Abstract
For several decades, the social investment (SI) state has been heralded as the saviour of the welfare state, while at the same time being criticised for being just another instance of neoliberal downsizing of the welfare state. Recently, efforts have been made to provide clearer conceptualisations of how to assess the existence and impact of SI. However, these attempts have hitherto mainly focused on the policy functions and instruments of the SI state. This article contributes to existing research by offering a novel analytical framework on the capacity needed by street-level organisations (SLOs) to implement the central policy functions of the SI state, and by elucidating how administrative reforms influence this capacity. The article applies the framework to the implementation in Denmark of Active Labour Market Policies (ALMPs) in local job centres. This case is considered an SI ‘flagship’ in terms of formal policies, while also having undergone multiple administrative reforms, which makes it highly illustrative for the central argument of the article – that the success or failure of an SI approach is not only determined by politics and formal policies. The empirical analysis reveals how the capacity to implement SI policies has been enhanced by administrative reforms; this has been done by giving job centres more room for discretion and enhancing their ability to make long-term investments and to promote integrated service provision across different service areas. However, at the same time the local job centres remain closely monitored and controlled through an external accountability performance measurement system.
Keywords
Introduction
Throughout recent decades, the idea of the social investment (SI) state has been viewed as a blueprint for a new social model, which is better equipped than both the traditional Keynesian-Beveridge welfare state model and its neoliberal counterpart to tackle the (demographic, social etc.) challenges of contemporary society (Esping-Andersen et al., 2002; Hemerijck, 2018; Morel et al., 2012). As a policy paradigm, the SI state thus treads a path between the Keynesian and the neoliberal model, endorsing neither the passive social insurance schemes of the former nor the curtailment of benefits and rights of the latter. Instead, the SI state is framed as being concerned with furthering citizen-capabilities and improving employability by ‘raising and maintaining human capital “stock” throughout the life course and easing the “flow” of contemporary labor market transitions’ (Hemerijck, 2018: 812).
Literature on the SI state can be divided into two distinct waves. The first wave mainly centred around whether a turn towards an SI state has taken place (Hemerijck, 2013) and whether this could be viewed as a solution to the challenges facing modern welfare states (Barbier, 2012; Hemerijck, 2017; Nolan, 2013). Sometimes these debates have been fuelled by a lack of conceptual clarity as to why SI policies – in the eyes of both proponents and critics – have come to represent a wide variety of both specific policy instruments and normative propositions. In recent years, a second wave of scholarly attention has thus moved away from discussions of the principles of SI towards efforts to provide more precise conceptualisation and operationalisation of how to measure its existence and impact (Andersson, 2018; Heitzmann and Matzinger, 2021; Hemerijck, 2018; Kvist, 2014; Plavgo and Hemerijck, 2021). This second wave of SI literature has mostly focused on what can be termed the politics of the SI state – i.e. electoral support, public attitudes etc. (e.g. Busemeyer et al., 2018; Marx and Nguyen, 2018) – or the formal policies of the SI state – i.e. the formal content of the policies underpinning SI (e.g. Kvist, 2014; Morel et al., 2009).
The current article is situated firmly within the debates of this second wave of SI literature, but it seeks to move the discussion forward by shifting the focus away from the politics and formal policies of SI towards the operational conditions of the SI state. The article thus puts forward the argument that the success or failure of the SI approach is not only determined by politics and formal policies. It is also dependent on ‘prevailing governance structures and institutional actors’ (Hemerijck, 2018: 819) and their influence on the ‘capacity of the SI state’. Drawing on the literature on policy capacity, we define policy capacity as the ‘set of skills and resources – or competences and capabilities – necessary to perform policy functions’ (Wu et al., 2017: 3). Unlike most of the policy capacity literature, we concentrate on the capacity of street-level organisations (SLOs) (Brodkin, 2011a) to implement policies in practice, rather than the capacity of governments or parliaments to enact SI policies. Consequently, we also shift the focus from the study of the formal policy reforms of SI (e.g. investments in life-long learning, expansion of childcare, etc.) to the administrative reforms determining the conditions under which policies and services are administered and delivered.
The article thus aims to move existing research on SI forward in two ways: 1) by offering an analytical framework that specifies the central dimensions of capacity necessary for implementing central policy functions of the social investment state and 2) by elucidating how administrative reforms influence this capacity of the social investment state.
Compared to the existing conceptualisations of SI, we redirect the focus away from formal policy reforms to administrative reforms, and from policymaking to implementation. First, we review how the existing literature conceptualises the SI state, through the lens of the policy capacity concept. This allows us to tease out three main theoretical components of the capacity to implement SI policies. Inspired by the burgeoning literature on the consequences of administrative reforms for SLOs (e.g. Brodkin, 2011b), we next take up the specific empirical case of implementation of active labour market policies (ALMPs) in Denmark, to elucidate empirically the link between administrative reforms and the capacity for implementing SI policies. Policywise, Denmark has for several years been known (and even heralded) as one of the clearest examples of a move from a Keynesian welfare state model towards an SI state (Barbier and Ludwig-Mayerhofer, 2004). This is especially vivid in the case of ALMPs: Denmark for decades has been one of the countries investing most heavily in activation policies (Breidahl, 2017), while also providing a model that stands out ‘from the rest in its emphasis on a long-term strategy and on human capital development’ (Lødemel and Trickey, 2001: 278). At the same time, Danish ALMPs have undergone several significant administrative reforms in the last two decades (Andersen and Breidahl, 2021; Larsen, 2013). The Danish ALMPs therefore provide an illustrative case for our theoretical framework; allowing us to empirically study a country and policy sector deeply aligned with the SI policy paradigm, as well as the influence of frequent and far-reaching administrative reforms on the capacity to implement SI policies.
