Social impact bonds and public service reform: back to the future of New Public Management?

Abstract This article explores where the increasing adoption of SIBs and outcomes-based contracting may lead public service systems—toward New Public Governance, or ‘back’ to New Public Management. We present analysis from the first significant longitudinal qualitative study of a major UK SIB focused on improving outcomes in the context of social determinants of health to analyze how the two governance logics manifest and interact across the SIB lifecourse. We find that while both governance logics were present at initiation, over time NPM elements strengthened while NPG elements weakened. Two inalienable elements of the SIB model—investor power and data requirements for contract management—appeared to drive this change. Our findings provide evidence that SIBs promote a retrenchment of NPM, rather than a transition to NPG or a hybridization of the two governance logics. Findings also show how NPM, rather than a transitional stage toward NPG, can prove the more resilient and dominant governance logic within institutional forms.


Introduction
Social impact bonds (SIBs) are a distinctive and growing element of contemporary public service reform, however their role-whether entrenching New Public Management (NPM) or facilitating a move beyond it-remains contested. Discussion has been polarized between critics who emphasize the limitations of SIBs as an extenuation of NPM and those who perceive it to have scope for enhancing collaboration, experimentation and learning in an NPG context. Importantly, SIBs not only embody governance logic, but enact it, cumulatively shaping future public service landscapes. SIBs may facilitate a transition to NPG, with NPM proving a transitional phase as Osborne (2010) predicted, or conversely may inculcate NPM and marginalize NPG elements over time. It may also be that NPM and NPG logics hybridize into stable configurations, as has been found with other initiatives (Wiesel and Modell 2014;Wollmann 2018).
While studies have explored the conceptual similarity of SIBs to NPM and NPG (Albertson et al. 2020;Dayson, Fraser, and Lowe 2020;Joy and Shields 2013;Ormiston et al. 2020), none have analyzed how the two governance logics interact over time. Because most SIBs are relatively recent phenomena, most studies rely on anecdotal accounts (Hajer 2020), secondary data (Chiapello and Knoll 2020;Ormiston et al. 2020) or snapshot information (Hevenstone and von Bergen 2020). By analyzing the development, operation and stabilization of a SIB longitudinally we seek to uncover the temporal dynamics of NPG and NPM in more granular detail.
We explore how NPM and NPG logics manifest and interact over time by analyzing three key functional distinctions between the two governance logics: inter-organizational relationships, performance emphasis, and governance mechanisms (Osborne 2006(Osborne , 2010. We apply this schema in a longitudinal qualitative analysis of a major UK SIB focused on reducing secondary care demand through a large-scale social prescribing intervention. These distinctions are explored through multiple perspectives of key SIB stakeholders at initiation, development, and stable operation phases, providing a holistic and evolutionary perspective of how institutional dynamics within SIBs facilitate a move toward NPM or NPG.

Literature: social impact bonds and public service reform
The classic SIB model involves an investor who provides working capital for a social intervention which, if it meets specified outcome measures, will generate a rate of return on investment. Since the first SIB launched in the UK, a growing cohort of investors, government organizations and other service providers have moved into the SIB domain, increasing investments while also tweaking aspects of the SIB model (Carter et al. 2018;Fraser et al. 2020;Wilson et al. 2020). The Brookings Institution estimates that by March 2022, 225 impact bonds had been contracted across 37 countries, totalling close to half a billion dollars in investment. Despite much heralding of SIBs as a significant innovation in the funding of public services some have considered SIBs an extenuation of New Public Management, the current of service reform which swept through Anglosphere in the late 1980s and 1990s (McHugh et al. 2013;Sinclair, McHugh, and Roy 2021). Hood's (1991) original conceptualization of NPM described seven doctrinal components: hands-on professional management, explicit standards and measures of performance, emphasis on output controls, disaggregation of public services, increased competition in service systems, the adoption private sector management styles, and a greater emphasis on discipline and parsimony in resource use.
SIBs combine NPM's focus on performance management with an increase in private sector involvement through social finance and external investor involvement (Edmiston and Nicholls 2018). SIBs also carry forward key elements of NPM's managerial emphasis and customer-oriented value system in the way that they use contracting and associated measurement tools to structure payments (Joy and Shields 2013;McHugh et al. 2013;Warner 2013). Like NPM reforms, SIBs are often justified as a mechanism for promoting an entrepreneurial approach to public management (Mulgan et al. 2011;Dowling 2017), while transferring the risk of innovation to private investors.
Unlike NPM reforms however, SIBs-in a UK context at least-are a public sector innovation rather than a private sector management technology, emerging in response to the distinctive characteristics and missions of organizations delivering services to citizens in need of care or welfare support centered around the UK Cabinet Office. Where cumulative NPM reforms have been claimed to have led to a fragmentation and layering of public sector functions as agencies were broken up or split into purchaser-provider dichotomies, SIBs were created to respond to precisely these environments. SIBs intend to catalyze cross-sectoral synergy, promote cohesive planning and design processes across boundaries and align incentives for collaboration (Carter et al. 2018).
Many authors have also discussed the affinity of SIBs with elements of the New Public Governance (Albertson et al. 2020;Joy and Shields 2013;Ormiston et al. 2020). Osborne (2006Osborne ( , 2010 defines NPG as a plural (multi-actor) and pluralist (multi-process) approach to governance driven by relational rather than contractual dynamics amongst organizations. SIBs embody elements of NPG's 'neo-corporatist' approach, carrying multiple heterogenous actors in long-term contractual engagement (Carter et al. 2018), which may permit the development of trusting relationships and a focus on inter-organizational working (Albertson et al. 2020).
Osborne positioned NPG as a descriptive frame for understanding and analyzing governance development, and also a basis for reappraising the relevancy of social initiatives in a public service landscape dramatically changed by many years of NPM-derived reforms. In early descriptions (Osborne 2006(Osborne , 2010, Osborne distinguished between NPG and NPM on their theoretical basis, the assumed nature of the state and governance, their contrasting value sets, and through key functional distinctions between the focus, emphasis, and mechanisms of governance. We consider in the following section how SIBs may interact with NPM and NPG across three key functional domains which Osborne described.

