Working with Complexity in the Context of the United Nations Sustainable Development Goals: A Case Study of Global Health Partnerships

Multi-stakeholder partnerships (MSPs) have become a major driver to attain the United Nations Sustainable Development Goals (SDGs). However, managing MSPs is difficult because of the multiple complexities they involve. We seek to contribute to a better understanding of how MSPs cope with these complexities by exploring the MSP scope. In our study of four global health MSPs, we find that a function-oriented scope in terms of focusing on a single intervention helped filter the relevant external and internal complexities, whereas an issue-oriented scope focused on addressing the health issue with multiple interventions magnified the complexities. As a result, the latter MSPs became overwhelmed and self-absorbed, while the former MSPs managed to remain outward-looking and sustain their collaborative energy and support. On this basis, we identify three mechanisms through which the MSP scope either helped or hampered the ability to cope with complexity, and we discuss the theoretical and practical implications for MSPs addressing the SDGs.


Introduction
The transition from the United Nations Millennium Development Goals (MDGs) to the Sustainable Development Goals (SDGs) has been considered instrumental in shaping and revitalizing a global agenda of social, environmental, and economic development (Van Tulder, 2018;Voegtlin & Scherer, 2017). The main lessons from the MDG era include a greater acknowledgment of the linkages across development goals and interventions to avoid silo approaches, as well as the need for more cross-sectoral decision making and solutions (UNDP & World Bank, 2016). Consequently, SDG #17, Partnerships for the Goals, calls for robust collaborative action-for example, in the form of multi-stakeholder partnerships (MSPs) that bring together multiple partners from the business, government, and civil society sectors MacDonald et al., 2019).
MSPs hold considerable promise for the prospect of addressing grand societal challenges. However, managing MSPs is not an easy task. The societal challenges underlying the SDGs tend to be dynamic, multifaceted, and interwoven, and they commonly unfold in complex environments (Dentoni et al., 2018;Van Tulder & Keen, 2018). Recent work on organizational complexity suggests that one way to master such external complexity is to match it with internal or collaborative complexity (Schneider et al., 2017), which, in the case of MSPs, means involving multiple and diverse partners. However, working with diverse partners is a difficult and complex endeavor as well because of differences in their goals, structures, and ways of working (Ashraf et al., 2017;Quélin et al., 2017;Sharma & Bansal, 2017).
Extant research on how to manage complexity in MSPs emphasizes adapted governance structures and processes (e.g., Bryson et al., 2015;Dentoni et al., 2018;Gray, 1989;Provan & Kenis, 2008;Van Tulder & Keen, 2018) and supporting relational approaches (e.g., Le Ber & Branzei, 2010a;Sloan & Oliver, 2013). While drawing on the 1 3 strategic alliance literature (e.g., Khanna, 1998;Oxley & Sampson, 2004), we suggest that the MSP's scope (broadly defined as what partners do together and what demarcates the partnership boundaries) presents an additional means for the MSP partners to cope with internal and external complexity. For instance, the scope of the MSP may help the partners observe and diagnose the external environment, thereby clarifying priorities and boundaries-for example, what kind of support to offer to beneficiaries and what to leave for unilateral action or other initiatives to act on. Likewise, the scope may help the partners develop the MSP's design in terms of which working groups or internal MSP divisions to have.
To the best of our knowledge, the MSP literature has not yet systematically uncovered the role of the partnership scope in relation to managing internal and external complexities, whereas the strategic alliance literature mostly considers the collaborative scope in terms of managing opportunistic hazards and attaining individual rather than collaborative goals (e.g., Khanna, 1998;Lioukas & Reuer, 2020;Oxley & Sampson, 2004). Consequently, we seek to explore the following research question: How does the partnership scope influence the way in which MSPs cope with external and internal complexities? Answering this question will help advance our understanding of how MSPs manage complexity. It also has practical relevance as a failure to manage an MSP's internal and external complexities could significantly diminish its capacity to contribute to the SDGs in terms of social, environmental, and/or economic impact.
To answer our research question, we studied four MSPs working in global health (and,thus,contributing to SDG #3). We chose global health as our empirical context because, on the one hand, the diverse landscape of health MSPs shows variation in MSP scope and, on the other hand, it has matured over time, with some MSPs even being considered role models for jointly addressing societal grand challenges (Sachs & Schmidt-Traub, 2017). In addition, health problems tend to be complex and interrelated with other SDGs, which provides a suitable context to investigate MSPs' coping with complexity.
Our findings depict the various internal and external sources of complexity that these MSPs had to cope with in their daily operations. With respect to navigating these complexities, two of the MSPs had a function-oriented scope (i.e., the partners addressed the need for a specific health intervention, such as distributing vaccines or developing a cure for a focal disease). The other two MSPs had an issueoriented scope (i.e., the partners focused on addressing one or more health issue(s) more generally). We found the function-oriented scope helped filter the relevant external and internal complexities, whereas the issue-oriented scope magnified the complexities. As a result, the latter MSPs became overwhelmed and self-absorbed, while the former MSPs managed to remain outward-looking and sustain their collaborative energy and stakeholder support. On this basis, our findings suggest that the MSP's scope influences the coping with complexities by (1) allowing for clear messages vs. risking mixed messages for diverse stakeholder groups, (2) providing clarity for vs. obscuring MSP structures, and (3) helping focus attention vs. diverting attention.
These research insights have two main theoretical implications. First, they contribute to the MSP literature by providing a more nuanced picture of the complexities that MSP partners experience in addressing the SDGs, thereby delineating their external and internal sources. Second, we elucidate the role that the MSP's scope plays in working with complexities and, thus, extend existing research on managing MSPs (e.g., Bryson et al., 2015;Dentoni et al., 2018;MacDonald et al., 2019) with a focus on the MSP scope. From a practical and ethical perspective, our findings indicate the capacity constraints of MSPs working with complexities to address the SDGs. Specifically, our findings suggest that by offering multiple interventions for one or more health issues, MSPs risk becoming self-absorbed and ineffective, which also highlights the limitations of what we can expect from single operational MSPs.

