Collaborative Learning Processes in Social Impact Bonds: A Case Study from the Netherlands

ABSTRACT

Social Impact Bonds (SIBs) bring together organizations from various sectors, with different organizational logics. Successful cross-sector collaboration, therefore, is challenging. In this paper, the collaborative learning process between partners of the SIB The Colour Kitchen (The Netherlands) is analysed. By analysing a good practice case, this paper shows (1) how collaborative governance theory helps to understand the dynamics of an SIB partnership, (2) provides insight into theoretical and empirical favourable conditions and actions for collaboration in an SIB, and (3) illustrates the complexity of SIBs and shows why such a normative proposition is still not widely practised.

KEYWORDS:

• Social Impact Bond
• collaborative governance
• collaborative learning
• SE partnership
• social enterprise

Introduction

By seeking partnerships, social enterprises are able to fulfil their social mission – and address societal problems (e.g. unemployment, environmental pollution) – more effectively (Austin 2000; Henry 2015; Choi 2015; Meyskens, Carsrud, and Cardozo 2010; Austin, Stevenson, and Wei-Skillern 2006). In the network society (Castells 2000), tackling societal problems requires collaborative efforts between private, public and third sectors (Koppenjan and Klijn 2004). Pressure on public funding and a trend of neo-liberalism has led to a retreatment of the public sector in social policy (Pollit and Bouckaert 2011). Given both developments, new modes of governance arise, in which governments are more dependent upon the involvement of different stakeholders from various societal spheres (Rhodes 1996; Pollit and Bouckaert 2011). One of these innovative partnerships involving social enterprises is Social Impact Bonds (SIBs).

An SIB is a contract between a public sector authority, an investor(s) and a service provider (e.g. a social enterprise) in which a commitment is made to pay for improved social outcomes that result in savings for the public sector. The repayment of investors depends upon the outcomes being achieved (Gustafson-Wright, Gardiner, and Putcha 2015). The first SIB was launched in Peterborough (UK) in 2010, where investors financed a programme combatting recidivism of ex-offenders (Disley et al. 2011; Fox and Albertson 2011).

SIBs require organizations from different sectors – each operating with certain organizational logics – to collaborate. Social enterprises, for instance, constantly move between different institutional fields and thereby face conflicting logics, by combining economic and social purposes (Doherty, Haugh, and Lyon 2014; Lehner and Kaniskas 2012). Collaboration in cross-sector partnerships is challenging and a breeding ground for uncertainty (Koppenjan and Klijn 2004). Organizations might differ regarding their commitment to partnership goals, stakeholders operate in line with their own organizational culture and agreeing on a shared purpose can be difficult. Furthermore, coordination is time consuming and good relationships develop only slowly (Koppenjan and Klijn 2004; Huxham et al. 2000).

Nevertheless, the interest for SIBs increases rapidly (Gustafson-Wright, Gardiner, and Putcha 2015). This raises attention to the question how SIB actors and networks successfully manage their collaborative effort. Since these partnerships are innovative and require organizations to look beyond their organizational horizons, actors are continuously involved in a collaborative learning process. Cooperation and learning, however, are often separated, while it is crucial to support interaction around complex issues in collaborative efforts. It is through interaction and reflection that underlying assumptions of stakeholders become visible and can be addressed (Koppenjan and Klijn 2004, 11). A deeper understanding of how organizations engage in a collaborative learning process creates valuable knowledge about the favourable conditions and actions that lead to sustained cooperation.

Therefore, this article answers the following research questions: How do stakeholders in an SIB engage in a collaborative learning process, and what are the favourable conditions and actions for successful collaboration in SIBs? In order to answer this question, an in-depth and good practice case study is conducted into the SIB The Colour Kitchen (Utrecht, The Netherlands) by analysing the design process and first year of implementation (daily operational activities, evaluative meetings). With ‘successful’, the research means that a variety of organizations collectively managed to sign an SIB, and that collaboration continues during the implementation phase despite several challenges. The paper does not answer the question whether this SIB has been successful in terms of impact. With ‘collaborative learning process’, the paper refers to continuous learning by organizations through facing and overcoming challenges in their partnership. Research shows that critical reflection is a decisive condition for the development of transformative learning processes in innovative collaborations. Collaborative learning questions assumptions, creates metaphors that facilitate new interpretations of on-going practices and creates new ways of making sense of the world (Sørensen and Torfing 2011, 859). In order to map the learning process and facilitate learning, the method of Learning History was used. This allowed for a close examination of the collaborative process, and through reflection and participation it supported the process of learning (Bradbury and Mainemelis 2001, 352).

This study is academically relevant. Research into collaborative processes in social enterprise partnerships (SE partnerships) (De Bruin and Lewis 2015, 7–8; Jug and Mendoza-Abarca 2016) and SIBs is scarce. Jug and Mendoza-Abarca (2016) conducted a literature review on cross-sector collaborations including social enterprises. They only found nine empirical studies, which stresses the need for further inquiry. De Bruin and Lewis (2015, 8) call for more research into social entrepreneurship as a partnership ‘in order to understand how cross-sector collaborations operate and challenges are resolved’. This study is the first to investigate collaborative learning processes within SIBs and recognizes the complexity of these partnerships and the likely challenges that occur.

Social entrepreneurship literature has spanned a period over 20 years (Short, Moss, and Lumpkin 2009), with much focus on its country-specific definition (see Hoogendoorn, Pennink, and De Vries 2010; Defourny and Nyssens 2010). But, there is far greater demand for ‘research that explores the effects of context, process and the impact of social enterprise’, because practitioners ‘will be better placed to develop informed policies and interventions which develop, support and sustain social enterprise…’ (Shaw and De Bruin 2013, 741). Therefore, with regard to societal relevance, the research first develops collective memory within involved organizations (Bradbury and Mainemelis 2001). Due to an emphasis on innovation in organizations, institutionalizing knowledge is often overlooked (Byrkjeflot and du Gay 2012, 105). Learning History creates written stories based on perspectives of participants (Bradbury and Mainemelis 2001, 342). This enhances institutionalization of knowledge because organizations can easily return to the small stories. Second, the study facilitates learning. Since the process of designing and implementing the SIB is brought back to life, participants are able to reflect on their experiences. In turn, this creates dialogues within and between involved organizations which allows learning to happen and paves the way for a shared desired collaborative future (Roth and Kleiner 1996). Third, since SIBs are complex partnerships and transaction costs for their development are high (Warner 2013), learnings from this article are useful for practitioners interested in SIBs and other innovative partnerships aimed at public problem solving worldwide.

The paper proceeds by providing insight into the configuration and principles of the SIB phenomenon, and explores how collaboration in SIBs can be understood from the perspective of SIB and SE partnership literature. Subsequently, the article argues that SIBs can be seen as ‘collaborative governance’ (Emerson, Nabatchi, and Balogh 2011). In the following paragraph, the paper explores and proposes the main favourable conditions and actions for collaborative governance according to theory. Hereafter the research design and methodology are described. The subsequent section presents the main results of this study. The paper ends with conclusions.

