Contested Social Impact Bonds: welfare conventions, conflicts and compromises in five European Active-Labor Market Programs

Abstract

Over the past decade Social Impact Bonds (SIBs) have attracted much public policy and management research interest and debate. This article draws on the Welfare Conventions Approach to explore the diversity of five SIB-financed Active Labor Market Programs in four European countries using comparative case study methods. We identify a tension between the requirement to align civic and financial interests in SIB-financed programs alongside a drive to reform public sector procurement in a more entrepreneurial direction. We suggest that the diversity of SIBs emanates from the political struggles in implementation processes stemming from a plurality of welfare conventions that actors need to align and compromise. SIBs are built within historically grown “institutional contexts” that are themselves on the move over decades of welfare state reform, and processes of marketization – and thus far from homogenous.

Introduction

Social Impact Bonds (SIBs) are a high-profile development in international public policy and management over the past decade furthering an outcomes-based commissioning approach to the design and delivery of public services (Edmiston and Nicholls 2018). Recent years have seen a welcome increase in detailed qualitative case study research into SIBs that emphasize the heterogeneity of SIB-financed projects (Cooper, Graham, and Himick 2016; Neyland 2018; Carter 2021; Lowe et al. 2019, Tse and Warner 2020; Fraser, Tan, and Mays 2021). SIBs are strategically ambiguous policy instruments which despite sharing some similar basic principles, demonstrate significant diversity both within and across national settings (Fraser et al. 2018; Maier, Barbetta, and Godina 2018; Williams 2019). Situating our findings within wider debates about public management and welfare state reform, this article asks the following research question: what are the reasons for, and the implications of such diversity found across SIB-financed projects?

Comparative welfare scholarship emphasizes diverse historical institutional welfare regimes (Esping-Andersen 1990), and national institutional trajectories derived from political-economic path-dependencies (Hall and Soskice 2001). Pollitt and Bouckaert (2017) refer to trajectories of modernization and reform, recognizing the importance of politico-administrative and cultural differences between nations as well as similarities in reform efforts to make government more efficient and responsive. Comparative welfare scholarship has been challenged by increased privatization, marketization and financialization across Europe in recent decades. Today welfare states combine diverse welfare organization principles defying clear-cut categorization (Bode 2008). The “transformational adjustment of welfare state arrangements” has been analyzed in terms of “drift, layering and recalibration” (Seeleib-Kaiser 2016:220), requiring an analytical lens attuned to the local flexibility and institutional contradictions (Seo and Creed 2002).

Competing institutionalist theories emphasize different institutional reasons for change or stasis in public management. For instance, rational choice institutionalism prioritizes fixed rationalist preferences, historical institutionalism prioritizes historical path dependency, sociological institutionalism prioritizes cultural norms, and discursive institutionalism, prioritizes ideas and discourse (Schmidt 2008). Theories of competing institutional logics (Thornton, Ocasio, and Lounsbury 2012) or conventions (Biggart and Beamish 2003) emphasize the contemporary heterogeneity of institutional constellations. The Welfare Convention Approach (WCA) draws on this tradition, using welfare convention theory to guide comparative research of SIBs (Chiapello and Knoll 2020). SIBs are themselves hybrid, multi-sited arrangements bringing together different actors, including social service providers, government representatives, and private investors to solve social problems through pay-for-performance contracts. In this article we apply the WCA to five cases across four countries – the UK, the Netherlands, Switzerland, and Germany to explore how five SIB-financed employment and training programmes emerge locally as moral justificatory compromises between social worlds that, corresponding to national-level differences, are institutionalized differently through processes of conflict and compromise.

A significant theoretical contribution of the paper is to highlight the inherently relational character of the WCA as an approach to comparative international public management scholarship of outcomes-oriented reforms such as SIBs. We demonstrate that for SIB-financed projects to be implemented, compromises between actors who are embedded in different conventionalist settings need to emerge. This means that the prevalence of different actor groups and their conventionalist differences become explanatory factors for the implementation of welfare reform programs. This is somewhat different from classic institutionalisms (e.g., Schmidt 2008), because it highlights the conventionalist formation of actors as an explanatory factor. It puts a premium on the negotiated legitimacy between welfare reforms and the dominant welfare actors within national economies. This approach helps explain why we see such diversity across SIB-financed projects.

The following section begins with a discussion of comparative research on welfare states and reform. It introduces the main tenets of the WCA and highlights its analytical potential for a comparative SIB study. Then the methods and findings from the five case studies are presented. Finally, we discuss the implications of these findings.

Comparing welfare states, SIBs and the WCA

Esping-Andersen’s (1990) comparative research identifying different welfare regimes characterizes the UK as a liberal welfare state with “charities, relying on voluntary resources” and “seen as key actors to solve market and state failures” (Defourny and Nyssens 2010:4) while Germany is characterized as a conservative welfare state with a “strong corporatist tradition of largely state-funded intermediate non-profit bodies that administer the social insurance system and deliver social services” (Grohs, Schneiders, and Heinze 2015:167–168). Despite similarities in their welfare state designs, Switzerland is often classified as a conservative welfare state and the Netherlands as hybrid though Hall and Soskice (2001) classify both countries alongside Germany as “coordinated market economies.” However, the last three decades have seen increasing liberalization and a shift from collective solidarity to individual responsibility and social spending reductions in the Netherlands in particular (Pollitt and Bouckaert 2017).

New Public Management (NPM) reforms, promoting market mechanisms, target regimes, and performance indicators have been important in all these countries over the past 40 years, though implementation in the UK remained closer to the marketized model while Germany and Switzerland (and to a lesser extent the Netherlands) maintained features specific to their own policy landscapes including a hierarchy of policy professionals in Germany and a system of multi-lateral dialogue and compromise in Switzerland (Pollitt and Bouckaert 2017).

The SIB emerged in the liberal welfare state of the UK and is less prevalent in so called coordinated market economies or conservative welfare states. The number of SIB-financed programs in the UK (89) compared with Netherlands (17), Germany (3), and Switzerland (1) appears to confirm the analytic validity of the national welfare regimes model to a degree.

