Business Models for Sustainable Finance: The Case Study of Social Impact Bonds

: Business models for sustainability (BMfS) are relevant topics on research agendas, given their orientation toward sustainability issues. However, traditional versions of these models are often ill-equipped at solving complex social problems. Cross-sector partnerships for sustainability (CSPfS) have been recognized as a new paradigm that mitigates the failure of traditional models. Impact investing, and social impact bonds (SIBs) in particular, represent an interesting ﬁeld of research in innovative business models for sustainable ﬁnance, even though the literature does not consider SIBs within this broader ﬁeld. We propose an exploratory study based on qualitative methods aimed at conceptualizing SIBs within the framework of BMfS and understanding how SIB collaboration varies across social sectors and geographical areas. Our study identiﬁes three different models of SIBs characterized by the different degrees of collaboration between actors: (i) SIB as a fully collaborative partnership; (ii) SIB as a low-collaborative partnership; and (iii) SIB as a partially collaborative partnership. Our ﬁndings are useful to policy makers and practitioners involved in the SIB design, suggesting that a fully collaborative SIB model may stand a better chance of achieving the expected social impacts. of how the SIB across social sectors contexts by identifying three different models of SIBs based on the different degrees of collaboration between SIB actors: (i) SIB as a fully collaborative partnership; (ii) SIB as a low-collaborative partnership; and (iii) SIB as a partially collaborative partnership. The explorative case study suggests that when the SIB is conceived as a low-collaborative partnership it may be exposed to intervention failure, whereas a fully collaborative model may reduce the risk of SIB failure through the design of collaborative processes.


Introduction
Over the last two decades, the business model concept has become increasingly relevant [1] with researchers and practitioners exploring and developing innovative formulas [2], including business models for sustainability (BMfS). Several academic studies have been published on BMfS, and industrial practices have explored the issue to identify sustainable business model archetypes [3]. Despite this growing interest, a very important research gap still exists, specifically with respect to "how sustainable business models function and are applied in the real world and what determines their success (or otherwise) in the market" [4] (p. 4581). The real world, especially in the post-crisis era, is characterized by great changes and even bigger challenges. In economic and financial systems, innovative (and alternative) forms of business and finance-based on collaborative dynamics and shared value [5]-are emerging. For example, innovative BMfS called cross-sector partnerships for sustainability (CSPfS) are promising solutions. The CSPfS are an important new paradigm able to address complex social issues. They are aimed at mitigating the failure of isolated governmental or social sector organizational actions [6][7][8]. Intuitively, multiple actor collaborations are a fundamental element in BMfS [9], though there are currently few studies in the literature.
Turning our focus to the financial industry, financial systems do not currently appear to fully embrace sustainability models [10] and are unable to support the achievement of sustainable development goals. However, interesting case studies of BMFS have been experienced, especially after the global financial crisis. Specifically, the last decade has developed innovative and alternative funding approaches, to leverage social impact [11], based on multiple aspects-going beyond risk and return profiles-such as sustainability, solidarity, collaboration, and social impact [12]. They can be labelled as business models for sustainable finance (BMfSF).
Literature focusing on the financial industry is even more scarce for BMfSF [13]. A very recent work [14] outlines the components of a social innovation business model-that constitutes a significant part of the financial system-where social value is generated through relationships between a number of key partners (e.g., intermediaries, beneficiaries, social enterprises, service providers, and supporters).
Social impact bonds (SIBs) embody many of these aspects, with the presence of cross-sector collaboration in shared value processes appearing as an interesting area to investigate. SIBs are one of the most recent innovative financial instruments to fund social programs (e.g., programs to help homeless people, or rehabilitate prisoners) by creating synergies between public entities, governments, social organizations, and financial institutions. In light of this multi-stakeholder involvement, SIBs can be seen as a "partnership between stakeholders" [15] (p. 731). Specifically, SIBs represent an innovative partnership for sustainability and can be defined as a public-private, collaborative, and outcome-based contract incorporating impact finance logics. These aspects render SIBs particularly interesting to study within the framework of both BMFS and CSPfS, and though they do appear in the literature on impact investing [16], they have yet to be analyzed within BMfS.
Thus, we propose an exploratory study based on qualitative methods aimed at conceptualizing SIBs within the framework of BMfS and understanding how SIB collaboration varies across social sectors and geographical areas. More in detail, we investigate the key collaborative elements of the SIB projects and the ways in which these innovative schemes-based on cross sector collaborations-may contribute to BMfSF, by tackling important challenges [17].
To the best of our knowledge, this is the first research investigating this issue. Therefore, it seems appropriate to use a qualitative research method that is based on multiple case studies to capture a wide range of information and create broad insight, which is useful for explanation building.
Our study contributes to BMfS and to CSPfS literature in several ways. First, we add to the literature on BMfS by analyzing evidence from the financial industry and theorizing the importance of underexplored collaborative dynamics of alternative finance schemes for alternative business models. Second, by combining insights from literature of BMfS and evidence from SIB case studies, we outline the boundaries and characteristics of a SIB partnership. Our hope is that this will encourage better usage of cross-sector collaboration approaches as a way to improve sustainability outcomes and to address pressing sustainability challenges. We also identify three different models of SIBs based on the different degrees of collaboration between the SIB actors. Finally, we provide insights useful to policy makers and practitioners involved in SIB design, with the suggestion that a fully collaborative SIB model may be more likely to achieve expected social impacts.
The remainder of the article is structured as follows. In the next section, we provide a literature overview; then we present our research design and findings; and finally, we conclude by discussing the contributions and implications of our research.

