Procedural tools and pension reform in the long run: the case of Sweden

ABSTRACT

Welfare state reform is understood to be risky, difficult and still ongoing. As such, there is a need for analytic tools that can aid understanding of how governments are able to overcome the various barriers they face in seeking change, and the challenges of managing complex and hybridised welfare arrangements. This article argues that the policy tools literature provides several promising avenues for doing so. Specifically, it suggests that procedural policy tools – increasingly recognised in policy studies – are often critical for mitigating risk of blame, sustaining coalitions and supporting the ongoing adaptation of social policy systems to changing conditions and information. This is especially critical where policy innovations entail new substantive relationships between states, citizens and markets, as initial assumptions about individual and group behaviour are likely to be flawed. Procedural tools can enable learning and recalibration in the pursuit of public legitimacy for changing arrangements. To demonstrate the argument, this article engages closely with a case study of public pension reform in Sweden in the 1990s. The case study finds that, initially, reformers made use of a politically insulated parliamentary working group to design a new system, one which attempted to automate elements of policymaking and transfer risk from the state to the individual. However, this goal has proved more difficult to realise than imagined, and in the longer-term, the government used procedural tools for the purposes of learning and recalibration, and in doing so has gone some way to re-establishing modified forms of state accountability.

KEYWORDS:

• Welfare state reform
• policy instruments
• procedural tools
• pensions
• Sweden
• blame avoidance

Introduction

Scholars have long noted the challenges involved in making and sustaining social policy. Changing economic, demographic and political conditions place welfare systems under pressure. And yet, existing arrangements are often sticky, with beneficiaries likely to mobilise in the face of explicit retrenchment. Those studying the so-called ‘new politics of the welfare state’ have sought to understand the strategic choices that reformers face under such conditions. The major advances in this literature have developed out of close consideration of the strategies and substantive policy choices that can be used to distribute benefits and losses in ways that head-off backlash, create opportunities for credit claiming and help overcome institutional hurdles.

At the same time, recent work in policy studies presents an opportunity for closer consideration of the role of ‘policy tools’ in welfare state reform. In particular, scholars have highlighted the importance of ‘procedural’ policy tools – those that shape decision-making processes, rather than outcomes directly. To the extent that procedural tools are considered by welfare state scholars, their role is largely seen as one of overcoming organised interests. More rarely has there been direct engagement with the vibrant policy studies literature on tools, nor has there been close consideration of the relationship between procedural and substantive tools.

In this article, it is argued that procedural tools may play a key role in mitigating blame and backlash, sustaining reform coalitions and supporting the ongoing function and adaptation of social policy systems in the long term. The argument is demonstrated by engaging a case study of public pension reform in Sweden. The use of procedural tools is important for pension reform not only because of the risk of political backlash but because pension systems often require recalibration in the face of uncertain future conditions. They are also increasingly complex, involving multiple ‘tiers’ for citizens to negotiate. Sweden’s reform is a valuable case, due to the innovative nature of its new system, the tricky political conditions navigated by reformers and the ongoing recalibration of several important policy instruments that has been seen in recent years.

The analysis highlights the difficulties inherent in using new policy tools to shift accountability from the state to citizens, even where carefully designed. However, it also demonstrates the possibility that procedural tools may act as mechanisms for learning, recalibration and even the adjustment of policy goals. In the Swedish case, this process has been anchored around the ongoing work of a cross-party parliamentary working group, but has also involved the reorganisation of government agencies in response to various challenges in individualising and automating the pension system. The process of adaptation is especially important in the context of designing robust social security systems, given increasing of policy mixes in social security, and uncertainty regarding the response of citizens to new forms of policymaking.

Policy tools and the welfare state in the long run

At the heart of much recent research on the welfare state are the twin observations that various demands, problems and crises drive governments to seek reform, and yet reform itself is difficult and risky. Exogenous factors, such as declining birth rates and shifts in the labour market driven by technological change and globalised trade (Hemerijck, 2013; Morel, Palier & Palme 2012; Morgan, 2013), as well as political ideas and endogenous problem pressure (Hannah, 2018), are theorised to drive change. And yet, change, whether to scale back government provision or to expand into new arenas, is argued to risk backlash from constituents who perceive benefits from existing arrangements (Pierson 1994; Starke, 2006). For this reason, scholars have long focused on the ways in which governments pursue ‘blame avoidance’, through strategies such as manipulation of procedure (such as delegation of decision-making to another body), perception (utilising framing and communication) and payoffs (such as providing sufficient compensation to affected constituencies) (Vis, 2016; Weaver 1986).