National welfare models remain useful as analytical heuristic tools but need to be considered in light of developments including the hybridization and combination of models (Bode 2008) particularly as “the political vocabulary structured by oppositions between state and civil society, public and private, government and market, coercion and consent, sovereignty and autonomy and the like, does not adequately characterize the diverse ways in which rule is exercised in advanced liberal democracies” (Rose and Miller 1992:174). Approaches pioneered by Esping-Andersen or Hall and Soskice are too broad to capture processes of hybridization and the combination of old and new welfare governance instruments within countries and within concrete welfare reform endeavors (Arts and Gelissen 2002). The more we analyze the details of public sector reform, the less homogenous and straightforward such reforming interventions appear. For this reason, research may focus on administrative dilemmas, nonlinearities, and inefficiencies (Pollitt and Bouckaert 2017). Still, research focusing on technical challenges, while valid, may underplay the political, social, cultural, and historic dimensions of these welfare interventions.

The Welfare Conventions Approach may remedy some of these issues (Chiapello and Knoll 2020; see also Batifoulier, Domin, and Rauly 2021). The WCA explicitly promises to analyze the political economy of SIBs and the societal conflicts inscribed into the technical details of implementation. It reformulates technical complexities as conflicts between historical and plural welfare conventions. In so doing, the WCA combines the historic and political perspective of welfare regimes research with an interest in the localized technical details integral to public sector reform. It emphasizes the historical, societal, and political dimensions of seemingly mere technical instruments of government. Financial and statistical instruments of control spawn their own redistributive and participatory effects and determine who speaks for whom, who pays whom and why, who controls whom, who can be helped and who may not qualify for help. The WCA underlines the combinatory and ambivalent character of contemporary welfare reforms, which, of course, are influenced by national institutional trajectories, but also emanate from local constellations, tensions, and initiatives. Welfare conventions assert a particular reason and a solution for a social problem. They proclaim where the responsibility for solving the social problem lies (with investors, philanthropists, state representatives, the community, etc.), what professional knowledge is important to monitor and control the process, and how the financial relations need to be designed (Chiapello and Knoll 2020).

The philanthropic welfare convention views the social problem as one of mercy and compassion. The solutions to the welfare problem are donations and charity. The civic welfare convention locates the reason for the social problem in missing human and social rights and equality. The answer lies in redistribution via taxation to ensure a well-financed public sector in the fields of education, health, and housing. The communitarian welfare convention considers solidarity and community self-help and mutualism to be the solution to the welfare problem. It proposes membership schemes, solidarity, a supportive community, and the organizational form of the cooperative, or the association, as solutions to the social problem. The full employment welfare convention draws upon macroeconomic knowledge about economic booms and recessions. State interventions and investment programs ensure the survival of economic industries and branches. Its welfare guarantee is organized around the creation of high-quality jobs and administrative support for adequate wage-levels (such as support for trades unions). The market welfare convention sees competition as the solution for the social problem. In the welfare, or public sector this often leads to “quasi-markets” (Le Grand 2011), such as voucher systems, or public tendering so that the most cost-effective solution can emerge. Many NPM reforms (Pollitt and Bouckaert 2017) since the 1980s have been based on this welfare rationale in combination with the entrepreneurial convention.

Nonetheless, the entrepreneurial welfare convention is different from the market convention in that it is not necessarily built upon fair competition and transparency, but rather on the idea of constant deal-making, risk-taking and a “handshake-mentality” that is dependent upon networks, high-profile contacts, and the entrepreneurial attitude of people heading projects. In the NPM era, public managers experienced the implementation of contractual management instruments, and ideas of “optimization” and “performance.” Within the entrepreneurial welfare convention, the welfare problem is solved by social entrepreneurs and the poor and socially excluded are helped to become more entrepreneurial.

The financial welfare convention diagnoses the problem as the lack of attractiveness of the social sector for profit-seeking investors. Its solutions render social problems investible and attractive not only for “mission-driven investors” (philanthropic welfare convention), but for “profit-driven investors” (Hochstädter and Scheck 2015). Preferred instruments are financial market technologies, such as social impact ratings – see e.g., the GIIRS technology (Barman 2020) and social investment funds (Hellman 2020; Bourgeron 2020). The behavioral welfare convention frames social problems as psychological in origin and – unlike the market convention – rejects the idea of rational choice and individual responsibility. The behavioral welfare convention seeks to gather knowledge of behavioral patterns to be able to design social interventions to influence and improve decisions subconsciously, also known as “nudging” (Thaler and Sunstein 2008).

The WCA suggests welfare conventions never exist in a pure form. They confront each other through reciprocal critique in everyday organizational and economic life and they appear to be compromised with other conventions in concrete institutions (Chiapello and Knoll 2020). Even though the entrepreneurial and the market convention might seem natural allies, as, for example, in classic labor market activation programs, their combination may not always be easily defended. Such insights help explain why existing research into SIBs has highlighted significant mismatches between the somewhat simple promises promoted by early SIB proponents and the more contested empirical reality of SIB implementation. For instance, research on UK SIBs has demonstrated how local government commissioners can find themselves conflicted between a civic-market compromise institutionalized in public sector tendering programs that are confronted by the collaborative (communitarian) or entrepreneurial deal-making attitude that is needed to get SIBs signed (Fraser et al. 2021). Such a constellation can be portrayed as “fraud” or “collusion” from a market fairness and transparency perspective (see Neyland 2018, who analyses the SIB as an anti-market device). Also, Williams (2019) shows how SIBs in practice have frustrated a lot of social impact-oriented charities and trusts (who might have been expected to champion the model).

