The Governance of Social Risks: Nurturing Social Solidarity through Social Impact Bonds?

ABSTRACT Despite having been around for a decade now, Social Impact Bonds (SIBs) – payment by result contracts funding social programmes – are still a niche instrument. Constituting but a fraction of the overall impact investment sector, they were expected to grow much faster and augur a new model of pursuing social policy objectives. Whilst this has not yet occurred, they nevertheless continue to benefit from a great degree of political support and academic interest. But outside of the practitioner-focused literature, the scholarship investigating SIBs has largely identified financialisation and the erosion of social solidarity as the main dynamics underpinning this development. This article argues that it is important to also attend to SIBs as expressions of transformations occurring within the design and pursuit of social policy objectives. By looking at SIBs as a form of governance of social risks, the article argues that SIBs nurture their own forms of social solidarity. Based on three distinguishing tenets of SIBs, three types of solidarities are emphasised: inter-temporal, cross-sectoral and risk-insurance solidarities. Whilst these can spur social inclusion, innovation and collaboration, the article discusses how they can also be spurious and can come undone.


Introduction
Towards the end of 2018, at one fell swoop, the UK government announced it would fund 22 largescale social programmes across the country seeking to provide services as diverse as work experience for the young, tailored health plans for older people or support for former alcohol and drug dependent people as they reintegrate into their communities (DCMS 2018). But instead of providing the money for the programmes upfront, the government announced it would fund them through an instrument called Social Impact Bond (SIB). Under this arrangement, the service deliverers undertaking the social programmes would receive their capital from private investors, who would, in turn, be repaid by the government, but only according to the degree of social impact achieved. If the programme failed to reach the minimum targets for social impact created, the investors stood, in theory, to lose all the capital invested (Social Finance 2009).
Far from disappearing from the market as some media coverage would have led to believe (Pybus 2017, Kay 2018, Murphy 2018, SIBs continue to make headway throughout the world. More than 150 projects have already been launched globally (with more than a fifth of that only launched since 2018) and more than 70 are currently in development. 1 Ever since David Cameron, the former UK Prime Minister, contended in 2013 that social impact investing 'can transform our societies by using the power of finance to tackle the most difficult social problems' (Cameron 2013), SIBs have benefitted from a great degree of both political support but also academic interest. Within the academic literature, SIBs have been the subject of papers stemming from fields as diverse as public policy, sociology, geography, anthropology or media studies. 2 But the verdicts coming out of these studies are mixed, with the lion's share of the literature construing SIBs as pernicious artefacts of processes of financialisation, which, some argue, render social policy into a new frontier of capital accumulation (Bryan and Rafferty 2014, Dowling 2016, Tse and Warner 2020 and the target population of SIB programmes into commodified investment propositions (Neyland 2018, Sinclair et al. 2021. While SIBs can be perceived as being products of financialisation, that is, of broader processes of increasingly valuing non-economic phenomena through the lens and instruments of finance professionals (Chiapello 2014), this article will make the case that, by gauging SIBs as particular approaches to governing social issues, this narrative can be meaningfully complemented by an account that also looks at what emerges in the wake of financialisation. This paper, therefore, scrutinises SIBs precisely as expressions of a certain form of the governance of social ills. Against accounts that see SIBs as winding back the welfare state and dissolving the bonds of social solidarity, it will look at how three tenets of this governance structure can open up new avenues for social solidarity, but also how these can be spurious and come undone.
The relevance of the notion of 'governance' for the social policy has been emphasised for some time now as reflective of a specific historical re-organisation of how social policy objectives are pursued, particularly through the co-ordination of various forms of public-private interactions (Osborne 2006, Newman and Clarke 2009, Rolfe 2018. The 'governance' model is said to be synthesising two major paradigms that dominated debates during the 1980s and 1990s regarding the appropriate providers of publicly funded (social) services: the government paradigm and the market paradigm (Frahm and Martin 2009). While the former has its source in the old Weberian model of a highly centralised, bureaucratic and hierarchical state system characterised by the domination of principles of impartiality, objectivity and the authority of public service professionals in the delivery of social services (Torfing and Triantafillou 2013), the latter drew on private business expertise and market principles in order to foster competition, non-governmental service delivery and costumer choice (Klijn 2012). Governance, on the other hand, preserves elements of both paradigms but also adds in a dose of network theory as the structuring principle of actor interaction (Frahm and Martin 2009). In other words, in the governance paradigm, there are multiple interdependent actors implicated in public policy (with the state and private actors sitting alongside NGOs, community organisations, volunteers etc.) and multiple policymaking processes (e.g. stakeholder participation) contributing to service delivery (Daly 2003, Osborne 2010. Some have argued that within this paradigm the state's role has shifted to that of ensuring 'meta-governance', that is, managing the sites of interaction and the relationships arising from the network by 'steering' and not 'rowing' (Gilbert andGilbert 1989, Newman andClarke 2009).
