The rise of cities in the welfare domain

Traditionally, the welfare state has been closely associated with the process of state formation itself, as pension systems, collective insurance and redistribution institutionalised solidarity between societal groups, thus cementing an inter-generational social contract (Keating, Reference Keating2001; Ferrera, Reference Ferrera2005). Today’s challenge in welfare politics increasingly lies in designing effective policy and governance arrangements in areas ranging from active labour market policies and social assistance to lifelong learning and vocational education, early childhood education and care, work–life balance policies or multi-problem families and social exclusion (Morabito et al., Reference Morabito, Vandenbroeck and Roose2013; van Berkel & van der Aa, Reference van Berkel and van der Aa2015; Chanfreau, Reference Chanfreau2022). Whilst traditional benefits and rights are primarily defined and managed at the central level, the shift to capacitating welfare services also implies a change in the locus of welfare politics from central state institutions to municipalities and cities as local welfare providers (Sellers and Lidström, Reference Sellers and Lidström2007).

Naturally, rescaling the welfare state to the local level leads to a reconfiguration of the actor constellation. Cities in particular have risen to prominence as frontiers of welfare fallouts and societal change. Challenges of labour market and skills shortages, segmented labour markets, technological change and social exclusion and deprivation often manifest earlier and more intensely in larger cities (Kazepov and Barberis, Reference Kazepov, Barberis, Kennett and Lendvai-Bainton2017). Cities experience spatial inequality and segregation, usually exacerbated by the influx of highly skilled labour in service sectors, pushing up housing prices and gentrifying city-centre neighbourhoods. In responding to these city-specific challenges, cities have become key sites for social policy innovation (Gerometta et al., Reference Gerometta, Haussermann and Longo2005). As such, we set out this paper with the basic assumption that Europe’s larger post-industrial cities, typically governed by progressive political forces favouring pragmatic, problem-based politics, are breeding grounds for new ideas and social activism. In theory, larger cities should have more institutional capacity to deliver on promises and the political clout to challenge national policy boundaries, enabling them to act as agents of change in welfare politics beyond local boundaries.

In this paper, we examine the role of cities – in this case, Amsterdam – and their innovative potential from the perspective of the social investment policy paradigm. The shift to services in the welfare state has gone hand in hand with the rise of social investment as an influential policy paradigm in academia and (international) policy circles (Hemerijck, Reference Hemerijck2013; Reference Hemerijck2017; European Commission, 2024; OECD, 2015). The overarching objective of social investment welfare provision is to enhance people’s opportunities and capabilities to manage social risks typical of post-industrial societies whilst maintaining high-quality employment levels necessary to support the ‘carrying capacity’ (or, tax base) of popular welfare programmes such as pensions and social security in times of ageing societies. The social investment policy approach relies on capacitating services to provide stepping stones to healthier and productive lives and early identification of problems to ensure ex-ante risk prevention rather than ex-post compensation.

On average, the literature on social investment maintains a strong state-centred bias, with policy change primarily explained from institutional legacies, power resources and electoral competition (e.g. Beramendi et al., Reference Beramendi, Häusermann, Kitschelt and Kriesi2015; Garritzmann et al., Reference Garritzmann, Häusermann and Palier2022). However, the territorial dimension is essential to making social investment work, as capacitating services are – for the most part – operated at the local level. This is where the social investment policy paradigm overlaps with the social innovation literature, which analyses new modes of collective action, governance and participation to provide answers to problems for which available solutions have proven inadequate (Klein et al., Reference Klein, Fontan, Harrisson and Lévesque2012; Oosterlynck et al., Reference Oosterlynck, Novy and Kazepov2020). Whilst there is no sharp conceptual distinction provided in the literature between the two strands, we conceive of social investment as a set of policy principles on how to organize the welfare state at large, whereas social innovation focusses mostly on civil society engagement, actor mobilisation and processes of social learning to produce new policy norms in the social domain, thus typically occurring at the local level. Social innovations may contribute to achieving a social-investment-oriented welfare state. Linking city-level social innovation to the broader social investment transformation of the welfare state will enable us to understand whether, how and why cities can be frontrunners in this broader transition, rather than studying them merely as isolated welfare delivery areas.