Social impact bond as collaborative governance

With the introduction of SIBs, a new method has been found to finance social outcomes via private investments. SIBs are a type of Payment by Results contract in which financial risks are transferred from public to private sector. If the project fails, there will be no public savings and no money flows back to the investor. SIBs have fixed time periods, and more or less fixed rates of return (Gustafson-Wright, Gardiner, and Putcha 2015). In 2010, Mulgan et al. (2010) expected that the number of commercial SIBs (involving a private investor) will increase, since work will be done on developing proven models of intervention and reliable partners with track records. The latter is the key of SIBs, which are based on proven intervention methods, for instance, the re-entry programme that reduces recidivism in Peterborough (UK) (Disley et al. 2011).

There is no blueprint of an SIB configuration (Arena et al. 2016). Generally speaking, however, most SIBs include (1) a public sector organization (e.g. ministry, municipality); (2) an investor (e.g. philanthropist, private bank, foundation); (3) a service provider (e.g. a social enterprise); (4) an intermediary organization; and (5) an evaluator. The intermediary organization usually coordinates the stakeholders and/or governs the money flows. The government sets the terms of the arrangement, and investors pre-finance the programmes by transferring money to the project implemented by the service provider. Investors are usually non-profit organizations or foundations, which allow them to bear a higher risk and have an interest in creating social returns. Evaluators monitor the development of the intervention, and calculate the return that will be paid (or not) to the investors (Warner 2013; Nicholls and Thomkinson 2013).

Arena et al. (2016) conducted a review of 22 SIBs worldwide, in order to analyse their configuration and to gain understanding to what extent they diverge from the reference model in Peterborough (UK). The latter model consists of an intermediary organization which obtains the operating funds, and contracts the service provider. Arena et al. (2016) found that the reference model differs from most other SIBs. For instance, risks are often shared between stakeholders instead of transferred to the investor(s). Furthermore, the innovativeness of SIBs can be questioned since most SIBs are an extension of existing interventions, and prescribe in their contract the specific intervention. Next, Arena et al. (2016) argue that the current trend of SIBs does not really go beyond traditional public procurement. This might be explained by the ecosystem in which SIBs operate, where, for example, social value is still not recognized everywhere and the prevailing approach to public procurement (lowest price versus quality) hinders innovation.

Social enterprises have been cooperating with organizations from different sectors for a long time. These can be public sector organizations (Terjesen, Bosma, and Stam 2016; Korosec and Berman 2006; Gawel 2014), conventional businesses (Di Domenico, Tracey, and Haugh 2009; Huybrechts and Nicholls 2013) or non-profits and NGOs (such as foundations and citizen initiatives) (Austin 2000). Through collaboration, resources can be exchanged that lead to larger social impact (Meyskens, Carsrud, and Cardozo 2010; Choi 2015). SIBs, however, are far more complex collaborations since they include more than two collaborating partners from different sectors. How can these partnerships be understood from the perspective of social entrepreneurship literature?

Research into (collaborative learning processes within) cross-sector partnerships involving social enterprises is still scarce (Jug and Mendoza-Abreca 2016; De Bruin and Lewis 2015). The same applies to research that analyses SIBs. Several scholars empirically explored (the emergence of) SIBs. The focus of these articles primarily concerns uncertainties regarding technical details (Fox and Albertson 2011), challenges and possibilities in outcome measurement (Jackson 2013), SIB configurations (Arena et al. 2016), opportunities and concerns for the development of SIBs (Warner 2013) and the comprehensive RAND evaluations of the HMP Peterborough SIB (Disley et al. 2011; Disley and Rubin 2014; Disley et al. 2015).

Although the RAND reports mainly focused on the evaluation of the instrument, several key findings relate to collaboration. For instance, it was found that there has been a history of good partnership working and that the time spent with relationship building between organizations positively contributed to develop processes and procedures of collaboration and coordination (Disley and Rubin 2014, 29–30). Furthermore, creative thinking by the governmental organization (Ministry of Justice) was important in designing the SIB because the government had to innovate in order to enable the SIB model to be put in practice (Disley et al. 2011, 53). Besides, Disley et al. (2015, 38) found that the director coordinating the various service providers played a vital role through developing, maintaining and coordinating relationships between service organizations and other agencies. Lastly, Disley et al. (2015, 8) point at the value of discussing details of practical manners upfront because this provides clearance in processes and procedures.

Warner (2013) explored how SIBs can be theoretically understood from the perspective of New Public Management (NPM). She explores, among other elements within NPM, the applicability of Public–Private Partnership (PPP) literature for the analysis of SIBs. Compared to PPPs, however, SIBs consist of a multitude of partners and relatively short-term contracts, whereas PPPs can spend over decades and consist mostly of fewer partners. Thus, PPP literature is not applicable for the study of collaboration in SIBs. Warner (2013, 309) also argues that SIBs are ‘the intellectual descendants of NPM's emphasis on markets and performance management’. This view contradicts the argument by Fox and Albertson (2011) who argue that SIBs provide more room for the service provider to determine their strategy. These two views highlight the ambiguity and uncertainty about where to place SIBs in the academic debate.

This paper argues that SIBs are innovative and complex partnerships (in financing social outcomes), which requires multiple organizations from different sectors to act beyond their own institutional boundaries. This view therefore relates more to Fox and Albertson (2011) and post-NPM by emphasizing the inter-organizational approach and responsibility to solving public problems. Acting beyond institutional boundaries also implies that organizations can often find themselves galloping in the dark regarding their role and the roles of others in an SIB cooperation. SIBs are part of an emerging family of partnerships aimed to address societal problems, through financing social outcomes and Payment by Results agreements.

As outlined above, existing research into collaborative learning processes within SIBs and SE partnerships is scarce and does not grant the tools to seek for patterns. Emerson, Nabatchi, and Balogh (2011) developed a perspective on analysing the dynamics of cross-sector collaborations. Their framework – collaborative governance – provides insight into the favourable conditions and actions that support collaborative learning processes in complex partnerships aimed at public problem solving. Thus, collaborative governance can be seen as a conceptual map for exploring and situating components of SIBs. In this research, collaborative governance functioned as a theoretical lens to search for patterns in the data and to structure the literature and empirical findings.

Several authors developed conceptual frameworks of collaborative governance (see Ansell and Gash 2008; Emerson, Nabatchi, and Balogh 2011). Although frameworks overlap, authors differ in the starting point of their framework. Ansell and Gash (2008, 544) define collaborative governance as ‘a governing arrangement where one or more public agencies directly engage non-state stakeholders in a collective decision-making process that is formal, consensus-oriented, and deliberative and that aims to make or implement public policy or manage public programs or assets’. The definition by Ansell and Gash (2008) includes the government as the key actor of these partnerships. This definition is too narrow for studying SIBs because these are based on a more equal footing, by rejecting vertical NPM relationships (Fox and Albertson 2011). Therefore, the definition by Emerson, Nabatchi, and Balogh (2011, 2) is used, who emphasize the ‘multi-partner governance’ aspect of collaborative governance-based on more balanced powers. They define collaborative governance as ‘the processes and structures of public policy decision making and management that engage people constructively across the boundaries of public agencies, levels of government, and/or the public, private and civic spheres in order to carry out a public purpose that could not otherwise be accomplished’. Besides their definition, the framework is appropriate for this study because it incorporates other and similar frameworks grounded in empirical studies (e.g. Ansell and Gash 2008; Bryson, Crosby, and Stone 2006; Koppenjan and Klijn 2004) and is therefore the most up to date study on cross-sector collaborations aimed at public problem solving.