This article seeks to build on these insights, emphasising a specific dichotomy between the state's 'stepping back' as the principal provider of welfare whilst 'stepping into' the network of welfare actors emerging after its withdrawal. Against the literature that assumes that SIBs mirror a process of rolling back the welfare state (Joy and Shields 2013, McHugh et al. 2013, Dowling 2016, it will argue that the governance model that SIBs reproduce in fact enhances the ability of the state to 'govern at a distance', particularly through the focus on managing (social) risk. At the same time, based on three tenets of SIBsdiscounting of social risk, alignment of interests and de-risking governmentit will argue that SIBs nurture their own forms of social solidarity. In particular, three corresponding types of solidarities are emphasised: inter-temporal, cross-sectoral and riskinsurance. Seeing SIBs through this lens allows both for an endogenous form of critique that grasps them primarily as a form of the governance of social risks and for the teasing out of openings and opportunities for social solidarity and emancipation that they may engender (Table 1).
The paper continues as follows: first, it will briefly discuss SIBs' reception in the academic literature, highlighting how this reception conceals specific shortcomings and why it would useful to move beyond the financialisation narrative and understand SIBs as attempts at nurturing social solidarities in a different manner. It then proceeds to discuss three distinguishing tenets of SIBsthe discounting of social risks, the alignment of interests and the de-risking of governmentand to highlight how reconceptualising SIBs in this fashion uncovers particular consequences for (social) risk governance that are not normally captured by construing them as products of exogeneous dynamics.

SIBs beyond the Financialisation Narrative
First mentioned in a piece in Economic Affairs (Horesh 2000), SIBs still lack an agreed-upon definition. Social Finance UK, the institution that launched the first SIB and coined the term, defines SIBs as 'a new contracting and financing mechanism … [that] seeks to drive significant non-government investment into addressing the causes of deep-rooted social problems with returns generated from a proportion of the related reduction in spending on acute services' (Social Finance 2009, p. 2). The Commissioning Better Outcomes Fund, a fund set up by the Big Lottery Fund to finance and support the development of SIBs, defined them as 'a type of Payment by Results (PbR) contract, where the finance needed to make the contract work is provided by social investors rather than by service providers' (TBLF 2014, p. 1). The UK Department for Digital, Culture, Media and Sport defines SIBs as 'an innovative funding model, where a socially-minded investor provides up-front funding to an organisation such as a charity or social enterprise to deliver a service. Once this service achieves results, the government will make payments and the social investor will be reimbursed' (DCMS 2018). Other organisations involved in SIBs may emphasise a different dimension of the programmes, depending on what they see as the central mechanism or objective of a SIB, but it can be argued that at least three aspects common to many extant SIBs distinguish them from other, conventional social programmes: the discounting of social risk through financing preventative programmes that tackle future social issues, the alignment of interests through involving a wide variety of stakeholders in the design and delivery of programmes, and the de-risking of government through offloading funding and delivery risks to said stakeholders.
That said, part of the problem with defining SIBs is that even the term itself, Social Impact Bond, is misleading. SIBs are not 'bonds' in the financial sense, given that the interest (i.e. return on the bond) is not fixed but contingent on the performance of the programme funded. This misunderstanding creates the impression that SIBs are simply, or primarily, a funding mechanism meant to attract private capital seeking to make a profit through the delivery of social programmes. This conception was reinforced by the initial enthusiasm surrounding the potential for the creation of a secondary market in SIBs, where investors would be able to buy and sell SIBs as any other financial asset traded on capital markets (Salamon 2014, Nicholls andTomkinson 2015). Given all the unforeseen barriers, this enthusiasm has now been dialled down, but SIBs are still suffering from a 'branding issue'it has even been suggested that proponents and issuers of SIBs should prioritise the search for an alternative term that 'better describes SIBs as contracts for outcomes with a social purpose' (Wooldridge et al. 2019, p. 7).
However minor, this branding issue might even play a role in the SIBs' reception in the academic literature. Outside the world of policy practitioners, the lion's share of the academic literature construes them as being part of wider processes of financialisation (Joy and Shields 2013, Dowling and Harvie 2014, Dowling 2016, Harvie and Ogman 2019, Tse and Warner 2020, Sinclair et al. 2021. In this view, a SIB is what the name advertises: a bond, that is, a new financial instrument employed in the Alliance-formation and social citizenship social policy sector, indicative of 'the language of finance […] invading public policies' (Chiapello 2014, p. 30) and 'financial ways of calculating […] becoming more pervasive socially' (Bryan and Rafferty 2014, p. 895). This bondwhich in some cases is described even as a derivative, whose value is derived from the performance of the underlying service provider in achieving the agreedupon outcome (Lilley et al. 2019)creates new opportunities for capital accumulation for financial investors who now target social policy programmes and thereby aid in the financialisation of social policy. In this context, it is assumed that financialisation 'collateralises the social' through the packaging of social programmes into financial instruments and 'disciplines' the recipients of the programmes by exposing them to the pressures of financial market competition (Bryan andRafferty 2014, Lavinas 2018). In doing so, it 'colonises the social' through the imposition of financial forms of valuation upon non-financial activities (Chiapello 2014). But looking at how 'the social'i.e. social impact/outcomesis gauged through the various valuation processes available to investors and service deliverers reveals that it is not entirely clear how far these financialised instruments actually travel within 'the social'. Even measurement frameworks like Social Return on Investment (SROI), constructed explicitly to mirror the Return on Investment (ROI) performance measure (which is widely used in the private sector to gauge an investment's profitability) but targeted at capturing extra-financial value creation relative to investment, only seek to monetise social outcomes based on stakeholder preferences, and not necessarily investor's (Arvidson et al. 2010). Indeed, according to Social Value UK (2019), a fundamental principle of SROI is to engage stakeholdersparticularly the target populationin the early stages of the impact project in order to inform what gets measured and how it will be included in the accounting for value process. Furthermore, when it comes to monetising social outcome production and calculating the SROI ratio towards the end of the impact project, it is again the target population that is supposed to take primacy (NEF 2009). While some financial proxies are easier to identify and calculate (e.g. the proxy for an improvement in mental health could be the cost of counselling sessions), others rely more on capturing value from the subjective perspective of the target population (e.g. asking how much a person would be willing to accept in compensation for an increase in crime as a proxy for the monetary value of reducing crime, or how far would a person be willing to drive to access a community centre and use travel costs as a proxy for the monetary value of improving access to local services).