The blame avoidance concept has been central to our understanding of the lack of radical retrenchment once expected by analysts, and of the more subtle means used by government to navigate constrained circumstances (Bonoli, 2012). However, more recent literature has questioned the centrality of the simple understanding of blame avoidance – as resulting from the prospect of organised backlash to policy change. For one, the link between policy change and electoral punishment is not as well established as might be expected (Wenzelburger, 2014), with governments successfully claiming credit in some cases of welfare retrenchment (Nelson, 2016; Slothuus, 2007). Recent literature in policy studies also suggests that welfare state scholars use an overly narrow conception of ‘blame’ (Hinterleitner, 2017). Christopher Hood’s (2010) work, for example, suggests that blame avoidance strategies are related not only to elections but a range of potential points of accountability for government. Policies can succeed or fail at various stages of the policy process, and have consequences for various actors and relationships, such governing coalitions or factions within parties (McConnell, 2010). As Leong and Howlett (2017) suggest, the pursuit of credit and blame may be more accurately described as matter of pursuing or maintaining legitimacy, rather than only electoral success.

This broader focus on longer-term legitimacy has begun to be reflected in contemporary work on the welfare state. The work of Alan Jacobs (2011) is a key example. Jacobs argues that, contrary to the dominant scholarly view, the long-term consequences of complex social policies do in fact shape the politics of reform. For Vis and van Kersbergen (2013), governments are responsible for the basic function of a social insurance system, and need to be seen responding appropriately where consensus around problems develops. For Häusermann (2010), successful reform is often as much about the engineering of sufficiently broad coalitions as it is about avoiding backlash. Various authors have discussed the ways in which moves towards individualisation, marketisation and ‘delegation’ of welfare programmes and services have resulted in hybridised and sometimes technically complex new arrangements that require ongoing management from government (Morgan & Campbell, 2009). Finally, recent work has sought to build on the work of Jacobs (2011) to examine the longer-term resilience of major reform (Béland et al. 2020).

All of this work is valuable, and advances understanding of the strategic dilemmas facing policymakers. However, it is not without limitations. This is particularly the case in relation to understandings of the specific tools and instruments that governments actually use to maintain the legitimacy and function of key programmes. Where detailed work has been done, this is largely in relation, timing and the role that distribution of benefits and costs can play in avoiding blame and claiming credit (Nelson, 2016; Slothuus, 2007; Wenzelburger et al. 2020) or a more general ‘institutional capacity’ to insulate decision-making from external pressure (Jacobs 2011). Comparatively less attention has been paid to procedure, and the ways in which procedural and substantive tools may interact in the longer-term management of complex social programmes. In the next section, I will outline how the welfare state literature may draw upon existing work in policy studies to better account for the role of policy tools.

The value of a policy tools perspective

The policy sciences have long paid attention to policy tools and instruments (Hood, 1983), defining them as ‘the techniques through which governments generate, evaluate, and implement policy options’ (Capano & Howlett, 2020, p. 1). As part of the ‘new design orientation’ (Howlett, Mukherjee, & Woo, 2015) in policy studies, a more recent revival of interest has advanced understanding of the dynamics of policy choice in ways that have largely not been integrated into scholarship on the welfare state.

Attention to tools and instruments has not been absent from the welfare state studies (Jensen, Arndt, Lee, & Wenzelburger, 2018). However, there is reason to think a synthesis of these two literatures would be valuable. For one, much theoretical work on policy design builds upon empirical cases in areas that face similarly constrained policymaking environments, such as energy transitions (Rogge, Kern, & Howlett, 2017). Second, recent work on American health care policy has acknowledged the potential role of substantive and procedural tools as part of an adaptable policy ‘mix’, over the life over a social policy (although this work has largely focused on policy ‘feedback’ and resilience to direct attack, see Béland et al. 2020).

The tools and instruments literature can contribute to understandings of the welfare state in two main ways. First, through typologies and schemas of instrument types and policy components (Vargas & Restrepo, 2019), and consideration given to the mechanisms by which tools may impact target groups (Capano & Howlett, 2019; Lascoumes & Gales, 2007). Second, by outlining the types and effects of procedural tools, those that shape the policy process rather than impacting citizens directly (Howlett & Rayner, 2007; Lang, 2019). While consideration of the procedural is hardly absent from welfare state studies (Jacobs 2011; Allern, Bay, & Saglie, 2013; Fredriksson, Blomqvist, & Winblad, 2012), it requires further theorising as a core element of welfare policy design. Third, through work on policy ‘mixes’ and complexity, scholars have investigated the relationship between sets of instruments and between instruments and goals (Howlett & del Rio, 2015). Consistency and coherence in instrument choice and alignment, supporting clearly identified objectives and goals, are understood to be important. However, doing so is clearly challenging in practice, especially where new is repeatedly layered is upon old (Carter, 2012).