The next section outlines the analytical and theoretical framework, followed by a description of the research design and methodological background. Then follows an empirical analysis of how administrative reforms have influenced the capacity for implementing SI. The last section summarises the central findings and discusses their broader theoretical and empirical implications.
Dimensions and determinants of the capacity to implement SI policies
In recent years, important contributions have been made to refine the conceptualisation and operationalisation of SI as a policy paradigm – i.e. to define the shared normative and cognitive framework that informs SI policies (Andersson, 2018; Heitzmann and Matzinger, 2021; Hemerijck, 2017, 2018; Kvist, 2014; Plavgo and Hemerijck, 2021). One of the most influential contributions is Anton Hemerijck's definition of three policy functions of the SI paradigm: (1) raising and maintaining human capital ‘stock’ throughout the life course; (2) easing the ‘flow’ of contemporary labour market transitions; and (3) upkeeping strong minimum-income universal safety nets as social protection and economic stabilisation ‘buffers’ (Hemerijck, 2018: 812).
While the ‘buffer’ function aligns well with traditional policy instruments of income-replacing benefits, the two other functions are presented as necessitating novel policy instruments that break with both the Keynesian-Beveridge and the neoliberal policy paradigm. Hemerijck highlights how the function of raising and maintaining stock requires ‘capacitating social services’, while the prime examples of a policy instrument easing the flow of labour market transitions is that of ‘active labour market policies and job matching’ (Hemerijck, 2018: 817). Similarly, other recent conceptualisations of SI have highlighted related policy instruments concerning human capital and employability development (Heitzmann and Matzinger, 2021), proactive and integrated social services (Andersson, 2018) and early childhood care and education (Plavgo and Hemerijck, 2021). These conceptualisations have all contributed insights into both the policy functions and the more concrete policy instruments of the SI paradigm. This literature has, however, been far less concerned with defining the capacity – i.e. skills, resources and capabilities (Wu et al., 2017) – necessary to perform these policy functions.
To enrich this part of the discussion, we bring in the concept of ‘policy capacity’ (Mukherjee et al., 2021; Newman et al., 2017; Peters, 2015; Wu et al., 2017). Although the literature on policy capacity mainly focuses on the capacities for policymaking at the central level of government, a few more recent studies have broadened the scope and application of the concept. Hence, Wu et al. (2017) define policy capacity in terms of three sets of skills/competences (analytical, operational and political) and across three levels (individual, organisational and systemic), thereby providing a framework that can be applied to all stages of policymaking – from agenda-setting to implementation and evaluation. Consequently, the framework allows for more contextualised considerations about whether and how the importance of certain types of capacity varies according to the specific level of policymaking and the specific policy function at hand (Howlett and Ramesh, 2016; Mukherjee et al., 2021).
In this paper, we direct attention toward the analytical and operational capacity. Wu et al. define analytical capacity as capacities that ensure that ‘policy actions are technically sound’, while operational capacity is defined as ‘capacities [that] allow resources to be aligned with policy actions so that they can be implemented in practice’ (Wu et al., 2017: 5). Both the capacity to analyse and judge the soundness of actions and the resources available to implement such actions are thus paramount in enabling implementing agents (such as SLOs) to execute SI policies.
Regarding the question of ‘capacity for what?’, we furthermore need to define in more detail ‘what skills and competencies are required to be able to address, or at least mitigate, the given problem at hand’ (Bali and Ramesh, 2018: 334), i.e. to implement an SI approach. We hereby exclusively focus on those dimensions of analytical and operational capacity that specifically support the main policy functions of SI, rather than dimensions related to state capacity or good governance more generally (Fukuyama, 2013), such as the existence of professional and competent staff (Decarolis et al., 2020, Nistotskaya and Cingolani,2016). These more general dimensions of capacity are of course equally – if not more – important for the functioning of any state. However, the aim of this paper is to conceptualise the dimensions of capacity uniquely related to the implementation of the main functions of an SI state. 1
Three dimensions of capacity for implementing SI policies
By applying the lens of policy capacity to the existing literature on the formal policy instruments and functions of SI, we highlight three main dimensions of capacity for implementing SI policies: significant discretion; resources for making long-term investments; and cross-organisational integration and collaboration.
Capacity for discretion
Existing conceptualisations of the main policy functions and instruments of the SI state all highlight the normative commitment to enhancing citizens’ freedom to act (Hemerijck, 2018: 823). Consequently, SI polices are said to generally favour a capacitating and highly interventionist approach to at-risk citizens, to build up their human capital and/or avoid human capital deterioration. This is often defined in contrast to the more regulatory measures favoured by both the traditional Keynesian-Beveridge and the neoliberal welfare state model. However, it is one thing to design and adopt policies with the explicit aim of ‘human capital development’ (Heitzmann and Matzinger, 2021), such as early-childhood education or vocational training for the unemployed. It is quite another thing to have sufficient capacity to implement such policies in a way that also enhances the human capital of the citizen. The schism between the policy and practice of interventionist human-processing measures was already noted by Bo Rothstein (Rothstein, 1998) in his seminal book on the moral and political logic of the welfare state. According to Rothstein, there will often be a trade-off between the interventionist capacity and the steering capacity of the welfare state: The more interventionist the goal of the policy, the less the state is able to centrally stipulate and control the implementation of the policy. For a policy to succeed in changing the behaviour, motivation, competences etc. of the citizen, it needs to be tailored to the specific circumstances of that individual citizen. In other words, the success of interventionist policies is much more context-dependent than more regulatory policies such as, for example, the provision of a flat-rate pension scheme (Rothstein, 1998). To achieve the interventionist goals of the SI state, the implementing organisations need significant discretionary room for designing, prioritising, and delivering services in accordance with the needs of the individual citizen. Decentralisation of resources and decision-making to local SLOs is thus a critical operational component of the SI state (Hemerijck, 2018).