On the other hand, to the extent that the social is 'remade', ethnographies and other qualitative studies of SIBs on the ground have sometimes argued that these programmes work to remake the target population into neoliberal subjectivities, specifically following Foucault's notion of 'entrepreneur of the self' and what he calls 'neoliberal governmentality' (Foucault 2010). For instance, the second SIB launched in London (The London Homelessness SIB), analysed by both Andreu (2018) and Cooper et al. (2016), was found to function by bracketing off pernicious social structures as a cause of homelessness and operating on the behaviour of the individuals targeted, following the neoliberal prescriptions of activating them as entrepreneurs and reintroducing them into the economic cycle as a way of 'curing' homelessness, which is construed as a failure in entrepreneurialism.
The underlying assumption here is that the move towards neoliberal entrepreneurialism of the self and individuality is undermining the solidarities that form the basis of the welfare state (McFall 2019), understood as the set of policies that allow the provision of top-down redistributive policies while embedding markets in society and ensuring de-commodification (Polanyi 1944, Esping-Andersen 1989. This narrative permeates many of the critical analyses of SIBs (e.g. McHugh et al. 2013, Dowling and Harvie 2014, Rosamond 2016, Harvie and Ogman 2019, Rosenman 2019, Sinclair et al. 2021. While the issue of the welfare state de-commodifying society is contested by some to begin with (Jessop 2002, Panitch andGindin 2013), recent studies have confirmed that welfare practices shifted significantly particularly since the mid-90s when there was a move towards what has been called the 'social investment state' 3 (Esping-Andersen 2002, Kersbergen andHemerijck 2012, Busemeyer et al. 2018). In this context, the neoliberal critique that welfare states of the Bismarkian type undermined the work ethic and created dependency was integrated into their functioning, transforming their way of framing problems and identifying solutions.
In particular, there was now an emphasis on self-investment via constant learning, orientation to the future and fostering collective wellbeing through individual success (Jenson 2012). Active employment is prioritised: social policies aid non-working people get back into the employment cycle, although they are expected to build resilience themselves by investing in a suitable stock of human capital throughout their life. Similarly, individuals are expected to provide for their own pensions (in the form of defined contribution plans and a multi-pillar system mixing pay-as-yougo and fully funded approaches) and in some cases healthcare. This, together with the mass investment culture developing as a result of the advent of the retail investor (Ott 2008), refashioned individuals into entrepreneurs taking control of their financial future and perceiving themselves as 'project bearers' who strive to become worthy of investment, that is, rated agencies (Feher 2018).
SIBs, as a development within social policy, tie into these neoliberal dynamics and are to an extent a continuation of them, but they also go beyond, nurturing particular forms of relationality that cannot always be captured in the notion of the entrepreneurial self. Indeed, far from excluding the capacity to foster social solidarities, SIBs function by reterritorialising (Deleuze and Guattari 2013) them into new configurations with different intensities. What makes this process possible is the particular type of governance model that SIBs employ, to which the paper turns now. It rests specifically on three pillars, as mentioned above: the discounting of social risk, the alignment of stakeholder interests and the de-risking of government. Each pillar has its own particularities but is also fraught with dangers.

The Discounting of Social Risk
Observing how business valuation changed at the turn of the twentieth century from being based mainly on liquidation or replacement value to being based on expected profitability, John Commons argued that this increasing preoccupation with the future ushered in an epochal change that would ultimately engross more and more domains of human activity and would change how the present was perceived and inhabited (1934). Otherwise known as 'futurity' (Atkinson 2009, Palan 2015, Beckert 2016, this phenomenon is central to how SIBs operate. Indeed, SIBs are marketed as being oriented towards future social issues, whose development is countered through preventative programmes (Social Finance 2009, TBLF 2014. This is claimed to be superior to direct government provision, which, the argument goes, is reactive (responding to a crisis when it occurs) and runs on siloed budgets and interventions (Mulgan 2011). By engaging in preventative programmes, SIBs would not only ward off future issues, but would also produce cashable savings for the central government. SIBs' proponents thus argue that by preventing reoffending, for instance, they eliminate the cost of accommodating offenders in prisons; similarly, by helping rough sleepers find stable accommodation, they eliminate the costs of publicly funded shelters or the healthcare costs associated with homelessness. Cashable savings would thus arise from the difference between the projected government expenditure on particular public services and the overall cost of an SIB programme, including the estimated impact of the intervention, savings to the commissioner and the financial return investors receive (Social Finance 2011). In the aforementioned London Homelessness SIB, for instance, the cost of homelessness was based on a number of factors like the estimated costs of temporary accommodation, reconviction costs, police time, job seeker's allowance, A&E and ambulance usage, as well as psychiatric hospital usage, and was averaged at £37,000 per person and £24.2m per cohort.