The Complexities of Managing Multi-stakeholder Partnerships
MSPs are formed by bringing together various organizations from the public, private, and nonprofit sectors to tackle pressing social, economic, and environmental challenges (Selsky & Parker, 2005;Sloan & Oliver, 2013). The power of partnerships lies in utilizing the synergies these actors can create by combining their different resources, knowledge, and networks. This enables them to come up with more comprehensive and innovative solutions to address grand challenges (Koschmann et al., 2012;Quélin et al., 2017) than a single organization could (Waddock, 1988). As a result, MSPs are seen as important drivers to reach the United Nations SDGs, which call for more comprehensive approaches and the mobilization of various organizations around the world (Van Tulder & Keen, 2018;Voegtlin & Scherer, 2017).
Although social partnerships are generally acknowledged as promising and potentially leading to economic, social, and environmental changes in the long and short term for individuals, organizations, and society (Austin & Seitanidi, 2012), successfully managing and maintaining them is challenging (Caldwell et al., 2017;Ungureanu et al., 2019). Therefore, it is little wonder that many still fail to achieve their goals despite all the effort and resources invested in making partnerships work (Kolk et al., 2008). The challenges encountered are exponential in the context of MSPs, given that the multitude of partners and regions involved entail greater complexity than bilateral or local partnerships do Sloan & Oliver, 2013).
Previous research has grouped the relevant external complexities into two main dimensions. First, MSPs try to tackle 'wicked' problems  such as human rights issues (Baumann-Pauly et al., 2017) or environmental sustainability (Clarke, 2014), which involve uncertainties, ambiguity, and the dynamic interactions between different elements (Dentoni et al., 2018;Gray & Purdy, 2018;Waddock et al., 2015). Second, not only are the problems that MSPs target complex but so is the environment in which they work: It is often turbulent and requires attention to be paid to various institutional elements (Bryson et al., 2015;Gray & Purdy, 2018;Van Tulder & Keen, 2018). For example, the institutional environment may decrease the MSP's ability to function (Moog et al., 2015) and may trigger conflicts between actors from the Global South and North regarding how to best adapt the MSP's approach to the issue (Reinecke & Ansari, 2015). Moreover, an approach that works in one country may not be the best way to address the same social issue elsewhere (Rein & Stott, 2009). Therefore, MSPs might have to adjust to the institutional context in which they are embedded (Vurro & Dacin, 2014).
To deal with external complexity, organizational complexity scholars highlight the need to respond with internal complexity, such as setting up new task forces or new divisions, adding more structural measures, and engaging in collaborative approaches (Daft & Lengel, 1986;Galbraith, 1977;Schneider et al., 2017). In the context of MSPs, this includes the involvement of various partner organizations but can also imply increasing the number of operational projects and MSP structures. However, these measures add another level of complexity to the management of MSPs-a layer that is internal.
Internal complexity in terms of diversity of stakeholders and approaches is the distinctive feature of MSPs (Austin & Seitanidi, 2012;Gray & Purdy, 2018), and the extensive research on MSPs focuses on how to harness this diversity. One of the most important challenges concerns the ability to balance unity with diversity (Saz-Carranza & Ospina, 2011) and dialoge with confrontation (Koschmann et al., 2012;Ospina & Saz-Carranza, 2010), thereby aligning the partners' different interests, worldviews, and ways of working. For example, the difficulty of reaching consensus among actors (Gray & Purdy, 2018) who might have conflicting goals (Battilana et al., 2017;Quélin et al., 2017) has received significant scholarly attention.
In investigating MSPs' intersection with complexity, we focus on operational MSPs that differ from the widely studied standard and regulation-centered multi-stakeholder initiatives (MSIs) (for a review, see de Bakker et al., 2019). Although MSPs and MSIs have similar features (e.g., they include multiple stakeholders and target societal grand challenges), collaboration in MSPs centers on jointly implementing operational activities like services, products, and campaigns and builds on constant partner interaction and coordination (Stadtler, 2016). This operational function makes their tasks even more challenging than MSIs.
In conclusion, many studies have shown that MSPs have to cope with multiple sources of complexity. Drawing on organizational studies, one may further deduce that successfully coping with the external complexities underlying the societal issues that MSPs address requires matching them with greater internal complexities (Schneider et al., 2017). However, our knowledge of how these two levels of complexity simultaneously influence MSPs is still nascent. To address this research gap, we study the MSP scope. Despite it being an important characteristic of MSPs, their scope has not previously received much research attention.
We posit that the MSP's scope is relevant to observing how MSPs cope with multiple sources of complexity for at least two reasons. First, the scope defines the boundaries of the partnership and guides what the partners could and should do together. Thus, it may determine how much complexity they need to work with. Second, the MSP's scope may serve as a bridge to translate external complexity into internal complexity and vice versa. For instance, the MSP scope might determine which partners to engage with (internal) to work in a new country (external), or might influence the leveraging of MSP employees' different expertise (internal) in addressing different donors expectations (external).