Capacity for long-term investment
The second critical feature, which is a consistent theme in the SI literature (Andersson, 2018; Hemerijck, 2017; Kvist, 2014; Morel et al., 2012), is the normative commitment of welfare policies to a life course perspective. This perspective is twofold, as it refers to the life course both of the individual and of the whole population. Regarding the former, policies are intended to target citizens at an early stage (either in their life as a whole or in their contact with the welfare state), to prevent social marginalisation and decrease dependency in the longer run (Kvist, 2014). Regarding the population as a whole, policies are geared towards inter-generational redistribution, seeking to enhance the level of employment within the working-age population so that the workforce can finance universal care for children and the elderly – referred to as the carrying capacity of the welfare state (Hemerijck, 2018). In both cases, the formal policies of an SI state have a distinctively long-term perspective geared towards proactive investment to reap future rewards.
Viewed through the lens of policy capacity, the future-oriented temporal perspective of the formal policies (Andersson, 2018) also needs to be backed by capacity for long-term investment in local SLOs. This dimension of capacity is thus closely related to fiscal resources, as SLOs will be unable to implement a proactive and long-term approach if such resources are tied to, for example, pre-specified target groups or a single budget year. Therefore, the capacity for long-term investments is first of all contingent on SLOs having the resources and the formal backing by law to build up fiscal deficits in the short run (e.g. by allocating resources from future budget years to the present), to save resources in the longer term. Secondly, the ability to make long-term investments is not only a matter of economic resources and formal regulations (i.e. operational capacity). It is also contingent on SLOs having the analytical capacity (Wu et al. 2017) to calculate expected returns on investments, as well as the resources to monitor and assess whether these long-term expectations are met. In cases where SLOs are providing public services, this means that governments also monitor and asses their performance according to such long-term goals, rather than, for example, measuring SLO performance on their short-term output-production. The point is that performance measurement creates incentives for SLOs to act in accordance with what is being measured: i.e. long term actions need long-term measurements.
Capacity for service integration and collaboration
The third dimension of our framework is related to the aim of policy coherence. Hemerijck argues that the simultaneous commitment of the SI state to the policy functions of stock, flow and buffer enables these different policies to reinforce each other and create a ‘life-course multiplier effect’ (Hemerijck, 2018: 817). By this, he means a virtuous circle where an investment in, for example, human capital development (stock) also eases labour market transitions (flow) and increases the carrying capacity of the welfare state (buffers). From an SI perspective, this ideal makes the ‘coherence of policies important’ (Andersson, 2018); for this reason, he argues that different policy areas – such as education, health and employment policies – should not be understood or designed in isolation from each other. Translated to the level of implementation, such formal policy commitments hinge on the capacity of the SLOs implementing the policies and delivering the services to integrate other policy areas and/or collaborate with each other. This capacity can range from organisational integration of two formerly separated service areas under the same management, to more informal agreements on collaboration between relevant organisations.
Furthermore, the capacity for service integration and collaboration is especially important in service areas targeting citizens with a variety of problems (e.g. poor mental or physical health, lack of education, social marginalisation etc.), a situation which necessitates inter-organisational collaboration (Andersson, 2018). Therefore, it is detrimental to the human capital enhancing goal of SI if the relevant services are organised in a highly fragmented or siloed way. 2
The relation between administrative reforms and capacity at the level of implementation
Continual waves of administrative public sector reforms have swept over the Western world in the last 30–40 years (Christensen and Lægreid, 2010; Reiter and Klenk, 2019), with substantial implications for the SLOs responsible for providing welfare services (Brodkin, 2011b; Considine et al., 2015; Fording et al., 2011; Fossestøl et al., 2015).
Since the seminal work of Michael Lipsky on street-level bureaucrats (Lipsky, 2010), two central features of policy implementation have come to the fore: 1) there will always be significant discretionary space for the implementing agent; and 2) the use of this discretionary space is always determined by both formal policies and operational conditions. Evelyn Brodkin has put this rather succinctly, using the analogy of a game in which the implementing agents and agencies are the players, with ‘formal policy that establishes the playing field and governance that sets the rules of the game’ (Brodkin, 2011b: 258). Basically, this means that formal policy reforms establish the boundaries and overall goal that SLOs must work within and toward – for example by defining the target group and the central policy instruments. Within this playing field, the SLOs – and the employees in these organisations – have ample discretionary room for choosing the means and path to reach this goal. However, this discretionary space is not unlimited, as it is bound by the rules set by administrative reforms. These rules do not determine the actions of SLOs, but they do insert a choice-calculus that renders certain actions more likely than others (Brodkin, 2011b). Administrative reforms do this in two broad ways. Either they stipulate directly how services are organised – for example by establishing or removing organisational boundaries or through detailed regulation of organisational processes – or they incentivise certain ways of organising and delivering services – for example by economically rewarding the use of certain types of programmes or by monitoring performance by reference to specific output or outcome goals.