Futurity is present in the fact that these numbers are a projection of the estimated costs occurring over a five-year period, assuming a 30 per cent reduction in the number of rough sleepers and discounted at the HM Treasury Green Book 4 rate of 3.5 per cent (Social Finance 2012a). The rate, also known as social discount rate (SDR), is routinely used in policymaking, and is meant to reflect the value that society currently places on the future social costs of an issue like homelessness (Arrow et al. 2013, Groom andHepburn 2017). The use of an instrument like SDR is based on two assumptions: first, that society has a time preference (i.e. a dollar today is worth more than a dollar tomorrow, or the costs of homelessness are higher today than in the future), and second, that society will be wealthier in the future than in the present, which is claimed to lead an implicit reduction in social ills (e.g. homelessness) as future generations get wealthier and wealth presumably trickles down (HM Treasury 2018). Viewed from a different angle, the SDR weighs future societal benefits to current societal costs (that is, net present value). But what this implies is that while in a PAYG scheme like unemployment or pension benefits, it is the current generation that is bankrolling the resolution of current social issues based on current costs, in an SIB it is still the current generation that is paying (via the central government reimbursing investors if the project is successful and based on the outcome achieved) but what the funding goes to is the resolution of future social issues based on their assumed present cost. In other words, the dimension of futurity that undergirds SIBs works also to forge intertemporal solidarities between the present and future generations.
Intertemporal solidarities are also expressed through the previously mentioned notion of cashable savingsthe idea that forestalling future social issues produces benefits for the present as well. Cashable savings are one of the main pillars of the raison d'être of an SIB and contractors are required to demonstrate in their pre-feasibility assessment that an intervention produces clear cashable savings to the commissioner and/or to other public sector bodies (Cabinet Office 2013). Although some doubt that SIBs can actually produce cashable savings or that these can be gauged and measured with a convincing degree of certainty (Edmiston andNicholls 2018, Harvie andOgman 2019), SIBs still produce estimates of cashable savings and some even attempt to demonstrate achieved cashable savings. For instance, Birmingham City Council's Step Down Programme, an SIB which commenced in 2014 and was aimed at bringing young people out of residential homes into foster placements, estimated a total of nearly £2m in savings over three and a half years (Plumridge and Sebba 2018). By foregoing government-funded residential homes and remaining in stable accommodation over 52 weeks in foster homes, it was estimated that each young person saved an average of over £50,000. Although it is not very clear to what extent these outcomes are actually secured for the long term or if that was even within the scope of the programme (or indeed to what extent impact permanency is a feature SIBs take or can take into account at all), it is nonetheless reasonable to assume that cashable savings can be calculated relatively confidently in this case.
While intertemporal solidarities can be considered a distinguishing feature of SIBs, there are nonetheless numerous problems with the assumptions underpinning this type of futurity. Take the idea of the 'wealth effect': for one, the Green Book assumes that there will be an average annual growth in per capita consumption of two per cent, even though the Office for National Statistics (ONS) has estimated a growth of 1.7 per cent for the two decades between 1996 and 2016 in its timeseries of quarterly national accounts. 5 The SDR has been in use since 2003 but has never been adapted to take into account the post-Brexit downward revisions of future growth or the uncertain global economic climate which might affect it. More importantly still, the SDR is silent on the distributional effects of the presumed growth in wealth. This seems particularly salient given the increasing evidence of a widening in inequalities which has been developing since the 1980s and has been reinforced in the post-crisis period (Dorling 2015, Piketty 2017. One of the consequences of this process is particularly the increasing vulnerability and precarity of the people from the bottom of the social ladderin other words, a likely increase in social risk. In this context, a 30 per cent natural reduction in rough sleeping presumed in the Homelessness SIB seems like a particularly adventurous assumption. Furthermore, the idea of an SDR itself is problematic. The higher the rate, the more it means that people place a lower value on the future than on the present. In a recent survey of 200 economists, more than three-quarters of the people said an SDR of two per cent is more appropriate given the wealth effect, the time preference effect and the real risk-free interest rate (Drupp et al. 2018). The use of a higher SDR might make it harder to pitch a business case for investing in future social ills. But most of all, the contested nature of the SDR only reinforces its socially constructed nature, based on plausible but dubious assumptions, and whose power only rests on the facticity (Power 2021) it acquired by being used repeatedly in practice. The SDR is not only performative (MacKenzie 2008) in the sense that it creates a reality that does not 'exist', it actually creates a reality that is fraught with problematic assumptions, and the expectations it relies on are not simply neutral fictions (Beckert 2016) but are value laden and pernicious, often placing a higher weight on present benefits than future costs. As Palan says, 'in the society dominated by futurity, it is no longer possible to have any clear indications of "real" value of any asset' (2015, p. 381). The 'real' value of future social issues is in fact a political matter decided upon by the negotiation between the actors involved in a SIB, an issue to which the next section turns.
