Negotiating a new Swedish model: Employment transition agreements and the struggle over redundancies

This article is aimed to contribute to our knowledge regarding employer and employee preferences about employment security and unemployment income protection as well as the degree of neoliberal change during the last decades through the lens of collective bargaining. It charts the institutional–historical development of an Employment Transition Agreement (ETA) between the bargaining cartel for white-collar unions, PTK, and the organisation for private employers, SAF/SN. ETAs are a form of institution through which Swedish trade unions and employer organisations give employees added protection in the event of redundancies, mostly in the form of added income protection but also matching services and shorter training programmes. Drawing from archival material and published statements in newspapers, the article engages with what the organisations wanted from such agreements during negotiations and how this shifted during the decades. The results show that the agreements have given employees added unemployment income protection while at the same time giving employees greater flexibility during collective lay-offs. It also shows how the agreement initially was more focused on employment security and proactive investments in skills. But when the balance of power shifted in favour of employers, PTK had to give up any such ambitions.


Introduction
Employment Transition Agreements (ETAs, omst€ allningsavtal), previously referred to as Employment Security Agreements (trygghetsavtal, ESAs), are collective agreements through which trade unions and employer organisations in Sweden provide added protection for employees in the event of dismissals. Currently, such agreements cover 90% of the Swedish labour market and provide support in the form of added income protection and matching services (Andersson, 2018;Walter, 2015). These agreements also fill an important role in negotiations over collective lay-offs, helping employers to increase flexibility vis-a-vis Sweden's nominally strict employment protection laws by enabling exceptions from the seniority-principle (or 'first in, last out' principle) regulated in the Employment Protection Act (Lag (1982:80) om anst€ allningskydd or LAS for short). This article charts the institutional-historical development of one such ETA for more than 40 years: the agreements between the white-collar union's Council for Negotiation and Cooperation (Privattj€ anstemannakartellen, PTK) and the Swedish Employers Association (Svenska arbetsgivaref€ oreningen, SAF), since 2001 the Confederation of Swedish Enterprise (Svenskt n€ aringsliv, SN)' from its inception in the early 1970s until today -and how the function and purpose of these agreements have shifted within a changing political and economic context. By highlighting how the PTK/ SAF/SN agreements have been related to the seemingly fundamental conflict between employers and employees over flexibility and employment security, what Emmenegger (2014) has labelled the flexibility-security nexus, the study will provide insights into the role of preferences and power on collective bargaining.
In a broader research context, this article contributes to the discussion over employer and employee preferences and institutions in capitalist economies. Is employment protection (such as job security regulations) and unemployment protection (such as unemployment insurance) the result of employers and their interest organisations making concessions to labour, either through collective agreements or government regulations and welfare schemes (Emmenegger, 2014)? Or are such institutions also vehicles through which employers can overcome some of the problems of underinvestment in skill and re-training that are thought to be almost inevitable in a market economy (Estevez-Abe et al., 2001)?
The article also contributes to the debate on institutional change from a historical perspective: more specifically, the debate as to whether there has been a shift towards a more neoliberal trajectory of change in industrial relations (Baccaro and Howell, 2011), or if the continued existence of collective agreements such at ETAs is an example of institutional resilience against such a development (cf. Martin and Swank, 2012).
The article focuses, as stated above, on one specific ETA that was created after negotiations between PTK and SAF/SN. Through a study of unique and detailed archival material, the article charts the preferences and strategies of both employer and employee organisations regarding employment security, unemployment protection and flexibility. Our research questions relate to the bargaining process between employer organisation SAF/SN and employee organisation PTK. First, what did these actors want to achieve, respectively, through these agreements? Second, how did these ambitions shift over time with the political and economic context? Third, can changes in the agreements reflect a shift in preferences or shifts in the balance of power on the labour market?

Theoretical framework and previous research
ETAs are collective agreements between employer organisations and unions that provide security and services beyond what is provided by public welfare when employees are threatened by redundancy (Jansson et al., 2018;Lindellee, 2018;Walter, 2015). Since many researchers studying welfare state institutions and labour market regulations focus on the interests of employers and employees in shaping institutions and regulations to align with their preferences, a similar framework could arguably be applied to the study of ETAs.
Some argue that institutions and regulations on the labour market are formed in a conflict between two opposing interests, where most labour market regulations and public or occupational welfare schemes are seen as the result of victories for the employees and their organisations (Korpi, 1981). Emmenegger (2015), for example, argues that there is an insolvable conflict of interest between capital and labour with regard to job security regulations due to the restrictions such regulations entail for employers' control over the workplace. The same reasoning would reasonably hold for employment security through collective agreements. Others argue that business interests can see the benefits of and support intuitions that overcome market failures and collective action problems, particularly the risk of underinvestment in skills (Hall and Soskice, 2001;Mares, 2003). Proponents of this latter viewpoint have even argued that 'social protection stems from the strength rather than the weakness of employers' and that 'the only way to encourage workers to carry a substantial part of the costs of firm-specific training is to increase job security and/or reduce the insecurity of job loss' (Estevez-Abe et al., 2001: 181, 151; see also Wood, 2001: 378). Furthermore, employers' organisations in coordinated market economies will use strong employment protection as a form of 'benign' or even 'beneficial' constraint to keep potential defectors in check (Streeck, 1997). Others take a more evolutionary approach, arguing that even if the employers do not necessarily see the benefits from the start, and might even initially heavily oppose it, they adapt their preferences and start to see the benefits of many welfare institutions and labour market regulations (Hall and Thelen, 2009).