Governance mechanisms
First, NPG and NPM are differentiated through their means of control and purposeful action, what Osborne (2006) describe as the 'mechanisms' of governance. NPM adopted a contractual basis which presumed divergent interests between providers and suppliers of public service, and in line with Public Choice Theory (Buchanan and Tullock 1965), that public servants' innate selfinterest required the regulation and inspection of behavior. NPM relied on binding contracts structured using results-linked incentives to motivate behavior, often with detailed key performance indicators (KPIs) attached to ensure enforceability. Contractual relationships were structured along agonistic principal-agent dichotomies based on Agency Theory (Jensen and Meckling 1976), wherein a powerful actor (the principal) must limit the autonomy of a subsidiary (the agent) through monitoring, inspection, and results controls. SIBs by design involve results-linked targets and detailed contracts between participating actors, and so at a surface level conform strongly to this NPM logic.
In place of creating detailed binding contracts to facilitate inter-organizational working, NPG instead advocates building enduring inter-organizational relationships by fostering trust and relational capital (Osborne 2006(Osborne , 2010. Conceptually, SIBs provide structures for relating which are long-term, which build trust and enable adaptation and experimentation through managing relational dynamics of governance actors (Albertson et al. 2020;Carter et al. 2018). Contracting for outcomes rather than processes and outputs theoretically enables innovation and experimentation toward high-level goals which may take years to achieve (Fraser et al. 2018). However, Osborne's identification of preferred suppliers and trust-based working as a mechanism for contracting sits uneasily alongside the reliance of SIBs on complex results-linked contracting arrangements. This element might create further purchaser-provider dichotomies rather than foster trust and decentralized accountability relationships (Hevenstone and von Bergen 2020).

Inter-organizational dynamics
The second distinction between NPM and NPG concerns the nature of inter-organizational relationships. NPM understands competition as the best route to improve services, and thus promotes a beneficial competitive dynamic between organizations. Responding to its decentralized and polycentric governance context, NPG places emphasis on the quality of relationships between organizations and fostering collaboration (Osborne 2010, p. 97). While accountability relationships are stipulated on results or hierarchies within NPM, under NPG accountabilities are mediated recursively and horizontally amongst a broader range of inter-dependent partners (Torfing and Triantafillou 2013).
SIBs are sometimes set within an NPM discourse-they inculcate a market discipline through inculcating a focus on results, which in turn should enable organizations to win contracts and compete successfully for funding (Muñoz and Kimmitt 2019). However, SIBs may also be positioned within a NPG-related discourse of collaboration, providing a model for cross-sectoral partnership which brokers a broad range of expertise from commissioners, investors, delivery organizations and intermediaries (Albertson et al. 2020;Ormiston et al. 2020).
The incentive structure accompanying outcomes-based contracting is argued to align the divergent interests of these parties, reducing risk for providers and commissioners while providing a return on capital for investors (Fraser et al. 2018). However, some consider that uneven power dynamics between investors and subsidiary 'provider' organizations prohibit genuine collaboration (Joy and Shields 2013;Ormiston et al. 2020). As Budd (2007) noted, there is also a tension between the need to manage decentralized networks of organizations and the need for organizations to conform to audited targets, a key feature of SIBs. The often-onerous performance information systems required for outcome-based governance have also been argued to create barriers to meaningful inter-organizational collaboration (Jamieson et al. 2020).

Performance emphasis
The final distinction lies in what Osborne calls the 'emphasis' of NPM and NPG. NPM particularly in its early iterations, promoted a form of intra-organizational output control and an emphasis on process efficiency (Hood 1991). NPG involves a shift in the emphasis of performance from intra-organizational efficiency and outputs toward inter-organizational processes and outcomes (Osborne 2010). This involves a strategic reorientation from internal to external effectiveness and the development and maintenance of relationships with key external actors.
SIBs align with NPG's performance emphasis with outcomes functioning as both the object of service design and, in the classic model, the trigger for investor repayments, providing a context for both financial and non-financial outcomes to operate together (Linderm€ uller, Sohn, and Hirsch 2020). An outcomes-focus is argued to drive a focus on citizen needs and public value  Osborne (2006Osborne ( , 2010 (Carter et al. 2018), while also providing a mechanism for tackling goals which cross traditional institutional boundaries (French and Mollinger-Sahba 2021). However, even in this domain SIBs express NPM logic. Authors suggest SIBs offer little opportunity for citizens or service users to influence SIB aims or shape service encounters (Maier and Meyer 2017;Warner 2013), a feature necessary within NPG to co-create value for differentiated citizens (Torfing and Triantafillou 2013). Strictly speaking, payment is attached not to the achievement of outcomes, but to the demonstration of outcome data. This is a critical distinction, since instead of promoting client-focused value creation, outcome incentives can promote perverse incentives for gaming, myopia or the forgery of performance information (Bevan and Hood 2006). SIBs involve multiple partners and provide a potential apparatus for inter-organizational strategic activity. Compared with alternative contracting arrangements like service-level block grants, SIBs are held to promote information sharing and free providers to innovate beyond their immediate boundaries (Carter et al. 2018). Others warn that SIB contracts may have become adopted for coercive intra-organizational management practices as providers become compelled to 'stay on track' (Edmiston and Nicholls 2018) and produce outcomes data for payments (Jamieson et al. 2020). In this view, deepening links and information exchange between agencies may not lead to a continuous process of information exchange and mutual influence amongst actors as with NPG, but instead a functional disaggregation and specialization of functions as with NPM.