At the same time, there are a number of ongoing areas of study in the tools literature that could be furthered through engagement with cases of policymaking in the welfare state (Capano & Howlett, 2020). In the following section, several of these will be addressed through a case study of pension reform in Sweden during the 1990s. First, the analysis will investigate the relationship between procedural and substantive tools. Procedural tools may play various important roles in shaping the behaviour of policy actors, such as aiding collaboration or compromise among actors and groups with diverse interests or determining the means by which policy is evaluated and recalibrated over time (Howlett & Rayner, 2007; Lang, 2019). However, there is a need to better understand how procedural tools both may shape the choice and support the ongoing functioning of substantive policy tools. Second, there is a need to further investigate decision-maker behaviour in ‘designing, discovering, and deciding upon tools in mixes’ (Capano & Howlett, 2020, p. 4), and the extent to which their decisions are based on instrumental reasoning, pre-existing ideas or an effective understanding of target groups (Colebatch, 2018; Weaver, 2014). Finally, the tools and instruments literature is increasingly interested in the temporal dimensions of policy change, especially the capacity to imbue policy with characteristics of robustness or resilience and adaptability (Capano & Woo, 2017; Jordan & Matt, 2014, p. 233), which, building on Pierson (1994) and others’ work on the ‘new politics’ of the welfare state, is an increasingly important theme in welfare state studies, especially in relation to pension systems.

Pension reform in Sweden

Pension reform is a highly suitable case for examining the relationship between procedural and substantive, beyond the fact of its long-term impacts. Pensions are an area in which there have been meaningful policy innovations over the past several decades, despite the risk of backlash, to the extent that even several long-established systems have partially or fully shifted between regime types (Weaver, 2010). These reforms have often entailed a complex mixes of financing and provision that require ongoing management by both governments and citizens (Bonoli, 2000). Among such examples, the Swedish case stands out. Not only did it largely precede a wave of multi-pillar reforms across the rest of world (Orenstein, 2008), the system that resulted has been described as ‘one of the most radical pension reforms in the OECD’ (Anderson, Blomqvist & Immergut 2008, p. 182) and has been seen as an example to emulate by policymakers in other nations. And yet, despite the radical shift, the formulation and implementation of the new system did not result in meaningful backlash or organised opposition. While several scholarly works have examined have examined the factors that enabled initial agreement (Loxbo, 2009; Lundberg, 2005; Marier, 2008a), an ongoing process of discussion and reform around pensions in Sweden over the past two decades provides an opportunity to revisit the case. In the following sections, I will explore the role played by procedural tools in establishing this policy mix and their use in supporting the ongoing functioning of the pension system (summarised in Table 1).

Table 1. Policy tools in Swedish pension reform

Policy tools	Classification	Purpose	Mechanism	Enablers/barriers	Long-term results
Committees	(a) Procedural (authority)	Agenda-setting, policy formulation, monitoring and recalibration	Consensus; control over information; broad mandate	Legitimacy; timing	Ongoing success in enabling consensus and recalibration
Automatic balancing mechanism	(a) Substantive (treasure) (b) Procedural (authority)	Financial sustainability; insulation from short-term politics	Automating benefit adjustments in line with economic conditions	Means of subversion always available	Partial subversion: adjustment and ad hoc policymaking in response to economic crisis
‘Orange envelope’ Agency collaboration and reorganisation	(a) Substantive (nodality) (b) Procedural (authority & organisation)	Supporting function of other substantive instruments – private investment funds, flexible retirement age – to ensure individual choice and adaptation of pension system to exogenous change	Public information campaigns; regular provision of detailed information regarding pension funds; creation of new public agency fit for purpose.	Incoherence with design of substantive instruments; other policy areas not in support; long-established social norms	Failure to encourage wide-scale choice; ongoing investigation and recalibration to adjust goals and pursue active citizen engagement through other means; attempts at agency coordination and eventual reorganisation.

