Performance management in social impact bonds: how an outcomes-based approach shapes hybrid partnerships

Purpose – This paper investigates how outcomes-based performance management (PM) regimes operate in the partnerships known as social impact bonds (SIBs), which bring together partners from the public, private and third sectors. The findings are analysed in the light of the different cultural world views of the partners. Design/methodology/approach – Published evaluations of 25 UK SIBs were analysed by a qualitative multiple case study approach. This study of secondary sources permitted the analysis of a wide range of SIB partnerships from near contemporary accounts. Findings – Outcomes frameworks led to rigorous PM regimes that brought the cultural differences between partnersintofocus.Whilepartnershipsbenefittedfromthe varietyofviewpointsandexpertise,thedifferences in outlook simultaneously led to strains and tensions. In order to mitigate such tensions, some stakeholders conformed to the outlooks of others. Practicalimplications – Theneedtoachieveapredefinedsetofpayableoutcomesembedsa “ linear ” viewof intervention and effect on the SIB partners and a performance regime in which some partners dominate. In designing accountability systems for partnerships such as SIBs, commissioners should consider how the performance regime will affect the interests of all stakeholders. Originality/value – This study adds to the cultural theory literature which has rarely considered three-way partnershipsembodyinghierarchical, individualist andegalitarianworld viewsandhowperformanceregimes operateinsuch partnerships.Three-waypartnerships arethoughtto berareandshort-lived, butthisempirical study shows that they can be successful albeit over a predefined lifespan.


Introduction
Measuring public sector performance Publicly funded agencies are held accountable for the quality of their performancebut this quality is rarely easy to assess.It is no longer possible to assume that "inputs-equal-outputs" as was traditionally the case (and still is for certain UK services such as defence and police).Instead, a wide variety of performance indicators reflect outputs and, increasingly, outcomes.
The move towards measuring outcomes reflects a desire to understand the effectiveness of services (Bovaird and Davies, 2011).But each type of measure has its own strengths and weaknesses (Hood, 2007).Inputs (numbers of doctors, teachers, etc.) are relatively straightforward to measure but do not assess the effectiveness of a service.Outputs (such as hospital treatments) can be measured in the short-term but are only a weak proxy for effectiveness and are subject to perverse behaviours such as threshold and ratchet effects (Bevan and Hood, 2006).Outcomesimprovement in well-being or prosperity of the beneficiariesmight seem to be a better way of judging service effectiveness.Outcomes, however, are often difficult to attribute to the intervention itself as they can emerge over long timescales and are themselves subject to gaming behaviours such as "cherry-picking" and "parking" (Fox and Morris, 2020;Carter and Whitworth, 2015).

Outcomes-based commissioning and social impact bonds
The public sector has also seen a shift not only towards measuring outcomes but also towards paying exclusively for outcomes (outcomes-based commissioning or OBC).OBC is particularly favoured by governments during periods of austerity as it aims to transfer the financial risk of underperformance away from the public sector (Bovaird and Davies, 2011;Lowe, 2017).OBC was particularly emphasized in the UK Government's Open Public Services White Paper (HM Government, 2011).While OBC is attractive to government, it is clearly risky for providers.Under OBC, provider organizations must fund the service without knowing whether their costs will be repaid.A mechanism intended to overcome this resistance, particularly amongst local and charitable providers, is the social impact bond (SIB) (Ronicle et al., 2014;Gustafsson-Wright et al., 2015;Carter et al., 2018;Fraser et al., 2018).In a SIB, an investor provides the upfront funding and bears the financial risks of underperformance.The investor is repaid (with interest) by the commissioner if outcomes are achieved.Though, as we will see, the transfer of risk is rarely so clear-cut.
The typical outcomes intended by SIBs are improvements in the life chances of individuals with complex social needs, such as reduced reconvictions of ex-offenders and stable accommodation for homeless people (Social Finance, 2012;Carter et al., 2018).Outcome indicators are central to the performance management (PM) regimes of SIBs as they are normally the sole criterion whereby payments from the commissioner are unlocked.Nevertheless, the whole premise of outcomes-based PM has been questioned, given that "outcomes are emergent properties of complex systems" over which individual public services have limited control (Lowe and Wilson, 2017).
The first SIB was launched in the UK in 2010 and, since then, over 150 have been implemented of which about half are in the UK [1].Worldwide, SIBs have received considerable support from governments and international organizations.The UK Government set up the Centre for SIBs in 2012 as a team within the Cabinet Office (later transferred to the Department for Digital, Culture, Media and Sport).The Cabinet Office also co-funded the Government Outcomes Lab at the University of Oxford to conduct research into this type of commissioning and to offer advice and support to potential and existing commissioners (Carter et al., 2018;Lowe, 2017).And, to encourage the adoption of this funding mechanism and reduce the risk to commissioners, a number of pooled "outcomes funds" [2] have been set up to subsidize all or part of the outcomes payments.Despite their increasing prevalencethough still representing a tiny fraction of UK public spending (see Floyd, 2017) -SIBs remain a controversial approach to public service reform, given the potential for substantial returns for private investors from public funds (see e.g.Fraser et al., 2018;Lowe, 2017;Neyland, 2018;Tse and Warner, 2020;Warner, 2013).SIBs are also criticized for "individualising" social problemstackling the symptoms rather than the structural causes of poverty and inequality (Cooper et al., 2016;Lake, 2016;Morley, 2019).