It is a consistent finding within research on administrative reforms that rule-setting through incentives – rather than direct regulation – has become increasingly prevalent, as traditional hierarchical steering has been supplemented by new forms of more ‘hands-off’ governance (Hood and Peters, 2004; Osborne, 2010; Reiter and Klenk, 2019). The continual waves of administrative reforms have thus introduced a plethora of new governance instruments and arrangements that rely less on direct command and control and more on directing action through norms, nudging, economic rewards etc. However, these instruments and arrangements are often additional to rather than superseding existing governance arrangements, leading to an increasingly hybrid nature of public sector governance (Christensen and Lægreid, 2010; Polzer et al., 2017). Consequently, because the various instruments and arrangements are often guided by logics grounded in different governance paradigms, such as New Public Management and New Public Governance, they may impose divergent demands on SLOs (Fossestøl et al., 2015; Trappenburg et al., 2022).
To return to the analogy proposed by Brodkin, the rules of the game of the SLOs may therefore be internally inconsistent. Specific administrative reforms – and the governance arrangements promoted by them – may thus simultaneously enhance and impede the capacity for implementing SI policies. Whether and how this is the case is the empirical question which we will now turn to.
Research design and methods
In the introduction we outlined our central motive for focusing on ALMPs and the Danish case: it has for a long time been emphasised as a strong SI state ‘flagship’, due to the many policies enacted since the 1990s to simultaneously raise the stock of the unemployed, ease their flow into the labour market and maintain a relatively high level of income support as a buffer (Hemerijck, 2018). Simultaneously with these formal policy reforms, a whole plethora of administrative reforms have targeted the SLOs responsible for implementing the ALMPs. Thus, the Danish case is highly illustrative of the multiple ways in which administrative reforms can influence the capacity for implementing SI policies.
Existing studies of the administrative reforms of the Danish employment services in recent decades highlight two distinct waves of reforms. The first wave culminated in the creation of municipal job centres in 2007 and generally sought to enhance central government control over the implementation of ALMPs (Larsen, 2013). As part of a major reform of the Danish public sector, ‘job centres’ were established in the Danish municipalities in 2007. This created a new structure for employment services, with ‘one-stop shops’ performing services (activation schemes, visitation, job guidance and job plans) for all unemployed people. Prior to this reform, the employment services were divided between the state and municipalities. The latter provided the employment services for unemployed people receiving social assistance (people without unemployment insurance), while the former serviced those receiving unemployment benefits (those with unemployment insurance).
From their inception, the job centres were tightly controlled, with a host of instruments regulating and monitoring their organisational structure, their economic priorities and their performance (Larsen, 2013). In the following years, the centralised steering system came under increasing scrutiny and criticism for keeping job centre activities so tightly regulated that they had no leeway to perform their central function of helping the unemployed individuals into work or education. This critique was especially pronounced in relation to the services targeted at the most vulnerable unemployed, with health or social problems. Following this first wave of reforms, a second wave of administrative reforms has taken place since the beginning of the 2010s. Many of these have had the intended goal of loosening the grip on the job centres and giving them more autonomy to design and deliver services as they see fit (Andersen 2021; Caswell and Larsen 2022).
Empirically we draw on a systematic document analysis as well as interviews with managers and administrative staff of four local municipal job centres in Denmark. The document analysis is based on a structured examination of administrative reforms of the employment services since the early 2000s, allowing us to identify how administrative conditions have been modified over time. We draw on an archive of documents consisting of the text of every administrative reform targeted at the Danish employment services between 2001 and 2021, as well as supplementary guidelines on how to implement these administrative reforms. The policy documents can all be found on the official website of the Ministry of Employment (bm.dk), while the guidelines are drawn from the website of the National Labour Market Authority (STAR.dk), responsible for monitoring and guiding the implementation of policies in the job centres.
The four municipal job centres were selected based on their participation in a larger four-year (2020–2024) research project (XXXX, anonymised). Managers and administrative staff were interviewed, allowing us to uncover routine patterns prevalent across the four job centres. Consequently, the job centres were not selected based on notions of most/least similar design, although the selection allows for some variance (in terms of size, geography, local unemployment rates etc.).
The semi-structured group or single interviews with job centre managers at different levels, as well as with internal administrative staff responsible for analysing and monitoring job centre data, have all been conducted by the authors. The managers were chosen according to two main criteria: 1) they had autonomy to influence the organisation of their given unit(s) – i.e. they were not purely line-managers 2) the level and responsibility of the managers were comparable across the four job centres. We were able to interview the top manager of the whole job centre (covering every type of benefit recipient), the manager responsible for job centre administration (budgets etc.) and middle managers responsible for specific target groups. A total of 22 managers and 8 administrative staff-members were interviewed through 30 single-person interviews and 4 focus group interviews across the 4 municipalities. All interviews were conducted in 2021 and were afterwards transcribed and systematically coded in NVivo.
How administrative reforms influence the capacity for implementing SI policies
The following analysis applies the three dimensions of the capacity for implementing SI policies, to elucidate how the administrative reforms of the last decade have created a growing discrepancy: job centres are both incentivised and disincentivised to apply an SI approach to social and employment services. We analyse how this discrepancy, simultaneously enhancing and limiting the capacity for SI, has come about, as well as its consequences for the implementation of an SI approach within the job centres.