Enabling cooperation and consensus – the role of committees

The existing Swedish pension system, known at the ATP, had begun in 1960 and the debate over its establishment was among the most divisive in modern Swedish political history (see Hagen, 2013; Heclo, 1975). The model preferred by the Social Democratic government would only triumph following three separate reform commissions, an inconclusive national referendum and an election. By the 1980s, however, it had become clear that the ATP system was soon to face financial difficulties, the result of demographic and economic changes and faulty assumptions made by an earlier generation of policymakers (Sunden, 2006).

Investigatory committees have long played a key procedural role in Swedish policymaking, helping to overcome its propensity for minority government (Petersson, 2016). This is perhaps even more so for pensions, with a series of major investigations having informed the development of the system (Hagen, 2013; Heclo, 1975). It is therefore not surprising that they were once again utilised to develop a replacement for the ATP. However, the composition and operation of investigatory committees are somewhat flexible according to the government’s purposes (Petersson, 2016). This flexibility allowed the repurposing of a very familiar procedural tool, through largely informal means, for an unprecedented policy change – one that broke through earlier disagreements to build lasting cooperation between the parties.

The first of the three committees was established in 1984, and released its final report in 1990 (Pensionsberedningen, 1990). It was composed of both politicians and members of interest groups, and was tasked with reporting on the state of the pension system, rather than forming consensus regarding an alternative (Pensionsberedningen, 1990, p. 12). The investigatory committee’s work set the ground for reform. Media coverage peaked with the release of its final report and proponents of reform were given substantial amounts of coverage in major media outlets (Hannah, 2018, p. 537). It also allowed the parties and interest groups to recognise the issues with the system – especially regarding its financial sustainability (Pensionsberedningen, 1990) – without having to deviate from long-held party orthodoxy, such as the Moderate’s preference for private alternatives (Margit Gennser, personal communication with author).

Following 2 the 1991 election, at which a centre-right coalition took power, a new committee was established to develop reform, which most parties now agreed was needed. Several unusual features of the committee contributed to an eventual agreement upon the outline of a new system. First, in a highly unusual step, the new Minister for Health and Social Affairs, Bo Könberg, took it upon himself to lead the committee rather than delegating to an independent expert or state secretary. Critically, unlike the 1980s investigation, it took the form of a parliamentary working group, excluding representatives of interest groups. According to Könberg (interview with author, 15 October 2015), this was entirely deliberate:

My aim was to have only … people in the room who were responsible and perhaps willing to negotiate … [In the 1980s committee] there were more people in the room looking at the people who were going to decide and take the difficult decisions than there were [decisionmakers] themselves.

Second, it included politicians with a high degree of policy expertise. In addition to Könberg, Anna Hedborg of the Social Democrats and Margit Gennser of the Moderates were very familiar with the policy details, having trained in economics (Gennser) and worked closely on pension issues as a state secretary (Hedborg) (Marier, 2008b, pp.525–526). Accordingly, the group environment has been described as having a ‘seminar-like’, rather than overtly political atmosphere (Marier, 2008b, p. 524).

Third, the party representatives were given license by their party leadership to negotiate openly with the other representatives. Moreover, according to an economist who assisted both the investigation and the working group (interview with author 14 October 2015), each had ‘enough power … and acceptance in their own party to go back to their party committees and say “now we’re thinking about this”.’ Committee members were largely shielded from having to justify their approach to grassroots party membership and the broader electorate, to the extent that the Left Party representative, who sought a wider public debate, was sidelined from deliberations (Loxbo, 2009, p. 249). For the Moderate Party representative, it was this approach that allowed the ‘the old Swedish tradition of consensus’ to be ‘reborn’ in the reform process (Margit Gennsser, personal communication with author). Könberg’s (quoted in Anderson & Immergut 2006, p.375) aim in constructing the working group this way was to ensure it would rest ‘on a foundation that corresponds to 85 to 95% of the Parliament and the Swedish population’. This meant that any blame or backlash would be distributed among the major parties, rather than focused on a governing coalition.

After the election, which returned the Social Democrats to government, a third committee was established, sharing some crossover in membership from the working group (including Könberg). The major task of this ‘implementation group’ was primarily to translate the general principles agreed on prior to the election into a more detailed design (Ministry of Health and Social Affairs 2009) and to maintain agreement among the five parties that had joined the recommendations of the 1991–1994 committee. This third committee been described by its members as having procedural legitimacy akin to an independent government body or a parallel cabinet for pension issues (Loxbo, 2009, pp. 256–257). According to a state secretary who worked for the committee (interview with author, 19 November 2015), the aim was to have agreement of ‘five to zero in every single paragraph’, and to prevent more difficult to resolve issues from spilling out into the public arena. As before, channels for debate and backlash were limited by the committee process.