Performance management in social impact bonds
This paper presents an empirical investigation of the effects of implementing an outcomes framework in hybrid partnerships.In SIBs, the partnership is between the commissioner, the (social or private) investor and the (usually non-profit) service provider.SIBs are hybrid in the sense of bringing together stakeholders with competing institutional logics (Greenwood et al., 2011;Skelcher and Smith, 2015;Johanson and Vakkuri, 2017).UK SIBs thus form an ideal set of cases for the investigation of PM in such partnerships having a unified rationale and a limited range of structures operating across a variety of policy areas.
The paper sets out to answer the following research questions: How does an outcomes focus shape the working and achievements of a SIB partnership?What do SIBs reveal about PM in hybrid partnerships?
While it can seem "almost heretical" to question the advantages of collaboration (McLaughlin, 2004, p. 103), the contrasting outlooks of diverse partners can lead to tensions.Organizations collaborate in order to achieve goals that they cannot reach individually, with each partner bringing skills, resources or expertise which the others lack (Thompson et al., 1990;Brandsen and Pestoff, 2006).Nevertheless, collaborations can be difficult and time-consuming to set up, can consume considerable organizational resources and are sometimes short-lived (Huxham and Vangen, 2005;Wegrich, 2019).Partnerships can, indeed, have more symbolic than practical effect (Dixon and Elston, 2020).Collaborations between diverse organizations face particular challenges in implementing PM regimes owing to stakeholders' competing goals (Minassians, 2015;Grossi et al., 2017;Skelcher and Smith, 2017).

The culture of collaborations
Although "organizational culture" is often invoked to explain the success or failure of collaborations, "culture" is often vaguely defined and typologies abound (see e.g.Scott et al., 2003).Mary Douglas's "Cultural Theory of Risk" offers a typology which has been described as "demonstrably complete" and offers a theoretical account of why particular combinations of values, preferences and ideas are combined (Douglas, 1982(Douglas, , 1987;;Thompson et al., 1990).Douglas identified four distinct world views based on the dimensions of "group" and "grid.""Group" describes the level of social cohesiveness and "grid" describes the extent to which the organization is based on rigid rules.The combination of grid and group leads to four distinct types, three of which are classed as "active" (hierarchical, egalitarian and individualist) and one as passive (fatalist).These types exist in mutual rivalry and interdependenceno single type dominates permanently as each has characteristic weaknesses or blind spots (Wegrich, 2019).Alliances form in order to temper the extreme manifestations of each type.For example, Thompson and colleagues argued that "the establishment" in developed countries is an alliance between individualists and hierarchists: market-oriented individualists benefit from stable property rights and the rule of law, while the hierarchy benefits from entrepreneurship and wealth creation of a competitive market (Thompson et al., 1990, p. 88).Cultural theory (CT) has been used to study social relations at different levels and across various policy fields (see, for example, Gross and Rayner, 1985;Verweij et al., 2006;Dean, 2017;Lodge and Wegrich, 2011;Manzi, 2007;6, 2004;6 et al., 2006).In The Art of the State, Christopher Hood applied the theory to public administration (Hood, 1998).A typology of public organizations based on this framework is shown in Table 1.
How do the parties in a SIB partnership fit into Douglas's typology?A plausible expectation is that the partners fit most naturally into the three "active" quadrants.The passive (fatalist) quadrant is omitted (for now) as only active organizations care enough about their goals to build alliances (Thompson et al., 1990;Weare et al., 2014).The systematic IJPSM 34,3 differences between governmental, market and voluntary/non-profit sectors are well established (for a review, see Herranz, 2008) and CT provides a theoretical basis for these differences.The SIB commissioner, generally a local or central government body, is often constrained by bureaucracy and procurement rules and is likely to be the most hierarchical.The service provider, in contrast, is typically a non-profit organization from the voluntary, charitable or social enterprise (VCSE) sector.The VCSE "world view" is likely to be more egalitarian and less rule based and hierarchical than that of the government commissioner.The private or social investor is likely to be the most individualist, seeing themselves as bringing "market discipline" to the partnership (see e.g.Cohen, 2011).
This paper draws on CT to interpret the contrastingand sometimes conflictingunderstanding of performance indicators in three-way hybrid partnerships.CT has been used to describe different regulatory approaches (Hood, 1998;Hood et al., 2004;Patel, 2007;Weare et al., 2014;Lodge et al., 2010).According to this literature, attitudes to PM are expected to be as outlined in Table 2.The empirical section, to which we turn next, analyses the PM of UK SIBs in the light of this framework.

Methods and data sources
Evidence of stakeholders' views and experiences of PM was drawn from the published evaluations of UK SIBs listed in the Appendix.The reports were analysed qualitatively under the following headings: "outcomes and payment structures", "outcomes development", "financial risks", "performance management systems" and "evidence of success", employing a deductive approach in which CT informed the analysis.Changes over time of PM regimes and interorganizational relationships were recorded and the impacts of the outcomes-based approach on front-line workers and clients were also explored where evidence was available.
The reports were written by independent evaluators to assess the development and implementation of the SIBs.Several of the evaluators were originally commissioned to provide quantitative evaluations but for reasons of data availability, confidentiality, cost or the absence of a matched counterfactual group, plans for quantitative impact assessments were not realized (see e.g.R05, p. 16; R22, p. 12; R16, p. 6; R17, p. 49).Only three quantitative analyses were produced for the SIBs in this study, which are described separately in the