The Role of the Scope in Managing Complexity
Seeking to better understand the role of the MSP's scope in managing complexity, we build on the strategic alliance literature, which has drawn close attention to the collaborative (or "alliance") scope. The alliance scope is broadly defined as the range of the alliance's functional activities (Lioukas & Reuer, 2020;Oxley & Sampson, 2004;Smith et al., 2014). Determining an alliance scope is an important decision the alliance partners have to make since it defines the core and limits of the collaborative endeavor. For instance, in the case of research and development (R&D) alliances, scope decisions define whether the partners collaborate only during the R&D process or stretch the boundaries of the alliance to also include manufacturing and/or marketing activities (Oxley & Sampson, 2004). Previous research has addressed two important implications of the alliance scope.
First, narrowing the alliance scope helps control the alliance partners, better monitor individual and joint activities, and consequently prevent opportunistic hazards and free-riding (Oxley, 1997;Reuer et al., 2002;Rivera-Santos & Rufin, 2010). For example, it is widely accepted that by narrowing the scope, the partners might prevent potential information leakage and better protect their sensitive assets from each other, such as specific technologies in R&D alliances (Oxley & Sampson, 2004).
Second, expanding the alliance scope also has its benefits, such as facilitating cross-functional collaboration (Lioukas & Reuer, 2020;Oxley & Sampson, 2004). However, an extended alliance scope is portrayed as requiring frequent and complex monitoring and coordination (Albers et al., 2016;Gulati et al., 2012;Rufin & Rivera-Santos, 2012) and may complicate the sharing of outcomes, such as in the case of property rights (Lioukas & Reuer, 2020). Consequently, partners-especially in international alliances-tend to limit the collaborative scope (Lioukas & Reuer, 2020) as it is difficult to monitor geographically distant partners. Likewise, when the alliance involves critical process uncertainties, such as in R&D alliances (Oxley & Sampson, 2004), or in business and NGO partnerships (Rivera-Santos & Rufin, 2010), partners tend to limit the alliance scope. A tighter alliance scope, in turn, requires less complex contracts, decreases hierarchies, and promotes trust between the partners (Oxley & Sampson, 2004).
On this basis, the strategic alliance literature has analyzed the role of scope in avoiding opportunistic hazards and has delineated its implications for the alliance's governance. However, we know less about the role of the collaborative scope in the context of MSPs, as well as the multiple complexities this context entails.