Summary
At a conceptual level, our discussion has shown SIBs embody a dualistic conceptual identity, able to take on elements of either governance logic. Table 1 describes how the contrasting characterizations of SIBs feature within the three key functional distinctions between the NPM and NPG.
Importantly, since the cumulative effects of service reforms shape the governance landscape and the available policy responses, NPM and NPG can be considered governance destinations as well as governance logics. A significant question is which destination-NPG or NPM-the proliferation of SIBs might lead to in practice. Almost all literature discussed has focused on whether elements of NPG or NPM are embodied by SIBs rather than enacted by SIBs, and none shows how the two governance logics interact across the SIB life course. Since this question concerns the management practices and impacts of SIBs it has an innate longitudinal dimension, since that the relationship between NPM and NPG within SIBs is fluid rather than fixed and the logics may either blend or diverge over time.
Osborne argued that NPM was a transitionary stage on route to a steady state of NPG, with the effects of NPM-fragmentation, layering and functional specialization-necessitating the embrace of NPG. SIBs may emerge as one element of this transition, enabling public sector landscapes to renew and update their public service systems to meet the demands of contemporary service reform. Conversely, SIBs may reaffirm and further entrench NPM values and mechanisms replicating the macroeconomic effects of NPM and resisting Osborne's hypothesized transition. Both eventualities presume that NPM and NPG operate as polarizing mechanisms, where in other studies the two have been found to coexist (Torfing and Triantafillou 2013) and indeed to hybridize into stable organizational forms (Wiesel and Modell 2014). Our study therefore tackles a central research question of relevance both to the SIB and broader governance literatures: do SIBs lead over time to a reaffirmation of NPM, a transition to NPG, or to a hybridization of both governance logics?

Methodology
To answer our research question, we present findings from a longitudinal qualitative study of a major UK SIB health-based SIB focused on improving the Social Determinants of Health (SDH) for a place-based cohort of patients with a range of chronic conditions in a Northern City in the UK. The SIB brought together a range of partners into a seven-year programme: investors who put forward the working capital; four service providers who were contracted to deliver the social intervention comprised of link-workers working with service users to improve their wellbeing; a special purpose vehicle (SPV) which provided operational leadership and managed relationships with providers; GP practices through which referrals to the service are produced; and a local state-run health commissioner, the eventual outcome payer.
The SIB had two key payment-linked outcomes (1) primary care referral rates, programme recruitment and ongoing wellbeing improvements and; (2) reduction in the number of hospital visits (measured through comparison with a control group in a neighboring locality). The first outcome was measured through a shared performance information system incorporating a tool developed by Triangle Consulting called the Well-being Star (MacKeith 2011). The Well-being Star is used to help manage long-term health conditions by measuring and tracking self-reported scores across eight outcome areas associated with individual wellbeing. Working capital was provided by two large social investors, one of whom were involved in contract design, and supported by startup subsidy from two external funders. Payment for operations from years 1-3 were based on outcome (1), however after year 3 would transition to evidenced financial savings to the local state-run health commissioning body through outcome (2).

Data collection
Time is an often-neglected element of research in public administration and service reform (Pollitt 2008), evidenced in a SIB context by the lack of significant longitudinal analysis in published academic literature. Qualitative longitudinal analysis is appropriate where exogenous change, critical events or evolutionary dynamics are important facets of phenomena of interest (Van de Ven and Poole 1995), and therefore relevant to SIBs, which often take many years to develop and stabilize.
Based on a temporal bracketing strategy (Langley 1999), our findings identified three distinct critical periods in the SIB's development: Stage 1: development (2011)(2012)(2013)(2014)(2015), Stage 2: initial operation and maturation (2016-2017), and Stage 3: stable operation (2018-2019), allowing us to analyze how NPM and NPG governance logics manifest and interact across the SIB development cycle. We draw from 31 semi-structured interviews across these three data points spanning all key partners (design consultants, investors, commissioners, funders, providers, and staff from the special purpose vehicle). Participants (described in Table 2) covered all key leadership and operational roles in the SIB. In Stages 2 and 3, all participants were re-interviewed, with the exception of design consultants and the government SIB funder whose operational role terminated at Stage 1. By interviewing the

Interviewee
Interviews at stages SPV Board Chair S1, S2, S3 SPV Board Member S3 SPV Manager S1, S2, S3 Provider #1 S1, S2, S3 Provider #2 S1, S2, Provider #3 (exited at S3) S1, S2, S3 Provider #4 (exited at S3) S1, S2 Investor S1, S2, S3 SIB development consultant #1 S1 SIB development consultant #2 S1 SIB development consultant #3 S1 Performance Information System Manager S1, S2, S3 Government SIB funder S1 Commissioner #1 S1, S2, S3 Commissioner #2 S1, S2, S3 same participants, we could compare the interpretation and valuation of key events in the SIB timeline within individual participants and between actors over time. An incoming SPV board member was additionally interviewed at Stage 3. It was not possible to reach Providers 2 and 4 at Stage 3, owing to unavailability and exit respectively. The lack of significant alterations to viewpoints and affiliations to NPM or NPG over time amongst SIB actors generally and between Providers in particular, strongly suggest this would not have significantly affected study findings. Interviews lasted between 1 and 2 hours, and focused questions on the SIB's development and participants' experiences and perspectives. Interview questions in all three rounds centered on the three issues distinguishing governance logics in [Table 1]: the interrelationships between different actors within the SIB, and the role and value of contracting arrangements and payment mechanisms. Interviewees were asked about the function and value of these processes, enabling a comparison of interpretation of shared events between SIB actors. In rounds 2 and 3, interviews also probed the effects of key changes in the SIB process (in stage 3, for example, changes in payment mechanisms for providers, and the switch to outcomes payments for the SPV). This allowed us to affiliate justifications and value-assignments with NPM and NPG governance logics and monitor how the inter-related across the SIB life course (although we found little change in overall affiliation to NPG or NPM over time