Social impact bonds
Findings section.The analysis described here is a form of qualitative evidence synthesis (Noyes et al., 2018) or qualitative secondary analysis (QSA) (Largan and Morris, 2019), taking published qualitative evaluations as the empirical material.This research strategy was chosen to include evidence from the largest possible number of UK SIBs through a multiple case study approach (Cresswell, 2013).This approach was chosen over a research design in which new interviews were obtained for the following reasons.The published evaluations allow a range of near contemporaneous opinions to be obtained from the earliest days of SIB development across the entire sector in the UK.Several evaluations revisited the same SIB or group of SIBs at various stages during their life cycle, providing insight into the development of stakeholders' attitudes to the PM regime.
One limitation of QSA is that the analysis relies on the quality and reliability of the published sources (Scott, 1990;Noyes et al., 2018).The evaluations were commissioned for a variety of purposes and undertaken by different teams.They are therefore potentially subject to a range of selection and interpretation biases which can be compounded by the biases of the secondary analyst (Largan and Morris, 2019, pp. 71-73).In order to mitigate such issues, reports of the same SIBs by different evaluators were triangulated where possible and compared with external sources where available.Indeed, the relative heterogeneity of the sourcesthough most used similar methodologies and reporting structuresmitigates the risk of common source bias (George and Pandey, 2017) and gives greater confidence in the findings when similar issues are mentioned by diverse evaluators across a variety of cases.
The secondary analysis recorded the range of views found in the entire corpus of relevant sources.Detailed notes were made for each source and relevant quotations and descriptor words and phrases were collated.The sources were analysed under the headings above and cross-checked until "saturation" (i.e.no new relevant evidence was found) (Saunders et al., 2018).The analysis was supported by NVivo software.The deductive approach involved deliberately seeking examples that were incompatible with cultural theory as well as ones that conformed to it.No source mentioned CT or related concepts and thus the sources may be considered unbiased with respect to this theoretical framework.
All existing UK SIBs were considered for inclusion in the study.SIBs were included if they met the following criteria: (1) the SIB was completed or had been running for at least three years by the time of data collection in 2019; (2) at least two independent evaluation reports were available for each SIB (alone or as part of a group); ( 3) the evaluations included first-hand interviews with stakeholders from one or more of the SIB partner organizations and/or clients.Here, "independent" means that the report was commissioned by the commissioner and/or a central government department from an academic or commercial institution.The reports were not written by employees of the SIB itself or any intermediary organization involved in development or implementation of the SIB, although one was written by a provider organization.Background information from other sources is cited separately in the text.
A total of 25 UK SIBs met all three criteria and are listed in Table 3.This represents all but six of the UK SIBs launched before 2016.The 22 qualitative evaluations of these SIBs ("the sources") are referred to in the text as R01 to R22 (see Appendix).Several sources assessed the SIBs in groups based on the name of the outcomes fund or other grouping (as shown in Table 3).Where that was the case, individual SIBs were not always uniquely identified.

Outcomes and payments
SIBs were paid for "outcomes" ranging from long-term outcomes evaluated against a counterfactual group to early results which largely reflected process indicators such as recruitment to the programme.Thus, we see at one extreme the Peterborough SIB, where IJPSM 34,3 Table 3. Social impact bond names and sources payments were made on the basis of the number of reconvictions after a year post-release relative to a control group (R03).At the other extreme, the Essex multisystemic therapy (MST) SIB was reimbursed from the start of the intervention for "days of care avoided" by children at risk of being removed from their families (R15).Most projects, however, offered payments for a "rate-card" of long-and short-term achievements, as illustrated in Table 4. Thus, the Fair Chance Fund (FCF) SIBs for young homeless people received payments for clients' assessments followed by accommodation, education and employment outcomes More controversially, those SIBs also included payments for "reconnections" (the repatriation of homeless non-UK nationals) (Cooper et al., 2016).The two health SIBs had contrasting metrics."Ways to Wellness" was paid mainly for the reduced use of hospital and emergency health services compared with a counterfactual group (and a smaller proportion of payments based on subjectively improved well-being), while "Worcestershire Reconnections" measured reductions in loneliness on a standard scale, assuming that this improvement would lead to reduced health-care costs (R12, pp.30-32).The Department for Work and Pensions (DWP) Innovation Fund (IF) SIBs for young people at risk of becoming not in employment, education or training included payments for improved behaviour and attendance at school as well as for educational qualifications and entry into employment, higher education or training (R18).Table 4 illustrates how the payable outcomes in three sets of SIBs differ from performance metrics in comparable services funded in other ways.Of course, traditionally funded services are also concerned with outcomes and indeed may be assessed on them, as with examination results or employment metrics.Typically, however, the types of services offered by SIBs would, if they were provided by other means, be funded on the basis of inputs such as the number of posts or outputs such as levels of service provision.It should also be noted that not all SIBs have even approximate comparators.For example, the Peterborough SIB was aimed at short-sentenced prisoners, who at the time received no statutory supervision on release from prison.One major distinction between SIBs and other arrangements is that SIBs typically have a range of targets aimed at supporting clients towards outcomes which will benefit themand the wider societyin the long-term.This also means that SIBs often combine outcomes which would otherwise be the responsibility of several agenciesfor example, housing, health, education and employment.The table illustrates both strengths and weaknesses of the outcomes model.Payment for outcomes can generate operational flexibility as the level and type of service provision is not predetermined.However, outcomes can also depend on external factors such as the availability of jobs or housing, meaning that payments can become decoupled from the effort and resources expended to achieve them.