The following paragraph outlines the main favourable conditions and actions for collaborative governance according to theory. In order to present these coherently and consistently, the framework of Emerson, Nabatchi, and Balogh (2011) serves as a basic structure. They identified a small number of dimensions (system context, drivers, collaborative dynamics and collaborative actions) consisting of components operating in a nonlinear and interactive way to produce actions, followed by outcomes and adaptation. After each dimension a proposition derived from Emerson, Nabatchi, and Balogh (2011) will be stated.

Proposed favourable conditions and actions

System context

Ansell and Gash (2008) and Sørensen and Torfing (2011) speak of starting conditions that either facilitate or discourage cooperation. Emerson, Nabatchi, and Balogh (2011), however, argue that these factors should be seen as a surrounding three-dimensional space because external factors can influence collaboration at any time (e.g. an election). This study supports this argument, especially since the collaborative process in an SIB spans over multiple years.

The system context represents legal, political, environmental and socioeconomic influences that create opportunities and constraints and influence the collaborative process. The literature distinguishes many favourable conditions, of which three main factors will be outlined here. First, collaborations are more likely to succeed when existing networks are already in place before their initial formation (Bryson, Crosby, and Stone 2006; Koppenjan and Klijn 2004). Prior relationships imply that partners have interacted in positive ways in the past, thereby creating initial trust and lowering transaction costs. The same counts for prior general agreement about the problem that the partnership will address, the second favourable condition (Bryson, Crosby, and Stone 2006, 46). Third, political dynamics (Emerson, Nabatchi, and Balogh 2011; Ansell and Gash 2008) may result in policies and regulations that support the emergence of partnerships between organizations or provide incentives for (individual) organizations to work together. These conditions also have negative sides, e.g. when prior cooperation failed it leads to distrust or political dynamics create unfavourable conditions for collaboration. The potential of favourable conditions in the system context is unlimited. However, it is clear that contextual elements can influence collaboration at the outset or over time. This leads to the first proposition:

Proposition 1:

The system context creates opportunities and constraints for collaboration, and influences the dynamics of the partnership at the outset and over time.

Drivers

From the system context, conditions are shaped that create the drivers for collaboration. Drivers are different from the system context, because drivers form the impetus for collaboration which would otherwise not unfold. One or more of the following four drivers are necessary for a collaboration to begin (Emerson, Nabatchi, and Balogh 2011, 9–10). First, leadership secures resources and brings organizations together. According to Ansell and Gash (2008), facilitative leadership is the most important driver of collaboration, by setting and maintaining ground rules, creating trust or involving stakeholders. Second, consequential incentives refer to internal or external drivers for cooperation. Such incentives are consequential because they, for example, represent salient issues for organizations or there is severe time pressure. Third, interdependence, when organizations are unable to accomplish something on their own they have to collaborate (the ultimate consequential incentive). Interdependence as a driver relates to findings from Henry (2015), who found that the SE partnership in her study was able to emerge because partners could not solve the social problem on their own. Lastly, uncertainty challenges organizations to manage wicked problems in collaboration. Through sustained cooperation in networks wicked problems can be better addressed (Koppenjan and Klijn 2004). This leads to the second proposition:

Proposition 2:

One or more of the drivers of (1) leadership, (2) consequential incentives, (3) interdependence or (4) uncertainty are necessary for collaboration to begin. The more drivers present and recognized by participants, the more likely a partnership will be initiated.

Collaborative dynamics

The form and direction of collaboration and action are shaped by the drivers and system context, the development and effectiveness are influenced over time through the collaborative dynamics (Emerson, Balogh, and Nabatchi 2011). Collaborative dynamics are hard to represent given their nonlinear character (Ansell and Gash, 2008, 550). Successful collaboration requires elements within principled engagement, shared motivation and capacity for joint action to interact (Emerson, Nabatchi, and Balogh 2011).

Principled engagement is about reaching a shared goal and theory of action. Principled engagement can be reached through an iterative social learning process consisting of four elements (favourable actions). First, there needs to be a discovery of individual and shared goals, values and concerns. Second, a process of definition helps to reach consensus on the purpose, objectives and concepts of the collaborative governance. This can be done through articulating common purpose and agreeing on certain concepts people use to describe or discuss pitfalls and opportunities. Third, deliberation – reasoned communication – is crucial for successful engagement between stakeholders. Reasoned communication involves, for example, listening to other's perspectives, ability to have hard conversations or expressing honest disagreements. Fourth, principled engagement is concerned with the process of making joint determinations about procedural decisions (e.g. setting agendas, assigning working groups) and substantive decisions (e.g. agreeing on action items). Research shows that, in general, consensus building increases the quality of determinations (Emerson, Nabatchi, and Balogh 2011). Related to both determination and definition, Koppenjan and Klijn (2004) argue that shared agreement about the problem definition in networks is important. Principled engagement reinforces and creates shared motivation and builds a capacity for joint action (Emerson, Nabatchi, and Balogh 2011, 13).

Shared motivation is a self-reinforcing cycle of mutual trust, mutual understanding, internal legitimacy and commitment (Emerson, Nabatchi, and Balogh 2011). First, trust has proved to be a favourable element in reducing transaction costs, stability in interaction patterns, innovation and knowledge exchange (Koppenjan and Klijn 2004). Meyskens et al. (2010, 448) found that SE partnerships are mostly based on levels of trust between individuals rather than between organizations. Second, and in turn, trust enhances mutual understanding, which is about the ability to understand and respect the position of other actors, even when parties do not agree. A mutual understanding is reached when participants respect others’ positions. This enables stakeholders to reveal themselves to others and be appreciated. Third, mutual understanding generates interpersonal validation and cognitive legitimacy. Thus, when individuals experience trust and have mutual understanding, they feel legitimized and motivated to collaborate. Fourth, a sense of commitment towards the process and each other is able to emerge which allows participants to cross their organizational, sectoral and jurisdictional boundaries. In sum, repeated interactions through principled engagement foster trust, mutual understanding, legitimacy and commitment, thereby generating and sustaining shared motivation. The latter enhances and sustains principled engagement in a virtuous cycle (Emerson, Nabatchi, and Balogh 2011, 13–14).

Lastly, actors must generate a capacity for joint action that did not exist before. Four elements create the potential for effective action. First, procedural and institutional arrangements (informal and formal rules that structure interactions over time). For instance, in order to effectively manage perceptions, rules of the game and goals of stakeholders, Koppenjan and Klijn (2004, 203–208) propose to make use of independent intermediaries who are in charge of process management. These facilitators create and support interaction between parties by exploring mutual gains, organizing meetings (drawing agenda), providing meta-communication or introducing rules. Second, leadership can be essential on the outset (as a driver) or during the process. Moreover, leaders may be required on multiple levels and during different moments in time (resolving conflict, implementing actions). Third, successful collaboration requires partners to share (generated) knowledge, the currency of collaboration. Fourth, (scarce) resources should be shared and leveraged in partnerships. These may include, for example, budget support, time and administrative assistance. The necessary levels for the four elements are determined by the purpose and theory of action of the partnership. Both principled engagement and shared motivation will stimulate the development of a capacity for joint action (Emerson, Balogh, and Nabatchi 2011).

The factors described in this paragraph positively influence collaboration; this also implies that the absence of these elements decreases the likelihood of collaborative governance (e.g. no commitment or no reasoned communication). The described actions and conditions within collaborative dynamics lead to the third proposition of this study:

Proposition 3:

The quality of collaborative dynamics depends on the productive and self-reinforcing interactions among (1) principled engagement (discovery, definition, deliberation, determination), (2) shared motivation (mutual understanding, internal legitimacy, shared commitment) and the (3) capacity for joint action (institutional arrangements, leadership, knowledge, resources).