Enhancing the capacity for implementing SI policies
The capacity for discretion
As the delivery of employment services was decentralised to municipal job centres in 2007, the Ministry of Employment sought to enhance accountability, that is, to make sure that services were implemented in accordance with the intentions of the elected politicians. Consequently, job centre autonomy to design and prioritise services was initially kept rather limited, through a web of detailed regulations concerning the structure, resources, and content of employment service delivery. These initial administrative conditions turned out to be highly suited to implementing a workfare approach, concerned less with the enhancement and maintenance of human capital and more with incentivising the unemployed to leave the employment system as quickly as possible (Larsen, 2013). Consequently, these highly detailed procedural regulations directly impeded the implementation of an SI approach, which requires precisely the opposite operational conditions: ‘ample backing and discretionary policy space to regional and local authorities’ (Hemerijck, 2018).
Although some initiatives to increase organisational and fiscal autonomy were enacted during the first half of the 2010s (cf. the following sections), the detailed procedural regulations on the exact frequency and content of the available activation measures and client-caseworker meetings were not relaxed until 2018 as part of the so-called ‘de-bureaucratisation reform’. This abolished several of the formal regulations on how and when client-caseworker meetings and activation schemes had to take place after the first six months of the benefit period. The officially stated intention of this reform was to enhance ‘the freedom of municipalities to adapt the services to the individual unemployed’ (Beskæftigelsesministeriet, 2018). By relaxing centrally defined requirements on what job centres must do, the intention was to improve the centres’ ability to deliver more holistic and individualised services. These intentions clearly mirror the arguments of proponents of SI in favour of enabling a ‘high level of professional discretion attributed to decentralized levels of public administration for tailoring person-centred service provision’ (Hemerijck, 2018). The argument is that the choice of which services to invest in and deliver is highly context-dependent, as citizens need different forms of support to build up and/or maintain their human capital. According to the ideal of SI, the implementing agents (i.e. local job centres) should therefore not be bogged down by highly detailed and centrally defined procedural regulations, but instead they need ample scope for prioritising and tailoring services at the local level.
The capacity for long-term investment in services
According to the SI ideal, it is an important operational prerequisite that the implementing organisations have sufficient fiscal resources to invest in services in the present, in order to reap the rewards in the long run. However, such investments are difficult for SLOs to make if their budgets are tied to the delivery of pre-specified services to pre-specified target groups within the current budget year.
Up until 2015 this was very much the case for the Danish job centres. Although the local municipalities run these centres, the costs of benefit payments and service provision are not solely paid by the municipalities themselves. Instead, the state reimburses a certain share of the costs, to shelter the individual municipal economy from a sudden rise in the level of unemployment. However, the state does not reimburse the full costs, as it seeks to maintain incentives for the municipal job centre to do everything in its power to get the unemployed into work. Before 2015, job centres were incentivised to keep benefit recipients active, as the reimbursement rates (the share of the benefit costs covered by the state) were substantially higher for benefit recipients participating in job-focused activation schemes (such as job training). Municipalities were thus incentivised to invest their resources in producing certain activation schemes, while the actual effects – for both client, job centre and society – of these investments were of less importance. This changed markedly with the reimbursement reform of 2015, which decoupled reimbursement rates from specific activation schemes by, instead, gradually reducing them from 80 % to 20 % according to the length of the individual's time in the unemployment system. The formal intention of this reform was to ‘strengthen a municipal focus on results’ by ‘securing a greater economic incentive by the individual municipality to prevent long-term unemployment’ (Økonomi og Indenrigsministeriet, 2015: 1).
Several of the managers interviewed in our empirical material acknowledge and elaborate on how the reimbursement reform in 2015 encouraged job centres to focus on longer-term results, as resources were no longer tied to specific activation schemes. ‘It gives us the opportunity to look at the whole of the municipal economy (…) Meaning that we are able to do that which we believe has the best effect on the overall benefit-costs’ (Job Centre manager).
The manager argues that the reimbursement reform enables job centres to shift their priorities away from cutting their own costs in the short run – by continually producing the standardised activation schemes with the highest reimbursement rates – towards investing heavily and proactively in the services with the greatest potential for cutting overall costs for the municipality in the long run.
These new avenues for municipal resource allocation have also been accompanied by efforts to support the analytical capacity of the municipalities, enabling them to monitor whether their investments also lead to the expected future gains. The Ministry of Employment has thus developed several business cases for municipal investments in employment services, along with a support tool that each job centre can use to measure the effect of their investments. As described in a letter to each municipal job centre from the then minister of employment: The goal is that you – as a municipality – can use the support tool to calculate and develop investment cases for citizens who are outside the labour market, with the intention of judging whether a case is economically feasible in the short or long term (Beskæftigelsesministeriet, 2019).
The importance of having the capacity to assess and monitor potential and actual return on investment is also highlighted by the members of the municipalities’ administrative staff whom we interviewed: When we invest in certain programmes or target groups, these investments are first to be presented to us in the management information team before being presented to the local politicians. Because then we can make sure that the investments are made in a way where we can measure them [their results]. So it is now inscribed into these investment projects – when they are approved [by the local politicians] – how we are going to measure their results (Municipal data analyst).
The data analyst explains that the job centre has become acutely aware of the need for close monitoring of the long-term outcome of local investments; this is why the municipality is investing substantial resources in enhancing the analytical capacity to monitor this outcome. We identify similar analytical upscaling across all four municipalities in order to enhance their capacity for long-term investments.