Whilst the literature on local social investment is still in its infancy, there are a few exceptions. Focussing on German active labour market policies, Sabel et al. (Reference Sabel, Zeitlin, Quack and Hemerijck2017) argue that decentralized experiments can lead to the creation of capacitating services that are both more effective and cost-efficient than standardized practices. Kazepov and Cefalo (Reference Kazepov, Cefalo and Kazepov2022) argue that whilst the local level offers great potential for innovative bottom-up solutions to social challenges, the local level is not necessarily always preferable for social service design, as it may also reinforce territorial inequalities and fragmentation. Kazepov and Barberis (Reference Kazepov, Barberis, Kennett and Lendvai-Bainton2017) add that decentralised service provision may also lead to unstable innovative practices if small-scale initiatives are not scaled up, whilst the multiplication of actors leads to coordination challenges. To make local social investment work sustainably, Scalise and Hemerijck (Reference Scalise and Hemerijck2022) argue that vertical complementarity between the local and central level is of crucial importance.

In understanding what vertical complementarity between local innovation and central steering may look like in optima forma, we can draw inspiration from the experimentalist governance literature (Sabel and Zeitlin, Reference Sabel and Zeitlin2008). Experimentalist governance defines a set of basic governance principles to foster structured learning from diversity by formulating broad provisional framework goals at the central level and allowing for local autonomy for experimentation but with peer review and periodic revision of shared goals and metrics based on learning from implementation practice. Furthermore, as argued by Scalise and Hemerijck (Reference Scalise and Hemerijck2022), vertical complementarity also implies that decentralisations are not meant as cost-avoidance strategies and should avoid misfit between central-level social security rights and benefits and local activation and support practices.

Given the low empirical basis on local social investment, we treat these insights as inspiration for our otherwise inductive endeavour into the policies and politics of city-level social investment and innovation. In understanding whether, how and why cities may act as frontrunners and innovators for broader social investment transformations, this paper focusses on the case of Amsterdam. The paper is structured as follows: section 2 will substantiate our choices regarding city selection and research approach, followed by our three policy case studies. Section 3 will focus on the role of Amsterdam as a crisis manager in the case of teacher shortages; section 4 will focus on experimentation in the field of labour market activation; and section 5 will focus on Amsterdam’s new role in tackling energy poverty with the Amsterdam FIXbrigade. In section 6, we will make the case that Amsterdam has risen in importance throughout various welfare domains, and the city acts as a frontrunner and innovator in the Dutch welfare state. This rise in importance should not just be seen from its endogenous progressive politics, but also because the city acts as a stopgap, addressing gaps left by national welfare policies. As such, from the perspective of vertical complementarities, we point at several serious shortcomings and obstacles.

Case selection and research approach

The Netherlands is a typical example of today’s hybrid welfare models. It transitioned to a service-based welfare state from the 1990s onwards, but with a mix of public and private delivery and a mix of generosity with targeted provisions and strict conditionality. Constitutionally, it is considered a decentralised unitary state (Andeweg and Irwin, Reference Andeweg and Irwin2014), with municipalities responsible for a wide variety of tasks but without the fiscal autonomy of Nordic social investment vanguards. Roughly 60–80 per cent of municipal budgets depend on state funding, with temporary and earmarked budgets (Allers, Reference Allers and Polman2023) Yet, Amsterdam has robust own revenues from property tax, parking fees, land revenues and tourism tax which are on par with state contributions (Gemeente Amsterdam, 2024). Sweeping decentralisation reforms in 2015 have endowed municipalities with the responsibility for delivering social services including activation, youth care, social inclusion and long-term care (Sabel et al., Reference Sabel, Zeitlin and Helderman2023). Ever since, there has been a surge of local social investment experimentation. For example, Nijmegen set up an experiment in loosening conditionality in social assistance (Betkó et al., Reference Betkó2020); Arnhem eradicated all private debt of 50 vulnerable households (NOS, 2024); or The Hague introduced 16 hours per week of free early education for 2.5 year olds (Gemeente Den Haag, 2023). All these cases have in common that municipalities have taken matters into their own hands out of dissatisfaction with the results of national-level policies. The largest and most influential of these experimenting cities is Amsterdam.

Amsterdam has an extensive local policy agenda with distinct social investment goals and links to other domains such as the climate transition (Gemeente Amsterdam, 2022). The city has been governed by progressive political forces, thus contrasting the overall right-wing orientation of central governments At the same time, the city also suffers from many of the key traits distressing European cities in general, including housing price surges, labour and skill shortages, gentrification and social exclusion, particularly of migrant groups (van de Werfhorst and van Hest, Reference H. and E.2019). In this paper, we treat Amsterdam as a paradigmatic case, expecting it to provide the right conditions to study how a city can be a leader in broader welfare state change. Due to the high endogeneity of welfare politics, we prioritised internal validity over external representation by examining three case studies across different welfare functions in the same city. Finally, we refer to both ‘innovation’ and ‘experiment’ in discussing policy initiatives in our case study. Whilst the former refers to novelty in introducing policy norms, we use a narrower understanding for experiments, referring to initiatives intentionally designed as learning exercises, ideally including a control group.