Analysis
We frame our analysis following two components of our research question: How do NPM and NPG governance logics manifest across these three functional distinctions identified? How do these logics interact across the SIB lifecourse? Our longitudinal analysis was designed to capture the evolution of events and changes of views and perceptions of all key actors. This enables us to examine a deeper level of analysis linking events and perspectives to provide a more rounded analysis of the SIB development process (Van de Ven and Poole 1995).
At each of the three data points, we organized data using descriptive codes to map any new developments or changes in perspectives of actors at the three functional distinctions. We analyzed data in Nvivo to explore how governance logics were embodied within key events (e.g., providers dropping out in Stage 3) and linked these to perspectives of key actors at key points (e.g., investor perspectives on provider drop out at Stage 3). We analyzed the evolutionary dynamics within the SIB, looking for instance regarding any changes in perspectives, relationships amongst stakeholders, or interpretations of key events over time. We then adopted a thematic coding approach wherein different descriptive codes were attached to either NPM or NPG at either stage.
Our analysis followed an abductive approach, moving back and forth between inductive (datadriven) and deductive (theory-led) insights (Gioia, Corley, and Hamilton 2013). Our initial coding round was inductive, allowing a nuanced development and interpretation of interactions across NPM and NPG domains, allowing us to later abstract our findings around the three aforementioned functional distinctions. We finally brought together codes to characterize aforementioned functional distinctions between NPM and NPG as either co-existing (i.e., both NPM and NPG logics present and stable across time periods), reinforcing (one logic enhancing the other), in tension (one logic diminishing the other), or segregating (one logic emerges as dominant while the other is squeezed out).

Findings
Following standards in qualitative longitudinal analysis, we report findings over 3 stages of analysis: Stage 1 (2011-2015) captures the development and immediate initiation of the SIB as relationships were formed and contracts set in place. Stage 2 captures the first year of implementation, as initial operation and scaling of the SIB took precedence. Stage 3 captures the SIB in its stable implementation, having fully developed and transitioned to commissioner outcome payments. We present our findings narratively across these three stages whilst referring to examples from the raw data within.

Stage 1: complementary governance logics
Governance mechanisms: reinforcing logics The initial development process focused on the design of complex contractual arrangements between SIB partners. Initial payments to the SPV derived from contracts held with the commissioners, investors, and external funders. Providers were sub-contracted by the SPV with payment based on a base payment, and an additional variable payment whose rates were determined by performance on a range of 18 KPIs. The payment structure for providers was designed such that satisfaction of these KPIs was essential to covering operating costs and thus ensuring their financial viability.
The emphasis on contracts and KPIs as the basis for inter-organizational relationships suggested a strong alignment with NPM, with clear principal-agent divisions between funders and the SPV, and the SPV and provider organizations. However, we found a contrasting NPG logic underpinning contract design. Prior to the SIB formation a pilot project bringing together key actors-commissioners, delivery organizations, and supportive infrastructure (e.g., primary care and community sector organizations) cemented relationships among key stakeholders prior to SIB formation. Contract design was not a control tool, but an approach to stabilize preexisting relationships and secure finance to scale a partnership endeavor As the SPV board chair emphasized: One of the issues I suppose is the recognition that actually ( … ) the current competitive commissioning model isn't necessarily the best model in terms of social value. We need a more collaborative model, particularly if we're going to make best use of all available local assets. (SPV Board Chair) In this way, the preexisting relational context was carried alongside into the SIB structure. This intent was evident in shaping many early actions and decisions in Stage 1, significantly in recruitment of partners. Consideration of value alignment and likely trust informed selection decisions including the recruitment of the investor and the four service delivery organizations.
In this sense NPM elements were present but were seen as complementary to NPG elements of relational working and trust between partners. However, while this held for SPV and provider organizations, investors were singled out as holding a potential conflict of interest, functioning as both SPV board members and as capital investors, which was seen by some as a conflict of interest. In the contract design, significant reserve powers were stipulated by investors, including to remove key SPV staff and board members. While seen as a recourse in dire straits and justified based on the financial risk taken, this fed a mistrust of investor power by several interviewees.
Inter-organizational dynamics: co-existing logics We found broad recognition of interdependencies in the design of the SIB at Stage 1. Investors had contracting expertise but had never worked in clinical settings. Providers had preexisting experience with social prescribing, but neither this group nor commissioners had engaged in outcomes-based contracting. The complementarity of skillsets was a key feature of SIB design, including in the tendering of provider contracts.
We found contested rationales for the decision to appoint four providers, rather than one single agency, or for the SIB SPV delivering the service directly. For investors and certain SPV board members, this choice was rooted in NPM logic, providing an opportunity to test out four competing models and improve practice through a competitive dynamic amongst providers: I think that is, to an extent, incentivised through quite ambitious targets that we all agreed. We knew that the only way we can achieve them is by working really hard. We don't really sleep … being able to meet really aggressive referral targets, which were aggressive, we knew they were, has been great. (Investor) The provision of a standardized suite of KPIs, through its developing performance information system could lead to identification of poor performers.
The SPV Manager and certain board members aligned instead with an NPG logic, considering four providers would bring diversity and pluralism of experience, and provide opportunities for improvement through practice-sharing and collaboration: I think my view in terms of, you know, if you want to scale up a SIB it is much better if you can identify local assets, local people who have trusted credibility within the local system than parachute in ( … ) big national organisations who will not necessarily identify the best people within the patch. So having that local drive and that local understanding, I think is important. (SPV Board Chair) A bi-monthly 'provider forum' involving the SPV Manager and providers was designed as a learning and sharing space for providers. The SPV sought to safeguard collaborative dynamics and would not discuss provider performance publicly for fear of creating a 'league table' situation.
Performance emphasis: reinforcing logics The SIB was designed with a focus on reduction of secondary care, and this outcome-orientation guided the development process, in line with NPG. However, the SIB would first need to reach a scale and duration of contact necessary to achieve impact. The performance management and payment structure of the SIB was therefore designed to transition from a process to an outcome focus over time. The interim outcome of Triangle Consulting's Well-being Star was chosen as a proxy with an assumed direction of travel to its ultimate outcome, though interviewees expressed uncertainty over this link. Providers were insulated from this risk, with improvements in Well-being Stars being tied to the SPV rather than provider payments. With outcome measures playing less of a role, providers were managed in contract through a range of intra-organizational process measures, consistent with the intra-organizational focus of NPM. This was again justified as reaffirming an NPG logic-their insulation from outcome risk was seen to remove perverse incentives for 'gaming' improvements, and to free providers to focus on value-adding relationships with clients: In terms of figuring out how to create the structure that really aligns incentives amongst everybody and how do you then, again, incentivise everybody in the right way to maximise those outcomes that we're targeting? That was certainly the part that we worked really closely together on and we were able to bring all of our experience from the other projects that we have. (Investor) The Well-being Star system was cited as helpful by some providers, facilitating value-creating conversations and focusing on user-determined outcomes, in line with NPG. However, providers noted a significant administrative burden arising from prerogatives to capture data, and some practice-distorting effects were evident. As one provider described: So, you spend a lot of time trying to work out what it is you're supposed to be doing because you're not too sure how many Well-being Stars need to be done this week. Which is half a day's work, when you could have had half a day seeing your patient or half a day doing your Well-being Star. (Provider #1) Each provider could point to times in which this data burden would squeeze outpatient time and room for a client-focus within NPG, however there were hopes that once the performance information system was further developed, that this would ease.