While some Social Democratic party members and labour unions eventually tried to mobilise against elements of the new system, especially the investment component (Anna Hedborg, interview with author, 18 November 2015), party leadership remained wedded to the agreement reached in 1994 and supported the implementation group (Lundberg, 2005). Following implementation, the working group has been maintained, in an attempt to continue cooperation between the five parties to the 1994 agreement. While there have been several more severe clashes over policy (and threats to leave or attempts to disband the committee), the working group has remained a quasi-permanent feature of the Swedish pension landscape (Weaver & Willén, 2014). As will be discussed in the section on recalibration, the role of the committee has transitioned again – to the monitoring and evaluation of the system and as the main approval point for further change. At the same time, in taking on this shifted role, it has remained a key defender of the system and has helped safeguard its longer-term resilience.

Substantive instruments and supporting procedures

This section will focus on the role played by procedural tools in relation to three elements of the new system designed to maintain its ongoing function: the automatic balancing mechanism (ABM, also known as the ‘brake’), the flexible retirement age and the individual investment account ‘premium pension’. Collectively, these substantive instruments serve to individualise the pension system, creating a closer link between work-life and investment choices made by workers and the benefits they would eventually receive. This was an explicit aim of the pension committee, closely linked to the understanding of the issues with the old system that had emerged from the initial investigatory committee. For Könberg (interview with author), it was important to have ‘as much of the thing that you are forced to pay connected … to what you get for it’. For others on the committee, the question of ‘who should be paying for who?’, needed to be answered with ‘you should save for yourself’ (Hedborg, interview with author).

However, this was not only a matter of normative policy preference. The pension group’s goal was also to shift focus away from the ‘the dominant perspective of previous reform efforts’, namely the welfare of the elderly, towards political and financial ‘stability’ and ‘sustainability’ (Pensionsarbetsgruppen, 1994, p. 85). While in line with broader trends regarding the individualisation of social policy and investment, the shifting of risk to individuals also mitigates the extent to which government has to directly reduce or increase pension contributions and benefits, at least in ways that are easily identifiable to workers or beneficiaries. As such, the likelihood of backlash, blame and electorally punishment should theoretically be reduced in the longer term.

This has been enabled, with some degree of success, by a mix of substantive and procedural tools. The first example is the ABM, which may be considered to have both procedural and substantive elements (see Table 1). The need for an ABM is partly due to the fact that the ‘notionally’ defined contribution tier (the largest element of the system, funded by a 16% payroll tax) emulates a pre-funded, defined contribution scheme, but remains funded on a pay-as-you-go (PAYG) basis. Current tax revenue still pays for current benefits, but contributions are recorded in a ‘notional’ individual account. The balances of these accounts vary in line with estimated life expectancy of cohort, average wages and chosen retirement age. However, because the NDC component is PAYG, it is still vulnerable to demographic change and economic conditions. For example, if the workforce shrinks, average wages may grow faster than total contributions to the system (Sunden, 2006, pp.139–140). When such imbalances happen, the brake temporarily reduce the level of indexation applied to the NDC accounts (a substantive intervention) until balance is restored (Settergren, 2001). In theory, this is done without government involvement, the intention being to avoid conflict between short-term political incentives and the longer-term functioning of the system (a procedural mechanism). To further mitigate the temptation for government to act, the indexation changes happen over the course of several years and, as conditions improve, indexation increases beyond the usual rate, meaning that absolute losses are unlikely (Ministry of Health and Social Affairs 2009, p. 76).

The operation of the smaller individual private investment accounts and the flexible retirement age can also be seen as mitigating government accountability for the impact of changing socio-economic conditions on pension benefits. In the case of the flexible retirement age, the idea is that workers will choose to work longer as life expectancy increases. For the individual accounts (funded by a 2.5% payroll tax), the risk of changing economic conditions is transferred to the individual, who is responsible for choosing between a range of investments funds and risk profiles. Thus, increases or reduction in benefits are no longer uniform. Moreover, as the value of an investment fund is constantly in flux, changes in pension benefits as against wages are not necessarily even identifiable by beneficiaries.

However, while it is the case that government’s role has changed, its new responsibilities, particularly in administering the private investment component are, if anything, more challenging. For one, choice requires information – especially with hundreds of different investment funds available, and especially when imposed on a citizenry not accustomed to this degree of choice (Stefan Oscarsson, Senior Adviser for Pensions, Ministry of Social Affairs, interview with author, 19 November 2015). The government has therefore consistently sought to shape citizen behaviour through communication, particularly through yearly ‘orange envelopes’ that contain information about their accumulation of benefits and investments (Nyqvist, 2016).