Methods
To address this research gap, we carried out an inductive study of four global health partnerships and adopted a comparative case study methodology (Eisenhardt, 1989). A comparative approach allows for replication (Yin, 2003) and, in our case, a close observation of the differences and similarities regarding the role the MSP scope plays in dealing with complexities. In this section, we discuss how we collected and analyzed our data.

Research Context and Case Selection
To investigate the role of the MSP's scope, we chose to focus on global health partnerships for at least three main reasons. First, access to appropriate health services is an important pillar of development on which the capacity to participate in education, employment, and economic welfare depends. As the SDGs emphasize, health issues as mentioned under SDG #3 are closely related with other social problems, which makes them inherently complex, yet critical to be addressed. Second, targeting global health deficits requires collaboration between different sectors (Sachs & Schmidt-Traub, 2017), which gives rise to complexities at different levels. Third, owing to the high number of MSPs in global health and diversity in the respective MSP landscape, this focus allows us to observe MSPs with different types of scope.
To identify suitable cases, we built on a 2004 report mapping global health partnerships (Carlson, 2004) that focused on advancing the health of deprived communities in low and middle-income countries. We screened the 75 mentioned MSPs and purposefully sampled MSPs, first, based on the following criteria: (1) early formation in the MDG era, (2) being a MSP with multi-stakeholder boards and operations, and (3) differences in MSP scope. Specifically, in screening the different MSPs, we recognized that some MSPs focused on a specific health intervention whereas others sought to address a specific health problem or set of health problems based on multiple interventions.
Second, using these three criteria, we chose four MSPs that showed the same formal structure with a multi-stakeholder board and a secretariat to coordinate the MSP activities and that were all headquartered in the same country. Thus, the chosen MSPs displayed similarities in four main criteria and differences regarding their scope. For the sake of anonymity, we refer to these MSPs as Cata-P, Innov-P, Shape-P, and Vital-P. Table 1 introduces the four MSPs with their key characteristics.

Data Collection
We used multiple data sources to understand the MSPs in light of our research question (see Table 2 for an overview). We screened 61 partnership documents, carefully read 12 articles on the MSPs published in medical journals, and studied 22 videos that outlined the partnership goals and structures. These documents were particularly helpful in understanding the MSP stories. To get more specific insights into the role of the MSP scope and complexity, we conducted 38 interviews with 37 people. We spoke to employees at the MSP secretariat in particular, given that they formed the coordinating hotspot where different internal and external complexities intersected. Moreover, we analyzed anonymous employee reviews of the MSPs' working environment from the Glassdoor website, which included additional critical perspectives.
We collected data between May 2016 and December 2019. The interviews were our richest sources of data. We spoke to 11 people from Cata-P, 7 from Innov-P, 9 from Shape-P, and 10 from Vital-P. All the interviews were recorded and transcribed. The average length of an interview was approximately one hour. We used a snowball approach, and after each interview, we asked the interviewees whether they would add anything or suggest other people to interview. For each MSP, we sought to interview people from the human resources division, external relations, and those coordinating the operational implementation in different countries. The diversity of the resulting perspectives, together with the insights we got from archival documents, allowed us to triangulate the data. Moreover, four of the interviewees had worked for more than one of the MSPs studied here and could provide us with some comparative accounts between the MSPs.
We started the interviews by inquiring about the kind of complexities that the interviewees experienced and how they tackled them. These questions gave us a broader picture of the MSPs. We then asked what especially helped or challenged them in their ability to cope with the manifold complexities.