Negotiations
The evaluation reports described complex and lengthy negotiations to agree the outcomes and payment mechanisms.In most cases, this was a technocratic process in which there was little or no opportunity for service providers (let alone clients) to influence the chosen outcomes.For the Peterborough SIB, for example, Social Finance (an intermediary organization specialising in SIBs) [3] and the Ministry of Justice defined the payment criteria, with service providers and prison authorities only involved much later.For the Innovation Fund SIBs, the commissioner (DWP) invited potential service provider and investor partnerships to tender for a selection of predetermined outcomes from a rate card.FCF service providers were also invited to tender for outcomes from a rate card, not all of which were found to be appropriate for the client cohort (R21, p. 59).Similarly in the Essex SIB, the omission of operational staff from the initial decision-making was associated with later difficulties regarding referral criteria and responsibilities (R15, p. 27).
Some SIBs adopted a more cooperative approach to agree outcomes, though this brought its own complications.The London Homelessness SIB went through many rounds of "competitive dialogue" between the commissioner (the Greater London Authority, GLA) and shortlisted candidates (R07, p. 40).Potential providers also had to recruit investors who expected complex and detailed outcomes projections, even though providers knew that these forecasts were unreliable, e.g."a stab in the dark" (R07, p. 45) or "a blank sheet of paper" (R08, p. 18).And while some providers found that the process "professionalised our approach to mapping and predicting outcomes" (R07, p. 46), it was costly and time-consuming for providers whether or not they were ultimately awarded a contract.
Overall, these experiences illustrate the complexity of SIB negotiation and design processes (FitzGerald et al., 2019).The interests of investors, often represented by an intermediary or investor-owned special purpose vehicle (SPV), often dominated.Nevertheless, even some investors were dismayed by the complexity and length of the negotiation process, attributing it in part to the political salience of SIBs (R21, p. 59).

Outcome-based commissioning and financial risk
The design of outcomes is inseparable from considerations of financial risk.A SIB is meant to transfer financial risk from the commissioner and service provider to the investor.This rationale was often explicitly stated in evaluation reports, but in practice the transfer of risk was rarely so clear-cut.
The closest example of this "ideal" was the Peterborough recidivism SIB.In this case, investors paid service providers the full cost of the service through a combination of capital funding and fees.Repayment remained uncertain until the end of the evaluation which in this case was about seven years after the initial investment.This may have been "ideal" from one perspective but this level of risk to investors was only possible because the investors were mainly charitable foundations interested as much in the social outcomes as in financial returns (R02).
Most SIBs mitigated investor risks in two main ways.First, as already mentioned, payments could be made for early results.These would be more accurately described as output or process indicators but are nonetheless called "outcomes" in the evaluation reports.The use of such indicators transfers risk back to the commissioner, who risks paying both for the early results and for ongoing services to the same clients (R05, p. 92).Second, some SIBs imposed a payment-by-results (PbR) regime on service providers, thus transferring risk from investors to providers.For example, finance was withheld from providers who missed contractual targets in three of the "Trailblazer" SIBs (R05).These strategies could be combined.For example, DWP IF investors paid for only the first six to 12 months of the service, after which the projects were expected to fund themselves through ongoing outcomes payments (R18, p. 70).FCF SIBs had a similar arrangement (R22, p. 78).
Occasionally, however, providers were able to argue that PbR was inappropriate for their involvement.In Worcestershire Reconnections, PbR was considered an unacceptable risk for the charitable provider, which would encourage focussing on "easier" beneficiaries (R10, p. 13).The lack of financial risk to providers was also seen as "central" by investors in the Peterborough SIB (R02, p. 6).In most cases, however, passing some financial risk to providers was "expected to promote more effort and better performance" (R05, p. 106).
The design of outcomes affected the distribution of risk.If outcomes were too easily met, investors would be repaid even if few clients benefited (R05, p. 92).Attribution is also important herethe extent to which the intervention caused the outcome.This was considered to be an issue in the Essex SIB: the rates of children entering care in the county had reduced substantially by the time the SIB started but the payment criteria were by then contractually fixed (R15, p. 27).Similarly, DWP SIBs were judged a qualitative success but there is no evidence that their clients did better than a matched group of disadvantaged young people (Salis et al., 2018).
Investors also made use of flexibility in contracts in order to maximize repayments.In the DWP IF SIBs, for example, "[s]everal projects were still 'tweaking' their performance [numbers] to maximize outcomes even in the final few months of their contracts" (R19, p. 32).Similarly in the FCF, outcome numbers were re-profiled towards goals that the projects were more likely to achieve (R22, p. 3).Commissioners of the Trailblazer SIBs also found that IJPSM 34,3 intermediaries frequently challenged outcome metrics on behalf of investors.Although investors were nominally willing to lose their investment, in practice this was rarely found to be the case (R05, pp.118-121).