For case selection, we follow Hemerijck’s (Reference Hemerijck2017) conceptualisation of the social investment welfare state as revolving around three mutually reinforcing functions: (1) raising lifelong human capital ‘stock’ development; (2) easing the work–life balance ‘flow’; and (3) providing adequate and inclusive, but non-status-confirming income protection and social safety net ‘buffers’. In practice, policies may serve multiple functions, but the conceptualisation allows for the provision of a cross-section of the welfare state. As such, in terms of stimulating human capital stock, we examine how the city responds to teacher shortages affecting segregation in the school system. Second, we discuss how Amsterdam tries to guide long-term unemployed individuals towards shortage sectors to stimulate labour market flow. Third, regarding buffers, we look at the problem of energy poverty, which also allows us to bring in the climate transition angle.

In terms of research approach, each case follows a similar structure. We first examine the context and emergence of the policy problem, followed by analysing the city’s response, the novelty and effects of the response, the ability for scale-up or further diffusion and the institutional fit with national policies. Our analysis is based on extensive document analysis, covering national and municipal legislation, policy briefs, evaluation reports, government press releases, parliamentary and local council debates and reputable news sources, alongside twenty-five expert interviews with city-level and national policymakers, practitioners and experts (see Annex 1). Semi-structured questionnaires allowed us to explore key themes whilst ensuring a consistent focus across all interviews. We also used probing techniques to falsify our assumptions on each case, thus sharpening the analysis. By triangulating between multiple data sources and cross verifying the findings, we aimed to maximise the reliability and validity of our conclusions. To ensure high ethical standards, interviews were kept anonymous and we agreed not to use direct verbatim quotes. Our interview material, therefore, serves to contextualize the cases, but we cite publicly available sources as much as possible.

Stock: the city as crisis manager of teacher shortages

Education is central to any successful social investment strategy to support the productive capacities of citizens and break the intergenerational cycle of disadvantage. Education has traditionally been one of the strengths of the Dutch economy. The Dutch welfare state has internalised the learn-first approach and in 2024 boasts the EU’s lowest school drop-out and not in education, employment or training (NEET) rates (Eurostat, 2024). However, the Netherlands has also been plagued by dropping PISA scores and, in terms of reading quality, now ranks as Western Europe’s worst performer (OECD, 2022). There is ample evidence of increasing stratification and segregation in the education system (Onderwijsraad, 2019). An important factor in this steady deterioration is the growing teacher shortage (ibid.).

With close to 10,000 vacancies open in primary education and more than 16,000 in secondary education (Wester et al., Reference Wester, de Wilde, Jellicic, de Moor and Vankan2021), the shortage of teachers has become a nationwide emergency. Cities – particularly Amsterdam – acted as agenda-setters as the crisis manifested itself first in the major cities. Higher housing prices, more segregated schools and a higher volume of vulnerable yet work-intensive children drove teachers out of the city. In 2022, the shortage in the five big cities reached 14.9 per cent for regular and 24.6 per cent for special-needs education (Onderwijsraad, 2023). Schools have had to fall back on 4-day school weeks or teachers without official qualifications. The shortage is most severe in neighbourhoods with the most socio-economically disadvantaged children, especially those with a migration background, who need guidance above all (Adriaens et al., Reference Adriaens, Elshout and Elshout2022). The higher the number of students at risk of falling behind on basic skills (hereafter: risk-children), the lower the number of teachers available, thus deepening segregation (Reches & van Spijker, Reference Reches and van Spijker2023).

The Ministry of Education became aware of the shortages in 2013, as budget cuts reduced the inflow of teachers whilst retirements increased the outflow. The response was to facilitate lateral entrants (transitioning from another sector) into the teaching profession and introduce new entrance examinations to enhance teacher quality (Ministerie van Onderwijs, Cultuur en Wetenschap, 2013). In practice, these exams were seen as factual memory tests that did little to develop didactic skills and further discouraged entry, especially for teachers with migrant backgrounds in urban areas (Hagoort et al., Reference Hagoort, Schaap and Heijmans2017). As such, national policies aimed at prevention but did not recognise the growing local emergency.