Collaborative actions

Depending on the specific context and aim of partnership actions may include enacting new policy measures, deploying staff, enforcing compliance or monitoring implementation. Emerson, Nabatchi, and Balogh (2011, 17–18) argue that actions are more likely to be successful if partners share a theory of action and the collaborative dynamics (e.g. leadership, trust, deliberation) generate capacity for joint action. Therefore, the fourth proposition is as follows:

Proposition 4:

Collaborative actions are more likely to be implemented if (1) a shared theory of action is identified explicitly among the collaboration partners and (2) the collaborative dynamics function to generate the needed capacity for joint action.

Emerson, Nabatchi, and Balogh (2011) also describe two additional dimensions: impact and adaptation. This paper views on-going collaboration in the implementation stage (collaborative actions) as the impact and success resulting from the collaborative dynamics.

Research design and methodology

Providing in-depth knowledge is the key advantage of single case study research (Flyvbjerg 2001). Moreover, because not much is known about SIBs and SE partnerships, it is highly relevant to conduct explorative and in-depth research into the favourable conditions and actions for collaborative governance in SIBs. Although there are no generalizable answers in case studies to making partnerships work, recognition and understanding of these factors provide a platform for participants and practitioners to reflect, learn and create desirable action in their particular situation (Huxham et al. 2000, 353). The following paragraphs present the case of this study, outline the research design, and elaborate on the collection and analysis of data.

Case selection and description

In order to investigate which favourable actions and conditions contribute to successful collaborative governance, a ‘good practice case’ of collaboration in SIBs has been chosen, since organizations managed to close a deal. In addition, collaboration is seen as successful in the implementation phase (evaluative meetings, daily operational activities) because cooperation remains, despite several challenges. However, this does not imply that this has been the best practice, because there are indications for improvements that will be presented in this paper. In this study, the second SIB in the Netherlands is analysed. This partnership was established in April 2015. Partners agreed on a four-year contract. The case was selected because it was the only recently closed SIB in the Netherlands. At the start of this research, the SIB in Utrecht has been in operation for half a year. This made it easier for participants to reflect on the design stage. Moreover, insights could be gathered from the first decisive moments and experiences in the implementation phase.

Rabobank Foundation (part of Rabobank Group) together with Start Foundation (social investor) pre-finance €734.000, in a project implemented by The Colour Kitchen (a social enterprise). In contrast to the reference model in Peterborough (UK), this SIB does not have an intermediary organization that governs the money flows. Investors rather transfer money directly to the service provider. The Colour Kitchen educates and coaches – during four consecutive years – 252 youngsters without any basic educational qualification.¹ As a restaurant, The Colour Kitchen educates the clients (cook, host/hostess and catering assistant) which should lead to jobs for the youngsters in the hospitality industry. The social enterprise has several restaurants in the Netherlands and provides catering for multiple organizations. The involved stakeholders agreed on three main performance agreements on percentages on graduates, paid jobs after fulfilling the programme and keeping hold of a job. If these performance targets are met, the Municipality of Utrecht pays back the investment to the investors and a potential bonus if results are better than expected. Stakeholders initially agreed on a maximum annualized return ranging between 2% and 6%. The municipality is able to pay back the investors by its savings on social welfare (estimated €800.000). During the programme, participants receive social welfare payments. However, it is expected that participants will find a job more quickly compared to a reference category. Society Impact functioned as an intermediary organization and Pimbaa developed the societal business case underlying the entire SIB structure and conducts evaluative research (Start Foundation, 2015, 22–25). The risks of failure and extra savings in the SIB are allocated among the participating organizations.

Research design

One of the aims of this research is to let SIB participants learn from their experiences. Critical reflection is important for the development of transformative learning processes in innovative collaborations (Sørensen and Torfing 2011, 859). Moreover, facilitating interaction between participants supports learning processes (Koppenjan and Klijn 2004, 11). Therefore, and in order to analyse the collaborative process, a methodology originating from action research (Lewin 1951) has been used: Learning History (Roth and Kleiner 1996). Learning History is able to facilitate learnings by means of (inter)action and reflection and ‘catalysing synergy between scholarship and practice’ (Bradbury and Mainemelis 2001, 341).

The method measures and assesses organizational learning surrounding-specific events (e.g. organizational change) within a single organization. This research, however, concerns six directly involved organizations related to one specific event: their collaboration in the SIB. The method helps SIB participants to create an organizational infrastructure to support learning by bringing the history of the project to life (Bradbury and Mainemelis 2001, 341). Bringing the history to life and (collaboratively) reflecting on it is important because it unravels the underlying assumptions or causes of action (mental models) of participants. This so-called ‘double-loop learning’ ultimately leads to the creation of dialogues between people, using a storytelling perspective. In order to organize collective memory of learnings, Learning History produces a document which portraits a jointly told tale. The document functions as a mirror by telling the story from the perspective of the participants, and by making contradictions discussable (Bradbury and Mainemelis 2001, 352). The jointly told tale is not included in this article.

Data collection process

Data were gathered using different methods, which leads to triangulation and therefore increases the validity of this research (Krefting 1991). This study analysed the design stage of the SIB and the first year of implementation, because collaboration is needed within the implementation phase (evaluative meetings, daily operational activities). The research process can be distinguished into four phases.

First, the research involved participants in an early stage through three preliminary interviews. All three participants played an important role in the SIB and by having these exploratory interviews the researcher was able to (1) focus the study, (2) discuss the aim, most pressing topics and methodology, and (3) gain access to the other key persons in the SIB partnership.

Second, by speaking to the 12 key persons of this SIB, the researcher was able to gain insight into different experiences of those directly involved. During the interviews, the ladder of inference (Roth and Kleiner 1996) was used in order to reveal pure observations and work towards assumptions and abstract rationalizations of participants. The core of each interview was to ask participants to pinpoint their key notable moments in the collaboration on an empty timeline. They were asked why they chose these moments and what happened (e.g. what did you experience?). In addition, the interviews focused on factors of success and challenges in their collaboration and functioned as a way of reflection. Four external experts were interviewed, which provided insight into more general dynamics of SIB collaboration and reflection on the specific case. All interviews (see Table A1 for indicative job titles, Appendix) were semi-structured, which allowed participants to tell their story, while remaining some controlled by the researcher (Bryman 2012, 471). The average interview lasted well over an hour, with the shortest taking 10 minutes (interviewed for the second time) and the longest running nearly 3 hours.

In the third phase, a focus group was held consisting of six participants (from five organizations). The focus group bridged individual experiences, in order to come to a better understanding of their collaboration. Participants of the focus group said to be grateful for the meeting because their normal joint meetings were merely focused on ‘doing business’. The focus group provided them the opportunity to reflect on their partnership in a comfortable setting. Besides these primary sources of data, documents served as secondary data (websites, press releases, the societal business case and the ‘collaborative agreement’) throughout this study. These provided information about attitudes of organizations towards the SIB and background information.