Business Models for Sustainability (BMfS)
The concept of a business model for sustainability (BMfS) is relatively recent and its theorization is at an early stage of development [18][19][20][21]. Abdelkafi and Täuscher [18] indicated the conceptualization A range of CSPfS have emerged to address the evolving nature of social problems [44], the extent [45,46] and complexity of social and environmental issues [46,47], and the failure of actions from isolated governments [6][7][8] and social sector organizations [8]. In fact, CSPfS have been promoted as "social problem-solving mechanism among organizations" [48] (p. 79). Van Tulder et al. [43] hold that a CSPfS is the answer to uncovered economic, social, and environmental needs by providing social goods, such as clean water, health, or education [37,47] through collaboration [49]. A more holistic definition of a CSPfS is provided by Waddock [36] (pp. 481-482) who defined a CSPfS as "the voluntary collaborative efforts of actors from organizations in two or more economic sectors in a forum in which they cooperatively attempt to solve a problem or issue of mutual concern that is in some way identified with a public policy agenda item". These definitions are based on a common concept of collaboration that implies sharing the responsibility [50], and thus, a non-hierarchical position the actors [51]. By contrast, the definition of Waddock [36] specifically emphasizes CSPfS issues in the public agenda, while others more generically refer to the need of meeting public issues. The United Nations [52] Global Sustainable Development Goals include the CSPfS (Agenda 21 in 1992) as a new tool for fostering economic, social, and environmental development [53].
CSPfS involve a wide range of actors from different sectors [54][55][56] with different purposes that can bring different resources to the partnership. On one side, governments find an alternative way to produce public goods-a typically governmental prerogative-by involving firms [57] and social sector organizations [58]. On other side, firms-especially large corporations-find a way to generate a "long-term competitive advantage" [43] (p.2), while social sector organizations gain ways to finance social issues [46]. In fact, Kolk et al. [59] found that CSPfS produced effects at three different levels: macro level, meso level, and micro level. CSPfS have positive effects on sustainability issues, for instance, reducing poverty or improving environmental protection (macro level effects). At the same time, CSPfS produce reasonable effects on the actors involved in the CSPfS, for instance, increasing firm reputation and providing funding inflow for social sector organizations (meso level effects). Finally, CSPfS affect customers (or beneficiaries) and employees through the so-called micro level effects.
Austin and Seitanidi described contributions that actors provide the CSPfS [16] (p. 933), distinguishing between tangible and intangible resources, and sustained that "the more the partners are willing to deploy their distinctive organization-specific resources, the greater will be the potential for value creation". Specifically, social sector organizations, because of the wide knowledge of social problems, may play a relevant role in market and beneficiary identification or may support the legitimacy in a civil society [60]. More in depth, Yan et al. [46] reviewed the role of NGOs in CSPfS and grouped them into three main classes: enabling roles (service provider, capacity builder, and consultant), coordinating roles (bridge and mediator), and change-facilitating roles (initiator, convener, advocate, leader, and innovator).
Seitanidi and Crane [61] identified the following types of CSPfS: public-private partnerships, public non-profit organization partnerships, private non-profit organization partnerships, and finally, tripartite partnerships. The strategic collaboration between firms and non-profit organizations for corporate social responsibility purposes is also called a social alliance or business non-profit partnership [62].
CSPfS may also differ in terms of size, geographical extension (local, regional, global), timing (short time-frame or long-time frame) and by arrangement (voluntary or mandated) [47].

BMfS and CSPfS in the Financial Industry
Business models created to address sustainability issues are gaining increasing interest in the finance industry, thus we can identify specific models. On one hand, the financial industry plays a pivotal role in sustainable development and in creating shared value for businesses and society. On the other hand, actors involved in the financial industry are looking for new pathways for sustainable value, and this renders BMfS extremely relevant. A recent study by Yip and Bocken [13] focused on BMfS archetypes in the banking sector, including innovative sustainable financial products, such as green bonds, crowdfunding, socially responsible funds, and impact investing [13].
Impact investing is a relevant issue to analyze [16] aimed at achieving measurable social and environmental impacts alongside financial returns [63][64][65][66][67]. Within impact investing, collaborative models emerge as ways to meet relevant social issues that would otherwise be left unaddressed. Austin and Seitanidi [16], on the other hand, stated that it is the partnerships or collective models in general that have taken on a new light thanks to impact investing. Within this framework, SIBs take center stage as the most collaborative impact model.