However, it is harder to see how agreements or regulations that concern employment protection could be perceived as beneficial by the employer (Emmenegger, 2019). Public pensions mean that companies do not need to provide them (Swenson, 2002), social welfare relieves them from paternalistic burdens, and labour market policies extend the supply of labour (Martin and Swank, 2004), while vocational training increases the number of skilled workers (Thelen, 2004). Restrictions with the purpose to provide increased employment security, however, reduce the employer's control over the workplace (Emmenegger, 2014).
In the Swedish case, the political struggle over the Employment Protection Regulation illustrates this. The current law for employment regulation, LAS, has been highly contested by employer interest groups since its inception in 1973 (see institutional context below for an explanation of the meaning of the law in the context of Swedish industrial relations). Swedish employers argue that the regulations impede their ability to re-organise their production and keep essential personnel in case of redundancies. Unions, on the other hand, argue that the law is paramount for the security of their members, even if it appears that most unions preferred employment security through collective agreements until the 1970s (Nycander, 2010: 24-28). This gives the impression of an insolvable conflict hinging on opposing preferences concerning employment protection. However, previous researches on ETAs, particularly Forsberg (1996), have argued that there generally was a common understanding between SAF and PTK before the 1990s regarding the goals and functions of the ESA and the foundation TRR (short for Trygghetsrådet, The Security Council) that administrated schemes and programmes funded through the agreement. In the 1990s they started opposing it. She argues this could be attributed to a generational 'changing of the guards' at SAF. Those who had taken part in developing the agreement retired and were replaced with a new, 'tougher' generation (Forsberg, 1996: 97-99;cf. Johansson, 2000). Other, less personal and more external or structural causes have been put forward to explain SAF's move away from centralised bargaining and cooperation through bipartite or tripartite institutions during the 1980s and 1990s. Such causes have been changes in the mode of production, increased international competitive pressure and the globalisation of trade and financial market. This would be an example of a punctuated equilibrium when SAF walked out from peak-level bargaining and corporatist mediation (Johansson, 2000) and headed towards a neoliberal trajectory of change (Baccaro and Howell, 2011). In a similar vein, B€ ackstr€ om (2005) argues that it was through the new agreement in 1994 when it was renamed and rebranded, from an employment security to a transition agreement, that a seller-buyer situation emerged (cf. Engblom, 2018), where unions sold employment security, in the form of exceptions from the seniority-principle, for increased unemployment protection, in the form of TRR schemes and programmes. There seems to be a conundrum here, where the employers have not changed their preferences towards Employment Protection Regulation, but, rather, have changed their views on an institution that before the 1990s was explicitly aimed at providing increased employment security.
To understand why employers might have supported ESA/ETAs that were created to provide increased security for employees while they at the same time opposed employment protection regulations, it is important to distinguish between employment security and unemployment (income) protection. Unemployment protection means protection from income reduction due to unemployment while employment protection is defined as different forms of institutionalised employment security (Estevez-Abe et al., 2001: 150). Compared to the situation in Germany, Swedish employers are much less involved in the training and education of the workforce. Vocational education has developed to be the primary responsibly of the state (Busemeyer and Trampusch, 2011). Employer involvement in education and training is often denoted as an important reason behind firms' preference or acceptance for employment security. This would imply that employers in Sweden might accept solutions that provide unemployment protection, especially as it might facilitate flexibility at the firm level, but that they should strongly oppose employment security, regardless if this is forced upon them through either government regulations or concessions in collective agreements. We, therefore, hypothesise that employers should be opposed to and work against employment protection, but that they could potentially see some benefits with different forms of unemployment protection. This is particularly the case if the responsibility to provide for unemployment benefits or further training and education is provided by the government, not the individual company. Employees, on the other hand, are expected to be positive to both employment security and unemployment protection and could be prepared to forgo one for the other, or sacrifice salary increases for added protection, if they perceive the deal as a whole to be more beneficial. How important these are in collective agreements for unions depends on the perception of the urgency of the need for such benefits compared to others, such as salary increases. Higher levels of unemployment might for example increase the interest in such solutions.
An alternative hypothesis with regard to employers is that employer organisations accept solutions that entail increased employment security if such an agreement provides a partial solution to the collective problem of underinvestment in skills. As already mentioned, earlier studies of ETAs have suggested that this was the case before the 'neoliberal turn' that SAF took in the early 1990s. That is, the changes in the collective agreements for employment security or transition between SAF and PTK were not only or even foremost a result of changes in the balance of power but also the result of a shift in employers' preferences. However, earlier Swedish studies regarding ETAs (such as B€ ackstr€ om, 2005;Forsberg, 1996) have not been clear regarding the role and the changing positions of the key stakeholders and have not relied on archival material from either employer or employee organisations.

Material and method
The article uses an institutional approach within the tradition of historical institutionalism that views human beings as both norm-abiding rule followers and selfinterested rational actors and focuses on the study of the gradual changes in institutions (for example Mahoney and Thelen, 2010). The use of a single case in a historical context gives opportunities to explore aspects of the negotiations between employers and employees not observable through macro or quantitative approaches and also elucidates how shifting cultural, economic and political contexts can change institutions over time. The case here is the agreements between SAF/SN and PTK. The reason why the article focuses on this particular agreement is twofold. It is one of the oldest and largest foundations. It has also functioned as the blueprint for later agreements in other areas of the labour market, including the TSL (Trygghetsavtalet SAF LO) agreement between SAF/ SN and the blue-collar workers belonging to the Confederation of Labour (Landsorganisationen, LO).
The article draws extensively on unique archival material from both PTK (at TAM-arkiven, referred to as TAM/PTK/in references) and SAF (at Centrum f€ or N€ aringslivshistoria, referred to as CfN/SAF/). The material consists of protocols from collective agreements, congresses, internal memos and minutes from meetings between the two organisations. This helps us reconstruct what the sides wanted to achieve and their preferences during the negotiations in the 1970s, 1980s and 1990s. For the period after the new millennium, the article mostly relies on statements made by representatives for PTK, associated unions and SAF/SN and associated employers' organisations in newspapers (DI, DN, GP, NT, SvD) and union periodicals (Kollega, Arbetet, Ingenj€ oren).

