement, in which a system of policymaking eroded gradually, left employers and employees with a more limited set of options and a higher level of uncertainty. Although the social partners so far have escaped increased economic risks in the sick pay scheme, it is therefore questionable whether we should necessarily interpret their new alliance and their conflictual media strategies as positive strategic choices that indicate strength and influence.
Finally, and in continuation of the second point, the results thus arguably question the role that corporatism and group strength have played for the resilience of the sick pay scheme. I have de-emphasized resilience in my argument in order to focus on the mechanisms underlying the somewhat surprising characteristics of this case. When concluding though, it is hard to avoid discussing the implications that the case has for how to understand the lack of reform. One possible interpretation is that public opinion appears to be a key obstacle to reform. Instead of fixating on the policy status quo, the conclusion above stresses the success of governments in transforming a system of policymaking. An important element in this transformation has been the attempts to bypass groups and apply hierarchical government authority, which in the end is based on an electoral mandate. From this perspective, the positions and strategies of the social partners, and even the new veto point in the shape of a neo-corporatist pact, might be less consequential than we think. Governments have reformed the process of welfare policymaking and would likely have succeeded in reforming the content as well -if only public opinion was favourable. This is of course a counterfactual which we obviously cannot observe. For now, it is impossible to separate the 'will of the people' and 'special interests', since the two align in this case, which again points to the need for further exploration in a comparative context. However, if this interpretation has any merit, the 'world's most generous sick pay scheme' will be retrenched once the framing is right and electoral risks are sufficiently low -no matter the positions and strategies of employer and employee groups.