The operation of this substantive instrument has required support from procedural interventions. In particular, the government has engaged in an ongoing process of reorganisation and collaboration between agencies. For example, in 1998, responsibility for pensions was divided between two agencies. While the non-investment components of the system would remain with the existing Social Insurance Agency (SIA), a new government agency was created to manage the investment component – Premiepensionsmyndigheten (PPM) or the premium pension authority. According to Nyqvist (2016, p. 73–75), who conducted extensive interviews within both agencies, the SIA was viewed as not fit for purpose, and it was believed that a new agency would be able sidestep perceived inefficiencies. A 2006 review of pension administration also states that a new organisational form was favoured in light of the need to manage and regulate a large number of independent investment funds (Pensionsadministrationsutredningen, 2006, p. 34), and to differentiate the social insurance and individual investment components of the system (Pensionsadministrationsutredningen, 2006, p. 39). The new agency then took a far more individualised approach to communication with citizens, emphasising the degree to which the new system required them to make active choices (Nyqvist, 2016, p. 98–99).

Despite these attempts to align substantive and procedural goals and instruments, and in doing so develop sufficient capacity, in practice automating and individualising the pension system has proved challenging. With regards to the ABM, it was first activated following the 2008/9 financial crisis. Because the ‘buffer’ fund used to payout benefits is invested, its value declined during the downturn, necessitating a reduction in the indexation of benefits to keep the system in balance (Schoyen & Stamati, 2013). However, with the 2010 election nearing, a change was made to reduce the ABM’s sensitivity to economic fluctuations, thus reducing the size of benefit cuts in the short term (Schoyen & Stamati, 2013). That the change was agreed to by the parliamentary working group, and that there are legitimate technical reasons for making such a change, does soften any judgment of political manipulation (Settergren, 2010). However, it remains the case that government intervention in an ‘automatic’ mechanism was partly driven by the proximity of the election and a desire to avoid blame, and possibly even claim credit with retirees (Weaver, 2016). In addition, the government further undercut the purpose of the ABM, which is to ensure a balance between inflows and outflows, by enacting a series of tax cuts for retirees to soften short-term losses.

Campaigns to have Swedes make active choices on retirement age and the premium pension have also been of only mixed success (Weaver & Willén, 2014; Stefan Oscarsson, Senior Adviser for Pensions, Ministry of Social Affairs, interview with author, 19 November 2015). A 2013 report from a parliamentary inquiry into the retirement age found that Sweden still had a ‘de facto normal retirement age of 65ʹ, and that the number of hours worked by older people were not increasing anything like at pace with increases in life expectancy (Pensionsåldersutredningen, 2013, p. 23). There remain strong social norms in around retirement at 65 in Sweden, supported by widespread beliefs that an early retirement will contribute to health and quality of life, and the fact that occupational pensions tend default to retirement at 65 (Pensionsåldersutredningen, 2013, p. 24, 32). Similarly, only a small percentage of working-age Swedes have made active investment choices for the premium pension. Instead, most choose to leave their money in the publicly managed ‘default’ investment, which is now by far the most popular investment choice for new entrants into the pension system (Sunden, 2006). This has been partly attributed to the sheer complexity of information that citizens are required to digest to understand their retirement position and the choices facing them (Pensionsadministrationsutredningen, 2006, p. 50–51).

Ongoing recalibration

While efforts to individualise and automate the pension system have narrowed one window of potential risk and blame, the complex mix of procedural and substantive tools created to do so has opened up new areas of risk and accountability for the Swedish government, and has required careful evaluation and ongoing recalibration to maintain core functions and public legitimacy. This process has largely been enabled by the consensus-based committee process. Following the experiences of the first use of the ABM, in 2015 the pension committee agreed to further ‘soften’ the operation of the brake, aiming to reduce the degree of fluctuation of benefits (Regeringskansliet 2015). Similarly, while Swedes have not made use of the flexible retirement age to the extent originally envisaged, after several years of investigation and discussion through the pension committee, the Social Democrat-led government, supported by the five parties of the working group, recently made the decision to increase the minimum retirement age in 2020 and again in 2023.