Additionally, the promise of the SIB to create an open, flexible, collaborative, and innovative atmosphere (a philanthropic-communitarian compromise) – as put forward by Carter (2021) – is frustrated by a structure that is rigid and bureaucratic and serves to generate revenue streams for investors (financial-civic compromise). In such an institutional world, charities and trusts are reduced to their role of de-risking investors (and/or the state) and are not as involved in the creative and collaborative process they would like (Williams 2019). From other comparative SIB studies (Carter 2021; Tse and Warner 2020; Fraser et al. 2021) we know that SIBs can take on multiple forms, and they are constantly evolving due to conflicts and critique on the ground as well as central government infrastructures such as rate card-schemes or tax exemptions for impact investments. For instance, Williams (2020a) suggests SIBs must overcome localized “valuation struggles” and align quite diverse interests – to an extent that may make them a rather unwieldy welfare technology. As the WCA emphasizes agency and relationships between different actor groups, it may help to show how such valuation struggles unfold within hybrid welfare institutional landscapes in different national welfare states.

In this article we apply the WCA to study SIBs as an example of complex, multi-actor welfare state reform. We use the WCA to analyze how SIBs combine and align diverse welfare conventions in concrete settings. In so doing, we seek to understand the diversity of SIBs through political and historic welfare struggles that emerge via local implementation within diverse international institutional settings. We explore the extent to which locally negotiated SIBs challenge national welfare trajectories; alongside how national institutional specificities influence local SIB constellations when implemented. Our overall aim is to explore how SIBs survive or fail in specific institutional settings and to test the WCA as an analytical heuristic for comparative research designs.

Methods

This qualitative case study (Eisenhardt 1989) comprises of five SIB-financed Active Labor Market Programs (ALMPs) in Europe, three in countries leaning more toward the liberal welfare state type (UK and Netherlands), and two toward the continental European type (Switzerland and Germany). The five interventions target young people (3), refugees (1), and vocational workers (1) to offer supports like job search, counseling, and retraining to improve educational and employment outcomes. The programs range from small (n = 20) to large (n = 1,300). All five are multi-year interventions. The case study characteristics are outlined in Table 1 below.

Table 1. Case study characteristics.

Case	Period of activity	Actors	Programme aims	Targets	Invested capital
UK	2015-18	Not for profit provider; National and local government co-commissioners; Four investors; Some intermediary involvement in set-up; commercial sector evaluator	Deliver support for over 1,000 disadvantaged youths aged 14-20 to avoid becoming NEET (not in education, employment or training). Work closely with local schools. Test the benefits of a PbR approach tied to social investment. Further SIB experimentation and knowledge.	11 targets overall, some ‘hard’ such as achieving qualifications, securing employment and training. Others ‘soft’ such as improved attitude, attendance, and behavior at school. Delivered via centrally administered rate card with a different value for each proven target achieved.	£1.25m
Netherlands 1	2013-15	For-profit provider; Local government commissioner; Three external investors; Some intermediary involvement in set-up; commercial sector evaluator	Deliver support to over 150 youths aged 17-27 to foster an entrepreneurial spirit, stimulate a new mindset and reduce dependence upon social assistance. Also, to experiment with the SIB model and create a financial return on investment.	Reduction in expected social assistance days of benefit using a Cox Hazard Ratio model estimating with 40 years of historical data. There is a flat payment of 40 euros assuming a baseline reduction of 154 days and then an additional payment of 15 Euros for every “additional” day reduced, again assuming a 154-day reduction—to a max of 210 days reduction.	1.1m Euros from investors (including provider at around 40%)
Netherlands 2	2016-19 (terminated early in 2018)	For-profit provider; Local government commissioner; Two external investors; Some intermediary involvement in set-up; commercial sector evaluator	Deliver support to over 130 local Dutch unemployed residents (of any age) with work across the border in Germany. Language and skills classes to be delivered in support of this goal.	Reduced days spent on social welfare using a treatment group compared to a control group that received no support.	1.1m Euros from investors (including provider at 5%)
Switzerland	2015-2020	Not-for-profit provider; Local government commissioner; one investor; no intermediary involvement; commercial sector evaluator	Deliver support to over 240 refugees to integrate into the local job market. The original plan was to do this through ‘supported employment’ i.e., place then train, but this changed to a more conventional train then place model.	Sustainable, long-term integration into the labor market, triggering social inclusion of immigrants in the local Canton. Targets for numbers included in the programme, targets for sustainable employment, and completion of training.	2.7m Swiss Francs
Germany	2013-15	Two not-for-profit providers; Local government commissioner; Four investors; Some intermediary involvement throughout; commercial and academic evaluators	Support around 70 young people below the age of 25. The participants were hard-to-reach and should have neither completed nor enrolled in an apprenticeship programme, not attend school, are unemployed, have neither had contact with an employment agency nor participated in and agency’s programme 2 years before establishment of contact by the project.	Place at least 20 youths into work or an apprenticeship (subject to social insurance contribution) for more than nine months.	250,000 Euros

Most interviews were conducted face to face but there were a small number of telephone interviews when this was not possible. Most interviews were conducted in English, though several interviews were conducted in German or Dutch depending on the preference of each informant. In addition to the interview data, we also draw on documentary data from the sites to learn more about the interventions and wider aspects of each project. For instance, most SIB-financed programs underwent an evaluation and wherever possible we draw on these documents to further our understanding alongside our interview and quantitative data. It was not possible to conduct observational analyses of these programs. We were not granted access to the loan agreements or contracts between parties in this research project. We therefore had to explore the nature of the contractual requirements through evaluations, audit reports, and interviews. Some informants shared costs and financial information with us on a confidential basis so it can inform our understanding but not be identifiably published (Table 2).

Table 2. Qualitative interview informant details.