Institutional context
There are a couple of contextual factors that have to be taken into consideration with regard to the development of the ESA/ETAs between PTK and SAF/SN. First, it is important to be aware of the basics concerning Swedish industrial relations. These have been characterised by high degrees of self-regulation and by solving conflicts through collective bargaining rather than government interference and regulation (this norm is often referred to as 'the Swedish model'). Second, and very important in connection with the agreements under scrutiny, is the Law for Employment Protection (Lagen om anst€ allningsskydd, LAS) and its relation to industrial relations in Sweden. Before the 1970s, there was no government legislation concerning collective dismissals. Instead, these were regulated through collective agreements (Bengtsson, 2006). For example, in the foundational main agreement from 1938 (the so-called Saltsj€ obadsavtalet), it was stated that in case of redundancies the needs for the production of the company concerning the skills and suitability of the labour force as well as the workers' justified concern for employment security should be taken into consideration (Nycander, 2010: 15-16).
By the late 1960s, however, the difficulty for particularly older workers to re-enter the workforce was perceived as such a social problem that universal employment protection was needed (Calleman, 1999: 68-69). In 1973 parliament voted for the first iteration of the Employment Protection Law, LAS (SOU 1979:7;Prop.1973: 129). LAS established the seniority-principle (that is, those with the longest period of employment should be dismissed last during layoffs) when an employer gave notice of redundancies, giving priority to employment security for older employees over the firm's needs for flexibility. Only through comments to the draft regulation was SAF able to include in the legislation that it should be taken into consideration if senior employees had adequate skills for the jobs that remained after layoffs when priority listings were established (Calleman, 1999: 70-73). That is, exemptions from the seniority-principle could be made if the senior employee was unable to fulfil any of the roles that remained at the workplace.
With its focus on employment seniority, the letter of the law is comparatively strict in its restrictions on managements' rights to dismiss employees. However, LAS is semi-dispositive (Fahlbeck, 1984). The term dispositive means that it is possible to make deviations from (parts of) the statutory law through collective bargaining. In LAS this prerogative is given to national unions. National unions can, in turn, if they wish to do so, delegate this prerogative to unions at regional, company or shop level. This means that deviations from the seniority-principle in the establishing of priority lists for collective dismissals due to redundancies are allowed if the union, after negotiations with the employers, agrees to it. Exactly how common deviations from LAS's seniority-principle are in practice is not known, but most studies have indicated that deviations are quite common, and especially so for white-collar employees. A joint survey study by SN and PTK in 2011, for example, showed that in a vast majority of cases employers and local union representatives had agreed on which employees would have their contracts terminated (although this includes cases when they both agreed to follow the seniority-principle). TRR was involved in 80% of these cases (Rudberg and Hedlund, 2011). Thus, LAS restricts the autonomy of the social partners while at the same time enabling them to disregard the letter of the law and find their own solutions if they are able to find common ground.

Initial agreement
The original idea for the first ESA was devised by Ingvar Seregard from the Swedish Union of Clerical and Technical Employees in Industry (Svenska Industritj€ anstemannaf€ orbundet, SIF) and first chairman of PTK (1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985). He presented the idea during negotiations with SAF in the late 1960s, but they were at the time only able to reach an agreement on AGE (avgångsers€ attning), a supplementary unemployment benefit for employees 40 years or older that had been employed at the same firm for at least 5 years (Forsberg, 1996: 53;SOU 1973:56, 152-153). (AGE was inspired by the lump-sum severance pay AGB (avgångsbidrag) agreed between LO and SAF in 1964.) During collective wage bargaining in 1973, SAF and several white-collar unions -SIF, CF (Sveriges Civilingenj€ orsf€ orbund, the Swedish Association of Graduate Engineers), HTF (Handelstj€ anstemannaf€ orbundet, the Union of Commercial Salaried Employees) and SALF (Sveriges Arbetsledarf€ orbund, the Swedish Association of Supervisors' and Foremen) -that would within the year coordinate themselves into the wage bargaining cartel of the PTK, strike a deal for an ESA and create the non-profit foundation TRR. TRR was, and is, a council jointly controlled by the employer organisations and trade unions that handles the administration and the different schemes and programmes that are paid for through the agreement. The board of TRR consists of 12 (originally 10) persons where SAF and PTK have 6 representatives each. The board directs TRR with the help of a working council that consists of representatives from SAF and PTK and TRR's CEO. Member companies of SAF were already paying the equivalent of 0.25% of their wage costs to finance AGE and this fee was now increased by a further 0.25 to cover the costs for AGE and TRR (Forsberg, 1996: 65). These fees would otherwise have been used for salary increases.
The initiatives from the white-collar unions for both AGE and the first ESA can be explained by perceived needs to increase their members' security in the face of increasing unemployment, structural change and disenchantment with the public employment services. For decades, white-collar unemployment had been an almost unheard-of problem. By the end of the 1960s, however, unemployment levels started to increase, even if the levels were still low by today's standards (unemployment among members of SIF increased from 0.1% to 0.2% in the 1950s and 1960s to 0.6% in 1969 and 1.2% at the time of the crisis of the early 1970s) (Nilsson, 1985). Furthermore, representatives of white-collar employees had become disenchanted by the Public Employment Services. Active labour market programmes were mostly aimed at blue-collar workers, not white-collar employees and professionals. In the late 1960s, the Public Employment Services also disbanded offices that specialised in referring employment opportunities to whitecollar employees, much to the disappointment of their unions (Delander et al., 1991: 80-81). Union representatives were therefore looking for alternative ways of providing increased security for their members.