Social Impact Bonds (SIBs): Overview
SIBs are a relatively new type of payment by results contract; they are focused on outcomes where public entities partner with private investors to finance actions aimed at tackling social problems. These services are often delivered by social service providers, generally social enterprise organizations, typically over a period between three to five years, but sometimes longer [68,69]. A specialist SIB intermediary is also often involved and plays an important coordinating role. More specifically, investors provide the up-front finances and are repaid, including a pre-defined return on investment, only if expected social outcomes are achieved [70]. SIBs may be categorized as hybrid instruments incorporating classic bonds as well as equity characteristics [71]. In other words, in a typical SIB contract, financial returns may see an upper limit as in traditional bonds, but they are also related to performance as in the case of equity instruments [71].Given the emphasis on the social performance derived from such outcome-based contracts, typical SIB models include an independent evaluator responsible for measuring the impact of the intervention on the target population by adopting pre-defined impact evaluation methods [72]. SIBs are frequently adopted to focus on the prevention or reduction of challenging social problems, which generate actual or future savings in public service budgets. A large part of the outcome payment returns to the initial investors if the SIB is successful, and they are funded from the savings generated by improving social outcomes [73]. The world's first SIB was launched in England at the Peterborough prison in 2010 [74]. According to the Social Finance's online database of social impact bonds [75], another 120 SIBs have been promoted by governments worldwide since then, mostly in the United Kingdom, the United States of America, and Australia. Table 1 represents the SIBs launched since 2010 for each category of social issue funded. Academic contributions concerning SIBs derive from multi-disciplinary perspectives based on financial, social entrepreneurship, and public policy theory [76]. Fraser et al. [77] provided an extensive review of SIB literature by identifying three main narratives: "a public sector reform narrative located within broader theories of New Public Management (NPM); a private financial sector reform narrative located within broader theories of social entrepreneurship; and a cautionary narrative skeptical of public and financial sector developments such as NPM and social entrepreneurship, and thus of SIBs" (p. 5). The first narrative analyzes SIBs as an extension of outcomes-based contracting in public services [78]. The second sees SIBs as a new opportunity for the private financial sector to gain increased access to public funds [79] and, at the same time, as a mitigating factor against the anti-social, dangerous aspects of financial capitalism in [80]. The third narrative poses questions about the financialization of public service, deriving from the introduction of logics and normative assumptions of the private financial services sector in the delivery of social services [81].
Distinguishing features of SIB emerging in the literature may be summarized into the following main streams: (i) SIBs represent an innovative partnership between socially oriented investors, public commissioners, and non-profit service providers, often coordinated through a specialist intermediary to tackle deeply ingrained social problems [82]; (ii) SIBs are directed to produce improved social outcomes for society, generating public savings [83]; (iii) SIBs models produce financial risk transfer from the public sector to investors [84]; and (iv) SIBs financial returns depend on achievement of outcomes after a rigorous evaluation and measurement of it [84]. The interplay of the above described mechanisms is represented in Figure 1. commissioners, and non-profit service providers, often coordinated through a specialist intermediary to tackle deeply ingrained social problems [82]; (ii) SIBs are directed to produce improved social outcomes for society, generating public savings [83]; (iii) SIBs models produce financial risk transfer from the public sector to investors [84]; and (iv) SIBs financial returns depend on achievement of outcomes after a rigorous evaluation and measurement of it [84]. The interplay of the above described mechanisms is represented in Figure 1. As a partnership, the different actors in the system are crucial to the success of SIBs. Therefore, it is important to understand the different stakeholders' roles and motivations driving their involvement in SIB implementation. After defining SIB design, the typical SIB model works as follows. (i) A specialized intermediary raises funding among impact investors. The funding(ii) provides the working capital to one or more social service providers to scale evidence-based programs.(iii) During the implementation phase the intermediary coordinates all SIB parties, and social outcomes are delivered over a target population.(iv)Pre-defined outcome thresholds are measured by an independent evaluator and, in cases of success, the commissioner returns outcome payments to the intermediary, which repays the initial investors. As can be clearly observed, the financial and contractual mechanisms reveal some key characteristics typical of SIBs, different from other forms of outcome-based contracts. First, SIB outcome payments are based on the logic of "cashable savings" that are achieved through preventive social programs [86]. Additionally, the focus on pre-defined outcomes transfers the risk of project failure to the private sector [87]. Finally, SIBs encourage innovation [88] because, beyond the introduction of market forces and investor inspection into social policy, they are receptive to collaboration directed at the co-creation of collective social impact [89].
The lifecycle of SIB encompasses different stages that are mainly grouped under three areas: design, implementation, and evaluation. The design stage includes the following sub-phases: the definition of the social issue, the development of the social intervention strategy, the development of outcome metrics and control, the estimate of budget for service and the valuing of outcomes, the develop of financial model, and the selection of SIB partners. The implementation stage is the central phase of the SIB lifecycle and includes the delivery of the social intervention, the gathering of data on the delivered impact, and the sharing of performance information among partners. Finally, in the evaluation phase data gathering is completed, the impact evaluation is conducted according the predefined methodology, the results are reported to stakeholders, and, in cases of success, investors receive their investments plus a return. As a partnership, the different actors in the system are crucial to the success of SIBs. Therefore, it is important to understand the different stakeholders' roles and motivations driving their involvement in SIB implementation. After defining SIB design, the typical SIB model works as follows. (i) A specialized intermediary raises funding among impact investors. The funding(ii) provides the working capital to one or more social service providers to scale evidence-based programs.(iii) During the implementation phase the intermediary coordinates all SIB parties, and social outcomes are delivered over a target population.(iv)Pre-defined outcome thresholds are measured by an independent evaluator and, in cases of success, the commissioner returns outcome payments to the intermediary, which repays the initial investors. As can be clearly observed, the financial and contractual mechanisms reveal some key characteristics typical of SIBs, different from other forms of outcome-based contracts. First, SIB outcome payments are based on the logic of "cashable savings" that are achieved through preventive social programs [86]. Additionally, the focus on pre-defined outcomes transfers the risk of project failure to the private sector [87]. Finally, SIBs encourage innovation [88] because, beyond the introduction of market forces and investor inspection into social policy, they are receptive to collaboration directed at the co-creation of collective social impact [89].