Case	Providers	Government Commissioners	Investor/ intermediaries	Evaluators	TOTALS
UK	10	1	1	1	13
Netherlands (2)	4	1	2	2	9
Switzerland	6	2	1	1	10
Germany	2	2	3	2	9
TOTALS	22	6	7	6	41

We interpreted the data using the WCA as an analytical heuristic. We studied the conflicts and compromises at each site with a particular focus on the processes of alignment of diverse and distinct welfare conventions. Interview and documentary data were interrogated repeatedly to understand key emergent issues drawing on the principles of “constant comparison” (Glaser 1965). On a technical level, applying the WCA as an analytic device was challenging. There are eight different conventions – and there can sometimes be a degree of ambiguity about which convention best fits with empirical data. For example, both the communitarian and the entrepreneurial agenda allow for flexibility and situational adjustments of contractual designs; “being entrepreneurial” can mean quite diverse things. For instance, service providers may be forced to act more entrepreneurially as a disciplinary measure as part of a SIB, or they may choose to do so (see Cooper et al. 2016). The analytical categories of welfare conventions are thus not self-explanatory. They need to be contextualized to become meaningful. Organizations tend to be polyphonic – with some actors emphasizing different conventions to others from within the same organizations. This poses analytical and interpretive challenges in labeling an actor or organization “entrepreneurial” or not. These are situational questions that required iterative in-depth discussions by the authors of the paper about appropriate ways to aggregate an organizational “position” or “motivation” at a particular time – so too, which voices to sometimes prioritize over others. These processes involved writing, sharing, and debating detailed case study reports for each site amongst the research team, challenging our own analyses and assumptions, re-reviewing interview transcripts in detail and sometimes going back to informants to sense-check our developing interpretations.

Findings

UK case study

The UK public sector has been shaped according to the market and the philanthropic welfare convention over recent decades toward smaller government, austerity, and voluntary assistance, particularly under Prime Minister Cameron’s Big Society project that encouraged charities to fill institutional voids (Ferlie et al. 2019). ALMPs are designed to move social insurance recipients, or those at risk of moving onto social insurance, to encourage people into employment. These programs are commissioned through public tendering and have often used payment-by-results (PbR) contracts since the 1990s. PbR contracts combine market and entrepreneurial rationales to transform social service providers into businesses that compete in state-coordinated quasi-markets.

The UK is a highly centralized state (Pollitt and Bouckaert 2017), and the central government is the dominant driving force behind the dissemination of SIBs. Central government departments such as the Department for Work & Pensions, or the Department for Education subsidize SIBs through multi-million-pound outcomes funds. One of the key financial instruments – a classic compromising device in the sense of Boltanski and Thévenot (2006) – in the UK are rate cards that define outcome-focused price tags such as for “Improved attitude toward school, £700,” or “Sustained employment, £2000.” The state defines provision in terms of “outcomes” rather than interventions, echoing the entrepreneurial convention and NPM reforms of the 1990s, which are administered to “de-risk” investor capital by allowing the “front-loading” of payments (financial convention). The front-loading of payments became necessary in the UK because the early SIB model proved unattractive for investors, as they were required to wait for their payment with a chance of 100% loss at project close, as in the first Peterborough pilot. The introduction of rate cards allowed more regular and calculable streams of money, allowing SIBs to become a valuable financial asset (Williams 2020a). In the classic market model, the price of the good is determined by the marginal cost of production. By using a rate-card, the government entirely delinked market prices from the cost of production compared to quasi-market PbR contracts of the NPM era which still anchored payments around the cost of service-delivery. As part of the UK Youth Engagement Fund, the rate card functioned to:

Support the development of the social investment market, build the capacity of social sector organizations and contribute to the evidence base for SIBs. (Ronicle and Smith, 2020:4)

Youth Engagement Fund Aims, 2014

The rate card is a central-planning government tool that – one could argue – resembles in parts the logic of the full employment convention. The difference is, of course, the subsidies do not support high quality jobs, but the “SIB economy” (Williams 2020b) with its own professional structure, knowledge schemes, and investor profits.

In contrast to the non-UK sites in our sample, valuation struggles (Williams 2020a) are relatively absent at the local level in the UK case. This is because investor return rates, evaluation methodologies, and value for money calculations, are decided by central government a priori in a top-down manner and imposed as a condition of accessing outcomes funds. Whilst some local government and provider actors in our case critiqued the potential for financialization and investors profits (financial convention) inherent in the SIB idea, they viewed the SIB as a means to an end to provide the youth employment and training services needed (civic convention) in the absence of other funding options. One interviewee from the local government stated:

[This] project for us had a couple of objectives: one was it brought some additional money in to do something that we would probably otherwise want to do anyway… The other element… was if we get in there now, we [learn] how to do these things [in preparation for] the next round of money [that] comes out from government.

Local government commissioner, UK

In the case we studied, the central government funding structure incentivized the local government actors to think more entrepreneurially about how to access state funds. Rather than local government competitively tendering local services (market convention), these actors were instead encouraged to collaborate locally with promising providers and a local investor and to compete with other municipalities over central government money to secure funds that would not otherwise be available. Furthermore, the investor was enticed by high rates of return:

The interest rate was slightly higher than we would ordinarily charge anyway, so we put our money out at 6.5% in the main, and all the Social Impact Bonds we looked at were all offering between 7% and 10%… we've got the comfort of the higher interest rate.

Investor, UK

In fact, the final outcome-payments were significantly higher than anticipated, motivating this investor to share these with the service provider as an extra-contractual payment at the end of the project. This payment could be interpreted as a philanthropic or communitarian gesture (or even a sign of guilty conscience due to perceived overpayment), though it was framed in the language of the financial convention as a “financial bonus.” The financial success was in a small part based on practices like “creaming” and “parking,” that is the selection of less needy clients in the SIB project and the limiting of assistance to clients unlikely to achieve additional payable outcomes. Some front-line workers felt these practices went against their professional ethos, indicating a value conflict between the civic ethos and the entrepreneurial ethos. Some staff had to learn to detach themselves from their civic or professional ethos of helping participants as much as possible and apply a more entrepreneurial cost-benefit ratio calculation when working with their clients.