The capacity for cross-service integration and cooperation
In the summer of 2022, the Danish parliament passed a new law on ‘Holistic services for citizens with complex problems’, which is in many ways the culmination of a larger trend in administrative reforms towards promoting greater policy coherence (Andersen and Breidahl, 2021). With this law, citizens receiving cash or sickness benefits and who – beside active employment services – also benefit from social services (such as treatment for drug abuse or socio-pedagogical support), are granted the right to a holistic plan intended to: ‘…secure coherence between the services the citizen receives and that services are provided in an order which is meaningful in relation to the individual citizen's whole-life situation’ (Social- og ældreministeriet, 2022).
The law clearly mirrors the need of SI for complementarity and coherence between a broad spectrum of policies targeting the welfare state's stock, flow and buffer (Hemerijck 2018). Or to put it differently, there is a need for policy coherence (Andersson, 2018). In the case of the Danish employment services, the goal of policy coherence has figured prominently on the political agenda for the last decade. Early efforts were already made in around 2012–2014 to establish the necessary operational conditions for implementing more integrated and coordinated employment services.
With the 2012 reform of disability pensions and flexible employment, municipalities were obliged to establish a rehabilitation team consisting of representatives from municipal employment services, municipal social services, municipal health services and regional health services. The law stipulates that the purpose of the rehabilitation team is: … to secure interprofessional coordination and integrated services across administrations and authorities with a focus on employment and education so that the individual citizen – to the greatest extent possible – enters the labour market (Lov om organisering og understøttelse af beskæftigelsesindsatsen, 2014).
The requirement to create a rehabilitation team can be seen as a decisive moment, as it marked the first break with the single-purpose logic of the job centre organisation established in 2007, at which time it was stipulated by law that job centres had to be organisationally separate from traditional social and health services in the municipalities. The intention behind this organisational split was that the job centre should only focus on helping citizens to find a job, instead of tackling the social or health problems facing the most vulnerable unemployed. The job centres were thus actively shaped to discourage service integration and coordination between employment services and other policy areas. The introduction of rehabilitation teams with the 2012 reform marked a first clear break with this idea of service separation. However, the most important step towards enhancing job centre capacity for policy coherence came with the 2014 Employment Reform, which abolished the regulations upholding an organisational split between the job centre and the rest of the municipal welfare services. The municipalities were then suddenly free to decide whether to integrate social, health or other services into the same organisational structure as the job centre.
One of the four job centres in our study has recently (2021) used this freedom to integrate parts of social services (substance abuse treatment and social-pedagogical support) into the job centre organisation. Even though the individual social and employment services are still governed by different legal regulations and ministries, the job centre manager from the given municipality argues that this has greatly enhanced the capacity for cross-service integration – and thus the service coherence necessary for implementing an SI approach – simply because the providers of social and employment services are now physically situated nearer to each other under the same management: One of the most important tools for creating greater coherence [across services] is to inscribe service integration all the way out at the front desk of the individual caseworkers. By actually placing them together – close together – with colleagues [from the other service area]. Now the ones providing social-pedagogical support and the traditional job centre caseworkers are under the same manager (Job centre manager).
Limiting the capacity to implement SI policies
As outlined above, the administrative reforms enacted in recent decades have in several ways enhanced the freedom of the municipal job centres to: 1) prioritise and tailor the content and frequency of client-caseworker meetings and activation offers; 2) invest more heavily in services in order to gain long-term rewards; and 3) use local autonomy to restructure their inter- and intra-organisational boundaries. However, while such ample discretionary room for the SLOs is a necessary precondition for implementing an SI approach, it can make it more difficult to achieve the goal of holding job centres accountable, which has been a central concern of the Ministry of Employment since the creation of the job centres in 2007. As we will demonstrate below, the Ministry has dealt with this issue of accountability by upholding – and in some instances expanding – the detailed performance management system created in 2007, alongside the relaxation of the procedural, fiscal and organisational instruments regulating the job centres, thereby enabling continuing close monitoring of job centre performance.
On the face of it, governing through performance measurement should be ideally suited to an SI approach, since – compared to more regulatory tools – it leaves ample discretionary space for the job centres, as long as they deliver the expected results in the end. It is, however, a well-established finding within public administration research that PM systems tend to produce many of the same detrimental effects typically associated with detailed procedural regulations (Pollitt, 2013). This is even more pronounced when services are complex – and thus not easily measurable – and when the goals are set by external stakeholders (usually the government) without expertise in the service area. Many studies have documented synecdoche-effects (Pollitt, 2013), where easily measurable and manageable output goals come to define what counts as the most important performance goals of a whole organisational field. When the achievement of such easily measurable output goals is furthermore linked to high-powered incentives – in the form of punishment or rewards – governing through PM can often shift the focus of SLOs from solving the problems facing their clients to producing standardised services and easily measurable goals (Brodkin, 2011b).
These mechanisms are reflected in the Danish employment system, where a PM system primarily focused on detailed measurement of job centre outputs (i.e. the number and frequency of client-caseworker meetings and activation schemes) has been in place since 2007. Over the years, administrative reforms have made various amendments to the system, but the main element remains a biannual benchmark of all the municipal job centres according to one specific outcome goal: the number of actual benefit recipients compared to the number of expected benefit recipients (calculated using a number of socio-economic indicators for the given municipality). If job centres fall beneath the given threshold for two consecutive benchmarks, they are punished by being placed under increased ministerial surveillance for a period of nine months. If they fail to improve their results within these nine months, they are then placed under ministerial administration until their results are improved – thus removing the newly granted fiscal, structural and procedural autonomy necessary for implementing an SI approach. However, there is an important caveat to this system, as low-performing job centres can retain their autonomy if they live up to output goals concerning the share of clients partaking in frequent meetings with their caseworker as well as participating in employment-related activation schemes.