Method
The lifecycle of SIB encompasses different stages that are mainly grouped under three areas: design, implementation, and evaluation. The design stage includes the following sub-phases: the definition of the social issue, the development of the social intervention strategy, the development of outcome metrics and control, the estimate of budget for service and the valuing of outcomes, the develop of financial model, and the selection of SIB partners. The implementation stage is the central phase of the SIB lifecycle and includes the delivery of the social intervention, the gathering of data on the delivered impact, and the sharing of performance information among partners. Finally, in the evaluation phase data gathering is completed, the impact evaluation is conducted according the pre-defined methodology, the results are reported to stakeholders, and, in cases of success, investors receive their investments plus a return.

Method
Given the relevance of the topic and the exploratory nature of the work, using a qualitative research method [90]-based on multiple case studies (within-and cross-case analysis)-can appropriately capture a wide range of information and create in-depth insights into SIBs. A multi-case study analysis allows us to be rigorous and apply comparative logic [91][92][93] useful for explanation building [94,95].
Qualitative methods are still not frequently utilized in financial research, even if the post-crisis period has recently seen a birth or "renaissance". This is supported by a part of the academy that is questioning the dominant paradigm, by highlighting the need to adopt typical methods from the social sciences, to gain a better understanding of the real world [96]. In addition, several scholars agree on the usefulness of empirical research studies (and, particularly case studies) in the context of large challenges such as sustainability-related problems [20,97].
According to the best methodological criteria defined in the literature, a case study analysis is particularly useful for understanding unknown (or partially known), complex, and messy phenomena [98,99]. Gummesson [100] specifically noted that case studies offer the advantage of examining several key elements and the linkages among them, through multiple perspectives. Viewed in this light, the exploratory and inductive approach of our case study research appears to be important: empirical observations through inductive elements may contribute to describing and conceptualizing the phenomena [101]. In addition, case study analysis is particularly relevant to the theory-building process [91,[102][103][104].

Case Selection
For the purpose of this research, we adopted an appropriate sampling strategy, aimed at selecting a variety of heterogeneous cases [105], to ensure an adequate overview of the investigated issues [106] and to contribute to the theorization process [91].In the words of Eisenhardt [91](p. 537):"As Pettigrew [107] noted, given the limited number of cases which can usually be studied, it makes sense to choose cases such as extreme situations and polar types in which the process of interest is "transparently observable". Thus, the goal of theoretical sampling is to choose cases which are likely to replicate or extend the emergent theory".
Our sample consisted of four SIBs located in several geographical areas, tracking different social issues and characterized by the different stages of implementation. The distinctive and innovative features of the sample-to justify their choice-are presented in Table 2. Thus, the number of cases is appropriate for the exploratory stage [92]. Moreover, only cases able to provide rich information about the phenomena under study were included [91,108].

Data Sources
The analysis was based on various sources of information [95] that met the criteria of transparency and trustworthiness [109]. Specifically, we utilized the following multiple data sources: (i) public data provided by subjects not involved in SIB projects (e.g., news and articles published by magazines, academic works, and official reports published by governments, institutions, organizations, etc. . . . ); (ii) archival records made available by subjects involved in SIB projects in several ways (e.g., information available on websites of main partners, news, press releases, data, and reports disclosed by participants in SIB projects; legal, policy, technical, organizational, and internal documents). In addition, questionnaires were specifically designed (starting from the main results of the literature review), tested, and deployed (sent via e-mail in November 2018). Sometimes, several in-depth phone interviews were conducted with multiple key informants who were expected to be knowledgeable about the SIB projects (November 2018). Triangulation was realized using different data collection methods such as document analysis, interviews, and questionnaires [99,109].

years) Implementation
The project supports the integration of between 2500 and 3700 migrants and refugees into the Finnish labor market through the provisions of training and job-matching assistance.
The project target of refugee integration is the first of its kind in Europe and in the world; the SIB was funded not only by private investors but also from The Investment Plan for Europe and the European Fund for Strategic Investments (EFSI) and therefore represents the first experience in Europe of UE institutional investor engagement.
The SIB manager was chosen with a public tendering and, differently from traditional SIB scheme, it is not a specialized SIB intermediary but an impact investor; also, a Finnish institutional impact investor, SITRA, played central role in design phase of the SIB.

Concluded
First launched in Upper Austria, the issue addressed poverty and marginalization among women affected by violence by helping them to them to find a long-term job, which fulfils certain criteria.
The SIB represents the pilot in Austria; the social intervention funded is unique worldwide.
The SIB impact investor played the role of intermediary within the SIB and does so free of charge; the areas of activity and the financial framework were set by the public commissioner.