The UK case demonstrates an interesting turn from a quasi-market for social service providers toward a quasi-market for municipal projects, where local governments, charities, and service providers find common ground and collaborate to access central state money. The top-down orchestrated funding structure is a rather rigid administrative central-planning tool installing the state as an “outcome buyer” and subsiding the national SIB industry (Williams 2020a). This structure fostered solidarity on the local level, such as the sharing of financial gains between the philanthropic impact investor and the service provider. Local-level austerity imposed by the central government was circumvented by local government and provider actors accessing central government funds earmarked for SIBs. Outcomes funds and rate cards are imposed upon local government and nonprofit providers – directing them to deliver prescribed interventions through new partnerships which include investors who receive high rates of return at little risk. On the ground, the SIB creates a conflict between a civic and an entrepreneurial ethos handled by social workers in their daily professional life.

Netherlands case study 1

Dutch public management reforms through the 2010s emphasized an increased focus on outcomes delivered through networks and collaboration (Jilke, van der Voet, and Van de Walle 2016; Pollitt and Bouckaert 2017). SIBs fit very well into this landscape, and it is no surprise that the Netherlands is amongst the more active European SIB countries. In both our Dutch cases, local government actors were interested in experimenting with new approaches to labor market integration and fungibility between benefit and programming funds motivated SIBs as a long-run cost saving strategy. Despite this relative pro-SIB environment, in contrast to the UK, there was no preexisting SIB or outcomes fund architecture at the national level. This means that contractual details and the respective impact and business model and valuation struggles (Williams 2020a) need to be negotiated case by case.

The first Dutch case was a very entrepreneurial SIB in many respects. It was a municipal level project that centered around a small group of entrepreneurially minded actors influenced by international ideas about social finance potential and UK SIB learning. Together with an expressly entrepreneurial and profit-minded provider organization, local government actors sought advice from bankers, management consultancies, and social investors about how to reform traditional public service delivery. These actors spoke of existing misaligned incentives, inefficiencies, and inabilities to improve employment programs, and articulated the belief that local government culture ought to be challenged and improved through a focus on impact:

We want to change the culture in local government, a culture which is now still a lot based on input and throughput but not on output and impact… So, we want to use Impact Bonds to strengthen the culture far [to create better impact than] ever before.

Investor, Netherlands

Informants not only criticized the public sector for its lacking entrepreneurial attitude, but also for how large – privatized since the 1990s – exploitative employment agencies acted like “cowboys” and lamented a lack of trust between municipalities and ALMP providers.

The Dutch situation is thus a simultaneous criticism of failed market and failed civic welfare conventions. The SIB mechanism, with its call for evidence and empiricism, is presented as a corrective due to its focus on “outputs and impacts,” fostering entrepreneurialism and reestablishing civic qualities in the sense of a sound handling of tax money and public interest that was endangered by privatization and the market convention.

The provider organization was the largest investor of the SIB (40% of all funds) with the CEO investing their own pension as a mark of commitment. The provider suggested that the SIB model radically altered the power balance in terms of contract negotiation in the direction of providers with the investors at their side compared to conventional contracts:

You have a two against one situation [in SIB negotiations with local government] so it’s a whole different ballgame.

Provider CEO, Netherlands

The provider and the investor were empowered by the SIB process to negotiate quite favorable conditions that institutionalized creaming and parking (Hevenstone et al. under review). For example, they selected more work-ready clients through a larger guaranteed number of referrals as well as by not officially enrolling those who appeared to be less work-ready in the first few weeks.

Significantly, the evaluation model designed by a local analytics firm and used by the evaluators to monitor the program and allocate outcome payments was problematic. The control group selected from administrative data was not well matched to the client group; there was significant missing data for the client group making the model inaccurate; and prediction equations for counterfactual outcomes reported in auditing documents were incorrect. In aggregate, these errors likely over-estimated expected counterfactual benefit durations, thus over-estimating program impact and biasing contracts to higher payments:

We didn’t have [the best] data so… the vice mayor, our director, he said ‘well we will just work with something because otherwise you never get the perfect model’, we now have proof that the Social Impact Bond comes [from] all parties working together, getting their strength, and improving… but if [we] do [a] Social Impact Bond again, then there have to be huge improvements in the measuring model.

Local Government Commissioner, Netherlands

The evaluators continued to use the flawed and criticized model to the end, even though they were aware of those limitations outlined to them by the analytic firm and also aware of their own limitations in understanding and implementing the statistical model. This led to likely over-payment of the investors and the provider, who had invested both organizational and private funds in the SIB. All actors involved were aware of the civic accountability problems and framed the project overall as a “learning” experience to be avoided in the future. In this case evaluation methods were not neutral and the entrepreneurial spirit and desire to make this pilot project a success overrode civic and market standards of transparency and accountability, which are usually guaranteed by public tendering programs, characterized by tighter administrative control measures. These issues around measurement and payment represent a conflict between the civic and the entrepreneurial conventions. The civic convention requires objective impact estimates to know whether services are effective, but entrepreneurial creativity can undermine such approaches.

The conventions approach highlights how the three main actors – investors, providers and local government commissioners use the SIB to challenge the prior hybrid market welfare and civic convention dominance, altering the established market-led commissioning process of public procurement at the local level – delivering a SIB shaped principally by entrepreneurial conventions leading to high private profits for the provider, who invested their own private capital. The agency of the provider organization was enhanced through partnership with investors enabling them to collectively negotiate a contract that delivers significant financial returns and relatively low risk. As with the UK case, there was no competitive tender for the social service provider contract. This ran counter to standard procurement practices in the Netherlands (market convention) and was justified by the experimental and innovative status of the SIB (entrepreneurial convention).

Netherlands case study 2

The SIB valuation struggles (Williams 2020a) played out very differently in our second Dutch case compared to the first one. In this case the civic and, in some sense, the full employment convention prevailed over the entrepreneurial and the financial conventions. However, many of the motivating concepts identified in our first Dutch SIB held true for this one. For instance, senior local government figures wanted to change the modalities of service funding toward more results-based payments and instigate a change in welfare provision from a market and civic convention toward a more entrepreneurial and financial one in which investors and providers partner up and deliver services through pay-for-performance contracts.