In the following, we elucidate how the PM system – and the high-powered incentives coupled with it – impedes the job centres’ capacity for long-term investment in collaborative and tailored services, by incentivising them to prioritise resources and actions leading toward the stipulated outcome and output goals.
Incentives for short-term rather than long-term investments
The outcome goal of reducing the number of benefit recipients is aligned nicely with the overall goal of SI, to enhance the carrying capacity of the welfare state (Hemerijck, 2017; Morel et al., 2012). However, if we look at the timeframe for expected success, a discrepancy between the outcome goal measured by the PM system and the outcome goals of SI comes to the fore.
If the job centres are to avoid being placed under increased surveillance and eventually administration by the ministry, they must not underperform (i.e. have a higher number of benefit recipients than expected) for two consecutive benchmarks. This gives the job centres a one-year timeframe (as benchmarks are published biannually) to increase the number of benefit recipients leaving the system. Viewed from the perspective of job-ready citizens – for whom unemployment is typically only temporary – a one-year timeframe seems well suited to the goal of easing and shortening the flow between unemployment and work. However, as the goal of SI is also to build up the human capital of the whole stock of unemployed – including the most vulnerable unemployed who are also affected by other (physical, mental, or social) problems – a one-year timeframe seems less sufficient. In a choice between investing in drug treatment or socio-pedagogical therapy for the most vulnerable unemployed, or helping the most job-ready with their job search, the job centres are incentivised to consider the option with the greatest likelihood of leading to a successful job outcome within one year – that is, potentially leading to creaming and parking (Lipsky, 2010). Although it is not always a question of either-or, job centres are hereby incentivised to constantly keep an eye on the short-term outcome goal, even when investing in services for the most vulnerable unemployed.
The need for keeping a close eye on the benchmark is acknowledged by all the interviewees across the four municipal job centres. A data analyst explains it in the following way:
‘We are keeping an eye on it, because we want to make sure that we don’t get down into a really ugly spot’ [at the bottom of the benchmark] (Municipal data analyst].
This explanation is consistent with how the other interviewees approach these benchmarks.
Another municipal data analyst explains that it is also necessary to constantly keep track of performance in-between the biannual benchmarks, because these ‘always [have] a minimum of three months’ time lag so it is not very timely management-information’. More frequent and timely performance measurements are thus needed if management is to (pro)actively use these benchmarks to steer the job centre organisation away from the threat of sanctions: ‘One of the strengths of our own management-information system is that we can get a picture of the here and now. It is not the numbers of last month. For me it is totally uninteresting to look at those old numbers today’ (Manager).
Across the four case municipalities, management is thus presented with monthly or even weekly reports on the outflow of benefit recipients across the different target groups. The downside of such constant performance monitoring is that the timeframe for success is further shortened, as the relationship between actual and expected number of benefit recipients is expected to balance out on a monthly instead of a yearly basis. This further disincentivises job centres from making long-term investments in human capital-enhancing measures targeted at the most vulnerable unemployed.
Incentives for enhancing internal productivity rather than cross-service cooperation
Although the timeframe of the outcome goal discourages long-term social investments, the importance of this outcome-measurement system should not be overstated. The managers interviewed all perceive this outcome as almost completely dependent on factors outside the direct control of the job centre – such as the actions of individual benefit recipients or the availability of jobs in the greater economy. No matter the strategy and investments in services made by the job centre, success on this measurement of outcome will therefore remain inherently uncertain.
None the less, it is paramount for the job centres to be successful if they are to avoid being placed under ministerial administration. This is where the importance of the output goals comes in, as they provide the job centres with a way out of this predicament. Even continually underperforming job centres can avoid sanctions if they reach the expected number of client-caseworker meetings and activation offers. And unlike the outcome goal, performance on these two output goals is completely within the control of the individual job centre: ‘If you perform badly on the results [that is, the number of benefit recipients], then it is super super important, that you at least make sure to provide the activities that we need to do’ (Municipal data analyst).
Focusing on at least making sure to deliver on output goals is a consistent finding across the interviews in the four job centres. To steer the job centres with only the outcome goal in mind is very risky, which is why the production of the more manageable outputs becomes the dominant yardstick of success. Furthermore, the performance on these output measures is continually monitored and tended to by each job centre, so that they remain manageable and do not fall too far below the expected threshold. This creates a focus on internal productivity within the job centre – as a whole, within the individual sub-units and even for individual employees. And it is precisely this logic of internal productivity of each individual job centre sub-unit which runs counter to the goal of policy coherence deeply embedded within the ideal of SI.
Proponents of SI argue that SI goals can only be achieved through a combination of policies (e.g. social, ALMPs, care-policies etc.) (Andersson, 2018); for this reason, formal means of integrating and cooperating across sectorial, organisational and professional boundaries are paramount. In the Danish case, such cooperation can, for example, take the form of a partnership between job centres and NGOs for service delivery, or collaboration with the municipal social department to tailor services to the diverse needs of the most vulnerable unemployed. However, from the perspective of the PM system, such boundary-crossing activities are contrary to the straight and stringent lines of accountability running from the ministerial level down to the individual caseworker. By relying, then, on external partners – public or private – job centres sacrifice control over their output production, while still being held accountable for the said output production. A manager explains the dilemmas of cross-boundary collaboration in the following way: [The problem is] the time horizon. Because before you can reach the moment, where you have reflected together [with the client] and considered the situation of the client from a holistic perspective, I have a deadline where I need to send the client into activation tomorrow or else, I will not reach the given output-goals (Job Centre manager).