Traditionally, municipalities in the Netherlands do not have a formal role in education beyond school estate maintenance. However, from 2015 onwards, Amsterdam became increasingly active in the education domain. Children and school segregation have consistently featured as top priorities in coalition manifestoes, with initiatives focussed on vocational education academies, after-school activities, anti-discrimination activities, etc. (e.g. Gemeente Amsterdam, 2022). Particularly on the issue of shortages, the city could no longer ignore the alarmist tone of school boards, fed up with teachers breaking down in their offices after sending children back home because of a lack of teachers (Interviews 3, 4). The first municipal steps, outlined in the 2015–2018 Amsterdam Teacher Agenda, were relatively modest, allocating funds for teacher scholarships to enhance skills and school grants for quality improvement. As the shortage worsened, city-level activism was steadily scaled up. The 2019–2023 Amsterdam Teacher Agenda increased funding to facilitate entry, subsidised parking licenses and travel costs to incentivise teachers coming from outside the city, increased funding for training and introduced preferential treatment for social housing in Amsterdam for around seventy teachers per year. Whilst no silver bullet, mobility and lack of housing were considered some of the main obstacles for teachers (Haanstra-Veldhuis et al., Reference Haanstra-Veldhuis, Vadanescu and van de Wel2022; Interviews 1, 2, 4). More importantly, the city also managed the problem vis-à-vis the central government.

Amsterdam’s Deputy Mayor Marjolein Moorman and School Board President Arnold Jonk jointly lobbied the ministry to introduce a premium for teachers working in schools with a high share of risk-children, a targeted approach particularly beneficial for larger cities (Interview 3). This type of wage differentiation in a system of deeply institutionalised collective agreements is highly unusual, and initially, resistance was faced from the ministry and trade unions, who were worried about unequal treatment (AOB, 2021). However, the concerted lobbying efforts of the big cities, arguing that cities also face higher costs of living, together with Moorman and Jonk’s strategy of raising alarm bells in the media, proved successful. For the period 2021–2023, the ministry allocated funds and provided a degree of regulatory discretion for drafting local covenants between municipalities and school boards, including a salary premium for the 1,300 schools with the highest share of risk-children (Visser et al., Reference Visser, Zumbuehl and Schippers2024). Whilst the national average of the premium was around 6.5 per cent, in Amsterdam, the premium could be as high as 10.5 per cent of the annual salary (AOB, 2023).

The Amsterdam covenant included a voluntary agreement where participating schools committed to training lateral entrants, refraining from hiring new teachers if their shortage is small and refraining from recruiting teachers from other covenant schools where shortages are severe (MinOCW, 2020). Only schools that signed the covenant were eligible for the salary premium. The measure was made permanent with structural funding only 2 years after its introduction, with trade unions eventually supporting its inclusion in collective agreements. Official quantitative national-level analyses for the first 2 years concluded that it is too early to establish effectiveness (Visser et al., Reference Visser, Zumbuehl and Schippers2024). However, this national-level image did not correspond to the perception of some of our interviewees for Amsterdam, who saw a notable change for schools in disadvantaged neighbourhoods since the introduction of the measure (Interview 3). Confirming this perception, in 2024, the teacher shortage decreased for the first time since 2020, whilst the decline was strongest in neighbourhoods with more deprivation (Centerdata, 2024).

For the 2023–2027 Amsterdam Teacher Plan, the municipality expanded prevention, training and mobility provisions but also set up a new €6.5 million funding scheme to support management skills for school directors. Salary is an important element in fighting teacher shortages, but teachers also want more autonomy in teaching and less bureaucracy to prevent overburdening and increase work pleasure, for which skilled school leaders are of vital importance (Van der Heuvel and De Vroome, Reference Van der Heuvel and De Vroome2023).

Overall, we observe that whereas traditionally the city’s role in education was modest, it has steadily grown from focussing on quality education, providing extra funding, and supporting entry into the profession to acting as an agenda-setter and crisis manager. Compared with the role of the central government, the funding schemes are relatively small, but especially in its role of crisis manager and agenda-setter, Amsterdam was able to make a dent in an otherwise escalating situation. Amsterdam municipality showed to be a more responsive layer of government, battling – initially – against the tide. Yet, whilst wage differentiation to prevent segregation is novel, it is difficult to consider the measure as genuine social investment innovation. Social investment focusses on ex ante prevention of risks rather than ex post resolve. As admitted by Deputy Mayor Moorman, ultimately what is needed is a comprehensive and more radical national approach (Het Parool, 2023a).