Performance management
An outcomes framework is meant to give operational flexibility for service providers as they are judged solely on the end result.Paradoxically, however, providers were subject to rigorous and complex PM which was described as "onerous" and "intimidating".At times this impacted service delivery "[o]ne project manager described how some of her staff had felt that the constant pressure to achieve outcomes risked taking attention away from what clients needed, thereby 'threatening their professional and ethical integrity'" (R22, p. 80-81).
Commissioners also found PM requirements burdensome.For example, GLA employed a full-time post to monitor the two London Homelessness SIB contracts (R07, p. 62).This was comparable to the one full-time employee who monitored all other homelessness contracts for the council.Managing the Essex SIB resulted in unexpected costs for the commissioner that were estimated to be equal to the administration costs within the SIB (R16, p. 27).
The PM regime depended on the precise contractual structure of each SIB.However, a recurring configuration was the "special purpose vehicle" (SPV)a specially created organization with a central coordinating and managing role.The SPV managed the service providers and was responsible for reporting to the commissioner, investors and often to an additional oversight board.The central UK Government also played a part, sometimes as commissioner or through an Outcomes Fund and in any case via the monitoring role of the Centre for SIBs.Thus, there were multiple lines of accountability in which the SPV typically represented the interests of the investors.
After the launch of each SIB, investors were most concerned about referral numbers, putting commissioners and providers under pressure to increase recruitment.Once clients were recruited, other metrics became the focus of PM.Indeed performance indicators often proliferated, reflecting both paid outcomes and day-to-day case management data.PM often required new data systems and additional staff (R03, p. 6; R07, p. 62; R21, p. 57).Other organizations such as prison authorities and schools might be required to provide or enter datain some cases information that was not readily available (R03, p. 6; R07, p. 62; R21, p. 57).
Evidencing outcomes often required a great deal of effort.For example, in the London Homelessness SIB, tenancy agreements and payslips were often difficult to obtain from clients.The "reconnections" outcomes for repatriating homeless non-UK nationals required documentation from outside the UK.Evidence for the health-related outcome proved unavailable and this outcome was eventually paid by the commissioner regardless (R09).
The dangers of perverse incentives were recognized by all stakeholders.The Peterborough SIB was explicitly designed to avoid gaming behaviour.To avoid "cherrypicking", all eligible offenders were recruited and, to reduce "parking", all reconvictions in the year after release were counted (R03, p. 14).In the FCF "most interviewees were adamant that outcomes payments had not [. ..] moved their focus away from what was best for their clients" (R22, p. 81).Nevertheless, some FCF interviewees noted that there was less incentive to work with clients if no further outcomes were likely (R22, p. 81).In the DWP IF SIBs: "some form of participant selection was . . .seen by many investors and intermediaries as perfectly legitimate and indeed as an economic necessity because their primary focus had to be keeping projects successfully running"a view that was acceded to only reluctantly by providers (R18, pp.46-47).The DWP SIBs were also able to increase the number of participants and hence claim more short-term results than originally planned (R19, p. 40).Overall, however, the evaluations emphasized that both investors and providers focussed on meeting clients' needs whether or not they achieved the maximum financial return, with providers being particularly aware of the reputational risks of being seen to select "easy people" to work with (R05, p. 109).

Front-line workers
Front-line workers are central to the success of a SIB through their interactions with the clients on whose achievements the payments depend.Although some managers took steps to shield case workers from the pressure of targets (e.g.R05, p. 108; R19, p. 38), the evaluations gave ample evidence that the focus on outcomes had a major effect on the front line.The pressure to achieve outcomes and the associated evidential requirements left some employees feeling "burnt out" and in need of support (R22, p. 81).This pressure arose in two main ways.
First, as noted earlier, investors were anxious to increase referrals.Case workers expended considerable efforts in establishing relationships with referrers such as schools, GPs and housing authorities.SIB staff were surprised that their (free) service often required intensive "marketing" to persuade referrers to take it up (R13, p. 34; R18, p. 54).
Second, the outcomes focus affected the interventions themselves, often requiring case workers to coordinate a variety of external services on behalf of clients.This flexibility was valued by case workers but required sensitive interactions with other agencies, which in some cases saw the SIB as a competitor or a drain on their resources (e.g.R07, p. 64; R20, p. 43).SIB staff often had to act as advocates, advisors and guides for their clients and offer emotional as well as practical support.In many cases, the clients had complex needs and the SIB intervention was seen as a "last chance" (e.g.R21, p. 44).Notably, the outcomes focus led some case workers to challenge clients to achieve outcomes that they had not thought possible.For example, "It's helped key workers see they have to challenge young people as well as agree with them.Staff are now more likely to think in terms of working with a young person on a progression of outcomes and challenging as well as supporting them to take chances and do things" FCF Project staff (R22, p. 87).

This led to concrete achievements:
"I have never seen that amount of people with complex needs go into employmentand I don't know how we have done it.It's been a lot of pulling my hair out. . .24 young people have been into work for 26 weeksit doesn't happenit's higher than the Work Programme [government employment programme].This is the target never in a million years we would get into workit just shows what you can do" FCF Project staff (R22, p. 77).
Case workers, however, sometimes felt that their efforts were not valued if their clients' achievements did not match the paid outcomes.In the FCF, for example, the Construction Skills Certification Scheme was not one of the payable educational qualifications, despite being highly relevant to the cohort (R22, p. 35).

Clients
Where clients' views were sought, the evaluators found very positive views (though the representativeness of those views is not known).Participants of the Newcastle Ways to Wellness SIB were enthusiastic: "If anybody was asking me and they said, 'I've had a phone call off this Wellbeing,' I would say get there as quick as you can.Seriously" (R13, p. 26).Families who participated in the Essex MST SIB were also very positive, mentioning how the therapist "transformed" their relationship with their child (R17, pp. 38-39).Young people often built up long-term trusting relationships with their case workers (R21, p. 6; R19, p. 48ff).Clients often expressed regret at the service ending, either on their own behalf or because it would be unavailable to others in similar circumstances (e.g.R22, p. 26).