That the ESA was a response to PTK union grievances seems quite straight forward. When the risk of unemployment increased and the state was not delivering solutions, unions were prepared to address these issues through collective bargaining. Earlier research has not, however, mentioned much about the impetuses for why employer organisations in general, and SAF in particular, were prepared to accept such an arrangement and what their incentives were. Lars-Gunnar Alb˚age, chief negotiator for SAF during the 1973 negotiations, in a later interview claimed that everything went very fast and that they took the deal that was on the table (Forsberg, 1996: 65). It seems that SAF accepted the ESA without a clear idea of how it would benefit them. We would, however, argue that SAF soon found a suitable use for the ESA and TRR. We will get back to this, but first, we shall discuss the actual content of the 1973 agreement.
The agreement started with a couple of pages on what was referred to as 'common values' (gemensamma v€ arderingar), stating that rationalisations, instigated with the collaboration of employee representatives, are necessary for companies to be able to provide employment and security in the long run. The ESA was, the text stated, formed to alleviate the negative consequences of restructuring. First, companies should, in the collaboration with employee representatives, plan ahead and prepare their employees for new working conditions. Second, actions should be taken to ease the individual employee's adaption to new demands associated with these changes. Third, employers should strive to protect employees, even in the occurrence of unforeseen events. Layoffs should only be considered if and when all other feasible options of relocation and re-education within the company had been considered (CfN/SAF/F3aa:256, Trygghets€ overenskommelse mellan SAF och PTK, 13/1 1973, bilaga 1).
The 'common values' illustrate that the agreement intended to increase the employment security of employees. Why did SAF agree to this? Was it a beneficial constraint that SAF accepted to secure reinvestments in workforce skills? This seems not to have been the case. It is apparent from the comments that SAF sent to its members in March 1973 that there was disagreement from the start regarding the role and importance of the 'common values'. SIF had claimed that SAF and its members should abide by these common values. SAF, however, stressed in their internal communication that they disagreed and claimed that the 'common values' were not part of the collective agreement per se. That is, neither of the parties could accuse the other of breaching a collective agreement with reference to these values. To put it more bluntly, SAF did not consider the 'common values' to be anything they needed to conform to unless they perceived it to be in their own interest. Quite the contrary -already from the start SAF stressed that white-collar employees should be prepared to accept a large degree of both geographical and occupational mobility in order to facilitate employers' need for flexibility (CfN/SAF/F3aa:226, Kommentar till trygghets€ overenskommelsen, 22/3 1973). This is also evident from later remarks regarding SAF's practices. At the 1985 PTK-congress, for example, Inge Granqvist from SIF reminisced how the 'common values' had been thought of as revolutionary at the time, a way to proactively handle challenges associated with technological change and to reduce the need for redundancies and unemployment for older employees. Yet, he argued, it soon became apparent that the employers saw the 'common values' as little more than words on a piece of paper and they preferred to pay their way out of trouble, either with the help of AGE or early retirement, the former fully and the latter partly funded through TRR (TAM/PTK/AIa:3, Kongressprotokoll 1985, § 10). It seems reasonable to argue that the willingness of SAF to accept both AGE in 1969 and the ESA in 1973 could be related to the growing political pressure for increased employment protection that led up to LAS in 1974. Both agreements were targeted at providing increased security for the same groups on the labour market which the social democratic government wanted to protect through legislation. But by establishing an institution that provided further unemployment income protection for older employees, SAF nevertheless, intentional or not, fashioned an instrument that could turn out to be beneficial in further negotiations at both central and firm levels.
In the first years of the agreement, the foundation TRR and its funds were essentially used for three functions. It administrated the payment of AGE to older tenured employees that had been laid off. The sums were decided on a case-to-case basis but usually amounted to an average of 4-5 months of salaries. TRR was also involved in matching redundant employees with new employment opportunities. The establishing of this service was explicitly related to the fact that PTK was disappointed with the public employment services (Forsberg, 1996: 68-69). Finally, TRR also helped cover the costs for early retirement for older employees, usually in their 60s, that were judged unable to find new employment. In 1975, a good year with few redundancies, the yearly report of the foundation tells us that 70% of those that were supported by TRR received AGE, 18% were helped to early retirement and 12% were given support to find new employment, including the funding of introductory pay, internships, courses and related travel expenses. AGE stood for 86% of the costs (TAM/PTK/AIa:1, A˚rsber€ attelse 1975). Essentially, what TRR provided was thus mostly additional unemployment income protection for older, tenured employees that had been dismissed. Regardless of what the opening paragraphs of the agreement stated, it was not an agreement that to any significant degree further curtailed the managerial prerogative of employers or provided proactive investments in skills.

A revised agreement in 1976
It did not take long for both sides to suggest amendments to the agreement. During lengthy negotiations in 1975/1976, SAF wanted to reduce the fee, and increase salaries instead, as they thought that TRR had enough funds for the purpose of financing AGE, matching services, and contributions to occupational pension payments. PTK, on the other hand, saw the improved finances of TRR as a good reason to implement further schemes, services and projects in line with the 'common values' expressed in the agreement (TAM/PTK/A1a:1, St€ ammoprotokoll 1976). PTK wanted TRR to aid in stimulating and implementing plans for better and more proactive personnel planning at the firm level -efforts that would, they hoped, reduce the need for redundancies due to structural change and restructuring in the future (Brattn€ as, 1976). The hurdle, from PTK's point of view, was that the 1973 agreement meant that SAF had a veto on all such projects (TAM/PTK/ F4i:1, Tj€ anstem€ annen och syssels€ attningen inom den privata sektorn, 27/4 1976). PTK needed a way to force SAF and its members on the board of TRR to accept solutions that not only provided additional unemployment benefits and services but also increased employment security through proactive manmanagement and planning.