Data Analysis
A research protocol-including the data sources, research instruments and procedures, and researchers' role and responsibilities-was developed in advance [94]. The research protocol ensured transparency and trustworthiness in our work. In addition, it is worth noting that all data were evaluated and checked in each research phase by at least two researchers.
Given the fact that it was an exploratory study focused on a limited number of cases (based on a small data set), our research employed manual techniques for processing data. In this way, researchers were the principal actors in sorting, interpreting, and coding data; and the analysis took advantage of the creative process that characterized scholars' activities [110] regarding reflection on research questions, observation material, and conceptual frame [111] (p. 638). It is important to note that Strauss and Corbin [112] stressed the relevance of the creative aspect of grounded theory research, in which the theoretical sensitivity and analytic ability of researchers are central.
With regard to the data analysis process, we proceeded as follows. In the first step, two researchers (senior and junior) independently provided a detailed description of each case and the themes within them, drawing up key-concepts and their relationships. The questionnaires and interviews were developed and tested by two researchers (September and October 2018). The interview protocol included a questionnaire (composed of four sections and several open answers) and a semi-structured phone (or, alternatively, in-person) interview. The answers given in the questionnaires (and to phone interviews) were transcribed and analyzed independently by two researchers. Then, we carried out a cross-case analysis, to search for emerging patterns [91], by comparing key-elements, concepts, themes, and relationships [113]. The emerging findings were discussed in light of the existing theory on the BMfS, according to suggestions by Eisenhardt [91].
To ensure the process of data analysis was vigorous, an iterative coding procedure [112,114]based on techniques and tactics typical of analytic induction commonly used in qualitative research-was adopted. This procedure gradually revealed a final coding scheme (repeatedly subjected to adjustments through multiple rounds of coding and reflection) useful to structure our data and to deeply understand the examined phenomena.
The data analysis started with the observation, interpretation, and selection of materials. We produced a description of each case, which we continuously re-checked during their further interpretations. We carried out an in-depth analysis of our case studies by developing a system of categories for the case study comparisons. The observations were recorded through several kinds of transcription. The reading phase of the transcripts was characterized by creative moments of observation, interpretation, and selection of data in which researchers made connections between materials and provided coding. In this phase, great importance was attributed to "memos", which "complement and explain the codes that were found" [115] (p. 307). As stated by Flick [115] (p. 307), "coding includes the constant comparison of phenomena, cases, concepts, and so on, and the formulation of questions that are addressed to the text".
We firstly applied open coding and then selective coding [115]. Therefore, we grouped codes into categories to increase the comparability and understanding of materials and cases. To do this according to Miles and Huberman's [108] recommendations, two researchers (seniors) independently coded the data-identifying a range of sub-themes and major themes-and drew a data structure. Inconsistencies were discussed and solved by all the researchers, also in relation to the literature. An overview of our inductive coding structure is shown in Table 3.

Empirical Findings
In this section we contextualized SIBs under the BMfS framework, and we provided an understanding of how SIB collaborative aspects vary across social sectors and geographical areas.

The New Narrative of SIBs
Following the approach of Schaltegger et al. [21] and Yip and Bocken [13], we defined the value proposition, value creation and delivery, and the value capture of the SIB model ( Figure 2). This allows us to reinterpret the SIB model under the lens of BMfS. SIBs allow for the financing and delivery of social and environmental services (e.g., support the integration of prisoners, deliver education) to a target disadvantaged population (value proposition). Social and environmental services fill social and environmental needs of a target population, improving their living conditions (value creation and delivery). Finally, SIBs allow public bodies to obtain measurable social values and public savings, and they allow investors to obtain a measurable social value alongside a financial return (value capture).

Empirical Findings
In this section we contextualized SIBs under the BMfS framework, and we provided an understanding of how SIB collaborative aspects vary across social sectors and geographical areas.

The New Narrative of SIBs
Following the approach of Schaltegger et al. [21] and Yip and Bocken [13], we defined the value proposition, value creation and delivery, and the value capture of the SIB model ( Figure 2). This allows us to reinterpret the SIB model under the lens of BMfS. SIBs allow for the financing and delivery of social and environmental services (e.g., support the integration of prisoners, deliver education) to a target disadvantaged population (value proposition). Social and environmental services fill social and environmental needs of a target population, improving their living conditions (value creation and delivery). Finally, SIBs allow public bodies to obtain measurable social values and public savings, and they allow investors to obtain a measurable social value alongside a financial return (value capture).    Table 4 synthetizes value proposition, value creation and delivery, and value capture of the four SIBs under review. The SIB cases empirically demonstrated why and how they represented a BMfS. For instance, the New York City Adolescent Behavioral Learning Experience (NYC ABLE) case value proposition concerned the reduction of re-incarceration rates of adolescents. The SIB's value creation and delivery happened through moral reconation therapy, which was designed to improve social skills, problem solving, and impulse management. The value capture consisted of the number of recidivism bed-days avoided, which generated a maximum payment of USD 11.7 million and a public savings from USD 1 million to USD 20.5 million. The collaborative character of SIBs allows them to be included in a specific set of BMfS: the CSPfS. Specifically, under the CSPfS definition of Waddock [36], a SIB may be identified as the result of a collaborative effort from a range of public and private actors who act together to meet a specific social need. Figure 3 summarizes such a theoretical framework. exit the cycle of violence for good.
independence, allowing them to escape structures of violence for good.
1% per annum interest on the loans) Public savings: €1.8 million derived from multiple sector savings: healthcare, welfare benefits, and unemployment costs.
The collaborative character of SIBs allows them to be included in a specific set of BMfS: the CSPfS. Specifically, under the CSPfS definition of Waddock [36], a SIB may be identified as the result of a collaborative effort from a range of public and private actors who act together to meet a specific social need. Figure 3 summarizes such a theoretical framework.

SIBs and Cross-Sector Partnerships: How They Vary Across Sectors and Geographies
This section illustrates the SIB collaborative process and dynamics with the aim of identifying how collaborative dynamics vary across sectors and geographies. Results are presented along the following dimensions of analysis: (i) the processes and dynamics characterizing the design and structure of the SIB, the(ii) social dimension, and (iii) the financial dimension.