Nevertheless, this should not disparage the fact that, through focusing primarily on nascent or preventative programmes, SIBs expand the temporal horizon of the possible and spark discussions regarding the weight not acting in the present will have for the future. The notion of cashable savings indicates, for instance, that oftentimes it is more effective and less costly to focus on prevention rather than alleviation of acute problems. Through the discounting of social risks, SIBs may thus open up pathways for making the future more central in present considerations and reflecting on the real value of future social ills.

The Alignment of Interests
Part of the appeal of SIBs, according to proponents and practitioners, is that they automatically align the interests of stakeholders, particularly governments, investors and service providers around the achievement of pre-agreed outcomes (Social Finance 2012b, 2016). For instance, Bridges Ventures, a fund manager specialising in sustainable and impact investing, states that one of the distinguishing features of SIBs is that 'the contractual alignment of all three parties in delivering pre-agreed social outcomes directly aligns those outcomes with financial incentives, providing even greater assurance to investors that all counterparties are pulling in the same direction' (Bridges Ventures 2014, p. 11). In a simplistic model of outsourcing public services, the public-private contractual relationship is considered to be exposed to agency risk: that private actors are utility maximisers who act opportunistically and are unaccountable if costly monitoring mechanisms are not in place (Lambright 2009). Whereas some stakeholders in SIBs are mission-driven (e.g. NGOs and philanthropic organisations), others are driven by self-interested profit-seeking motivations (e.g. social enterprises, venture philanthropists, some high net worth individuals, impact investing banks). SIB programmes would then suffer from the same agency risks, if not for the fact that they are organised around the achievement of particular outcomes, which would presumably be demonstrated via rigorous outcome evaluations. By tying programme payments to degrees of impact achieved, the governmental agency contracting a SIB ensures that service deliverers, investors, intermediaries and other stakeholders involved have incentives that are aligned with the achievement of the maximum possible amount of impact. In line with the discourse of 'advanced liberalism' (Mitchell 2017), this network of actors is believed to work more efficiently in the pursuit of social value than bureaucratic administrations, which are perceived as siloed, heavy-handed and unimaginative. By fostering this network gravitating around measurable outcomes, SIBs therefore also build crosssectoral solidarities.
So far, there are very few SIBs in the UK that involve private capital coming from profit-seeking entities. The US, on the other hand, has made greater strides in involving market actors in SIB programmes. To take an example, the Utah High Quality Preschool Programme, an early childhood education SIB launched in 2013 for a duration of five years, sought to improve school readiness and academic performance among three-and four-year-olds (CNCS 2015). By doing so, it was expected that fewer children would be required to resort to remedial and special education services later on, and this would produce cost savings for school districts and other local government entities. The main investor in the project was Goldman Sachs, advancing a $4.6m loan to United Way of Salt Lake, the intermediary, who joined a consortium of other impact investors pushing the total investment to an approximate sum of $7 m. The repayment structure was organised as follows: if a child does not use special education services any time between Kindergarten and sixth grade, investors are paid 95 per cent of public savings plus an additional five per cent in interest payments. After this period, if a child continues studies without engaging special services, investors receive an additional 40 per cent of cost savings for every year services are not utilised. These financial incentives ensure that investors together with school districts and impact-driven organisations delivering the curriculum are aligned in maximising social outcomes produced.
The resulting cross-sectoral solidarities are thus expressed in the common aim of achieving increases in measurable social value, which is gauged through social impact metrics. The proliferation and insistence on metrics has been characterised as 'the credo of the new millennium' (Mitchell 2017, p. 15). The power that social impact metrics have accumulated can be grasped through their capacity to nurture a particular autopoietic field (Maturana andVarela 1980, Krause 2018). Autopoiesis is, in Luhmann's words, a 'general form of system-building using self-referential closure ' (1986, p. 172). Otherwise put, any system that is reproduced through elements of that system can be considered autopoietic. SIBs, for instance, display properties of autopoietic systems, in that the social impact metrics that are at their core produce a particular field of solidarities whose relationships they reproduce throughout the duration of the programme. The autopoietic dimension is important because the capacity of social impact metrics to focus attention and reproduce relationships guarantees to a certain degree that mission drift is prevented and producing outcomes for the target population is prioritised.