Lastly, participants were asked to comment on a concept version of the Learning History. Their comments were incorporated in the document. Hereafter, the Learning History was given back to the participants of the research and meant for Internal use only. The audio-taped interviews and focus group were transformed into verbatim transcripts and analysed with the assistance of MAXQDA 12. The analysis followed commonly used qualitative research steps of coding (open, axial, selective coding) (Corbin and Strauss 1990). These different steps mingled due to the iterative character of research (Gioia and Pitre 1990, 588).

Findings

The findings from the in-depth case study show multiple factors influencing collaborative learning processes in the SIB. By applying the chronological principle of Learning History, conditions and actions that contribute(d) to successful collaboration have been identified, as well as current challenges and potential improvements. The results are structured along several dimensions of the collaborative governance framework (Emerson, Nabatchi, and Balogh 2011).

System context factors

Three favourable conditions were found at the level of the system context: (1) societal developments; (2) existing networks and (3) prior agreement.

First, societal developments created a supportive environment for partnerships involving social enterprises to emerge. These developments concern the changing roles of different organizations. With regard to the local government, their changing role is partly caused by a new Dutch regulation (‘Participatiewet’). This regulation decentralizes responsibilities regarding social welfare to the municipal level, facing budget cuts at the same time. Municipalities in the Netherlands are currently discovering how to face challenges stemming from this new regulation. In general, the role of governments declines and they become more dependent on partnerships between public and private sectors for the creation of public value. In this case, the Municipality of Utrecht decided to actively engage with social enterprises in order to address unemployment. Participants mention an expert session held by the local government in June 2014 that invited social entrepreneurs from the city. During this meeting, the municipality presented their new policy on social enterprises and invited social enterprises to develop and hand in business cases aimed to provide jobs to people with a distance to the labour market. The government stated that they did not have any format for a business case, and their primary goal is to get people to work. The director of an involved department of the municipality described the meeting as follows:

For us, what happened at that meeting was impressive. Entrepreneurs reacted enthusiastic, something which civil servants did not count on. We expected to receive criticism, but they applauded for us. For some civil servants it was a turning point. (P1)

One of the directors from the social enterprise argued that the expert session was the starting point to collaborate with the municipality. The participant immediately called a civil servant:

I asked ‘do you have any formats?’ ‘No we don't he said’. That was very positive. You let an entrepreneur be an entrepreneur by not providing them a survey, but to hand in a business case from their perspective. (P2)

The expert session and mentioned conversation resulted in a first version of the business case, participants did not think of an SIB at that time. The changing role for the government has been extraordinary. Until then the municipality used to prescribe procurements, provide subsidies and make payments according to market demands. In its new policy, more attention is paid to Payment by Results and social enterprises. This finding relates to the favourable condition of political dynamics, because the changing role of the government created an environment supportive for collaborative governance (Emerson, Nabatchi, and Balogh 2011). Another societal development is the involvement of private banks into social returns, partly in order to improve their image and because they see opportunities in the social economy. For instance, the Rabobank Foundation is increasingly working with social enterprises and sustainability. Thus, changing roles causes organizations to work differently and become more open to collaboration. Subsequently, an ecosystem emerged supporting initiatives like SIBs.

Second, existing networks led to high levels of initial trust between stakeholders. For instance, both investors were already supporting and investing (in) the social enterprise. In addition, the local government was already delivering clients to the service provider on a smaller scale. Furthermore, several participating organizations were already exploring Payment by Results instruments like SIBs. For instance, Start Foundation already participated in the first SIB in the Netherlands, and Society Impact, Pimbaa and the Municipality of Utrecht were already jointly experimenting with some principles of SIBs (2013–2014). According to participants, these relationships were important because they set the basis for high levels of trust and lowered transaction costs by comparing this cooperation with other SIBs or partnerships. In addition, the track record of the service provider and initial confidence in the qualities of the social enterprise turned out to be important. The essence was to continue along the same path for the government. Both government and private investors argued that they wanted to make use of the track record, knowledge and skills (qualities) of the social enterprise.

Third, there has been prior agreement about establishing a partnership and problem definition. Although not all parties initially thought of an SIB, organizations had the incentives to start cooperating (scaling of business and seeking new business models, corporate responsibility, budget cuts, reducing unemployment rate more effectively). According to participants, prior agreement accelerated the process because there was no need to discuss whether every party supported the definition of the problem and the instrument of an SIB (see also principled engagement, processes of definition).

To conclude, Proposition 1 corresponds to the findings because several favourable conditions generated a supportive contextual environment for the SIB. The results confirm the expectation that existing networks play a major role in cross sector collaboration, because they incorporate high levels of trust (stable interaction patterns) and lower transaction costs (Emerson, Nabatchi, and Balogh 2011; Koppenjan and Klijn 2004). Besides, prior agreement was described as a proposed favourable condition by Emerson, Nabatchi, and Balogh (2011).

Drivers

Two drivers of collaboration formed the impetus to collaborate: (1) leadership and (2) interdependence.

No matter how determent participants were to start an innovative partnership, organizations were not taking steps forward. Stakeholders did not succeed yet in meeting up to discuss the possibilities of working together. After a few conversations between individual organizations, the social enterprise and one investor asked an intermediary to bring together the main parties in autumn 2014. According to several participants, there was a need to accelerate the process. Society Impact brought together the right persons at the right levels by demonstrating leadership. The intermediary quickly confirmed the prior agreement (see System context section) and used their network to bring together high-ranked employees of the organizations. By contacting these high-ranked positions the partnership gained legitimacy, it was easier to make decisions faster because they had a mandate and conversations could soon focus on action instead of gathering support. The following quote (investment manager, investor) illustrates the role of the intermediary:

They are really good in catalyzing these projects, in bringing people together. They brought together parties that normally would not join if they had to do it themselves. (P3)

The second driver, interdependence, plays a major role in this SIB. Since organizations depend on each other in order to reach strategic goals, they are unable to accomplish their goals on their own. Governments have retreated from social policy (budget cuts and neoliberalism) and are not ready yet to pre-finance programmes, so they seek for private investments. In this case, the social enterprise depends on government institutions for the delivery of candidates. Interdependence was also important during collaborative action and can be challenging as well. For instance, there might be tensions between service provider and government regarding the delivery of clients. The service provider might want to get ‘easy/profitable’ clients (cherry picking), but the government could send ‘difficult/less profitable clients’. Participants in the research were aware of these tensions and argued that trust is important in this respect. Thus, trust is based to a great extent on the interdependence between organizations: organizations need to trust each other because they are interdependent. Given this interdependence, collaboration occurs, remains and prevents deviant behaviour by stakeholders

In sum, leadership and interdependence have been the two major drivers of the SIB. This corresponds to the expectation (Proposition 2) because two of the four proposed drivers formed the impetus to initiate the partnership (Emerson, Nabatchi, and Balogh 2011).

Collaborative dynamics

The quality of collaborative dynamics in the SIB was enhanced through several factors. The following paragraphs show the importance of principled engagement, shared motivation and capacity for joint action.

Principled engagement

During the social learning process of the SIB, two favourable actions have been important in contributing to principled engagement: (1) transparent, reasoned communication and (2) creating a shared language. The findings also reveal that principled engagement occurred over time through three of the four main elements discussed by Emerson, Nabatchi, and Balogh (2011): discovery, definition and deliberation.