Stage 2: early stage NPM drift
Governance mechanisms: in tension Providers considered certain targets to be arbitrarily difficult-the 92% referral achievement target for instance was far set higher than rates achieved by other social prescribing interventions. The establishment of a bespoke Performance Information System (PIS) and regularization of reporting to the SPV board in Year 1 put into spotlight that three of the four providers were struggling to meet targets. Board meetings became more focused on the inspection of results consistent with a principal-agent structure, where under-performance prompted more active intervention. Reflecting on this, the SPV Manager stated: I hoped it would be enough to just say it verbally, they weren't closing the gap with the KPIs ( … ). I've been trying to build relationships and trust, but you can't take years to do that when performance is failing. ( … ) so I put them side by side and a lot of them found it really, really hard to see and thought it was overly aggressive and overly accusatory. (SPV Manager) In this way the strengthening of the requirements for performance information tightly coupled the principal-agent dynamics.
The SPV manager was put under pressure by investors to take a more interventionist approach with poor performers, which conflicted with the collaborative dynamics the SPV attempted to foster with providers. The SPV Manager attempted to reconcile these demands by acting as a 'buffer' between investor and certain board members, who felt providers were underperforming based on data they received, and providers, who felt data was hampering their service delivery. The SPV Manager described the tensions in being required to play two conflicting roles-as contract manager regularly holding performance-focused conversations with staff-and as a peer helping practitioners navigate a new form of working.
Inter-organizational dynamics: logics in tension Regular provider forum meetings enabled providers to share practice and negotiate the development and scaling process, and we found some examples of more formalized joint working, for instance the voluntary sharing of link workers between provider organizations to cover staff absences. However, in place of maturing and deepening collaboration, we instead found collaborative dynamics diminished over time amongst providers and between providers and the SPV. While interest in partnership and collegiality was expressed by all providers at interview, in general, the scope for improvement through learning and sharing, or closer forms of joint working was considered limited. This was put down partially to the design of the SIB-they had discrete caseloads (partially as an attempt to stave off competitive dynamics) and possessed limited system-shaping authority through their distance from decision making authority held by the SPV board. Providers felt that meetings became increasingly about contract management and process compliance rather than generative conversations about collaboration. As one provider outlined: One of the contracts we have is five times the size of [the SIB], and there's probably quarterly contract review meetings and then a couple of possible emails coming in for different things. And my first month, whole month, May, of being then more involved, I was seeing a volume of about twenty emails coming in to me directly, which obviously were generating conversations. And the link worker had about 24 or so emails, some of them were the same, some of them weren't, and it just struck me that ( … ) it was very unusual from my experience as both a commissioner and then as a provider for the intensity of the relationship. (Provider #3) Changing provider behavior to focus on measurement for the SIB was, the SPV manager said, an 'uphill battle from the start'. There was one exception to this, in that providers did collaborate to resist certain KPIs seen as counterproductive to the work. The most significant example of this was in the collective opposition to the KPI to complete a first Well-being Star and action plan 20 days from first contact, seen as arbitrary and unhelpful for the SIB mission. NPG-linked collaborative dynamics were therefore initiated to resist the imposition of NPM logic. However, provider resistance in this context was interpreted as obstinance, particularly by investors who felt they were underperforming. Consequently, longstanding concerns for providers, for instance the significant issue of lack of control over referral generation, remained unchanged.
Performance emphasis: logics in tension While providers were insulated from wellbeing outcome risk, the difficulty of KPIs linked to variable payments led to three providers struggling to cover operating costs. Providers reported a pressure to shift from value-creating conversations to focus on moving clients through KPIs, shifting the nature of data collection from improving to proving: So I sometimes felt that [link workers] were doing what was important to the people that were using the service and it was led by the people that were using the service but they didn't do the little things that trigger the payments, like make that final phone call to say where are you at now, so when I say gaming I don't mean that in an unpleasant or vicious way at all. I mean it just in terms of gaming because it is gaming a bit more than dotting the I's and crossing the T's, it's not necessarily for the benefit of the service user that you're making that final contact or that extra contact. (Provider #3) However, all felt that the user-oriented and value-creating nature of the service central to NPG logic had not been fully crowded out.
A second barrier to NPG logic was the unavailability and inaccessibility of data to providers. The link between interim outcomes and end outcomes was highly uncertain, with the SPV manager professing, 'it's going to be really hard to say what patients we're helping in what ways'. At the beginning of the programme service provider organization felt they could help with sensemaking how their service users were progressing toward the delivery of the programmes secondary care outcomes. However, some providers felt excluded from assessing secondary care data or client data even in the aggregate: I suppose the other bit that … when it was set up, that bit about the secondary [outcome] data was seen as something that they didn't need trouble us with, as providers. And for us it was, kind of, like, "Well, if that's ultimately what you want us to have an impact, then the more we can understand that, the more we can influence it in what it is we are offering to people," with that one. ( … ) So, naively [laughs], we thought there were more granular details we'd be able to work with. (Provider #1) The measures which the SPV and providers were reporting were also largely process-driven (e.g., numbers of service users recruited and assessed by link workers), rather than using pooled intelligence to improve the support of a long-term shift to service user's sustainably improving their wellbeing.