Flow: labour market activation in times of shortages

In 2022, more Amsterdam residents were employed than ever before (520,000), with unemployment at a record low of 4.8% (Onderzoek & Statistiek Amsterdam, 2023). At the same time, 8,500 residents are on unemployment benefits, 33,000 on (partial) disability benefits and 36,000 on social assistance amidst significant labour shortages in sectors such as education, childcare, healthcare, technical skills and ICT.Footnote ² Supporting these remaining groups in labour market entry is a challenging task. In a booming labour market, most of the easily employable inactive labour supply is already working. Of the remaining group, most lack the qualifications and skills to enter employment, with half coming from low-educated migrant backgrounds (CBS, 2022), and two-thirds citing illness or disability as the primary barrier to participation (Arbeidsinspectie, 2022). Clearly, not everybody can work. However, some groups remain unemployed due to a lack of support. For this latter group, municipalities have been testing out new approaches, with larger cities leading in innovation.

Since the introduction of the Participation Law in 2015, municipalities are responsible for activation and social support. The law allows municipalities the discretion to devise their own labour market integration approach, within certain bounds. Overall, the law favours a more restrictive workfare approach over capacitation as it emphasises self-responsibility and reciprocity, incorporating sanctions, job obligations, restricted access to sheltered workplaces and reduced benefits for shared households. Meanwhile, financial support for labour market reintegration has fallen between 2010 and 2018 from an average of €4,500 to €1,500 per unemployed (VNG, G4, G40, Viosa & Cedris, 2019). To incentivise municipalities to minimise social benefit dependency, they are allowed to retain any savings made on benefits. Often the logic of balanced budgets acts as an important obstacle to local social investment initiatives, as the costs are borne by the city, whereas benefits – better health outcomes or lower social security costs – manifest themselves at the national level (Van Olden and Van het Erve, Reference Van Olden and van het Erve2020). In responding to the incentive, municipalities initially focussed on the most promising groups, leaving the more complex – and therefore expensive – cases underserved (SCP, 2019). However, with further tightening of labour markets, municipalities can no longer afford this luxury, thus triggering the need to innovate in response to uncertainty about how best to activate the remaining group. Below, we focus on some of the main initiatives in Amsterdam.

An important innovation, contrasting national policy, concerns the approach to benefit recipients. In response to discretionary space within the Participation Law, progressive cities such as Amsterdam (but also Utrecht) have chosen to focus on facilitating basic needs and proactive guidance (Interviews 21, 22). Amsterdam proactively reached out to all unemployed individuals to offer personalised support from social workers based in neighbourhood teams. This approach contrasts with cities such as Rotterdam or Leeuwarden, which stress reciprocity over care by obliging benefit recipients to do community work. Interestingly, in these varying approaches, there is no clear winning strategy, as each has pros and cons, and much depends on the quality and caseload of individual social workers (Kremer et al., Reference Kremer, van der Meer and M.2017).

Following an experimental pilot phase in 2021, Amsterdam also introduced a supplementary earnings bonus for all social assistance recipients who secure part-time work. Under national rules, extra income would be deducted from benefits, disincentivising small side jobs that could lead to more substantial employment. With the earnings bonus, recipients get to keep 25%. A mid-term evaluation showed the outflow towards work amongst the experiment group to be twice as high as that of the control group (Onderzoek & Statistiek Amsterdam, 2020). The study also showed that the nature and intensity of counselling by municipal client managers are important predictors of success. Vulnerable target groups, such as single men, women in families and men with a migration background, particularly benefit from extra personal attention and trust.

In 2024, Amsterdam launched a trial in which 450 families on social assistance will receive an extra €150 a month for 2 years without any conditions. The Kansfonds and Stichting RCOAK, both donor-funded societal initiatives, finance the trial whilst the municipality organises the evaluation. Amsterdam had to set aside extra funds to compensate recipients for the automatic reduction in their social allowances when receiving a higher income, which would cancel out the effect (Custers, Reference Custers2023).

Amsterdam also has longer-running projects, such as the Amsterdam Work Brigade, which provides a 2-year work contract for those on extended unemployment benefits on the basis of the ‘place-first, then train’ principle, and a subsidised job initiative for unemployed workers with mental disorders, called VIP Amsterdam. A particularly interesting innovation in designing genuinely capacitating services is the recent application of the so-called Maatwerk Methode (customised approach), developed by the Institute of Public Values (Publieke Waarden, Reference Waarden2024). The Amsterdam Maatwerk Method involves cooperation between different parties, such as schools, youth assistance and other agencies, to collectively assess the needs of individual families and devise tailor-made solutions. The Maatwerk Method has been particularly successful in Utrecht, which provides tailor-made support for multi-problem families to solve challenging social assistance bottlenecks ‘one case at a time’ (Sabel et al., Reference Sabel, Zeitlin and Helderman2023). Utrecht became one of the only municipalities to combine high-quality care with budget sustainability, whereas most others struggle with deep budget deficits.