The results demonstrate that everything starts with transparent, reasoned communication. With regard to the discovery of goals (Emerson, Nabatchi, and Balogh 2011, 12), participants disagreed about the level of transparency in stating these goals during the first meetings (autumn 2014). However, analysing individual goals of stakeholders leads to the conclusion that individual strategic goals were best served through the common purpose. The latter being: getting youngsters with a distance to the labour market back to paid employment. Several interviewees argued that repeating this shared goal during those moments in which substance went into detail was important in emphasizing togetherness. This is what Emerson, Nabatchi, and Balogh (2011) mean with process of definition – articulating common purpose serves as a favourable action for collaborative governance. Furthermore, there has been reasoned communication between partners. For instance, in case of conflict actors communicated in a transparent way and settled their disputes. Thus, this finding shows that successful collaboration requires deliberation, through the ability of having hard conversations and expressing honest disagreements (idem).

The second favourable action within principled engagement has been the creation of a shared language. Interviewees argued that there have been some differences in interpretations of concepts such as ‘results’ or ‘return’, as a contract manager (municipality) argues:

You have to speak the same language when closing a SIB. The word ‘result’ for instance means something different for The Colour Kitchen compared to the Municipality, and the same applies to Rabobank Foundation and Start Foundation. (P4)

Again, transparent communication about what organizations mean when using these words has been crucial. In addition, it was found that the societal business case developed by Pimbaa bridged languages by defining concepts. Thus, processes of definition (agreeing on concepts) and deliberation (communicating about these) (Emerson, Nabatchi, and Balogh 2011) interacted and functioned as favourable actions within the partnership. Interpretation differences also resulted in a major challenge during the final process of designing the SIB (spring 2015). Almost all participants mentioned that differences in interpretations were clearly visible when lawyers entered the process, and that this slowed down the design process. For instance, lawyers of the investors initially wrote down that the municipality guarantees the loan provided by investors to the service provider. However, in legal terms, a municipality cannot guarantee for working capital. Therefore, and after conversations with government lawyers, they decided to skip this line. Nevertheless, the key is that lawyers represent the interests of their own client. Therefore, it is not so much an issue of conflicting languages but of conflicting goals and organizational logics, as one expert participant (attorney) argues:

Legally speaking SIBs are not so complex. It is complex because different organizations are involved. Banks are used to tightly drawn contracts, governments prefer to leave some room. You have to connect both. (P5)

Stakeholders in the SIB were able to overcome these organizational logics through leadership (see Capacity for joint action section).

Shared motivation

Interpersonal relational elements have been essential in the collaborative dynamics. Shared motivation reinforces engagement processes and the other way around. The data support the importance of (1) mutual trust and (2) (personal) commitment as main favourable factors.

Mutual trust is essential because it, for example, enables people to look beyond their own personal or organizational frames, creates mutual understanding and commitment. In this case, relational trust (Vangen and Huxham 2003) has been found as a crucial ingredient for successful collaboration. As mentioned in the section on system context, existing ties between organizations led to stable interaction patterns and trust. Participants mentioned they trust each other because they know what to expect from one another. A good example of relational trust is the chosen formal configuration for this SIB, which requires high levels of trust between organizations. The ‘collaborative agreement’ (not a ‘contract’) indicates a soft relationship. For instance, in comparison to the first SIB in the Netherlands, this SIB does not incorporate a separate foundation. This forms a risk for investors because they can be directly sued, instead of an intermediary foundation. Furthermore, the SIB does not prescribe clear client and contractor roles. This makes the construction less complex, but it also means that collaboration is based more on trust. The fact that organizations chose for a soft construction and managed to collaborate successfully indicates that there has been a great amount of mutual trust. As presented in the section on ‘drivers’, trust was also manifested through interdependency relationships.

Commitment plays a vital role in the collaborative dynamics of this SIB. The four key organizations (service provider, two investors and government) all experienced a strong commitment towards the partnership. Each organization wanted to close the deal as soon as possible. This commitment is especially visible on a personal level. All participants mention the use of the phrase ‘we are just going to do it’, which was used by several participants during negotiations in which substance went into detail, to emphasize togetherness or to solve problems. It confirms the commitment towards making this innovative alliance happen despite all the challenges they may face. Two quotes (investment manager, investor; director department, municipality) illustrate this personal commitment:

In the end people need to do it. I think that prestige played a role, during the process I thought ‘this is going to make it’. I think that the drive behind it was enormously helpful. (P6)

I personally make the case that we will do this in another way and in cooperation. And I take responsibility for that to organize it within my own organization. (P1)

Capacity for joint action

As the third component capacity for joint action refers to a collection of cross-functional elements that create the potential for taking effective action (Emerson, Nabatchi, and Balogh 2011, 14). It was found that three factors influenced dynamics: (1) resource management, (2) project management and (3) facilitative leadership.

First, organizations managed to utilize and exchange resources effectively. Actors differ with regard to their size, time, knowledge and skills. This implies that some organizations posit knowledge that others do not have. For instance, the local government was not able to conduct a due diligence on the business of the social enterprise, whereas one of the investing organizations has the knowledge to do so. By working together different resources could be utilized. At the same time, these differences imply dependency relationships. This can be a pitfall in SIB collaboration according to one participant. Since SIBs are innovative, not much knowledge is widely spread between different sectors. In this SIB, substantive knowledge and experience with SIBs were only present within Start Foundation and Pimbaa. The following quote by a consultant from research company Pimbaa illustrates how the risks of their knowledge position were mitigated, in which commitment (shared motivation) towards each other plays a role:

We were aware of our knowledge position. We sometimes said to each other ‘do they notice and do they really understand what they are doing?’ More or less, but we just make sure they understand and not abuse our power. (P7)

Thus, Start Foundation was well aware of their knowledge position and chose not to abuse their power by explaining carefully the proposed steps to design the SIB.

Second, stakeholders took joint responsibility for project management. Participants argued that governing roles and tasks were not clearly defined during the process. In the early design stages, Society Impact functioned as an intermediary by bringing actors together, arranging the first meetings and catalysing the project. After some time, other organizations were more and more communicating on an individual basis, the investors no longer saw the added value of an intermediary; this kept speed in the process and saved costs. In addition, the business case did not include the costs of such a role upfront; therefore, there were no negative financial consequences for the business case of the SIB. In the end, this led to a gradual and temporary (the intermediary organized the launch event of the SIB) disappearance of the intermediary in the design stage. The consequences of this change, however, were that participants described the role of who governs the meetings as ‘vague’ and the remaining organizations were looking at the municipality as the leader of the SIB. This raises the question whether SIBs require an intermediary organization to function effectively. No consensus was found between participants (director, social enterprise; contract manager, municipality):

I would not say that some problems could have been prevented by an organization that intermediates the process. But, I would not deny that such a role would have helped by critically looking at what we are actually doing. (P8)

They were the linking pin and organized meetings. It turned out that the investors said they could do it on their own. Suddenly we were ‘the SIB’ and had to organize meetings, that is not how it was supposed to be. (P4)

After organizations were brought together, project management was needed to organize meetings and implement deals (institutional arrangements, Emerson, Nabatchi, and Balogh 2011). In this SIB – despite governing challenges – the success partly lies in the fact that several organizations took responsibility to manage the process. However, Start Foundation took the lead by the introduction of a roadmap towards establishing the SIB because they gained experience from the first SIB in the Netherlands (see also resource management). Although most participants are satisfied with this role, some argue that project management should be one of the tasks of an independent intermediary.