Much like intertemporal solidarities, cross-sectoral solidarities are also fraught with limitations. In particular, there are two categories of problems, and both come from the fact that cross-sectoral solidarities are essentially generated through impact metrics. On the one hand, the decision regarding which measurement framework and particular outcome metrics will be employed in the SIB is taken as part of the initial negotiation of the original SIB contract. Some stakeholders are not included in this negotiation, and even concerning the ones that usually are implicated (public agencies, investors and service deliverers), it is naïve to assume that the interaction is on a level playing field. Governments may be driven ideologically to establish a track-record for social impact investing or to transform the nature of public policy more widely (Dowling andHarvie 2014, Harvie 2019), and thereby be more willing to make concessions to investors. This can sometimes result in perverse outcomes like 'cream skimming', that is, focusing on easy beneficiaries with a high likelihood of meeting outcome targets (Carter and Whitworth 2015). Non-profits, furthermore, are particularly disadvantaged, given that they face legal restrictions in financial reserves they are allowed to carry over yearly and employ them for long-term or 'speculative' investments (Joy and Shields 2013). There is also evidence that smaller providers are pushed out of the market because of the high transaction costs involved and their lack of capability in engaging with the market and dealing with SIB terminology and procedures, on top of the fact that investors oftentimes already have preferred providers (Wooldridge et al. 2019). Lastly, the lack of commissioner micro-management touted in PbR contracts is often replaced in SIBs with the higher degree of surveillance exercised by private investors (Edmiston and Nicholls 2018). Beside the pre-agreed metrics, these frequently require service deliverers to provide another layer of information regarding inputs, outputs and processual aspects. The structuring power of metrics is therefore undergirded by power relations that privilege specific actors at the expense of others.
On the other hand, there is also a profound issue with the nature of the outcome measurement process itself. It has already been mentioned how the facticity of numbers underlies some of the mechanisms operating in SIBs. This is particularly the case with regards to social impact, which is normally gauged via the meeting of particular performance targets, rather than impact evaluation. Many SIBs have never actually been evaluated (Fox and Morris 2019). Impact evaluations are undoubtedly costly, burdensome and particularly complex due to the technical difficulties involved in finding a comparator group for the target population and the lack of consensus on the methodological issues involved. Most evaluations done in the UK were funded by the government and are process evaluations based mostly on qualitative methods (Albertson et al. 2018). It is very difficult in these to attribute the weight that different elements had in the achievement of the outcome and establish what works and to what extent. The evaluation of the aforementioned Trailblazers SIBs, for instance, states that it appeared to encourage a stronger emphasis on demonstrating results than comparable non-SIB services but it is not possible to ascertain whether this was translated into better client outcomes. It was difficult to reach a clear verdict on the costs and benefits of SIBs in this field over the three years of the evaluation. (Fraser et al. 2018a) In the US, on the other hand, evaluations were almost exclusively based on Randomised Controlled Trials (RCTs), whereby two random samples from a population who might benefit from an intervention are identified, of whom one is allocated the intervention. The differences in outcomes observed are attributed through various statistical methods to the intervention. RCTs are considered the gold standard in various intervention schemes, but they are far from flawless. Recent studies have started to question the usefulness and even validity of RCTs applied to social interventions, particularly due to replicability issues and positivistic assumptions (Labrousse 2016, Deaton and Cartwright 2018, Bédécarrats et al. 2019. All in all, though, it is not really clear that meeting social impact targets necessarily reflects a real change that would not have been produced without the intervention itself. The social impact metrics that are the gravity force that produces and reproduces a SIB arrangement, therefore, create particular cross-sectoral solidarities, but that does not mean they are necessarily indicative of real social value added, or that they are not shot through with power relations which privilege particular actors at the expense of others.
Although lopsided, particular relationships are still forged during the lifetime of a SIB and new forms of collaboration may emerge. As the actors involved are driven by diverging institutional logics and interests, a balancing act between worldviews follows. The alignment of interests can thus result in mutual learning, through, for instance, the sharing of practices or simply the sensitisation to different life-worlds. Similarly, it may lead to the emergence of different lines of accountability, with actors driven, for example, by considerations for social justice alleviating other actors' finance-first instincts.

De-risking Government
The third aspect that is part of the core mechanism of SIB and is touted by its promoters as a paradigmatic shift in the design and delivery social policy objectives is the idea of risk transfer. Depending on the source, the risk in question can be intervention risk, implementation risk or financial risk (CAP 2011, Young Foundation 2011, Social Finance 2016. Social Finance (2012b, p. 22), for instance, writes that investors bear 100 percent of the financial risk in an SIB. This instrument allows government to only pay for those programs that are successful. Unlike some pay-for-performance contracts, non-profits are provided with upfront funding over the duration of the SIB. SIBs signify a new paradigm of public-private partnerships in the wake of the financial crisis, one that privatises the risks and shares the gains.
Given that payment to investors is contingent upon the achievement of pre-established performance targets and not inputs or outputs, financial risk is normally shifted from the public to the private sector. In the case of Public Private Partnerships or Private Finance Initiatives, for example, if the projects failed to pay out, the government can be left with a sizable bill. By contrast, an SIB triggers payments only when successful results are achieved (Roman et al. 2016). That means that in theory governments can engage with more complicated social risks that have no obvious, immediate solutions, given that the financial risk, as well as the intervention risk, is shifted to the investors. There is a question, however, as to why private investors would want to engage with such risky programmes. Presumably, a government would only commission a SIB if the degree of success of an intervention is questionable. But if private investors are similarly convinced that the intervention will fail, then they would also steer clear of the SIB. This paradox has elsewhere been described as follows: 'if both actors expect success, the commissioner ought not to use a SIB, and if either commissioner or investor expects failure they should not get involved' (Giacomantonio 2017, p. 52). A SIB project would only make sense if the envisaged success rate of the intervention is relatively unclear to both parties, and if the commissioner is particularly risk averse or drawn to the idea of addressing collective concerns through non-governmental action (Neyland et al. 2019).