Whilst designing capacitating services, such as the Maatwerk Methode, aligns with the moral philosophy of social investment by providing stepping stones for a healthier and more active life (Hemerijck et al., Reference Hemerijck, Bagadirov, Puertas Roig, M.A. and M.2022), a successful social investment welfare state also seeks to prepare rather than repair, thus preventing unemployment. In this field, we also witness the growing role of the city of Amsterdam. Amsterdam has a dedicated councillor for vocational training and labour market integration, with policies aimed at actively steering young people towards shortage sectors by capping studies with low employment and investing in transitions towards green and digital skills (Interview 1).

Overall, Amsterdam (as well as other progressive cities) has given a capacitating twist to an otherwise restrictive national activation policy. The Participation Law, a relic of the post-euro crisis period, was designed to address high unemployment and rising social benefit costs. However, when facing labour shortages, cities mostly opt for guidance and support instead. Experimental projects have demonstrated positive outcomes, albeit on a limited scale and lacking the experimentalist governance framework to allow for structured benchmarking and facilitated collective learning. Instead, evaluations were ad hoc and patchy. In general, the Participation Law has been heavily criticised for failing to achieve its goals because of its restrictive approach (SCP, 2019). Cities have become successful lobbyists in pushing for change whilst supporting their arguments with the outcomes of their capacitating approaches and experiments. In anticipation of change, some – including Amsterdam and Rotterdam – have simply refused to execute some of the harsher elements of the law, such as the deduction of social assistance for shared households (Bokhorst & Engbersen, Reference Bokhorst and Engbersen2024). In 2023, the government decided to reopen the law to remove some of the punitive edges, partly in response to city-level activism, a process still ongoing at the time of writing.

Buffer: addressing energy poverty with the Amsterdam FIXbrigade

Since the 2022 spike in energy prices following the Russian invasion of Ukraine, local governments across Europe have employed strategies to combat energy poverty (Eurocities, 2023). In the Netherlands, the price surge exposed pre-existing vulnerabilities related to energy affordability and housing conditions. Notably, energy costs in the Netherlands have consistently been amongst the highest in the EU in absolute terms, largely due to taxation (VaasaETT, 2024). Most households can manage these energy costs even during price surges, and for some, this can incentivise more sustainable energy practices. However, the implications are more severe for vulnerable households, where energy prices consume a significant share of disposable income, reducing funds for other basic needs or causing adverse health effects if heating is turned off due to financial stress. The problem is exacerbated when households live in dwellings with poor energy performance. Whereas wealthier households increasingly switch to (publicly subsidised) solar panels, old rental homes in vulnerable neighbourhoods risk falling behind. The resulting ‘energy poverty’ is associated with a host of societal problems, including mental health issues, degraded work and school performance, stress (Thomson et al., Reference Thomson, Snell and Bouzarovski2017) and respiratory illnesses such as asthma (Healy, Reference Healy2017).

In Amsterdam, 16 per cent of the housing stock has a poor energy performance, marked by energy labels E, F and G. The housing stock consists of many post-war reconstruction units, built as cheaply as possible, and therefore, of below-average quality. Most of these are concentrated in low-income neighbourhoods (Gemeente Amsterdam, 2022a). As such, 9 per cent of Amsterdam residents face energy poverty, defined as spending more than 10 per cent of their disposable income on energy bills (Mulder et al., Reference Mulder, Dalla Longa and Straver2021).

In response to the energy crisis, the Dutch government introduced an €800 lump-sum ‘energy allowance’ for low-income households. Municipalities were tasked with administering the measure and given some freedom to determine eligibility criteria. Amsterdam compensated a much larger group than national guidelines prescribed, whereas Rotterdam adopted a more frugal approach. More importantly for this discussion, an ‘energy poverty alleviation programme’ was funded to support municipalities in improving energy efficiency and resilience of vulnerable groups. This fund allowed Amsterdam to expand their local best practice in fighting energy poverty: the FIXbrigade.

In 2018, long before the issue of energy poverty gained national political traction, the community centre ‘Jungle’ initiated the FIXbrigade in East Amsterdam. This programme involves ‘fixers’ (teams of craftspeople) visiting residents’ homes in disadvantaged neighbourhoods to provide energy-saving advice and perform small-scale energy efficiency upgrades. Jungle predominantly recruits fixers amongst the long-term unemployed and migrant groups, providing them with a 6-month vocational training programme in executing tasks such as small-scale insulation, hydronic balancing of central heating systems and conducting infrared scans to detect heat leaks. These measures lead to energy and cost savings for residents, enhance residential comfort and contribute to battling climate change (Croon et al., Reference Croon, Maghsoudi Nia, He, Qian, Elsinga, Hoekstra, van Ooij and van der Wal2024). Meanwhile, the fixers also receive on-the-job training in green skills, becoming more self-reliant and gaining access to the labour market, which increasingly needs skilled workers for the energy transition. As a prime example of social investment complementarity, the initiative thus stimulates stock (skills), buffer (energy poverty) and flow (labour market activation) in an environmentally sustainable manner.