IJPSM 34,3
Few evaluations indicate that clients were aware of the link between outcomes and funding.But some clients were aware of the pressures: "There was also evidence from the final year interviews with young people in one FCF project that they had felt shelved once they had achieved all possible outcomes: 'The best thing about the project is that the staff that were employed were really caring, but the worst thing about the project was being so outcome driven for the business, definitely towards the end it took the support element away from people who needed it'" (R22, p. 81).Some clients of the London Homelessness SIB felt that money was not an issue: "Thames Reach don't do it for the money. . .they want to help" (R07, p. 87)."They don't treat you like they're trying to make money out of you" (R07, p. 88).However, case workers reported that some clients recognized the financial incentives:

Evidence of success?
Almost without exception, the qualitative evaluation reports concluded that each SIB produced benefits both for the clients and for the partner organizations.This might be a result of an unwillingness to report difficulties on the part of evaluators or stakeholders, a desire to justify the time and effort expended or a wish not to jeopardize future opportunities.Nevertheless, verbatim quotations from stakeholders were substantially positive so, assuming they reflect the balance of this evidence, we can conclude that the SIBs studied here provided something of value to most stakeholders.Positives were balanced with negatives, however, and all evaluations mentioned some difficulties such as tensions between partners or challenges in meeting recruitment numbers.While the qualitative reports were largely positive, the few quantitative evaluations that were carried out showed mixed results, as discussed below.
A frequent finding from the qualitative reports was that participation in a SIB enhanced the PM capacity of providers (e.g.R02, p. 41; R08, p. 22; R19, p. 31).Developing this capacity, however, required a sustained commitment to relationship-building within the partnership.Some of the most detailed discussions of how PM evolved were in the FCF evaluations."As relationships between providers and investors matured, both groups of stakeholders painted a picture of more productive and collaborative working, with each benefiting from the expertise of the other" (R22, p. 77).This understanding, however, only emerged towards the end of the three-year projects: "The intense data expectations, monthly board meetings and the problems inherent in the relationship, these were all more or less overcome towards the end of year two, and year three has been a lot better" (R22, p. 80)."All seven projects felt they had benefited from the additional scrutiny and push that investors brought through the SIB: 'it was useful in tightening up discipline, especially around the need to forecast outcomes better to plan repayment'" (R22, p. 79) and "[s]ome providers also felt that the commercial rigour and focus on outcomes had helped challenge a perceived complacency in the sector and changed how front-line workers were prepared to engage with the client group" (R22, p. 82).
A major advantage identified by providers was the long-term and flexible nature of the funding which could be deployed as needs arose.For example, the Essex SIB employed additional MST therapists and met staff training needs (R15, p. 6-7).Similarly, FCF providers "identified the ability 'to work creatively and flexibly. ..to do things differently within the budget'" (R22, p. 76).
While the qualitative evaluations were positive, the lack of quantitative impact assessments noted above is remarkable, particularly in view of the large amount of outcomes data collected.Only the Peterborough SIB, the London Homelessness SIB and the ten DWP IF SIBs were evaluated quantitatively and, of these, only the evaluation of the Peterborough SIB was designed into the payment mechanism.The reduction in reconvictions of the first cohort of released prisoners did not reach the payment threshold (Jolliffe and Hedderman, 2014).The second evaluation, by a different team, found that the SIB exceeded the payment threshold for both cohorts combined (Anders and Dorsett, 2017).The London Homelessness SIB was evaluated by comparison with a matched group (necessarily offset by a few years as the whole eligible cohort in 2012 was enrolled in the SIB).This evaluation, which was not linked to outcomes payments, found positive effects on rough sleeping, reconnections and entry into sustained accommodation, even though the SIB missed its numerical targets (DCLG, 2017).The evaluation of the ten IF SIBs showed generally negative effects of the SIBs (Salis et al., 2018).Compared to a matched group, the IF SIBs produced a lower proportion of participants in education, employment or training one year after starting the programme and a lower proportion obtaining national qualifications framework (NQF) level 2 and level 3 qualifications, though an increase in the proportion obtaining NQF level 1 qualifications.The authors caution, however, that imperfect matching and data unavailability could have affected these comparisons.None of the evaluations related outcomes to different types of interventions or made use of individual-level variables, making it difficult to learn from these results.
Another way of assessing SIBs' success is whether they were commissioned (or recommissioned) in the absence of central government subsidies.The prevalence of "outcomes funds" meant that local commissioners were usually protected from the full cost of outcomes payments.Only the Essex SIB was commissioned without any central government funding and it was not recommissioned.Referrers were similarly cautious.Although schools were often keen to continue the IF interventions, this "rarely translated into a commitment to fund the service themselves" (R19, p. 71).
Nevertheless, some influences persisted.Peterborough prison authorities commissioned several services based on the SIB interventions and some providers took up further SIB opportunities, such as with the Youth Engagement Fund [4].The FCF and London Homelessness SIBs led to improved partnership working between local authorities, housing providers and the charitable sector.Many service providers developed PM and forecasting capabilities which they used in other contexts.

Discussion
This study brings together evidence from almost all of the UK SIBs started before 2016.The first research question is answered by the finding that an outcome-based approach clearly played a large part in shaping the implementation and achievements of the SIBs in this study.The pressure to achieve outcomes and the associated evidential requirements resulted in pervasive and multilevel PM regimes.These regimes were shaped by the fact that, as one of the evaluations observed, while all SIB partners desired improved social outcomes, stakeholders had "different needs" from the outcome metrics: (1) "Investors need to be able to understand the [metrics] and the level of risk involved in their achievement.[. ..]; (2) Commissioners need to be confident in the robustness of the metrics, in order to make payments against them, and that they will not produce perverse incentives.In some cases, they also need to be confident they will generate cost savings that cover the costs of the outcome payments; (3) Service providers need to be confident they can deliver the outcomes and measure them" (R12, p. 18).