The premium pension has also seen several changes. First, the government introduced a simplified model of investor choice through the default fund (Barr, 2013, p. 37). Further changes to the premium pension are currently being made, following several scandals involving the investment funds (Palme, 2018). Much as in 1994, in 2017, the pension group came to an agreement on broad principles that would guide future reform. This agreement signalled a change of attitude towards the premium pension, based on the evidence of its operation thus far. In the agreement, the committee states that the pension system needs to function on the basis of ‘how people actually act’, rather than ‘how … they should act’ (Pensionsgruppen, 2017, p. 4). This involves offering enough to ‘satisfy … pension savers who are interested in making their own choice’, while recognising that the premium pension ‘is a social insurance scheme, for which the state must have clearer responsibilities’ (Utredningen om ett upphandlat fondtorg för premiepension, 2019, p. 26, 20).

Still, despite this shift in attitude towards the premium pension and retirement age, the government has not given up on its efforts to improve communication. For example, Nyqvist (2016) describes a year-long joint project between the SIA and PPM to better coordinate public communication. Following the 2006 review, which found information provision and coordination between agencies to be the primary issues in Swedish pension administration (Pensionsadministrationsutredningen, 2006, p. 49–50), a new agency was created to manage the entire pension system, the Pension Authority (Pensionsmyndigheten).1 The new agency has since developed a website that includes information regarding pension entitlements and investments from all sources, including occupational pensions that are not formally part of the public pension system (Boada-Penas, Settergren, Erland, & Poontavika, 2019, p. 18).

These efforts at recalibration appear to have had some success in building public legitimacy for this new model of social security. In a 2010 annual survey, only 31% of workers reported that they found the system ‘very’ or ‘quite easy’ to understand, with 41% finding it difficult. By 2018, these figures had improved to 49% and 21% (Boada-Penas et al., 2019, p. 23). In fact, levels of confidence in the pension system and the Pension Authority, self-reported understanding of pension issues and satisfaction with the information provided all increased during this period.

Discussion and conclusion

The contemporary welfare state offers various challenges for policymakers, such as increasingly complex policy arrangements and changing economic and demographic conditions. Through close examination of Swedish pension reform, this paper has made the case that the fortunes of social programmes, both at the initial stage of reform and in the long term, rely on well-designed interactions between substantive and procedural tools.

Three main findings emerge from the case analysis. First, it stresses the potential importance of investigatory and parliamentary committees or working groups (as also discussed by Stark & Yates, this issue). While the role of committees is well documented in public policy literature, the Swedish pension working group is unusual for a number of reasons. In particular, it has shown that flexibly construed procedural instruments may be of useful throughout the entire life of a programme. Given the fractious history of Swedish pension reform prior to the 1990s, the group has enabled lasting agreement that was unthinkable at the beginning of the reform process (Lundberg, 2005).

In contrast to established practice, the committee has excluded representatives of interest groups and continues to be comprised only of members of the Swedish parliament. In examining its continuing role, the paper goes further than previous analyses of the Swedish pension reform, which have generally focused on the role the pension working group played in agenda-setting and policy formulation (Marier, 2008a). From the perspective of more than 25 years post the initial agreement, the striking thing about the committee is not that a power over pension reform was devolved to a small group, but that responsibility for monitoring and negotiation remains largely in their hands.

Understandably, these continuing efforts to shield policy discussions from the realm of electoral politics, and hence shield the major parties from blame, have been met with criticism, with some prominent scholars (Weaver & Willén, 2014, p. 4) describing its function as ‘cartelistic’. While political conflict has largely been avoided, it is reasonable to ask whether the pursuit of stability has resulted in Swedish citizens having a greater degree of risk foisted upon them by a small group of elites looking to protect their own interests (Nyqvist, 2016).

Still, the second finding from the analysis suggests a less bleak outlook. The government, largely through the working group, has been committed to an ongoing process of learning and recalibration, especially with regard to the policy mix of procedural and policy tools and the validity of initial behavioural assumptions. Recalibration has been evident in various elements of the pension system, with major investigations into almost every element of the pension system. Among the changes have been the restructuring of government agencies, especially for the purposes of communication with workers and retirees, and the rethinking of behavioural assumptions around retirement age and investment choice. While many countries now have multi-tiered and notionally defined contribution pension systems, Sweden was among the first to move down this pathway (Anderson & Immergut 2006). Many of the behavioural assumptions on the part of the framers of the new system were therefore made under conditions of uncertainty, especially regarding the degree to which Swedish citizens would embrace choice and flexibility once given the option. As discussed, more recent investigations, working group agreements and policy changes have sought to recast the premium pension as a form of ‘social insurance’ – which the government must ensure is adequate – and frame ‘choice’ as effectively optional, rather than a central component of the system.