Despite its social investment potential, it took time for the FIXbrigade initiative to gain recognition and municipal funding. The municipality initially expressed scepticism about the initiative’s cost-effectiveness, primarily regarding the speed of implementation. The initiators were typical pragmatic doers, wary of bureaucracy and accountability. Energy efficiency, the benefits of home visits and job prospects for the fixers were difficult to measure, whilst costs were precise: €1.6 million for 1000 houses. Therefore, Amsterdam initially hesitated, leading the FIXbrigade to turn to the EU for funding. It was not until 2022 that the FIXbrigade received municipal funding from the ‘energy poverty alleviation programme’ to expand its work to other city districts. An official study on the FIXbrigade’s impact (compared with a control group) proved highly positive, showing a more than 21 per cent average reduction in energy consumption, enhancements in living comfort and physical health and even an increased propensity amongst residents to engage in neighbourhood volunteering projects post-intervention (Van der Wal et al., Reference Van der Wal, van Ooij and Straver2023).

The FIXbrigade soon transformed into a city-wide ‘phenomenon everyone was talking about’, receiving widespread praise from citizens in the local media and inspiring similar initiatives across the country (Interview 18). However, in Amsterdam, the initiative continued to battle against pervasive bureaucratic logic. Cost-effectiveness and speed of implementation continued to pose a dilemma for the city: whether to prioritise sympathetic social initiatives such as the FIXbrigade or opt for the ‘quick-and-dirty’ approach of private companies, particularly because the energy crisis required immediate action. The one-off and short-term nature of national funds for structural measures against energy poverty did not favour the FIXbrigade, leading most of the funding to go to private contractors. At the end of 2023, the FIXbrigade suffered an even greater blow. With the immediate urgency of high energy prices waning, it was widely expected that the fixers would win a major municipal tender for insulating nearly 7000 draughty houses in Amsterdam’s most deprived neighbourhoods. The left-wing coalition had extensively advertised the benefits of the FIXbrigade in their coalition manifesto. Yet, in an unexpected development, Energiebox, a social enterprise from Utrecht, won the tender with a proposal offering significantly lower costs.

The outcome ignited a vigorous debate within Amsterdam about the objective alignment of various municipal departments, the prioritisation of goals and the appropriateness of using large-scale public tenders to fund the scaling-up of community initiatives, considering the constraints posed by European tender regulations. The founder of FIXbrigade, expressing disappointment with the city’s decision, emphasised the organisation’s distinct advantage as a community centre in effectively identifying households at risk of energy poverty at a grassroots level, an ability they deemed superior to that of external entities (Het Parool, 2023b).

Despite local setbacks, the FIXbrigade has become a national best-practice example. Widely covered in national media, it emerged as a key local response to rising energy poverty during the 2021 energy price crisis. That October, the earlier-mentioned ‘energy poverty alleviation programme’ was launched, leading many municipalities to establish their own FIXteams on the basis of Amsterdam’s model (Croon et al., Reference Croon, Hoekstra and Dubois2024). In 2023, the Ministry of Justice established funding explicitly earmarked to train fixers in vulnerable areas nationwide (Rijksoverheid, 2023). A national FIXbrigade platform was established, which now provides standardised start-up kits, training and task lists for ‘master fixers’ to scale the model nationwide and even with the Eurocities network.

The FIXbrigade case underscores various governance challenges, particularly the EU-regulated public procurement processes favouring commercial and cost-effective entities over social enterprises with broader impacts. It highlights the limitations of temporary external funding that is restrictive in scope and the larger issue of the institutional void in climate policy. Whilst supporting grassroots initiatives such as the FIXbrigade is crucial, the Netherlands, with the world’s largest share of social housing (OECD, 2020), faces additional hurdles. Social housing providers were made independent in the 1990s, limiting oversight and leaving the government dependent on voluntary renovation agreements. Energy poverty alleviation is a relatively new focus for Dutch housing associations, which face regulatory obstacles that even unintentionally incentivise the allocation of the least energy-efficient social housing units to the most vulnerable households, thereby leading to energy poverty from the beginning of their tenancy (Croon et al., Reference Croon, Hoekstra and Dubois2024). Municipalities set ‘performance agreements’, but these are non-binding and mainly affect new dwellings, not existing ones. The national government faces difficulties in keeping housing associations accountable and ensuring interim progress. Additionally, resource scarcity, particularly workforce shortages in the construction and installation sectors, complicate large-scale renovation efforts (WRR, 2020). Amsterdam’s investments in training programmes for the energy transition face leakage effects as trained workers move to more affordable municipalities. This situation, set against the backdrop of decreasing national-level funding for labour market integration, reveals significant gaps in vertical complementarity and the broader challenges of steering large-scale renovation and improving resilience for vulnerable residents.