In this case, the local government was more cautious in securing the civic promise of the SIB having learned from our earlier Dutch case. The local government ensured the investors take on the risk of failure for the intervention, while enabling experimentation. This SIB promised a high rate of return to investors (financial convention), but this was balanced through a methodically rigorous evaluation regime that measured SIB cohort performance against a comparison group that was better matched (or perhaps even slightly biased in the reverse direction with the controls slightly more work-ready). Further, the local government ensured the service provider would not work with “low hanging fruit” (i.e., work-ready clients), to prevent unjust investor profits. The investor commented on this conflict of interest between the service provider and the local government:

[T]here was friction between [the local government] and [the provider]. [The local government] didn’t want to send [the provider] ‘low hanging fruit’ because it’s expensive, I think it’s their most expensive intervention on paper and you don't send people, you think will get a job anyway and [the provider] said ‘OK but… you are sending the wrong people.’

Investor, Netherlands

Furthermore, the project was conceived upon the idea of a flourishing economy in Germany, which could absorb skilled unemployed vocational workers from the Netherlands (full employment convention). The intervention was dependent upon the supply of Dutch clients with a reasonably high level of industrial qualifications to be referred into the program:

[W]e need technical people, electricians, builders, welders with a [good] schooling level… they have also [have] a driver’s license because you have to drive to work, they have to speak hopefully a little bit of German – when they don't speak that, we can manage that, we can do that – and they have to be… healthy.

Provider, Netherlands

Unemployed Dutch people were less motivated to move to Germany for a job as the Dutch economy picked up over time. This was used as a justification by local government actors to end the project earlier than planned with detrimental cost implications for the investors (including the provider). The referrals dried up and the quality of clients declined – exacerbating the “low hanging fruit” issue and making it harder for the provider to hit their targets. The local government commissioner was clear about their detached and financially cautious view of the SIB idea:

[W]e want to try new stuff with other people’s money… if it doesn’t work, we don't lose taxpayers’ money, if it works, great, we can either do it ourselves or continue the Social Impact Bonds or we found a new solution for a huge problem.

Local Government Commissioner, Netherlands

The commissioners here took the promise that SIBs should transfer the risk from the state to the investors seriously in contrast to our first Dutch case. Likewise, the provider organization was much less entrepreneurial in this case (the provider reluctantly investing 5% of the total investment of this SIB at the behest of the other investors). The investors bemoaned the lack of flexibility, better financial support for the project and inclination to make it a success – resonating with the entrepreneurial and financial conventions. In the absence of the kind of strong relationship between investor and provider seen in the previous Dutch case, the local government commissioner, with an understanding of an economy which might pick up or not – cognizant of the full employment convention – was relatively relaxed about terminating the project early suggestive of civic convention thinking.

The two Dutch cases show how the SIB evolves through conflicts and compromises between diverse welfare conventions within the same national context delivering very different actor constellations and negotiated rationales. The SIB, with its clear dominance of the entrepreneurial and the financial welfare convention needs to align with the existing institutional landscape and discourses. In the Netherlands, the SIB sought to solve the excesses of marketization and privatization of the 1990s and offer an alternative to classic civic social welfare provision. We thus saw the attempts to overcome the market rationale of fair competition (there was no public tender involved in both Dutch SIBs), as well as the promise to install an entrepreneurial spirit to solve welfare state issues. It seems that this more entrepreneurial convention jeopardized norms of transparency, fairness, and accountability. The promise of rigorous impact measurement, which in the end secures the civic taxpayer interest in a SIB set-up, may imperil the attractiveness of the SIB for investors. Thus, the civic-financial compromise of the SIB proved difficult to realize. Each Dutch case either disappoints the civic public interest, or the investor interest. It is difficult to align the two.

Switzerland case study

The Swiss political and administrative culture is often described as corporate (communitarian in the WCA typology) – with a high degree of civic-market compromise that prizes stakeholder collaboration between government and business searching for consensual solutions. Switzerland is federal and highly decentralized with a very responsive direct-democratic system (Dahmen, Bonvin, and Beuret 2017) and significant local political accountability. With respect to the WCA, what stands out is the communitarian and collaborative style that solved the valuation struggles (Williams 2020a).

Unlike our Dutch cases – that are characterized by a three-way collaboration between local government, provider, and investor – the Swiss case is a real collaborative “partnership” between local government, and single investor backed by Swiss companies that serve as potential employers for the refugee integration program. The social service provider was commissioned competitively after the key terms of the SIB had been negotiated by the local government and the investor. Competitive tendering draws upon the market welfare convention, which is considered important in this case (different to the other cases of this study) to select the best social service provider competitively.

The drive to bring the SIB idea to Switzerland came principally from one social investor. Conversations between the investor and the local government identified the topic of refugee employment as a suitable vehicle for SIB experimentation. There was a shared desire by the investor and the local government actors to place a Payment-by-Result element into the provider contract. Although this was small (1.5% either way) it highlights an aspiration to further entrepreneurial and financial incentives through the SIB in the social service sector:

For us it was also important that we do have these kind of bonus elements, even though symbolic, but the discussions were not symbolic at all, because very heavy discussions with the social institutions, whether they should move to more outcome based [contracting] and I think this discussion was already worthwhile and it also led to, well, to an adaptation of the contracts the canton [] has with certain institutions and they say ‘well… there should be some more outcome based parameters within a contract’.

Investor, Switzerland

Some local government actors were concerned that there may be public disquiet about the idea that investors might make a financial return “on the back of refugees” (civic and/or communitarian critique) and therefore negotiated a SIB in which any return to the investor was capped at a low rate (compared to the UK and Netherlands cases). However, the rate of return was rather attractive given the negative interest rates in Switzerland at the time. Still, there was a recognition that the level of risk for the investor was too low (in terms of the financial convention) leading a government actor to question if this really was a SIB as he understood it:

I have a financial background… for me, [the SIB] was exciting… it was something new… but I recognized at one point that it’s not really a Social Impact Bond, because … the risk [to the investor of any financial] loss is very small.