Stage 3: NPM dominance
Governance mechanisms: logics segregating By Stage 3 the SIB had reached a level of stable delivery, with payments drawn from commissioners from the achievement of secondary outcomes. For much of Stage 3, secondary outcome data showed no significant difference between the SIB intervention and control group. Following discovery of a data error however, it was revealed that secondary outcome data was indeed favorable to the SIB. It was against this backdrop of cautious optimism about SIB 'success' that interviews were conducted.
Providers noted the emphasis on control which emerged in Stage 2 had continued and deepened. Investors expressed frustration with the SPV's perceived reticence to impose control and undertake more hands-on performance management of providers in response to KPIs. The SPV Manager related: So, this is an example of something where the investor saw that starting to change in terms of not meeting the target, and we only had a couple of months where we hadn't met the target and eleven months where we had, but that to them was something that the team needed to act really quickly with and start to put pressure on the providers ( … ) So, we did ask the question of the providers, "What do you think is going on? Why do you think this has dropped off?" But to put undue pressure on them, we felt wasn't appropriate ( … ) At any rate, there became a point where the investors felt that as a staff team, we weren't responding to what the issues were with enough urgency. (SPV Manager) The SPV Manager felt powerless to resist investors because of their relative power, particularly their reserve contract power to remove key staff from the SPV board and staff team. While NPM elements had always been present, the buffering role played by the SPV to accommodate NPG, had by the end of Stage 3 been squeezed out.
The interventionist approach from investors was spurred by the availability of data to investors through the PIS. While the SPV manager could understand context alongside KPIs (e.g., whether deterioration was a blip or signaled a trend), appraising the relevancy of the KPIs to mission, the SPV board in line with NPM wanted corrective action taken such that performance standards were brought up to par. This development was also behind a significant structural change in the revision of the fixed payment to an outcome basis for providers, making payment subject to referral generation and incentivising long-term engagement. The earlier justification, of freeing providers from risk, was replaced by an intention to make providers take greater responsibility for their performance: Yes, we did shift the payment ( … ) basically the point being that really the focus for everybody, we want to make sure that all of the incentives are aligned towards really working really hard to keep the patients for a long time and to work with them for that period of long time. (Investor) Inter-organizational dynamics: logics segregating By Stage 3, opportunities for inter-organizational collaboration had diminished significantly. The provider forum which facilitated cross-organizational sharing and joint working diminished in frequency in Stage 2 and stopped altogether early in Stage 3. Two providers subsequently dropped out of the SIB, one early and one late in Stage 3, with their caseloads split amongst the remaining providers. One exiting provider in interview cited their ongoing struggle to cover costs of delivery, with both considering the data management burden and payment structure making the financial situation untenable. Dissatisfaction with contracts and KPIs were cited as secondary reasons for leaving, with the increasing control focus from SPV and providers, and the administrative burden restricting their freedom and autonomy in service provision.
For those aligned to an NPG logic (providers, SPV staff and certain SPV board members), the loss of two providers diminished the SIB because it lessened opportunities for collaboration and the diversity of practice. This move was welcomed, and even anticipated, by the investors and certain SPV board members, who saw this as arriving at a streamlined and evidence-based service better attuned to service delivery: I think we're not saying that all of our providers were quite good at building relationships with clients. Two of them were clearly less good at building relationships with practices and didn't necessarily see it as what they should be doing. ( … ) I think my view would be that we're now in a much better position and from the point of view of the executive team, contract managing two providers is easier than contract managing four providers. (SPV Chair) Performance emphasis: logics segregating By Stage 3, secondary outcome payments were financing the SIB, with early data (following correction of a significant and misleading data error) showing an improvement sufficient to cover projected SIB costs were it to be sustained over time. Delivery organizations too experienced a significant change with variable payments shifting to include measured improvements in wellbeing costs. Linking outcome data to resource allocation can at one level be interpreted in NPG logic as improving alignment to user value represented in the adoption of the Well-being Star as a relational approach to scaffold encounters between service user and link workers. As the SIB reinforced an NPM logic, the relational approach to service provision enabled by the Well-being Star earlier became institutionalized a technical means to deliver data to a performance control system. Reflecting on their experience with data collection and performance management over their years of service provision, one provider offered the following: I think that the entire performance management and project management of it is just an overwhelming industry. I think from start to finish the amount of money that has been spent just on the industry under wiring creation of the SIB itself, or the SIB style thing itself, just would not bear analysis compared to the amount of money that's spent on individual service users as result of this entire activity. (Provider #2) The relegation of providers to subcontractors rather than partners, present in investor and SPV board member positions at Stage 1, had also become the established structure by this point. The tight margins and process-orientation which providers contended prevented a user orientation were understood by investors and the SPV board as a positive and necessary component of overall SIB effectiveness. Even the remaining providers found that the target and payment structure left them with little creative agency to deliver on their organizational mission and values. Comparing their experience within the SIB with other funded service provision, one provider was clear it had created an additional burden: There is nobody that gives me money to provide a service that doesn't want some evidence of what I've done with it and how you've done it ( … ) [however the SIB] have been under that much scrutiny from the social investors that they are having to provide information more quickly and in more detail. (Provider #1) While the SIB engaged the energies of multiple actors at the initial phases, instead of leading to partnership and collaboration as with NPG, it had developed through functional disaggregation and specialization, as with NPM.