Conclusions. The politics of social investment cities: between stopgap and innovation hub

The case of Amsterdam as a local welfare actor shows that studying welfare state transformation, particularly the rise of social investment, from the perspective of national policies and politics alone, is insufficient. Within the polyarchic constellation that characterises welfare politics, the rise of cities as agents of change is a particularly interesting but understudied phenomenon. The case of Amsterdam confirms our assumption that cities can act as breeding grounds for social innovations as they devise pragmatic responses to more acute problem pressures, thus devising new policy norms which may instigate broader change. Through policy innovation and experimentation, Amsterdam emerged as a political force of significance in Dutch welfare politics and acted as a frontrunner in the broader social investment transition.

In each of our three policy domains, the role of the city has grown over time. Amsterdam was ahead of the curve in all cases and an initiator of broader change. The outcomes of innovations and experiments at the local level have the potential to shape cognitive orientations of welfare politics by showing what works and which national policy assumptions do not fit reality on the ground. Amsterdam acted as an agenda-setter when it came to wage differentiation to address the territorialised and segregating effects of teacher shortages, which were not recognised at the national level. Amsterdam introduced reduced conditionality requirements, unconditional cash experiments and personal guidance in activation policies, as it considered a punitive approach inapt to integrate the most difficult groups. Cities collectively helped push the reform of the national law higher on the political agenda. Amsterdam also supported, albeit reluctantly at first, the FIXbrigade – as a clear example of bottom-up social innovation – in tackling energy poverty and training long-term unemployed towards shortage sectors, a practice that has inspired similar initiatives throughout the country with a dedicated national funding scheme.

We explain the rise of the city of Amsterdam within the Dutch welfare state from three main factors. First, the city has been consistently ruled by progressive coalitions favouring social investment solutions. The city has the administrative capacity to deliver on social investment innovations and conduct and evaluate experiments, and it has the political clout to instigate broader change. Second, the decentralisation of 2015 has endowed municipalities with more tasks within the welfare state and some freedom in devising local solutions. Third, the city experiences welfare fallouts earlier and more acutely and responds in a pragmatic manner to changing social risks, thus acting as ‘a stopgap’, filling the holes and cracks left open by national policies. This last factor in particular bears potential wider theoretical significance that requires testing for other post-industrial large cities. Cities are at the frontier of welfare fallouts and societal change, and as such, offer a window into examining new challenges to welfare states, such as growing labour shortages.

Finally, in examining local social investment, we adopted – and refined – the conceptual framework of vertical complementarities, understood as a combination of organised learning to diffuse best practice, adequate financial support and institutional fit between welfare functions. This proved useful primarily in examining remaining shortcomings and gaps, but did not emerge as a necessary scope condition for success. In each of our cases, diffusion or upscaling of local practice occurred despite a lack of vertical complementarity. Even more so, it was the misfit between national frameworks and local practice that prompted local experimentation in activation policy. We did not come across experimentalist governance frameworks where local autonomy is deliberately granted in the face of strategic uncertainty but coupled with overarching framework goals and organised learning infrastructure for upward convergence. The 2015 decentralisations allowed for more autonomy but were also intended as cost-saving strategies, with ad hoc evaluations of local experiments and each city reinventing the wheel, thus risking fragmentation and territorial inequality. Whilst cities such as Amsterdam and Utrecht are making decentralisation work, others – primarily smaller municipalities – struggle financially (SCP, 2020). Another unforeseen obstacle lies in pervasive bureaucratic logic, where budgetary rules hinder local social investments. Whilst the costs are immediate and tangible at the local level, benefits are often realised at the national (budget) level, or trained personnel leave for opportunities outside of the city. Similarly, European tendering rules can prevent the build-up of sustainable partnerships with social initiatives.

Overall, whilst there are many studies on cities as local welfare actors and a wide literature on national-level social investment, studying the link between the two has proved a fruitful endeavour that brings new dynamics in welfare politics to the fore. At the same time, underlying concepts and scope conditions for success will require further refining on the basis of a broader empirical base (Hemerijck and Scalise, Reference Hemerijck and Scaliseforthcoming).