The third and the most important favourable action within capacity for joint action has been facilitative leadership (Ansell and Gash 2008). Leadership has been important to underscore togetherness (see commitment), acquiring mandate and to overcome organizational logics. Since SIBs are innovative instruments, they require actors to act beyond organizational boundaries. This implies being creative with existing organizational rules that hinder collaboration and acting with vigour. For instance, the director from the involved municipal department asked for internal legal advice regarding the signing of a Memorandum of Understanding:

I received nine pages of reasons why I could not sign that agreement. I thought about it for a few days, and I just tore up the document. But, with support from the top to continue and find other ways. (P1)

The example shows that organizational logics are not accustomed to SIBs, and leaders are needed to break through these boundaries. This requires looking beyond current legislation; allowing people to innovate by providing them room to do so and not rejecting anything that does not fit into regular ways of doing business; looking beyond the term of office of the city council (for the municipality). The internal organization of large stakeholders (government and bank) inhibits several norms and regulations that hindered the SIB. For instance, the local government faced challenges regarding the principle of arbitrariness, since they did not call for a tender. All participants mentioned the great vigour of several key persons during the process. This, in turn, was partly caused by the mandate (high level) and a large personal commitment as described earlier. However, the mandate itself also needed to be acquired through continuous engagement between civil servant(s) and alderman. Leadership was present and needed at multiple levels and organizations. For instance, representatives from Rabobank Foundation needed to collaborate with other departments and affiliated institutions (local banks), who were not used to participate in this new way of pre-financing. Furthermore, it is not Rabobank that invests in the SIB, it is the foundation affiliated to the bank. The foundation was not used to invest money in the Netherlands, and relies on donations from its local banks in the Netherlands.

In sum, principled engagement, shared motivation and capacity for joint action contributed to successful collaboration in the SIB. Besides, the findings showed that these dimensions and the elements within dimensions reinforce each other in an iterative process. Although not all elements within the proposed favourable conditions and actions of collaborative dynamics (Proposition 3) were present, the research identified a considerable amount of success factors.

Collaborative actions

The implementation phase mainly consists of operational activities conducted by the social enterprise (e.g. running the programme). Involved civil servants and employees of the social enterprise collaborate on a daily basis, and have evaluation meetings on the recruitment and delivery of clients. In addition, the research firm conducts evaluative research, and four times a year, main representatives meet to discuss the progress of the SIB.

Regarding the first year of implementation, it was found that favourable conditions and actions within principled engagement and shared motivation are important. Participants especially mention (historic ties of) mutual trust and transparent, reasoned communication. Investors and service providers argue that trust plays a role with regard to balancing freedom and control over the operational activities. The following quote illustrates how the service provider (director, social enterprise) perceives trust:

I do not have the feeling that someone is watching me over my shoulder to check whether everything goes well. (P2)

This experience by one of the directors of the social enterprise might be explained by the early stage of the implementation process since no emergent issues regarding the outcomes have been reported so far. During the focus group participants reflected on this. Investors argued that they are confident that the social enterprise is able to implement the project. However, investors also made clear that in case of failure – missing (repeatedly) multiple targets – they might have a larger interest in what the social enterprise exactly does in her operational activities. Transparency (deliberation), again, seems to be a key component to preventing conflicts. Interviewees argued that by keeping each other involved about the implementation helps partners understand what is happening and provides them the opportunity to give feedback.

The challenge that participants reflected on relates to the delivery of clients to the service provider, other challenges revolve around a complex VAT issue on services provided by the social enterprise. With regard to the overarching cause of these issues, the findings hint at the downside of commitment and leadership (vigour). Due to these favourable actions, details in designing the collaborative agreement were often skipped. For instance, the collaborative agreement does not incorporate important implementation elements (how to invoice, settling the performance targets). The following quote by a director of the social enterprise illustrates this:

Regarding implementation, everyone should have allocated more people, time and attention to the project. We were aware to choose for such a close deadline. We knew we were signing things that need to be discussed yet regarding further implementation. (P8).

Nevertheless, participants are confident that they are able to solve these problems. Another favourable factor for collaborative action has been the stable shared theory of action (Emerson, Nabatchi, and Balogh 2011, 17–18). Participants still attach great value to addressing the societal problem through the intervention by the social enterprise.

Besides positive factors there are indications for improvements. As soon as the contract was closed (April 2015), (1) multiple relocations occurred at several organizations and (2) the involved high-ranked civil servant who also relocated was not replaced by his successor. Several interviewees stressed their concern over these developments because they question whether the feeling of togetherness can still remain and there needs to be an accountable government official with decision-making power – who is directly involved – in the light of current challenges. Given the importance of leadership, commitment and stable interaction patterns – as pointed out earlier in the paper – both challenges put pressure on collaboration and the capacity for joint action. Thus, there is a need for (1) accountable high-ranked civil servants with decision-making power and (2) continuous commitment from all involved representatives.

To conclude, the findings correspond to Proposition 4 because it was expected that collaborative actions will be successful if there is a shared theory of action, and the collaborative dynamics generate the needed capacity for joint action. Although mutual trust and transparent, reasoned communication steer on-going cooperation, the absence of accountable leaders and continuous involvement might form a barrier for effective collaborative action.

Conclusions

This research conducted a good practice case study on the collaborative learning process of an SIB to enhance our understanding of the favourable conditions and actions for collaborative governance. The research highlighted a large variety of success factors, found indications for improvements and made linkages to theoretical propositions. The findings are summarized in Table 1. More specifically, the contribution of this paper has been threefold.

Table 1. Favourable conditions and actions for collaborative governance in the SIB.

Propositions	Discovered favourable actions and conditions
1. The system context creates opportunities and constraints for collaboration, and influences the dynamics of the partnership at the outset and over time.	1. Societal developments created an ecosystem for partnerships and social entrepreneurship to emerge. 2. Existing networks set the basis for high levels of initial trust, stable interaction patterns and lowered transaction costs. 3. Prior general agreement accelerated the design process.	Support, conditions shaped supportive environment and served as starting conditions (prior agreement, societal developments) as well as continuous factors (trust).
2. One or more of the drivers of (1) leadership, (2) consequential incentives, (3) interdependence or (4) uncertainty are necessary for collaboration to begin. The more drivers present and recognized by participants, the more likely a partnership will be initiated.	1. Leadership catalysed the project by bringing high-ranked representatives together. 2. Interdependence required organizations to start and remain collaborating.	Support, leadership and interdependence were found as important drivers of collaborative governance in the SIB. Addition to framework: interdependence may reinforce (forced) trust.
3. The quality of collaborative dynamics depends on the productive and self-reinforcing interactions among: (1) principled engagement (discovery, definition, deliberation, determination), (2) shared motivation (mutual understanding, internal legitimacy, shared commitment) and the (3) capacity for joint action (institutional arrangements, leadership, knowledge, resources).	1. Principled engagement: Transparent, reasoned communication through articulating common purpose; participants were able to have hard conversations and subsequently settle disputes (processes of discovery and deliberation). The creation of a shared language (processes of definition and deliberation) accelerated the collaborative process. 2. Shared motivation: Mutual (relational) trust is required because of soft structure in this SIB. Personal commitment to establish SIB, commitment creates e.g. togetherness. 3. Capacity for joint action: Effective resource management by stakeholders. Joint project management as solution to unclear governance of the partnership, kept pace in the process of negotiations and made clear what is needed to design an SIB (institutional arrangements). Facilitative leadership crossed organizational boundaries, obtained internal mandate and emphasized togetherness.	Support, different elements within principled engagement, shared motivation and capacity for joint action positively contributed to collaborative dynamics. Besides, the three dimensions of collaborative dynamics and the elements within the dimensions reinforced each other. Addition to framework: leadership in innovative partnerships can overcome organizational constraints.
4. Collaborative actions are more likely to be implemented if (1) a shared theory of action is identified explicitly among the collaboration partners and (2) the collaborative dynamics function to generate the needed capacity for joint action.	1. Mutual trust positively influences collaborative action by providing operational freedom for service provider. 2. On-going transparent, reasoned communication. 3. Shared theory of action. 4. Need for accountable government leaders with decision-making power. 5. Need for continuous involvement implementation stage; benefiting from stable interaction patterns.	Support, successful collaborative actions depend on shared theory of action and well-functioning collaborative dynamics, that generate capacity for joint action.