Local Government Commissioner, Switzerland

During the project run-time, local government and investors agreed to recalibrate initial valuation and measurement issues that improved the likelihood of ‘successful’ outcomes. We have described this as ‘institutionalized gaming’ elsewhere (Hevenstone et al. under review). This SIB explicitly targeted the easier cases:

In the non-SIB programs, we have a lower education level from the clients, you see. I think that’s the major difference… We take everyone … And in the SIB program clients have better education, better linguistic proficiency, and stuff like this. So, for me it’s absolutely clear why they have a higher outcome.

Provider, Switzerland

The selection of the better educated and more motivated refugees indicates the collaborative and cooperative style of the Swiss SIB through which protagonists aim to find common ground delivering a rather symbolic SIB (as noted by the Swiss local government commissioner above). Likewise, the initial shared commitment to rigorous measurement was softened over time so that the scheme would be considered successful. No one really bore a financial risk in this SIB configuration. Civic and communitarian morals led to a very low-stakes SIB that ultimately disappointed the investor – who had proposed to go much further in taking on greater risk for more reward (financial critique). Ultimately, in the Swiss case the decisive feature was the communitarian element enabling actors to manage emergent problems flexibly and based on reciprocity and trust between themselves, so that no one lost out.

Germany case study

The German welfare system has been classified as conservative (Esping-Andersen 1990), featuring some parallels with Switzerland in terms of corporatism and federalism. A series of welfare reforms in the 1990s were designed to increase marketization and competition in Germany (Bode 2008). Nonetheless, social service provision remains dominated by so-called “Freigemeinnützigen Träger,” nonprofit organizations (Grohs et al. 2015), which are politically independent, but state funded. In the last decades, the financial landscape in Germany for social projects has become hybridized, due to a Social Ministry-run European Social Fund-infrastructure, for example. Thus, the German welfare system is composed of array of diverse financing models.

The investigated case was meant to be “the first continental” SIB in Europe. There was therefore no national SIB infrastructure in Germany. The key agents advocating for SIB experimentation were philanthropic investors and an intermediary organization aiming to build the capacity of a domestic social impact investment network to challenge traditional institutional dominance of the state and the six German independent, state-funded welfare associations (“Freie Wohlfahrtsverbände”):

I do not want to have British circumstances [e.g., austerity] here in Germany in our welfare system, but it’s not so that in Germany everything is fine and in Britain everything is shit – no! It’s in between. And if another system has good aspects, why not [learn from] them? So, [at least] there should be a little bit more [of an] entrepreneurial point of view, because a good entrepreneur always is looking for good aspects from others to learn about.

Investor, Germany

This entrepreneurial rhetoric chimed with a desire on the part of the local government commissioner to use a SIB approach to pilot impact monitoring in service provision. In this case, the valuation struggles (Williams 2020a) took place amongst the investors, the intermediary, and the local government. Like the Swiss case – the provider was left out of these discussions:

The only negotiation which really took part was between [the investor] and [the local government]. And that was not really a negotiation. I think it was more a kind of bargaining…for us, [the SIB] was [just]another source of money.

Provider, Germany

The entrepreneurial push at the heart of this SIB experiment can be interpreted as a criticism of the civic-communitarian welfare state provision prevailing in Germany. This is a rather abstract and detached criticism not well anchored in the concrete institutional realities of welfare provision in this particular case. The social service providers that were recruited in this case, were familiar with European Social Fund-projects with competitive public tendering procedures, with a target figure-based control output performance system already in place. For the service providers, the motivation to participate was the promise of a “revolving fund” to be established in case of success.

The negotiations around this SIB exerted a great deal of energy in ensuring the criterion of “hard-to-reach” juveniles to make sure that the SIB would not focus upon the easier cases (low-hanging fruit) and would meet the requirement of “additionality.” Even though these criteria were defined from the beginning, it was necessary to specify the criteria and to sharpen how they were defined technically during the lifetime of the project. For example, there were discussions such as if a missing identity card could be replaced by a residence permit to verify for age, but also harder to solve issues, such as asking for a proof of the job center that these juveniles were lost and could not be reached by them adding increased burdens on the service providers in terms of extra paperwork. Furthermore, for the service providers, the target of 20 juveniles to be placed for at least 9 months in work or an apprentice scheme was reported to be somewhat artificial and restrictive from a civic social worker professional perspective.

Local government actors were determined that the SIB would not lead to high investor returns. The desire to experiment with a SIB generated significant conflict between the civic and the financial welfare conventions, in the sense that having a financial return on investment on social policy provision – whilst key for the investors, was resisted amongst local government actors and led to high risk, low return investment model:

[T]he entire risk in this case was on the investor side… Because in fact it was all or nothing. So, either you reach the 20 people in our case and if you only reach 19 then you get nothing. Charming on the one hand from a state perspective. On the other hand [for investors], difficult, of course.

Local Government Commissioner, Germany

Even though this was frustrating for the investors, like in the Swiss case, they accepted it in this instance to make the SIB happen. Interestingly, the rather low financial return was framed as an “inflationary adjustment” in official communications to avoid the impression of private investor profits in the social sector (showing an important role of the civic convention countering the financial convention in project design).

In the German case, the conventions approach highlights how the three main actors – investors, intermediaries, and local government commissioners initially collaborated closely to experiment with a more entrepreneurial approach to public sector commissioning. The role of the provider was secondary. Local government apprehension around the financial element of the SIB idea and a German ordo-liberal frugal state tradition led to a very high-risk, low reward SIB configuration that ultimately disappointed the investor – who had proposed to go much further in taking on greater risk for more reward – the civic welfare convention proved hard to challenge and the financial convention hard to establish.