In this way, the need for learning and recalibration highlights a third finding, which is the difficulty inherent in automating and individualising social policy. Following the implementation of the new system, the Swedish government faced different forms of, rather than diminished, accountability. In the case of the automatic balancing mechanism (ABM), while it theoretically reduced the need to active intervention in times of economic or demographic, ‘automatic’ has been shown to be something of a misnomer. The authority maintained by government over its appropriate calibration (the procedural dimension) ensures that the risk of blame remains, as does the temptation to claim credit by softening its effects during economic crisis (the substantive dimension). While the basis for how adjustments are made is pre-decided, the issue of indexation does not then disappear from the political realm. With regards to the premium pension, forms of government accountability became even more complex. Where citizens become responsible for individualised decision-making, government becomes responsible for providing the right mix of information and incentives and for developing a sophisticated understanding of target behaviour. Where these choices are funnelled through private investment funds, government therefore becomes responsible for their operation, and risks backlash when there is scandal. The result has been a significant amount of time and energy invested in the procedural tool of government reorganisation, as a means of enabling more effective communication and regulation.

From these case-specific findings, this article addresses several topics of ongoing research in both the policy tools and welfare state literatures, although its contributions are necessarily caveated by the reliance on a single case. Recent work on policy tools has sought to understand how tools may support multiple goals and play primary as well as supporting roles (Bali, Howlett & Ramesh, this issue). This article has emphasised the degree to which procedural tools enable substantive goals, especially when policymakers face political and policy constraints. For example, the individualisation of the Swedish pension system would likely not have been possible without a process of formulation that excluded key interest groups, especially blue-collar unions (Loxbo, 2009). Moreover, as Mukherjee discusses elsewhere in this special issue, what might be thought of a single ‘tool’ is often a ‘compound’ of substantive and procedural elements. The automatic balancing mechanism from the Swedish case is one example, as is the use of government reorganisation as a means of enabling communication with citizens. In both cases, changes in the procedural dimension were intended to help maintain alignment between means and ends, as new information about target populations became available, as views about agency capacity emerged or as policy objectives and political interests shifted over time. This speaks to Capano and Howlett’s (2020) question of how decision-makers decide upon tools in mixes, and how they balance considerations such self-interest, uncertainty and ideas. The Swedish case highlights examples of anticipation – of citizen behaviour and the political interests of other parties and interest groups – and learning, as well as reaction, in relation to the use of procedural tools. For example, the basic structure of the Swedish pension committee was in large part a consequence of Minister Könberg’s frustrating experience with its previous iteration. However, its role has been recalibrated over time, as policymaking moved from agenda-setting (see also Halpin, this issue) to formulation and implementation to ongoing management.

As such, the article has also highlighted the importance of procedural tools for the broader and ongoing pursuit of legitimacy and what Capano and Woo (2017) describe as ‘robustness’, or the ability to maintain core function in the face of uncertainty or disturbance. The skilful design of procedural tools within a policy mix may be said to enable learning and policy innovation writ large (see Demircioglu and Vivona, this issue). However, this would seem particularly important for contemporary welfare states. Welfare state reform is often characterised by hybridisation or ‘tense layering’ (Kay, 2007), especially in the context of marketisation and contracting out of services (Morgan & Campbell, 2009). It is well-known that path dependence can lead to the packaging of tools drawn from competing ideational paradigms, but the role of procedural tools in helping to stitch together potentially ‘discordant’ policy mixes has thus far been underacknowledged (Capano & Howlett, 2020, p. 6). While the Swedish case is somewhat unusual in its degree of ongoing cross-party collaboration, the question of how to maintain the longer-term legitimacy, coherence and robustness of hybridised and layered policy arrangements is one faced by all governments in the realm of social policy. As such, investigation of the use of procedural tools and, critically, how policymakers come to make decisions about their role in mitigating political risk and achieving substantive policy goals, would seem to be a valuable direction for future research on welfare state reform.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Australian Government [Award Number: Australian Postgraduate Award] and The University of Melbourne [Award Number: Arts Fieldwork Grant].

Notes on contributors

Adam Hannah

Adam Hannah is a Lecturer in Public Policy at the University of Western Australia. His research focuses on the role of political ideas in reshaping health care and welfare systems, and he has recently published on this topic in journals such as Policy Sciences, Public Administration, Policy and Politics, and the Journal of European Social Policy. 

Notes

² See Howlett (2000).

¹ Although according the 2017 agreement, some responsibilities for managing the premium pension will be devolved to another new agency (Pensionsgruppen, 2017, p. 5).