PTK had some success in its endeavours through the signing of a new agreement in 1976. The annual fee was kept at 0.5% and funds could now be used for further education or any other training that would widen the occupational skills of an employee -a support that would enable employees to take on another position within the company or enhance their ability to find other employment. It was now also possible for either of the partners to unilaterally initiate such projects, provided the other partner had full insight (CfN/SAF/F3aa:331, € Overenskommelse om vidgad anv€ andning av trygghetsfondens medel och vissa d€ armed sammanh€ angande frågor, 21/5 1976). This was seen as a major achievement at PTK since it reduced the veto power of SAF over the use of TRR funds (TAM/PTK/A1a:1, PTK St€ ammoprotokoll 1976, § 17).
This seemed like a clear PTK victory. Why did SAF, that only the other year wanted to reduce the fees to TRR, not only agree to keep it at 0.5% but also agreed to give TRR extended functions that were partly outside their direct control? To explain this, we must look at the negotiations over salaries and benefits of which the amended ESA was a part. These negotiations had taken 17 months from start to finish and had finally been solved through arbitration from a conciliatory commission (CfN/SAF/F3aa:331, Vad h€ ande n€ ar i avtalsf€ orhandlingarna SAF-PTK?, 21/5 1976). PTK's initial suggestion was that the fund should be split into two, with 0.25% going exclusively to extended purposes and projects aimed at increased employment security. At this time the position held by SAF was for a reduction of the fee to 0.3% where the remaining 0.2% should be used for salary increases within the agreement (CfN/SAF/F3aa:332, F€ orhandlingarna SAF-PTK om allm€ anna villkor: € oversikt € over nul€ aget, 10/5 1976). Hence, in 1976 SAF was prepared to accept higher salaries instead of paying fees to TRR, when their priorities seem to have been the opposite in 1973. Loss of their veto-power was, in the end, simply the best deal they could get at this time, the difference from PTK's suggestions being that then all funds could still potentially be used for added income protection. This is indicative of SAF's aversion to funding projects aimed at providing increased employment security that would be unavoidable if the funds for TRR were split into two parts.

Early retirements and further attempts at reform
Despite the hopes of PTK, it was for other uses than employment security that TRR funds were much more commonly used. During the late 1970s and 1980s, they were extensively used to finance early retirements. Contributions to occupational pensions surpassed AGE as TRRs largest expense in the early 1980s. This was part of a general development in the Swedish labour market. In the latter half of the 1970s, early retirement had become a common solution to deal with longterm unemployment. This solution had started to function as a valve, removing people from a workforce when the policy goal of full employment became harder to realise (Berglind, 1994). The government did not, however, provide cover for the decreased payments to occupational pensions that early retirement entailed. Here TRR filled an important role. In the 1970s, TRR provided 40% of the costs for continued payments to occupational pensions incurred for companies with redundancies. This made it more attractive for companies to offer early retirement to older employees, regardless of the union's priority listing, and was quite a popular proposition among older employees as well (TAM/PTK/F4i:1, Minnesanteckningar, PTK-T, 14/8 1978). The yearly fee to TRR was thus from their perspective akin to insurance that helped employers negotiate around the seniority-principle in LAS if older employees (that often were far down on priority lists following the seniority-principle) were prepared to accept early retirement. This was very convenient, and SAF was therefore dismayed when the government removed the possibility for early retirement due to labour market reasons in 1991 (CfN/SAF/F3aa:468, Untitled memo, 23/2 1990).
PTK was less impressed. They wanted to use the funds for proactive projects that would enhance the employment security for their older employees, not have them pensioned off. Employers' propensity in using TRR resources to facilitate the early retirement of older employees constantly encroached on using them for the purposes that PTK had originally envisioned. Concerns over the use of the funds to provide for further retirement were expressed several times at PTK meetings. Representatives argued that TRR must stop functioning as an institute for retirement (TAM/PTK/F4i:1, Minnesanteckningar, PTK-T, 23/11, 1976;14/8 1978). One concerned representative exclaimed that if this trend continued TRR would soon have assisted every 60-year-old in the country into an early retirement (TAM/ PTK/F4i:1, Minnesanteckningar, PTK-T, 26-27/1, 1983). These practices also threatened the financial sustainability of TRR. In September 1983 an emergency meeting was convened over the mounting costs. Calculations for that year showed that retirement provisions accounted for 450 of the year's total expenditure of 800 million SEK. Incomes from fees only amounted to 235 million SEK (TAM/ PTK/F4i:1, Minnesanteckningar, extra sammantr€ ade, PTK-T, 6/9 1983). Discontent over how employers used funds to get rid of older employees one way or the other started brewing among PTK unions and led to a long and animated debate at the 1985 PTK Congress. One SIF representative even accused the personnel of TRR (its CEO was presenting at the congress) of being the henchmen of the employers, forcing local union representatives to accept priority listings that suited the employers during collective lay-offs (TAM/PTK/AIa:3, Kongressprotokoll 1985, §14d). The PTK leadership had to act on this discontent, and during the latter half of the 1980s, when unemployment figures kept going down in an overheated Swedish labour market, they saw their opportunity. PTK wanted to use the good conditions on the labour market to push their positions forward (CfN/SAF/F3aa:468, Minnesanteckningar 13/8 1987;13/11 1988). In 1987 PTK contacted SAF to initiate negotiations over a revision of the agreement and presented several suggestions. Although PTK representative Sten-Olof Heldt stated that PTK was open for suggestions and did not commit to any specific demands, the overarching idea PTK presented was to include several of the sentences stated in the 'common values', or rather an updated version of them, into the agreement proper, together with several suggestions for prolonged notice of dismissal for employees older than 40 and some other smaller adjustments to the eligibility rules. He stressed that this would not incur any added costs and there would therefore be no need to raise fees (CfN/SAF/F3aa:468, Minnesanteckningar från f€ orhandlingsomgång, 12/6 1987). They wanted to put more emphasis on longterm planning, individual adaptions and increased security for the individual. SAF argued against the first part, claiming that the idea of long-term planning was an obsolete notion from the early 1970s, when it was thought that firms could have such long-term plans for their activities, including their man-management needs (CfN/SAF/F3aa:468, Minnesanteckningar, 4/12 1989).