Discussion and conclusion

The uneven dissemination of SIBs across countries, most of them being institutionalized in liberal welfare states (the UK and the US), demands public management scholarship to take institutional trajectories and political economies into account. However, national welfare approaches (Esping-Andersen 1990) struggle to account for the local and hybrid constellations of SIB implementation on the ground and classic institutional theories may overlook the relational aspects of SIB development within local settings. This article therefore used the WCA (Chiapello and Knoll 2020) as a middle-range theory to compare SIBs as an example of international public management reform. The WCA combines an interest for technical details and managerial complexities, which it views as shaped through struggles over questions of morality and justice defined by contradictory welfare conventions (e. g. the philanthropic, communitarian, civic, entrepreneurial, or financial). Western welfare states have undergone a high level of hybridization and feature diverse principles of welfare organization that co-exist more or less peacefully within countries. SIBs, aiming to insert an entrepreneurial and financial investor rationality into the world of social service provision, become implemented into these far from homogenous institutional worlds and must align with existing rationalities – thus developing in diverse ways.

The WCA highlights SIBs not as a necessary hegemonic project of simple diffusion, but as a difficult and contested process of implementation. The SIB is promoted as a criticism of the market reforms in the 1990s, particularly in the Netherlands, were it promises to solve the accounting problems and cowboy-like mentality of private service providers and the inertia and lethargy of the civic welfare state. As a solution it provides an impact measurement narrative clothed in the promises of an entrepreneurial culture. The consequence of this was an exacerbation of transparency and accounting issues leading to severe over-payments in our first Dutch case. We find a tendency in our comparative case study that competitive public tendering – indicative of the market welfare convention – is transformed by the SIB, presaging more networked, relational, and entrepreneurial approaches. At the same time, we show that SIBs that place a great emphasis on methodical rigor and real “risk transfer” between the commissioner and the investor, as in our second Dutch and our German cases, frustrate investors and endanger the SIB as a feasible instrument. These kinds of valuation struggles (Williams 2020a) between local governments and investors can only be prevented if the financial gains are kept low and real risk and returns are kept at a minimum level, such as in our Swiss case, or if a central government top-up fund takes the risk away from the local SIB partnership, such as in the UK case. In the UK case, high investor payment was not a threat to the local SIB collaboration. It rather led to a solidaric financial extra-contractual bonus from the community-oriented philanthropic investor to the local service provider. All in all, it seems that social service providers can gain from the SIB, which opens new funding lines and potentially also greater flexibility. But it may very well also lead to an increase in bureaucratic burdens, when commitments to methodical rigor are taken seriously, such as in our German and second Dutch cases. The biggest consequences of the SIB agenda potentially lie in the reorganization of funding lines between local governments and local providers (Williams 2020a).

Finally, we offer some reflections on the usefulness of the WCA. We found challenges working with the WCA. One is that the institutional world is complex and difficult to categorize within a simplifying schema of eight welfare conventions. This becomes visible, for example, in the ambivalence of the entrepreneurial category, which is in some cases an argument for methodical rigor and documentation, and in another constellation an argument for flexibility and creativity. One, of course, can explain these differences by the notion of the compromise (rigor stemming from a stronger role of a civic duty for tax payor money and flexibility from a financial need for return-on-investment, in some cases), but such a play with conventions also adds a layer of complexity to the research process. Nonetheless, this additional layer can be productive and thought provoking. We found, for example, the financial technology of the industry subsidy stemming from the repertoire in the full employment convention in the rate card technology driving the UK social impact investment market. It is interesting to see state subsidies emerging in the context of financialization (and not to support high quality jobs for the working class as defined in the full employment convention). Still, such a struggle with the interpretation of the WCA may also point to a disconnect between technology and discourse, which is argued to be closely aligned in the WCA approach (Chiapello and Knoll 2020). Furthermore, the focus on compromise and hybridization prioritized by the WCA may endanger losing sight of bigger reform trajectories such as financialization (Lake 2015; Tse and Warner 2020).

More positively however, the WCA highlights transformations such as financialization do not unfold in an unilinear and straightforward way. They evolve through conflicts, compromises and relational dynamics on the ground and may change their appearance over time. This perspective demonstrates how new tools and solutions emerge from the attempts to stabilize the core valuation struggle of the SIB – the difficult alignment of civic and financial interests (Williams 2020a; Maier et al. 2018). A strength of the WCA is that it provides a differentiation of broad concepts, such as “Capitalism,” “Neoliberalism,” or “Liberalism.” By drawing a distinction between the market and the entrepreneurial conventions, for example, it underlines how these may lead to different policy instruments and interventions with different outcomes and qualities for social service provision. It is an approach that reflects the developments of conceptual hybridization in welfare states (Bode 2008) and helps to link comparative public sector reform research focused on management complexity (Pasi 2014; Tan et al. 2021) within an understanding of political economy and historic fights over the organization of welfare. Ultimately, this offers some theoretically informed explanations for the diversity of SIBs within and across different countries.

Ethics

This study received ethical clearance from the ethics review panel at the London School of Hygiene and Tropical Medicine in 2019. Ref: 16119.

Acknowledgments

We thank the Swiss Network for International Studies for funding for this project. We also thank the anonymous peer reviewers for their very helpful comments and suggestions on this article.

Additional information

Notes on contributors

Alec Fraser

Alec Fraser is a lecturer in Government and Business at King’s Business School, King’s College London, UK. His research centers on the use of evidence in policy making and practice across the public sector. He has researched the development of Social Impact Bonds and Outcomes-Oriented Reforms in public service delivery in the UK and internationally over recent years.

Lisa Knoll

Lisa Knoll is a deputy professor of General Sociology at the University of Paderborn in Germany. Her research centers on economic sociology with a particular focus on sustainable and social impact finance. She is an expert in French pragmatist social sciences such as the French Convention School of Economics.

Debra Hevenstone

Debra Hevenstone is a professor in Social Work at the Bern University of Applied Sciences, Switzerland. Her research focuses on social policy design and quantitative methods. She has led projects examining unemployment insurance design and the family, an international comparison of Social Impact Bonds, and an analysis of part time work and the motherhood age gap.