SIBs are indeed sometimes risky for investors. For instance, the first SIB, the Peterborough Prison SIB, an ex-offender rehabilitation programme, was ended prematurely due paradoxically to the success of the first stage of the project (Barrow Cadbury Trust 2017, MoJ 2017). The roll-out of a nation-wide programme based on the structure of the Peterborough SIB eliminated the possibility to have a live comparator group against which outcomes are assessed and payments made, thus the SIB had to be terminated. The investors were paid for the first stage of the programme, but due to the commissioner reneging on the rest of it, investors lost the additional potential return. Nonetheless, for private investors, SIBs can be a good investment proposition, given that they benefit from various risk-mitigation instruments (Social Finance 2012b). For instance, in the aforementioned example of the Utah High Quality Preschool SIB, the JB and MK Pritzker Family Foundation contributed $2.4m in the form of a guarantee loan to mitigate the risk to Goldman Sachs's investment (CNCS 2015). In a different example, the Benevolent Society SIB in Australia, which sought to keep children with their families and avoid their entry into out-of-home care, was funded through two classes of bonds: a AUS $2.5m high-risk class subject to 100 per cent loss of capital if programme is unsuccessful, and a AUS $7.5m low-risk class that was capital protected with investors guaranteed their principal back and interest subject to performance outcomes (TBS 2013). Guarantees and tiered capital structures are becoming routine pathways for investors to decrease risk exposure.
The process of governments offloading interventional and financial risk might appear at first sight as an unlikely pathway of nurturing social solidarities. However, scholarship looking at the relationship between risk and governmentality has long pointed out that risk is at the core of contemporary forms of social governance (e.g. O'Malley 2004, Webb 2009, Baker and Simon 2010. Part of the wider transformations in government since the 1980s was an increase in pressure for making public services more accountable and transparent, particularly through increasing their capacity for self-analysis and for making them auditable (Power 1999). This reflexivity (Dean 2009), at the same time, does not simply mean more capacity for self-reflection, but it also entails a shift from governing socioeconomic processes directly to governing the governance of socio-economic processes. In delivering social policy through SIBs, for instance, governments offload risks back onto societal actors, therefore fostering an environment of grassroots 'responsible autonomy' and social citizenship, and thus governing through local sites of self-government such as communities, social enterprises, investors and the target population itself. Solidarities are nurtured therefore not through a centralised administrative state, but through networks of service provision, social activation and expertise, which are responsibilised by risk and governed at a distance through instruments such as performance management and impact targets. As a steward of the rights of the electorate (Lambright 2009), reflexive government governs the risks to taxpayers by essentially insuring itself and 'pricing away' financial and interventional risk to investors, and in doing so converting risks to the collective concern of social networks outside government. The government thus steps out of these self-governing networks, but only whilst extending its reach via contractual relationships and outcome targets and fostering the development of local relationships.
Offloading risk and thus nurturing solidarities, however, is not without problems. Financial risk, for one, is not always transferred from the public commission to private investors. For instance, in the London Homelessness SIB, it was one of the two service providers, Thames Reach, who took on financial risk (Go Lab 2019). In this case, Thames Reach invested its own capital alongside two loans from the Department of Health Social Enterprise Investment Fund and Big Issue Invest and a grant from the Monument Trust. It agreed to pay fixed interest rates to the two investors regardless of the outcomes received and maintain outcome payments from the Greater London Authority in case of success (though Big Issue Invest would receive a part of these as well). In doing so, the two investors did not assume any degree of financial risk. There are more SIBs structured on this Direct Provider model (Carè and De Lisa 2019), where service providers deliberately shoulder some or most of the financial risk, but even in the case of SIBs modelled on a partnership structure or with a Special Purpose Vehicle (SPV) intermediating financial flows and managing contracts the financial risk that service providers are exposed to is significant, particularly in the case of underperformance. For example, in the case of the Trailblazers SIBs which were not based on the Direct Provider Model, service providers were withheld payments when services did not perform as intended (Fraser et al. 2018a). This was also due to the flexibility that the SIB contractual relationships allowed, which meant that financial risk distribution was not fixed at the outset and was subject to negotiations. Lastly, as alluded above, a significant number of SIBs actually rely on public and/or philanthropic organisations' subsidies, which leads in some cases to private investors having guaranteed returns of at least 50 per cent of their invested capital (Pasi 2014, Warner 2017. In other words, risks are not always offloaded, and when they are, they might be unevenly distributed. This is problematic, because instead of nurturing solidarities in the form of creativity, collaboration and social innovation in the pursuit of social value creation, SIBs might actually hamper them. Indeed, because service providers can sometimes take on a disproportionate amount of risk, they might be tempted to avoid experimental service provision, and there is evidence that SIB programmes are biased towards established solutions with a long track-record (Fox and Morris 2019, Sinclair et al. 2021). Similarly, due to the SIB paradox, investors might not want to participate in particularly risky interventions.