Data Analysis
We used an iterative approach to the data analysis (Corley & Gioia, 2004). First, we drew on the archival (i.e., presentations, annual reports, and research articles) and interview data to write thick case descriptions of each MSPs in line with four questions: What do they do? How do they do it? Which complexities are involved? And why do these complexities come to the fore? After developing vignettes for each MSP and building on our insight that two of the MSPs focused on a specific health intervention while two others concentrated on health problems more broadly, we sought to understand in greater depth the role of the MSP scope and its implications for coping with the multiple complexities the MSPs faced. To this end, we grouped our data into codes around similar messages (Corley & Gioia, 2004). Using Atlas.ti software, we initially used open coding (Locke, 2001;Strauss & Corbin, 1990) to grasp the complexities involved and how they were managed. On this basis, these initial results pinpointed the complexities of addressing health issues, specific complexities that the transition from the MDGs to the SDGs brought about, and internal complexities arising within the MSPs. The open coding process yielded 137 pages of data incidents. In this process, we tried to remain as close to the informants' words as possible, which also gave us many codes. We created first-order codes to organize a vast amount of data (Van Maanen, 1979).
Next, we used second-order themes (Gioia et al., 2013) to develop relations between first-order concepts within each case and merged them into aggregate dimensions. We further compared first-order concepts across our cases, which helped identify similar sources of complexity and revealed differences in how the scope influenced the way they were handled. We went back and forth between empirical themes and relevant literature as we moved from the first-order concepts to the second-order themes (Eisenhardt, 1989;Gioia et al., 2013). Subsequently, we built aggregate dimensions by comparing the cases and examining the relationships between our second-order themes in iteration with the literature on MSPs (Dentoni et al., 2018;Van Tulder & Keen, 2018), SDGs (Voegtlin & Scherer, 2017), and (alliance) scope (Oxley & Sampson, 2004). The aggregate dimensions formed around the complexities, the MSP scope, the scope implications, and the different MSP outcomes. Our final data structure captured the cross-case comparison and comprised 19 first-order concepts, nine second-order themes, and four aggregate dimensions (see Fig. 1).   (14), Videos (7), Glassdoor website Webpage, Annual Reports (11), Information Leaflet (4),Videos (6), Glassdoor website

Findings
We illustrate our findings by first documenting the complexities that the MSPs faced and then elaborating on how the partnership scope influenced the way the MSPs coped with these complexities. Our findings show how a functionoriented scope helped filter complexities and, consequently, allowed two of our sample MSPs to stay focused on and attentive to the external environment. Conversely, the issueoriented scope that the other two MSPs had adopted tended to magnify complexities, which led these MSPs to lose focus and eventually become self-absorbed (see Fig. 2 for an overview of the emerging framework).

Sources of Complexity
Owing to the nature of their work, the four MSPs had to work with external and internal complexities on a daily basis. Our evidence emphasizes four main sources of external complexity: (i) the complex and (ii) the global nature of the targeted health problems, (iii) increasing donor expectations, and (iv) the shift from the MDGs to the SDGs.

The Complex (i.e., Interrelated and Dynamic) Nature of the Targeted Health Problems
The health problems that the MSPs addressed-initially under the umbrella of the MDGs and subsequently in line with the SDGs-were known for being especially complicated (Travis et al., 2004). Our interviewees described them as "very complex" (Q10) and interrelated with other social issues like "poverty, [where] you have developments [so that] you simply just can't have good public health outcomes" (X5). Shape-P's website explained the interrelatedness based on the design of the SDGs' multi-colored logo: "a circle showing how problems and solutions are interconnected." Similarly, Vital-P highlighted the complexity of their SDG-related health problems by pointing to "the magnitude and multi-dimensionality of the task" they faced (Annual Report, 2015-2016. Moreover, the targeted health problems were dynamic and evolving. Our interviewees highlighted that "[the problem] keeps changing; but also, the questions keep changing. When we speak about it in our team, the issue is different every year" (W8). This made it "very difficult to plan ahead because you don't know what outbreak [or] advancements there may or may not be" (X7). Because of their dynamic nature, the health problems involved uncertainty and, thus, were "not well understood" (Z4), full of ambiguities, and "moving every day" (Q9).