Processes and Dynamics Characterizing SIB Design and Structure
Identifying the social intervention represents the key element of a SIB. The main features of the SIB were built around this intervention. Identifying a service provider, the design of outcome metrics and thresholds, the choice of outcome evaluation, and identifying the target population were other key SIB features.

SIBs and Cross-Sector Partnerships: How They Vary Across Sectors and Geographies
This section illustrates the SIB collaborative process and dynamics with the aim of identifying how collaborative dynamics vary across sectors and geographies. Results are presented along the following dimensions of analysis: (i) the processes and dynamics characterizing the design and structure of the SIB, the(ii) social dimension, and (iii) the financial dimension.

Processes and Dynamics Characterizing SIB Design and Structure
Identifying the social intervention represents the key element of a SIB. The main features of the SIB were built around this intervention. Identifying a service provider, the design of outcome metrics and thresholds, the choice of outcome evaluation, and identifying the target population were other key SIB features.
In two of the four cases under review, the SIB social interventions were identified through a top-down decision flow. Only in the Australian SIB case was the social intervention identified through a call for social impact project to be implemented by a social benefit bond. In the case of the NYC ABLE project, the social intervention identification was built through a consultative process between the commissioner and the specialized SIB intermediary entity, MDRC.
In general, the presence of a collaborative process in the SIB design phase was much higher in the Australian SIB. In the NEWPIN SIB, the commissioning approach to SIBs appeared more open to partnership proposals than in the other SIBs under review. The main element explaining this prevalence of non-collaborative processes or not fully collaborative process-labelled as a form of rigidity in the commissioning approach-derived from national regulation on public procurement that imposed precise procedures on a public commissioner who was commissioning services in the public sector.
Finally, the governance structure of the SIB represents the contractual scheme around which the SIB regulated financial and outcome flows between different actors. In the cases we analyzed, the special purpose vehicle (SPV) played the role of SIB manager. In two cases, NYC ABLE and the Austrian SIB, the SPV was central in the network of the SIB. However, in the case of NYC ABLE, the SPV was owned by the specialized SIB intermediary, while in the Austrian SIB, the SPV was in an emanation of the SIB manager entity owned by the main SIB investor, Juva at Foundation. Also, in the KOTO case SIB, the role of the SIB manager was focused on the SIB investor (Epiqus), but without the presence of an SPV. In the Australian SIB case, we found two lines of contracts between the commissioner and, respectively, the service provider and the investors. In the NEWPIN SIB case, therefore, we saw the absence of a SIB manager entity and the SPV. Table 5 summarizes the above-described processes and dynamics.

Process and Dynamics in the Social Dimension
In a standard SIB model, the delivery of value is measured with one or multiple outcome metrics. The success of the SIB is measured against pre-defined thresholds along this metric. The evaluation of the level of outcome achieved through the implementation of the social intervention is captured by an independent evaluator.
Identifying the outcome metrics was performed collaboratively in the KOTO SIB and NEWPIN cases. In particular, in the NEWPIN case, one year after implementation, one metric of impact measurement was modified based on a dialogue among the partners because its original design created a distortion in the measurement of impact delivered. In the other two cases, the identification process was commissioner-led on the base of local/national accountability needs. Identifying the outcome thresholds reflected the same dynamics, except for the NYC ABLESIB case where some forms of collaborative decision-making flows between commissioner, service provider, and investors were present. It is interesting to note how collaboratively creating the outcome metrics contributed to a SIB design that more clearly suited the needs of the commissioner, service providers, and investors. Furthermore, outcome metrics identified in a collaborative manner represented a mitigation factor of potential principal-agent conflicts in the SIB design.
Finally, the presence of performance management systems in the SIBs was observed in the Australian and KOTO SIBs. It is interesting to note that the adoption of a performance management system produced a collaborative process by aligning all SIB actors toward the delivery and achievement of the expected SIB social value. The illustration of the findings on these dynamics are summarized in Table 6.

Process and Dynamics in the Financial Dimension
SIB financial flows can be divided into two main phases: the funding-where capital is raised through impact investors-and the outcome payments-where the principal is returned to investors plus a variable premium, depending on the level of outcome achieved.
The SIB funding scheme is the most differentiated element in this dimension. This variety may be due to national factors, to the typology of investors engaged in the SIB, as well as the technical instruments available in the domestic financial market. Specifically, in the cases of NYC ABLE and the Austrian SIBs, the SIB funding was designed around a prominent role of the public commissioner in terms of identification of investors most suitable for the first national SIB experience. The case of the KOTO SIB represented an innovative funding scheme, where, for the first time, the subscription of the SIB investment was also proposed to EU institutional investors (European investment fund). In the case of NEWPIN, the funding was carried by issuing a financial product similar to traditional bonds.
In the observed SIBs, outcome payments were influenced by many variables. For example, in the NYC ABLE case, the risk of SIB failure was the central element influencing the definition of the amount of financial returns. In the cases of KOTO and NEWPIN, the relationship between financial returns and outcome thresholds remained central to dimensioning the financial returns. In the case of the Austrian SIB, the public savings produced from the SIB implementation conditioned the definition of the financial returns.
Finally, another collaborative element was found in the presence of guarantee of capital. A guarantee represents a collaborative element since the SIB promoter participates in the possible losses, bearing the financial risk of not achieving the expected outcomes. In the cases analyzed, a guarantee was set up in two out of the four SIBs. This result was evidence of a close relationship between the (expected) public savings curve and the maximum outcome payments repayable to investors, in case of success. Moreover, the presence of capital protection was an incentive to successfully complete the SIB funding. Table 7 highlights results on the analysis of the collaborative processes for financial return generation.