Discussion: NPG and NPM logics as blending or segregating mechanisms within sibs
By exploring the evolution of events and perspectives over time across the three functional distinctions, the findings express a clear trajectory of the SIB from a mixed NPG/NPM model at Stage 1, a weakening of NPG elements in Stage 2, to an overarchingly NPM model in Stage 3. Table 3 reports the key factors which our abductive analysis reported across the three functional domains of NPM and NPG over time.
At Stage 1, 'combined logics', we found both NPG and NPM logics were present in guiding decisions around the design, recruitment, and early implementation of the SIB. NPG logic was a significant guiding factor for early decisions, spurring key events like the development of the provider forum, and guiding how providers and investors were selected for inclusion. The processfocus and administrative burden created by contract complexity, while set within NPM logic, were anticipated to reduce over time to allow NPG elements to deepen later. While NPM-oriented perspectives were evident from certain SPV board members and newly recruited investors (e.g., their competitive rationale for choosing four providers), at this stage NPM and NPG logics were blended, with NPM elements justified based on their ability to hybridize with and reinforce NPG characteristics.
By stage 2 'drift toward NPM', we find NPM and NPG logics, while still both present and coexistent, had begun to operate in tension. As the developing PIS enabled more detailed scrutiny of provider performance, investors and board members took a more interventionist approach to provider contract management and a clearer 'principal-agent' dichotomy developed between investors and the SPV. New NPG elements materialized in resistance to the strengthening of NPM logic, with the SPV Manager reconfiguring their role to preserve collaborative dynamics, and providers exerting agency to resist certain KPIs.
By Stage 3 'NPM dominance', NPG elements prominent at Stage 1 and Stage 2 had disappeared or significantly diminished, with NPM elements strengthened. The lessening of administrative burden expected from the development of the PIS did not materialize in practice for providers, who continued to be process-focused. NPM logic was by this stage motivating actions and justifying events-the exit of two providers was understood as a positive development by SPV board members/investors since it improved process efficiency. Significantly, our longitudinal analysis did not find any significant change in governance logic alignment over time amongst individual SIB actors, suggesting these changes resulted from dynamics within the SIB itself rather than changes of perspectives. Two significant elements of these dynamics were found. First, the relative power of NPM-aligned investors allowed them to dictate terms of engagement and ensure that their views held precedence. Measures taken by investors to reduce their own outcome risk-e.g., stipulating their involvement as SPV board members, or the specification of reserve clauses giving investors power to remove key staff, amplified their influence and so enabled NPM logic to influence SIB development more strongly over time (e.g., by fostering a competitive dynamic amongst providers, or enforcing a principalagent relationship between the SPV and providers). Some of these behaviors may have been driven by a view that the SIB was underperforming owing to initial signals from secondary outcome data, showing that NPG elements were tolerated so long as performance signals were positive. While perceptions of underperformance may have been an accelerant toward NPM, our data show that overall orientation toward NPM and support for tighter performance management structures remained even when secondary outcome data became favorable to the SIB and investor repayments. Indeed, this shows that NPG dynamics were tolerated so long as a positive direction of travel noted, Second, the contractual requirement for performance data created a focus on intra-organizational matters and process control. Data performed a crucial function in enabling payments to circulate and thus enabling SIB progression; however, this inevitably placed a significant administrative burden on both providers and the SPV to generate data as an integral part of ongoing service delivery. Despite the information system being envisioned as client centered, and interorganizational governance tool used to support improvements in the performance of link workers, service provider organizations and the wider programme, the tension between this and the need to satisfy contract requirements, set payment rates, and enable the production of evidence for in order for SIB payments to be made meant that it became in practice increasingly used as a management control tool (see Jamieson et al. 2020).
Both factors find support in previous SIB literature. Our findings support the positions of Joy and Shields (2013), Maier and Meyer (2017), and Warner (2013) who warned divergent financial incentives can diminish less powerful actors and undermine SIB claims to promoting collaboration. The investor in this case was a large and specialized social investor with a significant presence in the UK SIB market, and while other studies have noted similar managerialist attitudes amongst investors (Williams 2018) it could be that an investor with a different managerial and investment philosophy would have tolerated more uncertainty and resisted NPM logic. In our case, the scale of social investment required for population-level impact and uncertainty of the social intervention model brought significant capital risk and a limited choice of investors with sufficient experience and capital. Precautions could also be taken to address this by other SIB actors to resist investor power-e.g., countering investor demands, or seeking greater transparency by making contracts publicly available-however we consider that access to capital inevitably lends investors an upper hand in setting the terms of engagement and influencing decisions.
The complexity and costs associated with data demands from SIB contracts have also been noted by others (FitzGerald et al. 2019;Lowe et al. 2019;Sinclair, McHugh, and Roy 2021), as has the danger of this to divert the attention of activity into data production (e.g., Jamieson et al. 2020). It could be that a SIB with a more simplified structure (e.g., which handled the administration of the social provision in-house without contracted providers, or directly without an SPV) could ease the administrative burden we observed. However, this may also have resulted in a less diverse, and less innovative, intervention positioned even further from an NPG logic.