These negotiations would continue on and off until November 1990 but were severely impeded by the fact that the Association of Swedish Engineering Industries (VF, Verkstadsf€ oreningen), representing some of the largest export industries, had dropped off peak-level bargaining in 1983. VF no longer let SAF represent them unconditionally in negotiations that concerned employment contracts, such as prolonged notice of layoff for older employees. This was part of an increasing movement among employers away from centralised bargaining (De Geer, 1992). VF was also strongly against the notion of including texts from the 'common values' into the agreement due to the risk of claims for compensation that this would potentially incur. They were particularly concerned that this, in conjunction with the Codetermination Act of 1976, would give the unions a prerogative in interpreting the situation, threatening to further reduce management's prerogative over staffing needs (CfN/SAF/F3aa:468, Trygghetsavtalet SAF-PTK, letter from VF to SAF, 2/10 1987).
Notwithstanding beneficial economic and legislative circumstances, PTK was not able to achieve much headway in the latter half of the 1980s. SAF saw no apparent upsides to further reduce their control of what the funds of TRR could be used for. Even if this had been the case, their ability to negotiate such an agreement was constrained by VF's ambitions to reduce the central organisation's role in collective bargaining. Lower employment figures among those eligible for support from TRR did, however, mean that there were more funds available for proactive projects in the form of education and training for employees with out-of-date skills, and PTK could instigate such projects in line with the conditions in the 1976 agreement.

The worst of crises, rising unemployment and ESA becomes ETA
In the early 1990s, conditions changed radically. Sweden experienced a severe economic and financial crisis and rapidly rising unemployment that affected white-collar employees and professionals to a higher degree than before. The number of white-collar employees that were on the books at TRR increased from about 5,000 in 1989 to almost 50,000 in 1993. This entailed that TRR had to face rising costs at the same time as the government, due to financial strains, decreased replacement rates in unemployment and social insurances. The annual fee for companies, that had been reduced to 0.35% just a couple of years earlier, was temporarily increased to 0.95% for 1992 and again to 1.07% in 1993 to be able to cover rising costs (TAM/PTK/F6a:196, Protocols from negotiation, 23/2 1993). Proactive projects as well as education and training, which had been an increasing part of TRR's activities during the latter half of the 1980s, were reduced as funds had to be focused on financing payments for AGE and contributions to early retirement (Forsberg, 1996: 87-88).
On top of this, the Social Democratic Party lost the election in 1991, turning power over to a centre-right government. A government enquiry was promptly initiated and soon delivered suggestions that resulted in far-reaching changes to LAS, not least regarding the seniority-principle (Calleman, 1999: 88-93;SOU 1993:32). For a brief period, lasting 1 year, employers had to exempt two employees from priority listings (this was directly repealed in 1995 by the new Social Democratic government). The balance of power tipped over to the employers' advantage, who now had an opportunity to roll back some of the institutional changes and practices that PTK had pushed through during the previous decade. It was again the Association of Engineering Industries (just recently renamed VI, F€ oreningen Sveriges Verkstadsindustrier) that held a hard line against PTK, and SIF in particular, claiming that unions were too restrictive in accepting deviations from the seniority-principle.
In this context, SAF started to make the argument that the ESA was outdated and needed revision and PTK was forced to abide by it (Forsberg, 1996: 89-91; TAM/ PTK/F6a:196, Finansiering av Trygghetsrådet SAF-PTK, 11/1 1994). They set up a joint working group that debated how the agreement should be re-configured (TAM/ PTK/F6a:196, Finansiering av Trygghetsrådet SAF-PTK, 11/1 1994). But when PTK did not accept SAF's demand for giving them increased control over priority listings during collective lay-offs, the employers in January 1994 declared that they cancelled the ESA (TAM/PTK/F6a:196, Upps€ agning av Trygghets€ overenskommelsen m m, 21/1 1994). When they returned to the negotiating table the propositions from SAF were harsh, delivered at a time when LAS was under seemingly severe threat (TAM/PTK/ F6a:196, undated transcript from PTK-meeting). Among other things, they wanted to negotiate away priority listings per LAS's seniority-principle at all levels. If the firm and the local union could not agree on a priority list, laid-off employees would not get any support from TRR. If PTK did not accept these demands, TRR would be disbanded by the end of 1994 (TAM/PTK/F6a:196, F€ orhandlingar om Trygghetsvatal SAF-PTK). This put PTK under immense pressure to reach a new agreement or have thousands of their members lose AGE and TRR's support and services. PTK chairman Heldt thought that they had to accept it or lose TRR altogether; SAF was not bluffing (TAM/SAF/F6a:196, undated verbatim transcript from PTK-meeting, February 1994). After a couple of meetings, they managed to get SAF to step away from some of these demands with the threat of industrial action (TAM/ PTK/A1a:3, PTK St€ ammoprotokoll 1996, §9; TAM/PTK/ F6a:196, SAF suggestion, 17/12 1993). After several rounds of negotiations, they were finally able to reach a new agreement in August 1994.