A Theorization of How SIBs May Vary across Sectors and Geographies
Our empirical findings highlighted that collaborations between SIB actors varied across social sectors and geographical areas. Starting from this perspective, we identified three different models of SIBs characterized by different levels of collaboration.

SIBs as a fully collaborative cross-sector partnership
Our empirical findings demonstrated that SIBs may be characterized by a high level of collaboration between different actors. This was the case of the NEWPIN SIB, where collaboration between stakeholders emerged in the SIB design and structuring, social dimension, and financial dimension. Specifically, identifying the social intervention was realized through a collaborative process between the different actors, and the SIB design was marked by high degree of collaboration. The identification of outcome metrics and of outcome thresholds was collaboratively designed as well as the identification of the SIB funding scheme. The presence of a guarantee in the NEWPIN SIB was also encouraged by the commissioners, who were highly engaged to favor the funding of the first SIB launched in the state.
When we are in the presence of a such collaborative scheme, we can refer to the SIB as a fully collaborative cross-sector partnership.

SIBs as a low-collaborative partnership
When the collaboration between SIB actors isabsent in the design and structuring of the SIB and in the setting of social and financial dimension, the SIB takes the form of a non-collaborative partnership. This was the case with the PERSPEKTIVE ARBEIT SIB. In fact, the social intervention was identified through a top-down approach and the SIB design was characterized by a low degree of collaboration between the different actors. Furthermore, the identification of social and financial dimensions (outcome metrics, outcome threshold, and funding scheme) did not see a collaboration between SIB stakeholders. Finally, a guarantee scheme was not structured.

SIBs as partially collaborative cross-sector partnership
When the SIB is not following a fully collaborative model or is a low-collaborative model, the SIB occupied a "middle area" including a wide range of partnerships that presented a partially collaborative cross-sector partnership. The NYC ABLE and KOTO SIB cases took place in this "middle area". The NYC ABLE case showed a partially collaborative scheme of social intervention identification and a medium collaborative process for the SIB design. Meanwhile, the identification of outcome metrics happened through a top-down approach, as did identification of the SIB funding scheme. Thus, the only two elements showing collaboration between the actors in the NYC ABLE case were the identification of the outcome threshold and the presence of a guarantee scheme. By contrast, the KOTO SIB showed collaboration both in the identification of outcome metrics and in defining the outcome thresholds, with no collaboration in SIB design and structuring of the funding scheme. A guarantee scheme was also absent.

Discussion
The empirical study brings new insight into the BMfS literature by introducing SIBs to this conceptual framework. Specifically, SIBs may be considered under the lens of a specific BMfS: the CSPfS. Indeed, SIBs connect multiple stakeholders under at least one contract. They require an upfront transfer of funds to the service provider, and they return outcome payments to investors upon progress toward improved outcomes. How the collaboration may vary across social sectors and geographical areas is discussed below, looking specifically at the results and issues related to collaborative processes and dynamics in SIB design and structuring as well as the social and financial dimensions.
Collaborative and dynamic processes are key during the partnership building, design, and implementation phases of the project. Collaborative dynamics remain unpacked until all partners are identified. However, identifying the social issue of intervention, along with the measures and impact evaluation framework adopted, are also key elements.
In the cases investigated in this study, dynamics emerged that completely centered on the public commissioner, who establishes the social area of intervention and relative outcome metrics. In this way, the SIB partners receive the engagement in the SIB partnership with precise value focus boundaries. Such bottom-down dynamics may produce some form of rigidity in the nascent SIB. Specifically, it is interesting to note how such an approach set the groundwork for the failure of the NYC ABLE SIB. In other words, collaborative engagement in the identification of social issue intervention may improve the appropriateness of the intervention and create a better match between the target population and the intervention funded with the SIB. Strong commissioner motivation to engage in the SIB may be insufficient to ensuring project success if public actors do not have thorough knowledge of the most appropriate solution for the identified social issue. In this sense, in the Australian experience, the collaborative engagement of the social service provider produced a collaborative focus on the most appropriate intervention and selection of the relative outcome metrics. In this specific case, an evidence-based program appears as the best solution to be funded with a pilot SIB because partners may benefit from a historical baseline of solutions and impact that may mitigate asymmetric information distortions. The second reason explaining rigidity in the collaboration processes during partnership building arose from national legal context. National legal frameworks on public procurement can jeopardize a commissioner approach to the SIB. Therefore, partnerships for innovation are often subject to public tendering. However, what emerged was that in different contexts, public tenders may have focused over a specialized SIB intermediary, which then engaged the other SIB partners or, separately, over the single partners. Finally, the external context, especially taxes or other forms of financial incentives, remains equally important in different contexts and sectors.
With regard the collaborative achievement of the expected outcomes, we observed some differences between the SIB cases. In the NEWPIN SIB, one year after implementation, the collaborative focus between service providers and the commissioner produced a changed outcome metric with the aim to efficiently overlap the outcome evaluation with the social intervention. Such elements reveal the importance of collaborative dynamics, not only during the design phase, but also in the implementation stage. In this sense, sharing reliable and updated data between SIB partners may improve the effectiveness of intervention. Finally, the SIBs reviewed demonstrated greater focus on outcome measurement by adopting rigorous data collection and performance management approaches. However, it is important to distinguish such activities such as collection of management information from information collected to report robust evidence of effectiveness. In the analyzed cases, only the NYC ABLE project information was collected both for management information needs and to report how outcomes achieved were attributable to the SIB intervention by using a quasi-experimental outcome measurement method. There are various reasons for this, including differences in the social sector interventions that may influence collecting outcome data at the individual client level, over time.
Regarding the collaborative achievement of the expected financial results, SIB produces new rules to govern the delivery of social impact. First, these contracts distribute risk amongst partners in new ways. Through analyses of the SIB contracts in the four cases reviewed, we observed different SIB structures. In the NEWPINSIB, two different contracts regulated the relationships between commissioner and service providers and between commissioner and investors. In the NYC ABLESIB, the presence of a specialized SIB intermediary was directly related to the presence of a SPV, which was useful to concentrate financial transactions between intermediary and other partners. In the KOTO SIB we saw a model where the intermediary retained the central role. However, in this case the fund manager was an impact-investing company and, therefore, the financial flows in this SIB appeared closer to a social investment partnership. In the first model (NEWPIN), payments from the commissioner fed into the provider organization, and the commissioner made outcome payments to the investor. In the second one (NYC ABLE), instead of inputs flowing from investors, commissioners, and SIB specialist organizations directly to the service provider, they took place through the SPV. Finally, the figure of impact investors and SIB manager may overlap, as seen in the KOTO SIB case. This characteristic, though not a novelty in the SIB market, opens new considerations about the role of the SIB investors, and it and poses questions about whether such actors, as SIB managers themselves, enhance effectiveness compared to a specialized SIB intermediary.