Asymmetrical risk distribution is a challenge, but even so, SIBs still operate via the fostering of local networks, responsible autonomy and social citizenship. In the process of governing at a distance, alliances between actors are formed at the grass-roots level. Hierarchical power is watered down, and, in dealing with social risks, a reassembling of actors is promised whereby new forms of relationality emerge that may foster more participation and a form of social citizenship that may surpass contract culture and rely more on flexible rather than rigid norms as well as consent rather than discretion.

Conclusion
The Social Impact Investment Taskforce, an organisation set up during the UK's presidency of the G8 with the remit of 'catalysing a global market in impact investment', wrote that 'the financial crash of 2008 highlighted the need for a renewed effort to ensure that finance helps build a healthy society' (SIITF 2014, p. 52). Having first appeared on the agenda as the global economic system was grappling with the consequences of the financial crisis of 2007-8, SIBs have naturally been seen as indicative of the so-called 'social turn' of finance (Dowling 2016, Lilley et al. 2019. Amidst the flurry of studies analysing what went wrong in the evolution of the global financial system over the past three decadesthe period most commonly associated with the term 'financialisation' -SIBs were being perceived as mirroring some of the more noxious financial innovations identified as responsible for the destabilisation and eventually meltdown of the global financial system. But SIBs were not only derivative-like means of speculating on the performance of social programmes, but also investment propositions allowing private investors to accumulate capital from the commodification of society's downtrodden. While this reading of SIBs is useful in emphasising determinants of the emergence of SIBs as legitimate alternative means of pursuing social policy objectives, it has the limitation of not attending to SIBs as a form of governance of social issues and thus missing on the potentially deleterious consequences arising from that. For example, part of the problem with that interpretation, as highlighted above, is that it leads to assuming that SIBs erode the values engendered by welfare policies, particularly social citizenship and solidarity (Sinclair et al. 2014). SIBs are said to be individualising and atomising social inclusion policies that are part of organic processes and complex social interactions (McHugh et al. 2013) and framing societal problems as a lack of individuals' self-responsibility (Joy and Shields 2013).
Solidarity and social citizenship are indeed essential attributes of welfare states (Gelissen 2000, Baldwin 2008, Taylor-Gooby 2011, Arts and Gelissen 2012, but SIBs also engender their own solidarities. Based on three tenets of SIBs (the discounting of social risks, the alignment of interests and the de-risking of government), this paper has emphasised three: inter-temporal solidarities, cross-sectoral solidarities and risk-insurance solidarities. These solidarities are nurtured through the tools, interactions and structure of SIBs, but they can also be spurious and can come undone. They can lead to a lower valuation of the future, a metric-mediated power imbalance and an asymmetrical risk distribution. That said, shunning them and going back to a Bismarkian or Beveridgean type of welfare state is likely impossible, given how deep and pervasive the move towards the social investment state has been. SIBs tie into this development, but one should not be too quick to assume what this leads to. In as much as they reframe social programmes as investments, SIBs also promise a reassembling of actors and the role of government, and this leads to three particular openings that this paper has identified: SIBs expand the horizon of the possible, foster mutual learning and the balancing of diverging worldviews, and encourage alliance-formation and social citizenship. While this might not always pan out, it is important nonetheless to leave enough room to characterise what emerges. Therefore, advancing a perspective of SIBs as tools for governance of social issues is meant precisely to open up a new avenue for gauging the implications of this new instrument for pursuing social policy objectives. Future research could look into specific SIBs and analyse the extent to which solidarities are achieved and at what cost, particularly if SIBs are used as a legitimising strategy to wind down universal programmes with their own forms of solidarity. Notes 1. Database available at https://golab.bsg.ox.ac.uk/knowledge-bank/project-database/. 2. For an extensive review of the literature on SIBs, see Fraser et al. (2018b) and Tan et al. (2021). 3. Here, 'social investment state' and 'social impact investment' are not directly related. The former refers to how social policy is conducted at the governmental level; the latter refers to the financial investment mentioned beforehand. 4. The HMT Green Book is a guidebook providing guidance on how to appraise policies, programmes and projects that involve public spending. For more, see: https://www.gov.uk/government/publications/the-green-bookappraisal-and-evaluation-in-central-governent. 5. https://www.ons.gov.uk/economy/grossdomesticproductgdp/datasets/quarterlynationalaccounts.

Acknowledgements
The author is grateful to Andrea Mennicken, Mike Power, Peter Miller, Jose Bolanos, Finia Kuhlmann, and all the AC500 seminar participants in the Accounting Department at the LSE where previous versions of this article were probed. The author would also like to thank Anastasia Nesvetailova, Stefano Pagliari, Steffen Murau, and participants at the 32nd Annual SASE Conference and the 12th Annual Critical Finance Studies Conference for invaluable feedback, as well as all reviewers for their constructive feedback.

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

Funding
This work was supported by the UK Economic and Social Research Council under grant number ES/T008687/1.

Notes on contributor
Dr Andrei Guter-Sandu is an ESRC Postdoctoral Fellow at the Centre for Analysis of Risk and Regulation, Department of Accounting, London School of Economics and Political Science. He has a PhD in International Political Economy and works at the crossroads between financialisation, public governance, and sustainability.