IJPSM 34,3
This observation echoes the diverse uses of performance indicators commonly found in hybrid organizations (Agostino and Arnaboldi, 2017).SIB implementation brought these disparate requirements into focus.PM meetings were often the forum in which tensions surfaced, with data collection and referral numbers causing particular pressures.Nonetheless, all of the qualitative evaluation reports concluded that the performance regime had driven achievement of outcomes, sometimes well beyond expectations.
The few quantitative evaluations, as we have seen, were more equivocal.Their paucity is also telling.As noted by Chris Fox and Stephen Morris, comparative evaluation is rarely built into the payment mechanism for UK SIBs.Among the SIBs in the current study, only Peterborough and, to a lesser extent, Ways to Wellness did so.This contrasts with the USA, for example, where most SIBs make use of randomized controlled trials (Fox and Morris, 2020).Such methods have their limitations, but as Fox and Morris argued "[r]esults from a well-designed randomized experiment can give confidence to commissioners and investors that an investment has delivered the social outcomes intended".Without such a built-in mechanism, impact evaluation requires a separate initiative.Difficulties about who will pay for the evaluation, concerns about conflicts of interest and confidentiality, as well as practical limitations about identifying a suitable counterfactual can become insuperable.Why the UK has not built in more randomized evaluations is not clear.One reason may be the high cost of such evaluations as against the relatively small size of most UK SIBs (Fox and Morris, 2020).
In answer to the second research question as to how this study illuminates PM in hybrid partnerships, CT offers a way of conceptualizing and generalizing these findings.CT suggests that alliances form so as to mitigate the inherent weaknesses or blind spots of each world view.But alliances are unstable because each cultural type regards the others with suspicion and, over time, contradictions and tensions accumulate.Alliances between all three active world views, Thomson and colleagues contended, "are both rare and extremely shortlived" (Thompson et al., 1990, p. 92).
Can SIBs make this apparent impossibility work?Can a stable alliance form between hierarchists, individualists and egalitarians?In a study of Indo-French company alliances, Taran Patel suggested that three-way coalitions can indeed occur as the egalitarian world view helps to stabilize relationships between hierarchists and individualists (Patel, 2007).Similarly, Verweij and colleagues argued that complex policy issues can only be addressed by so-called "clumsy solutions", which take account of all four world views (Verweij et al., 2006).
The empirical findings from UK SIBs lead to four main observations.First, cultural differences generally accorded with our prior expectations of investors, commissioners and providers, as set out in Tables 1 and 2. This was evidenced both by what organizations revealed about themselves as reflected in the interview data and also by the perceptions of other stakeholders.Second, the partnerships benefitted from the differing expertise and viewpoints brought by stakeholders.Third, partnerships were nevertheless put under considerable strain by the different outlooks of stakeholders.And fourth, in order to mitigate such tensions, some stakeholders conformed and adapted to the views of others.
These observations played out as follows: focussing on each type of stakeholder in turn.Investors (or SPVs on their behalf) often acted in an individualist way, for example, making use of contractual flexibilities in order to ensure (and in some cases to maximize) their financial return.As shown above, they put pressure on commissioners to increase referrals and to renegotiate target numbers and on providers to achieve outcomes.Words such as "targets", "incentives" and "pragmatism" were used by or about investors (e.g.R05, R15, R19).The individualist outlook relies on competition and hence measurement to "keep score", but this focus on performance also has hierarchical aspects.Nevertheless, investors' outlooks were tempered by a willingness to learn about the cohort's needs (even accompanying a street outreach team on a night shift (R07, p. 43)) and to forgo maximum repayments in favour of the most appropriate outcomes (R19, p. 11).
Commissioners expressed naturally hierarchical outlooks focussing on due process and contractual stability, but evidence suggests that participating in a SIB diluted their hierarchical powers.They were concerned about procurement rules and possible conflicts of interest (e.g.R05, p. 86).Once they had committed to pay for outcomes, commissioners had fewer direct performance levers to pull than in a traditional service contract.In order to counter opportunistic gaming, therefore, commissioners tended to get closely involved in data validation, despite the increased management costs involved (R05, R16, R22).Commissioners expressed surprise at the inclination of investors to challenge the contractual terms (R05, p. 121) and were disconcerted by having their "shots called by a private investor" (R21, p. 65).
Provider organizations showed egalitarian characteristics, stressing the importance of meeting clients' needs rather than predefined outcomes.They were characterized by words such as "ethos", "needs" and "relationships" (e.g.R05, R08, R22).Most entered the SIB unused to rigorous PM and found the evidential requirements onerous.Nevertheless, providers were sometimes surprised by the achievements that an outcomes focus could bring, as noted earlier.In some ways, however, the outcomes framework aligned well with their egalitarian viewpoint.Being judged on outcomes rather than process meant that providers were empowered to draw flexibly on a range of services to meet clients' needs.In that way, SIBs allowed some of the "joined-up" cross-silo working that is often absent in the public sector.While providers fell most naturally into the egalitarian quadrant, they also found themselves adapting to "individualistic" expectations such as building capability in PM, business plans and outcome projections.And, some providers regretted that a more hierarchical ("topdown") approach to referrers was not possible (R18, p. 51).
While these broad characterizations of each partner type were remarkably consistent across the sources, viewpoints varied somewhat as demonstrated in the Findings section.Nor would we expect views to be identical, given the variety of organizations within each partner type [5].Providers, for example, range from national charities to small local enterprises.Some projects relied greatly on volunteers and others solely on paid staff.There was limited information about individual investors in the reports, but the profile of social investment in the UK makes it likely that investors were motivated as much by social as by purely financial considerations (Williams, 2020).Commissioners and cocommissioners included local authorities, clinical commissioning groups and central government departments, with different degrees of financial and hierarchical power visa-vis other partners.Given the variable amount of information about individual partners in the sources, it is beyond the scope of this study to detect systematic differences within partner types.It would be interesting to explore these aspects more fully via primary research.
Contrary to the predictions of CT, the evidence suggests that relationships within SIBs tended to improve over time.None of the SIBs in this study ended early (though as noted, SIBs that did not get off the ground are not included here) and many evaluations referred to initial tensions being overcome.This could be seen as evidence of an egalitarian world view prevailing, as all stakeholders spoke of the importance of long-term "relationship-building" and the importance of learning from others.But this resilience often seems to have been because one or more of the partners adapted to the interests of others.In particular, providers often adapted to the individualistic world views of investors.There were financial incentives for such adaptation (both in fulfilling the existing contract and in hopes for future funding) and also the reputational benefits of becoming more results-orientated and less "complacent" (R22, p. 82).And commissioners, as already noted, conformed at IJPSM 34,3 times to investors' pressure to renegotiate contractual details and to reconfigure outcome numbers.