The Problems' Global Nature Requiring Work Across Diverse Institutional Contexts
Common to the problems underlying the SDGs is their global and, hence, country-spanning nature. Tackling these problems requires working across different institutional and often developing country-based contexts. The four MSPs all operated in multiple countries-from 20 to 100. While only Vital-P had local offices in these countries, the employees of the other three MSP secretariats frequently traveled to the countries concerned when local coordination support or monitoring was needed. Operating globally, however, made the working conditions more unpredictable and, for "the interventions [to be] tailored to each country context," required both a holistic and a local understanding of the focal health problems (Shape-P, Annual Report, 2018). Our informants underscored how the problems' global nature became challenging in their everyday work. For example, "working with so many countries means there are many different aspects to [the MSP]" (X2). Furthermore, developing a local understanding was required 20 to 100 times, given that "every country [was] different, and you have to understand the landscape of each country (…) and work with a system that you have little control over," an interviewee explained (Z4).
In addition, the SDGs pushed to reach out especially to the deprived parts of the world. One Cata-P strategy document stated: "In low-income fragile states, […] the progress of the past decade has left too many behind." Working with unstable countries or regions also presented unique challenges, as another interviewee (Q9) highlighted: "There were so many times when I was supposed to have a call with them, and they didn't show up. Then I got a message saying, 'Sorry, there was a bomb in my neighborhood.' How do you react to that?".

Increasing Donor Pressures and Diverse Expectations
Another source of complexity involved dealing with MSP donor demands, including public and private donor organizations. One major challenge was the "shifting priorities of funding organizations" (Innov-P website). In addition, the priorities of different donors often differed and were difficult to match. For example, interviewee X5 stated that the MSP's job was "to [make] magic happen in very difficult and challenging environments: Comforting the donors that nothing [worrisome] is happening, while at the same time sending the resources to the countries." The donors asked for greater accountability, detailed reporting, and proof of where their money was being spent. "Pressure from taxpayers and donors often takes the form of wanting full visibility of expenditures," an interviewee (Z2) noted. Another interviewee (Q7) highlighted: "The donors are now like investors. Before, investors had a five-year horizon; now they're going to a two-year horizon. It's the same thing with the donors." However, ensuring that donors remained committed was an important criterion for the MSPs' sustainability. "They are the biggest influencers… they can bulldoze everybody else" and pressure the MSP in line with their donation interests, one interviewee acknowledged (W5).

The Shift from the MDGs to the SDGs
Having been established around the turn of the millennium, the MSPs in our study explicitly responded to the MDGs. The turn toward the SDGs then called for aligning approaches across the 17 goals (Hone et al., 2018) and paying attention to their interrelatedness. For the sample MSPs, this meant, for example, "a greater focus on gender equality and women's empowerment" (Shape-P, Board minutes, 2017) and "[making their programs] gender aware and, move towards gender-sensitive or -transformative programming" (Vital-P website). However, such a move required the development of a new set of expertise and capabilities and, as such, brought additional operational challenges with it.
Likewise, the SDGs "shifted the whole mentality" (W6) in the partnering space, including questions of where the global development aid was allocated. "How much money will go to our cause… the SDGs will define [that] to a great extent," commented an interviewee (Q7). This shift also had a threatening component: "I think as the MDGs shifted toward the SDGs, people realized that we are not necessarily going to be here forever" (X5). Speedy actions to reach ambitious objectives were called for to justify the MSPs' existence as this excerpt from Vital-P's annual report 2013-2014 illustrated: "The SDGs challenge us not only to reduce [this problem] but to end it by 2030… Our knowledge of what works has never been greater, but we need to implement it faster." In addition to coping with these external conditions, the four MSPs also faced internal sources of complexity, including the complex MSP (i) governance and (ii) operations, as well as (iii) the management of employees coming from different sectors.