Conclusions
Through a longitudinal qualitative study of a major UK-based SIB, we have explored how NPM and NPG governance interact across the SIB lifecourse. While many single case studies of SIBs exist, ours is the first to incorporate a significant temporal dimension, involving granular detail covering development, initial operation, and stable implementation, and encompassing all key SIB partners in its analysis. We offer two key contributions to the governance and public administration literatures.
First, our findings suggest that the proliferation of SIBs is likely to further entrench the value basis and governance dynamics of NPM, rather than facilitate a transition toward NPG. We found that while both NPG and NPM logics were represented within the SIB, internal institutional dynamics operated to reinforce and entrench NPM logic while diminishing NPG characteristics over time.
A single case study carries innate limitations on the empirical generalizability of findings and it could be questioned whether our conclusions reflect a particular-even unrepresentative-set of case conditions and variables rather than stemming from the SIB model. Our case was understood as a model of service commissioning which while typical of a UK SIB, differs to the experience in some other countries like, for example, the USA (Heinrich and Kabourek 2019) and Australia  where SIBs are more closely associated with attempts at social innovation and addressing market failure. We note the need to sensitize our conclusions in different contexts, particularly in different structural configurations (e.g., the presence of an SPV, or between contracted/in-house service provision) and roles played by key actors (e.g., with social investors who actively espouse NPG-related investment and management philosophies). However the resonance of our findings to noted analytical themes of investor power (Joy and Shields 2013;Maier and Meyer 2017;Warner 2013;Williams 2018), and data requirements for complex contracting arrangements (FitzGerald et al. 2019;Jamieson et al. 2020;Sinclair, McHugh, and Roy 2021), suggest our conclusions carry some theoretical generalizability.
Our findings we hope will spur exploration of a second dimension of the 'SIB effect': the impact which SIBs as instruments of service reform impart on the relationships, capacities and capabilities which underpin the health of public service systems. While provisional, our conclusions challenge the extent to which SIBs and outcome-based contracting can be asserted or marketed as enhancing NPG-related concepts like partnership, collaborative innovation and user value-orientation (e.g., Carter et al. 2018). This constitutes a significant problem for the field since NPG is often understood not as an optional governance logic but a necessity to tackle wicked social problems and address broader challenges of public engagement. Without significant adaptations to the classic SIB model, we suggest that SIBs may come to occupy a more limited and marginal role in public service reform than originally hoped for.
Our study does point toward some areas where useful remedial actions may be taken. One approach would be to adopt a relational approach to contracting as Carter and Ball (2021) recommend, with all parties agreeing to commit to shared working principles and actively working to build trust. Power imbalances between investors and other partners, particularly delivery organizations, could be improved by more prudent contract design which limits the power of investors to modify governance arrangements influence the use of committed capital. This however would likely be seen as a significant risk-one perhaps rather avoided by many investors.
On this point our article also carries relevance for other public service contracting arrangements where an external investor party may be absent (e.g., payment-by-results), and in the broader landscape of collaborative governance where the quality of inter-organizational relationships is paramount. Where social outcomes are targeted, we suggest inter-organizational relationships will inherently involve complex contracting arrangements (Brown, Potoski, and Van Slyke 2016). The significance of fostering relational quality through fostering cooperation, trust and mutual commitment in contractual relationships is recommended by authors drawing on relational contracting theory (Bertelli and Smith 2010;Chuang et al. 2020), which may provide a more effective basis for outcome contracting. In turn however, our findings suggest special attention must be placed on acknowledging asymmetries of economic and contract power in contractual relationships, which may prohibit the development of (or eventually overturn) cooperative and mutualistic relationships.
Our findings also carry a second contribution for the broader governance literature on public sector reform and transitions. Osborne (2010, 1-2) suggests "the time of the New Public management (NPM) has in fact been a relatively short-lived and transient one" and observes a natural transition of governance from a traditional bureaucratic model, to a market-based NPM model, to a plural and pluralist NPG model. While studies of other institutional initiatives have found stable hybridization of NPM and NPG in cross-sectoral partnerships (Wiesel and Modell 2014), we found these governance logics were strongly polarizing in the SIB. In contrast to Osborne's claim that NPM be understood as a transitionary governance logic, our findings show how NPM can prove the more dominant and resilient governance logic in hybrid forms like SIBs.
We note several areas for researchers to test, sensitize and extend our findings in future research. First, the UK has a particularly strong affiliation with NPM, and future research could explore whether these findings hold in countries where NPM has been less influential. Comparative analysis, exploring how NPM/NPG logics operate and interact in different sectoral and national contexts, seems a particularly appropriate methodology to test and refine our conclusions. Our findings may have been influenced by an early perception (faulty as it turned out) of underperformance relative to the secondary outcome measure, although the shift in orientation and behavior toward NPM both preceded and succeeded this period. The imposition of NPM during periods of underperformance shows a revealing faith in NPM-rather than NPG-as a driver of efficiency amongst SIB powerholders. Further research could also usefully explore the different management styles which may accompany succeeding and failing SIBs.
Future research may also be enhanced by drawing from alternative theoretical approaches. NPM and NPG represent different 'institutional logics', each representing historical patterns of behavior, values and belief systems. In this domain, research has explored as how organizations may reconcile seemingly incompatible logics (Siwale, Kimmitt, and Amankwah-Amoah 2021;Skelcher and Smith 2015). Researchers might also problematize the NPM/NPG binarism common in governance scholarship, analyzing the role which a financializing 'social investment' logic might play in surfacing different governance modalities in a social investment and outcome-based contracting context.
Finally, we recognize that SIBs and other outcome-based contracting methods such as payment by results have adapted to different contexts and now often diverge significantly from the 'classic' model in practice and vary considerably across countries. Researchers could seek 'outlier' cases which run counter to our findings in key respects of the NPG mechanisms outlined, zooming in on the micro-level management practices of SIB actors which may permit an NPG character to be significantly retained. Indeed, outcomes-based reform can play multiple roles in an administrative context including building trust and relational capacity through co-designing outcomes contracts (Carter and Ball 2021) or using outcomes for collaborative governance rather than performance management and contract enforcement . Given the arguments outlined in this article however, we suspect that such an achievement would involve a radical and definition-stretching departure from the classic SIB model.

Notes on contributors
Max French is a Lecturer in Systems Leadership at Newcastle Business School, Northumbria University. His research seeks to improve the management of complexity in public service systems, the development of academicpractitioner networks and research relationships, and the design and operation of outcome-based approaches to governance.