First, the paper showed that collaborative governance (Emerson, Nabatchi, and Balogh 2011) helps to understand the dynamics of SIBs. The study indicates that collaboration in SIBs is prone to several internal and external processes described by collaborative governance. Moreover, by presenting collaborative governance as a theoretical lens for the analysis of SIBs and SE partnerships, it adds another academic perspective on social entrepreneurship as a field of research. In Public Administration literature, the changing role of the public sector in creating public value has been described in great detail, with an increasing focus on network governance during the past two decades (see Rhodes 1996; Koppenjan and Klijn 2004). In social entrepreneurship literature, there has been less focus on collaborative processes within partnerships involving social enterprises (De Bruin and Lewis 2015; Jug and Mendoza-Abarca 2016). This study showed that collaborative governance can be used to analyse collaboration in SIBs and likewise emerging innovative partnerships, by looking at the factors and actions that support these partnerships. However, each SIB has its own unique collaborative dynamics. In their interaction, these dynamics determine the success or failure of a partnership.

Second, the research provided insight into the theoretical and empirical factors of successful collaboration in an SIB and potential improvements. The four theoretical propositions derived from the framework of collaborative governance (Emerson, Nabatchi, and Balogh 2011) were supported by the results (see Table 1). Societal developments, existing networks and prior general agreement within the system context created a supportive environment for the emergence of SIBs upfront and functioned as positive contextual factors during the process (Proposition 1). With regard to existing networks, Disley et al. (2014, 29–31) also found that a history of good partnership working positively contributed to collaboration in the SIB. This could mean that the relationship-building element in SIB partnerships is even more important in cases where organizations have not previously worked together. Two essential drivers – leadership (bringing actors together and catalysing the project) and interdependence – formed the impetus to collaborate (Proposition 2).

Insights into the collaborative dynamics revealed how principled engagement, shared motivation and capacity for joint action interacted in a self-reinforcing manner and contributed to effective dynamics (Proposition 3). The research adds one characteristic of leadership to the framework of collaborative governance (Emerson, Nabatchi, and Balogh 2011): breaking organizational logics in innovative partnerships. Leaders are needed who are able to look and act beyond their horizon. The importance of creative thinking was also found in the HMP Peterborough planning stage (Disley et al. 2011, 53). Therefore, empirical findings report – more and more – that working with SIBs means doing things differently than before. Leaders need to break existing barriers, but also invest time in building and maintaining partner relationships (see also Disley et al. 2015, 38). Another addition to the framework is the importance of (forced) trust nested within the driver of interdependence: actors need mutual trust because they depend on each other in order to reach their strategic goals and the shared goal of the partnership. An interesting element within the collaborative dynamics of this SIB has been the role of the intermediary. In this SIB, the intermediary was important for catalysing the project. After some time, however, the role of the intermediary gradually disappeared because investors no longer saw the added value of an intermediary, organizations were more and more communicating on their own and investors wanted to save costs. Furthermore, this kept speed in the process and the costs of an intermediary were not part of the business case. In part, this SIB was saved by the bell since one of the investors obtained knowledge and experience about SIBs (also a monopoly position) and set the institutional arrangements. Thus, this study suggests that intermediaries are not necessary per se in SIBs once collaboration has started, and that the chosen SIB configuration influences collaborative dynamics. However, the extent of independence of organizations managing project coordination can be questioned. Potential problems stemming from this independence can be mitigated by making upfront agreements about each other's roles and related costs. This finding relates to one of the HMP Peterborough learnings because it was found that details of practical manners (e.g. information-sharing) need to be discussed and agreed upon upfront (Disley et al. 2015, 8). In turn, the dynamics created subsequent collaborative actions. It was found that some of the elements described in collaborative dynamics remain important during the implementation, and that a stable shared theory of action contributed to on-going collaboration (Proposition 4). The findings emphasize, however, that collaborative action can be improved by reintroducing an accountable government representative with decision-making power and continuous commitment of involved representatives during the implementation phase. Therefore, findings confirm research by Meyskens et al. (2010, 448) who found that leadership in SE partnerships takes place on an individual basis. It is important to institutionalize leadership and associated institutions (norms, rules) within organizations in order to have sustainable cooperation. Interestingly, it was found that two major favourable actions – leadership and commitment – also had negative implications, which leads to the conclusion that these factors might serve as double-edged swords.

The third contribution of this paper is that it illustrated the complexity of SIBs, and shows why such a normative proposition² is still not widely practised. It was found that many elements within different dimensions of collaborative governance are needed in order to make complex collaboration in an SIB work. In turn, it indicates why the phenomenon is still not widely practised despite the amount of positive interest for the instrument. Since this explorative research is the first to analyse collaborative learning processes within an SIB, future research could investigate other SIBs. More work can be done to identify and explain mechanisms that drive SIB partnerships, and how different dimensions and elements relate to each other. Special attention could be paid to the importance of context (e.g. country, see De Bruin and Lewis [2015]), given the variety of SIB configurations (Arena et al. 2016) and its influence on collaboration.

Acknowledgments

The author thanks Albert Meijer and Peter Linde for their constructive feedback on earlier versions of this paper. Furthermore, the author wishes to thank two anonymous reviewers for their valuable comments and insights.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Compared to international standards, clients of the programme finished (parts of) high school and may obtained further diploma's under a certain level (Dutch: basic educational qualification (‘startkwalificatie’) equals a havo, vwo or mbo level 2 diploma).

2. McHugh et al. (2013) are critical on SIBs: they expressed their concerns by criticizing the mechanical model of cause and effect and thus the failure to grasp the complexity of social problems; warn for the unintended consequences for the third sector and a loss of public and democratic accountability because governments in the UK lose their control over the selection of service providers to an intermediary organization.

Appendix

Table A1. Indicative job titles participants.

Participant number	Indicative job title
1	Director department (Municipality of Utrecht)
2	Director (The Colour Kitchen)
3	Investment manager (Start Foundation)
4	Contract manager (Municipality of Utrecht)
5	External expert (attorney, law firm)
6	Investment manager (Rabobank Foundation)
7	Consultant (Pimbaa)
8	Director (The Colour Kitchen)
9	Matchmaker (consultant) (Society Impact)
10	Business advisor (Municipality of Utrecht)
11	Matchmaker (consultant) (Society Impact)
12	External expert (consultancy firm)
13	Regional manager (The Colour Kitchen)
14	Program manager social entrepreneurship (Rabobank Foundation)
15	External expert (manager SIBs, investor)
16	External expert (SIB analyst, investor)