Fatalism
The fourth world view, fatalism, has not so far been discussed.How relevant is it in the context of SIBs?Fatalists are defined as passive rule takers who do not value or choose their own way of life.A significant danger of SIBs is that the pressure to achieve paid outcomes can push two groups of stakeholders into fatalism.These are the front-line workers and the clients.
SIBs varied in the pressure that was put on the front-line workers, but in some (though not all) cases the outcomes framework led them to focus on clients more likely to achieve payable outcomes and to concentrate on the outcomes most likely to achieve a rapid return.Case workers felt that their work was not valued when it did not lead to payable outcomes.Thus, there is a risk that SIBs will reduce the professional autonomy of front-line workers, leading to demoralization.
The clients interviewed appreciated the services offered by SIBs, seeing their range and scope as something out of the ordinary amongst public services and valuing the dedication and persistence of their case workers.Thus, the SIBs did not lead to an obviously fatalistic outlook amongst interviewees (who did not include clients who left the programmes).Nevertheless, none of the SIBs appeared to have included clients in governance structures and none of the reports mentioned systematic ways of collecting clients' views or complaints.
The outcomes focus of SIBs, then, reinforces the tendency of the other world views to see the vulnerable in the population as fatalists, that is, as people to whom things are done (Thompson et al., 1990, pp. 94-95).If the main concern of the SIB is to generate outcome payments in order to "ensure financial sustainability" and to "grow the market", then the danger is, as Mildred Warner and others have argued, that clients' interests become secondary to the financialized outcomes that they represent (Warner, 2013;Tse and Warner, 2020;Lake, 2016;Morley, 2019).

Theoretical and practical conclusions
The theoretical contribution of this study is to apply CT to the three-way hybrid partnerships known as SIBs.It contributes both to the CT literature which has rarely considered three-way alliances and to the public management literature by seeking to understand the ways in which the culture of partner organizations impacts on PM.The study also has implications for the practical management of such partnerships.
As predicted by the theory, stakeholders from the public, private and third sectors tended to bring characteristic world views to the partnerships, though, as noted above, the diversity of partner organizations is reflected in a range of cultural viewpoints within these broad characterizations.We can agree with Patel (2007) that the viability of such alliances is not compromised by the inclusion of all three world views.Nevertheless, SIBs are still uncommon in the UK public sector landscape and are unusualperhaps uniqueamongst collaborations in having a well-defined lifespan.Thus, they may not wholly contradict the prediction that such three-way alliances are "rare and extremely short-lived" (Thomson et al., 1990, p. 92).
In many cases, SIBs' resilience can be explained by tendency of providers and commissioners to conform to the interests of the investors and it could appear that the individualist outlook won out.But this would be too simplistic.Each world view contributes something vital to the partnership, such as the commissioner's (hierarchical) desire to uphold the contractual terms and protect the taxpayer's interest and the provider's (egalitarian) insistence that the social goals remain the central focus.PM must somehow reconcile these competing interests.The fact that this did not prove easy confirms CT's prediction of "rivalrous co-existence" of organizations with differing world views (Patel, 2007).We should also note that the decision to participate in a SIB partnership demonstrates a pre-existing disposition to collaborate with organizations with different institutional logics.Thus, we would neither expect to find partners with extreme cultural world views nor ones that were wholly resistant to change.Early evidence suggests that SIB partnerships across the world have demonstrated flexibility in their response to the coronavirus crisis and have benefited both from the quality of the intraorganizational relationships and the in-built contractual flexibilities to maintain and adapt their services (Hameed and FitzGerald, 2020).
A practical lesson from this study is that while OBC offered the hope of management simplification, in practice PM imposed considerable administrative burdens across SIB partnerships.Moreover, the evidence suggests that the inherently outcome-based framework of SIBs imposed a traditionally linear type of PM, ignoring the well-known problems with such reductionist approaches (Bevan and Hood, 2006;Lowe and Wilson, 2017;Fox and Morris, 2020).The collaborative nature of SIBs may also be something of an illusion, as design and implementation processes tended to reinforce existing power asymmetries (Lowe et al., 2019;Morley, 2019).Initial negotiations between partners excluded the contributions of operational and delivery staff as well as potential clients and the wider community.This may change with initiatives such as the UK Life Chances Fund, which explicitly requires projects to have local VCSE and client involvement [6].The next generation of SIBs may reveal whether some of these lessons have been learned.
The definition of SIB varies between sources.In this study, all SIBs listed in the quoted sources are included.Two organizations (Thames Reach and St Mungo's) are listed under the "London Homelessness SIB".Here, they are counted as two SIBs as they worked under separate contracts with the commissioner, the Greater R22).Similarly, the two London Homelessness SIBs included outcomes such as employment and health, in addition to entering and sustaining accommodation (R07). ( hey ask me 'but you are going to get paid for that, you want me to go back [be repatriated], if you get rid of me you are going to get a payment'" (R08, p. 41).