Complex MSP Governance
The MSPs' defining characteristic was their multi-sectoral boards, which comprised representatives of the (e.g., pharmaceutical) industry, governments, international organizations, academics, and affected communities, with all of them "hav[ing] a voice" (X9). This diversity posed a major challenge, as expressed by interviewee W1: "In my opinion, the biggest challenge that [Shape-P] faces, especially in terms of governance, is probably the different interests because consensus drives the decision-making process. You have three, four cooks in the kitchen, and they all have their own agenda on how to do things." Important MSP decisions required approval from these diverse actors, but reaching consensus was challenging. An interviewee explained: "It's very difficult to reach decisions because some of the interests compete with one another. It can take a long time because you have to build consensus and do some horse-trading (…).

Complex MSP Operations
To implement the health programs, the MSPs relied on partner organizations from different sectors. This was often challenging in terms of coordination and alignment, specifically as the number of MSP partners ranged between approximately 150 and 400. For example, our interviewees underlined: "Partners have their own set of priorities. (…) Aligning these priorities is necessary to make the whole system work" (W2). To this end, the MSP partners and employees needed to design operational programs that "contribute some good to the [partner] without locking a country into an impossible situation with that [partner]. You don't want to introduce the wolf into the den" (W7).
Likewise, all of the partner organizations had to follow the MSPs' guidelines and work toward fulfilling the MSP goals. One interviewee highlighted: "[The big pharma companies] can be quite condescending. They come and pat you on the head like a small child and say, 'Thank you for your view; now go, and we'll just do what we want'" (Z6). Risks had to be managed, such as not being seen as "[the companies'] marketing department" (Q5) or "business development department" (X7). A similar challenge applied to NGO partners, whose doubts and different preferences had to be managed. An interviewee (Q6) explained: "Some [nonprofit] organizations are hostile to the idea of working with the private sector." For example, they questioned the companies' sincerity when becoming involved in social matters. As a result, coordinating and aligning among partners while working toward a common goal was a recurring challenge.

MSP Employees' Diverse Sectoral Backgrounds
In line with their multi-stakeholder nature, the MSPs hired people with different sector expertise to join their secretariats so that they could 'speak the language' of the different partner organizations, relating to the sectoral as well as the country language. This led to the MSPs secretariats uniting public health workers, former business employees, doctors, and employees from NGOs with networks all around the world. Hence, the secretariats were "quite diverse" (Z1). This meant that the employees brought different skill sets, knowledge, and priorities-a "different approach" (W7)and would be "doing different things" (X7). Such diversity was a source of creativity but could also lead to conflictsfor example, between "people in global programs and people delivering projects in the country" (X8).
An interviewee from Shape-P (W4) mentioned: "You have public health people with vast amounts of technical knowledge, in-country knowledge, and market knowledge, who may see the world one way. Then you have private sector people who see it in a different way. By tossing all these people together in a building, they're constantly fighting to be heard." This diversity was even greater for Vital-P because it had country offices: "People who are in the country feel they are part of the country team but don't really feel part of [Vital-P] overall" (Q3). So, it was difficult to manage those diverse employees who, in turn, worked in different contexts.
Generally speaking, the external and internal sources of complexity mentioned above challenged the four MSPs on an almost daily basis and made their work more difficult (see Table 3 for data excerpts).

The Role of the MSP's Scope
Turning to the role of the MSP's scope in coping with these different sources of complexity, we first depict the two MSPs that had a function-oriented scope. In line with such a scope, the MSP partners focused on addressing an intervention-related gap-a lack of vaccination-based prevention for Shape-P and a lack of drugs for Innov-P. Our findings suggest that a function-oriented scope helped these MSPs filter complexities and, thus, helped them function effectively and stay attentive to the external environment. Conversely, Cata-P and Vital-P focused on addressing health issues more broadly; as we will show, such an issue-oriented scope tended to magnify the complexities and led the MSPs to become overly internally focused.