Conclusions
In recent years, traditional business models for sustainability (BMfS)have shown to be weak at solving complex social problems. Cross-sector partnerships for sustainability (CSPfS) are seen as a new paradigm that mitigates the failure of traditional models, which are typically implemented by governmental or social sector organizations [6][7][8].
New and innovative models have also emerged in the finance field to answer the growing social and environmental challenges. Impact investing represents one of the most interesting BMfS [16] because of its value proposition of achieving financial return as well as a measured (or measurable) social and environmental impact. However, if the impact-investing business model is structured in a traditional manner, it may face similar limitations to the other traditional BMfS. These limitations may be overcome through CSPfS and, specifically, through social impact bonds (SIBs).
Using an explorative case study analysis of four SIBs, this research contributes to the existing literature of business models and social impact investments in several different ways. Foremost, to the best of our knowledge, this is the first study to position the SIB model within the conceptual framework of BMfS and, specifically, within the business models for sustainable finance (BMfSF).
Our approach derives from Schaltegger et al. [21] and Yip and Bocken [13] to identify the value proposition, value creation and delivery, and the value capture of SIBs, positioning the SIB model under the lens of BMfS. SIBs allow for the financing and delivery of social and environmental services (e.g., support the integration of prisoners, or deliver education) to a disadvantaged target population (value proposition). Social and environmental services fill social and environmental needs of a target population, improving their living conditions (value creation and delivery). Finally, SIBs allow public organizations to obtain measurable social value and public savings, while providing investors with a measurable social value alongside a financial return (value capture).
The collaborative characteristic of SIBs places them within a specific set of BMfS: the CSPfS. In fact, SIBs are characterized by collaborative efforts of a range of public and private actors who act together to meet a specific social need.
Our study provides a theorization of how the SIB CSPfS varies across social sectors and geographical contexts by identifying three different models of SIBs based on the different degrees of collaboration between SIB actors: (i) SIB as a fully collaborative partnership; (ii) SIB as a low-collaborative partnership; and (iii) SIB as a partially collaborative partnership. The explorative case study suggests that when the SIB is conceived as a low-collaborative partnership it may be exposed to intervention failure, whereas a fully collaborative model may reduce the risk of SIB failure through the design of collaborative processes.
Our early findings are useful to policy makers and practitioners involved in the SIB design and implementation in suggesting that SIB should be designed and implemented as a fully collaborative partnership, because the collaboration between all the interested parties may help create the desired social impact. Specifically, collaborative engagement, when identifying the social issue intervention as well as the other phases of structuring a SIB, may render the social intervention design and governance more appropriate. Similarly, collaboratively identifying outcome metrics and outcome threshold should improve overall effectiveness by mitigating the principal-agent conflicts that could arise during the SIB design stage. For the same reasons, the presence of a capital protection produces a type of collaborative participation that mitigates the risk of investors losing capital, in case the SIB fails.
Using a theoretical framework outlined in this paper, future research could expand understanding on the relationship between fully, low, and partially collaborative partnerships and SIBs to positively create social impact. Further evidence could be found by analyzing other SIBs, even in sectors that currently report low use of SIBs such as the environmental sector. Moreover, as the SIB market grows, it will likely drive the application of qualitative and quantitative analyses as well as the implementation of software-based qualitative analyses, given the large volume of data.
Scholarship would also benefit from more in-depth investigations into the design and implementation of SIBs and how they are influenced by factors such as national tendering regulation, the presence of a national structured social investment market, the nature of social intervention, and specific metrics and outcome thresholds for intervention funded with a SIB. For these reasons, future studies should approach SIBs by adopting multidisciplinary lenses and mixed methods useful for more in-depth explorations of SIB evidence within BMfS research.
Author Contributions: The Authors contributed equally to this research work.
Funding: This research was funded by Italian Ministry of Education, Universities and Research (MIUR), Project "An Italian platform for impact finance: financial models for social inclusion and sustainable welfare", Project Code: SIF16_00055.