In the new agreement, named the ETA, there was nothing left of the 'common values'. Instead, it was stated in §2 that the core idea of the agreement was that firms would continually pay fees to TRR and that these funds should be used in connection with redundancies, to give economic compensation to those that had been laid off, and support them in finding a new job. It also stated that it was a joint obligation for both parties to establish priority lists that would leave the firm with a workforce that would ensure that it could achieve productivity, profitability and competitiveness. According to LAS, it was the national union that had the prerogative to delegate negotiations over lay-offs to the local level, but in the new agreement, it was stated that these negotiations should be conducted at local or union club level. The term 'competence' was also introduced in the agreements. As noted earlier, the term 'adequate skills' had been included in LAS as a potential caveat from the seniority-principle. From now on the firms' continued need for competence should be taken into greater consideration when agreeing on priority lists, implying that just adequate skills and seniority should not be enough (CfN/ SAF/F3aa:13, F€ orhandlingsprotokoll, 17/8 1994). The fee for the new 1994 agreement was set at 0.35%, soon to be reduced to 0.3%. Most of the costs went to AGE and matching services as regulatory changes during the economic crisis had removed most of the possibilities for early retirement solutions and proactive projects and educational programmes had been reduced to only a few per cent of total costs at TRR (Forsberg, 1996: 91). This new agreement caused considerable rumination among PTK representatives.
SAF went on the offensive again a couple of years later. When tripartite negotiations in 1996 over reforming Sweden's labour laws, including LAS (Samarbetsavtal? summarised in Prop. 1996/97:16), led to nought (GP, 1996;SvD, 1996), SAF decided not to renew the agreement in June 1997 (GP, 1997a). SAF's, and particularly VI's, displeasure with the 1994 ETA was related to how union tactics had counteracted employer ambitions for decentralisation. In the 1994 agreement, it was specified that individual firms could, for an added fee of a further 0.2%, buy into a TRR package called extended services. These extended services consisted of further support to find a new job for those that had been made redundant; firms could, in essence, for the benefit of themselves and their employees retain some of the services that were not included in the 1994 deal. This was initially thought of as something that individual firms could opt-in for at their own discretion, but unions within PTK soon included demands that all companies should sign up for these services in industry-level collective agreements. Within 2 years, 95% of white-collar employees in PTK unions were covered by the extended services (TAM/PTK/A1a:3, PTK St€ ammoprotokoll 1996, §9). Much to SAF's annoyance, this essentially made the voluntary extended services quasimandatory. SAF also claimed that their members had to pay twice, both to TRR and then compensation on the side to individual employees that they did not want to retain, underlining that the primary function of the ETA was to increase the flexibility for employers and that the current deal was not delivering enough flexibility for the price they were paying. SAF again cancelled the agreement. Both journalists and PTK did, however, notice that there seemed to be disagreement within SAF. Statements by SAF's chief negotiator in archival material from the negotiations indicate that it was VI and the larger companies that wanted to 'maximise their freedom' (TAM/PTK/F6a:205, Minnesanteckningar, omst€ allningsavtal -f€ orhandlingar, 25/9 1997). But this ambition was not shared by all employers and many companies seemed in fact quite positive to the ETA and pleased with the services and support provided by TRR (TAM/PTK/F6a:205, F€ orhandlingarna om omst€ allningsvtal om olika alternativ inf€ or 1998, 27/8 1997DN, 1997bDN, , 1997cGP, 1997b). If SAF lacked full support from their ranks, this was not the case at PTK. They claimed that if the ETA was cancelled and not replaced, all priority listings when firms faced redundancies would in the future follow LAS and the seniority-principlewith no exceptions (TAM/PTK/F6a:205, Sammanfattning av SAFs f€ orslag betr omst€ allningsavtal, 4/11 1997). This would potentially make collective dismissals much more complicated for firms and they had to accept some solution short of disbanding TRR completely.
In December 1997 PTK and SAF reached a new agreement. The ETA was saved once again. Both sides claimed victory. Mari-Ann Krantz (SIF) reasoned that the continued existence of TRR was in no small part a result of what employers had not expected: the solidarity that union members had shown towards the unemployed. Anders Sandgren (VI), on the other hand, maintained that it was now possible for their members to choose different alternatives and that costs associated with lay-offs would be reduced (DN, 1997a;GP, 1997c). The fact that LAS was no longer under review for reform and unemployment was going down probably helped PTK in the negotiations. But it came at a price. The extended services were now gone, as they had been removed from the new agreement, and the annual fee was set at a modest 0.3%. where it has stayed ever since. Moreover, the opening paragraph of the agreement now stated that companies at which redundancy arises must be provided staffing conditions as optimal as possible for their continued operations (CfN/SAF/F3aa:14, € Overenskommelse om Omst€ allningsavtal, 28/12 1997). The 'common values' were nowhere to be seen and flexibility was to be prioritised before employment security. The mood at the 2000 PTK congress was far from jubilant as PTK chairman Holger Eriksson grudgingly acknowledged that these changes had been necessary in order to save the agreement (TAM/PTK/ A1a:7, St€ ammoprotokoll 2000, §1).

Recent developments
Since the turbulent mid-1990s, TRR and the agreement between SAF and PTK have not been under serious threat. This could at least in part be attributed to the fact that LAS -the lynchpin on which these agreements depend -also has not been under serious review for liberal reforms, at least until very recently. The ETA between PTK and SN has been the subject of re-negotiations at least twice during the last 20 years. Unionen, a white-collar union created through the merger of SIF and HTF in 2008, was concerned over the limited skill development received from their employers (Fahlberg, 2012). The incentives for SN to participate in negotiations were quite clear: to get rid of, or at least reduce the importance of, the seniority-principle in LAS, and give added weight to skills and competence in the establishment of priority lists (SvD, 2002;see also DI, 2011;DN, 2011;Kollega, 2011;NT, 2015). SN would, in turn, agree to double the fees to TRR. Unionen and SN were prepared to make a deal. Suggested changes in the agreement would give employers a stronger say over collective lay-offs, circumventing the local unions' influence on priority lists. However, rifts between Unionen and several of the other unions and some of their own local clubs made it difficult to reach an agreement (Arbetet, 2015;Ingenj€ oren, 2015). In January 2019 they were, however, able to make a deal that means that TRR can now supplement the government's financial aid for students (Arbetet, 2019).