The above quote illustrates a dilemma raised in several of the interviews, between doing what is most meaningful and effective for the client and doing what is measured by the PM system. The former emphasis may include services outside the job centre which are not explicitly employment-related, while the latter only includes employment-related activities offered by the job centre: The negative consequence may be that we – in order to live up to the output-goal – initiate an activity within the field of employment-services, which in reality is not super meaningful for the citizen. But because we then can make sure that we achieve the output goal, we do it anyway (Job Centre manager).
The above quote from a job centre manager in one of the municipalities highlights the risk of cooperating with the municipal social services within the existing PM system. As social service organisations are not held accountable for the output goals of the PM system, there is no guarantee that they will provide services within the timeframe that enables job centres to reach these expected goals. Cross-organisational collaboration therefore becomes either something you do at the risk of being placed under administration (because you fail to live up to the output goals) or – as is more commonly described in the interviews – something job centres only do when their internal output goals have already been taken care of. However, such a sequential approach to service delivery can hinder coherence between services: It means that we are stuck in a sort of redundancy, because when we have already made a good holistic approach – for example with housing assistance or extensive socio-pedagogical mentoring – then the client still needs to come and have a meeting with our caseworker in order for us to cross off the right boxes (Job Centre manager).
The manager's example illustrates the clash between implementing an SI approach and meeting the demands for external accountability. The job centre seeks to accommodate both goals, by using its capacity for cross-organisational service collaboration, while upholding a focus on internal production of standardised outputs. However, this leads to service overlap rather than complementarity – thus directly hindering the goal of policy coherence so central to SI.
Conclusion
The overall aim of this paper has been to contribute to existing research on the social investment (SI) state in two ways: first, by outlining – and empirically substantiating – a framework for studying the capacity necessary to implement the policy functions of SI; and secondly, by empirically elucidating how this capacity for SI is not only determined by formal policies but also highly dependent on the administrative conditions in which street-level organisations (SLOs) operate.
The article adds a new dimension to the operationalisation and measurement of the SI state (Andersson, 2018; Heitzmann and Matzinger, 2021; Hemerijck, 2018), by applying the concept of policy capacity to the study of SLOs. We demonstrate how the capacities of SLOs for judgment, long-term investment, and service integration are necessary preconditions for actually implementing formal SI policies. We do not argue that the three dimensions of capacity are the only operational conditions needed to support SI. Instead, this three-dimensional framework can be used to study SLO capacity to perform the main SI functions – as defined by the existing SI literature – in any policy area or setting where formal SI policies are adopted. Whether or not this capacity for SI is also transformed by SLOs to implement policies in accordance with the SI ideal is another question. We only tentatively hint at this by revealing how all the four municipal job centres have made changes to their organisational structure and procedures as a result of the greater autonomy afforded them in recent years – integrating (parts of) social services into the job centre structure or by making greater use of local investment projects (cf. section 4). These examples also confirm existing studies evaluating the early consequences of some of the recent administrative reforms, and which find that job centres generally have experienced and used their increased local autonomy to invest in and integrate certain services (Andersen & Larsen 2018). These empirical findings are of course not as suitable for general analytical use as the three-dimensional framework. Therefore, further research is needed to gain more knowledge of the relationship between the capacity to implement SI polices and the practice of SLOs in other policy areas.
The link between capacity for SI and actual implementation of an SI approach is made even more complicated by the empirical findings of the current article, which challenge our conception of the uniformity of such capacity. Our study shows that the many waves of administrative reforms have created a multi-layered patchwork of operational conditions that are neither fully internally consistent nor necessarily consistent with the formal policies targeted at citizens. On the one hand, the administrative reforms of the last decade have enhanced the discretion granted to job centres for investing in and integrating different services. On the other hand, the system of performance monitoring – introduced in 2007 to limit the discretionary space of job centres – remains largely intact and has, in some ways, even expanded. Consequently, the current administrative conditions in which the job centres operate are both enhancing and limiting the resilience of the social investment state. Our analysis shows that the job centres are forced to try to resolve an inherent discrepancy: they must seek to reconcile a performance monitoring system which incentivises them to prioritise standardised activation schemes, short-term goals and internal productivity, with formal policies which incentivise them to use their new organisational, fiscal and procedural autonomy to prioritise long-term investments in tailored and integrated services. While these demands may be especially contradictory in the Danish case, one would also expect to find co-existent and competing demands on SLOs (Røhnebæk and Breit 2021) across many other policy areas and countries, due to the many administrative reform waves that have swept across most of the Western world since the 1980s. Our findings thus suggest that future studies of the proliferation and consequences of SI policies need to be more attentive to both the internal consistencies of formal policies and operational conditions and the relationship between these.
Finally, the article also provides some tentative findings regarding the relationship between capacity to implement SI and the need to hold implementing agents accountable. The analysis shows that substantial autonomy for SLOs to tailor the content of services, to make long-term investments and to restructure their inter- and intra-organisational boundaries is an important precondition for implementing SI policies. However, the article also explains how a performance system used to enhance external accountability (Jakobsen et al., 2018) has directly impeded the ability of the job centres to use their capacity to promote an SI approach. Although this finding does not necessarily apply to external accountability measures in other cases, it does suggest a potential tension between the capacity to implement SI policies on the one hand and accountability concerns on the other. This begs the question of how to strike the right balance between centralised control and local autonomy in the quest for enhancing the capacity for SI, a question which should be considered by future research on capacity for SI.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: AP Møller Foundation.