At the same time, ETAs have become more common in the 2000s. An agreement between LO and SAF in 2004 and an agreement in 2012 for public sector employees have meant that a further two million employees are now covered by such agreements, bringing the coverage to 90% of the labour market. Concerns over unemployment risks as well as employers' interests in ways of negotiating for increased flexibility mean that these agreements are of interest for both parties -at least as long as LAS remains and gives central unions the prerogative for accepting deviations from the seniority-principle during collective lay-offs. The proliferation of such agreements since the 1990s could possibly be seen as (part of) a new social pact with increased emphasis on competitive corporatism (Rhodes, 2001), but in the tradition of Scandinavian corporatism, with little or no direct state involvement. It is, however, important to note that LAS and labour laws are again under review and suggestions for liberal reforms have been presented (SOU 2020:30). During 2020 the key stakeholders (LO, PTK and SN) have negotiated for an alternative solution to the changes presented in SOU 2020: 30 where ETAs would probably play an important part, but have not been able to reach a full agreement on this. PTK and SN, IF Metall and Kommunal, the two largest unions within LO, agreed in December 2020, but not other unions in LO. The matter is ongoing at the time of writing.

Conclusions and discussion
This article has charted the institutional trajectory of ESA/ETAs between SAF/SN and PTK from the 1970s until the present time. Our research questions were related to what SAF/SN and PTK wanted to achieve through these agreements. How has the content of the agreements and what TRR has provided shifted over time? Did changes in the agreements originate from a shift in preferences or can these changes be explained with shifts in the balance of power on the labour market? We hypothesised that employers would be more interested to grant employees added unemployment protection if it benefits firms in negotiating deviations from LAS's seniority-principle, thereby increasing their flexibility. However, an alternative hypothesis suggests that employer organisations might be prepared to accept increased employment security if this contributes to solutions of collective action problems associated with underinvestment in skills. Previous research on Swedish industrial relations in general and on ETAs, in particular, has often argued that there was a shift in preferences among employers and SAF in the 1980s and 1990s, away from centralised bargaining and corporativism. Unions, on the other hand, are expected to achieve greater employment security but are also willing to accept increased unemployment protection. What they prefer might come from their members but also depend on what they can get from their counterpart during negotiations.
The results of the article show that employers in SAF/SN have not changed their preferences regarding either employment security or unemployment protection during the period examined. They were always against increased employment security but could accept added unemployment protection during collective bargaining. Unions, on the other hand, had a strong preference for increased employment security beyond what LAS already provided. What they wanted was investments in the skills of older, tenured, employees so that they would not be made redundant by developments within the company. But, as the balance of power was shifting in the direction of employers in the early 1990s, PTK had to give up employment security in order to keep and expand added unemployment (income) protection for older employees losing their jobs. By the early 2000s, PTK unions had long given up their ambitions for increased employment security as it was no longer a viable option in negotiations with SAF/SN. They still promote skills investments over added income protection but acknowledge that such education and training often have to be conducted after being dismissed, not as a proactive measure. SAF, on the other hand, has never shown any interest in using the ESA/ETA as an instrument for overcoming the risks of underinvestment in skills (although that does not contradict the possibility of some companies acting in that way; see Ahlstrand, 2015, for an illustrative case). What did happen within SAF during the 1980s, though, was an increased preference for decentralisation of collective bargaining: more market, less central coordination. When their bargaining position was strengthened in the 1990s, they were able to push through changes. However, due to LAS stating that making exemptions from the seniority-principle was the prerogative of national unions, SAF was in the end forced to accept the continued existence of a centralised collective agreement rather than unilateral or bilateral solutions at industry or firm level.
In relation to previous research, the article makes a couple of contributions. First, these results speak against the notion in earlier research within the Varieties of Capitalism tradition, suggesting that employers in coordinated market economies can welcome or accept constraints to their managerial prerogative over whom to hire and fire if it contributes to solving collective action problems. SAF/SN have never approved of such constraints. This is quite likely related to the fact that employers in Sweden have only to a limited extent been involved in the vocational education of the workforce. The education and training of younger employees is the responsibility of the state. The same goes for older employees. In the case of this ESA/ETA, it has been unions that have been advocating investments in skills, since such investments strengthen an employee's value for the firm and on the labour market. These results are in line with the balance of power arguments that stress conflicting interests between employers and employees over the security-flexibility nexus (cf. Emmenegger, 2014) -both before and after the early 1990s.
This brings us to the second contribution. During the history of TRR these agreements between SAF/SN and PTK might give the impression of institutional resilience, but it is an institution that has been reshaped in order to facilitate labour market flexibility. There are nevertheless reasons not to over-emphasise the extent of a neoliberal institutional conversion. The practices of TRR changed less from the 1980s to the 1990s compared to changes suggested by the written agreements. Due to the influence of SAF, the benefits and services provided through TRR already functioned that way, even if PTK strived for something different. At the same time, the continued existence of LAS in its present form has given these institutions a certain degree of resilience to change. A better label for this institutional conversion might be what Thelen (2014: 14-15) has called embedded flexibilisation, where flexibility is introduced in a context that collectivises the risks associated with unemployment, rather than a wholesale neoliberal trajectory of change. Institutions such as ETAs could potentially be part of the solution to the challenges of modern labour markets, but this potential is hampered by the inherent conflict of interest between employers and employees.