Function-Oriented Scope Helping Filter Complexities
Innov-P and Shape-P followed a clear demarcation of what they set out to do as a partnership (e.g., being "focused on a singular [intervention] issue" (Z2) and "hav[ing] a straightforward mandate" (W8)). In its MSP documents, Shape-P repeatedly highlighted its focus on vaccinations: "to protect people's health by increasing access to immunization in poor countries" (Progress Report, 2015) and "protecting people's health by increasing the equitable use of vaccines in lower-income countries" (Progress Report, 2016). Similarly, in its 2016 annual report, Innov-P underscored its focus on developing drugs for a focal disease: "develop next-generation medicines for tomorrow that will contribute to the eradication of the disease." These excerpts show how a function-oriented scope clarified the boundaries of the collective endeavor by limiting it to a specific intervention. We identified three ways how the function-oriented scope influenced the way in which external and internal complexities were addressed.
(1) Allowing for clear messages about the MSP. First, the function-oriented scope made it possible to design clear In this way, the intervention-based internal interdependencies helped align employees and partners, giving them the impression that they were pulling in the same direction. Likewise, the simplicity and clarity of structures allowed the MSP actors to sharpen their view on external developments, such as capturing the complexities inherent in the targeted health problem, adjusting to local challenges, and responding to changes such as the shift to the SDGs.
(3) Focusing employee attention. Lastly, a function-oriented MSP scope helped identify MSP priorities and acknowledge boundaries in line with the focal intervention. Thus, the scope worked as a frame protecting the MSPs from stretching their boundaries beyond the available capacities. An interviewee from Innov-P explained: "We have to challenge everybody to assess whether this is something within the scope of [Innov-P] or something outside the scope… Then we either have to get in touch with our partners or rethink or re-approach it to ensure that we stay within the scope" (Z4). In addition, the function-oriented scope provided a lens to make sense of the complex health landscape, including which dynamics, related health problems, and partners were central: "There are things besides vaccination that might impact immunization projects. But then, it is always the same outcome we are looking for: getting more kids vaccinated" (W1).
Based on these three implications, we find that a function-oriented scope helped filter the complexities the MSPs needed to work with. Specifically, it helped them stay focused and not stretch their boundaries too far, which would mean multiplying the complexities. With the scope acting as a complexity filter, the MSP actors could remain outwardoriented and effectively implement their collaborative work. This helped maintain the collaborative energy without feeling overwhelmed. Our interviewees highlighted that they were happy in their working environment; for example, an online employee interview from Innov-P put it this way: "I am honored to be part of this organization." Moreover, the function-oriented scope helped attract and sustain donations. For example, the donations for Innov-P increased 8% from 2016 to 2019, and Shape-P saw a steady growth of 10% during the same period.

Issue-Oriented Scope Magnifying MSP Complexities
Compared with Shape-P and Innov-P, Cata-P and Vital-P focused more broadly on addressing a set of health problems, thereby using multiple interventions. For example, Cata-P's interventions ranged from the distribution of prevention material (e.g., bed nets, tents, and information material), diagnosis equipment and tests, and treatment via the distribution of drugs. Vital-P's interventions ranged from influencing markets to developing and scaling new products for various target groups. Next, we discuss three ways how such issue-oriented scope affected the coping with external and internal complexities.
( We're a very complex organization to work with" (X5), "I think the [Cata-P] model is complicated" (X11), and "getting too heavy" (Q6) and "so much change-in structure and in teams" (Q4). The more complex internal structures made it more difficult for the diverse partners to find the right contact person. It further directed governance and operational attention to the development and management of internal structures while obscuring a broader view on the targeted health issues. The pursuit of several interventions simultaneously focused on one or several health problems caused structural duplication and risked internally dividing the MSP. Our informants complained about having "silos" (X8) and "not being aware of what happens in other departments" (X2). They even mentioned competition arising between different parallel departments.
In the case of Vital-P, the MSP structures were constantly adapted to include new activity areas that eventually led to a new structural design. One of our interviewees put it as follows: "I think the biggest change […] that made it difficult was the implementation of [the new] structure. [Vital-P] has been struggling to work through [it]. I think it's good, but how it works and where the accountability lies is still a bit blurry. You have the global programs that would be like silos" (Q4). Moreover, the MSP worked with additional spinoff structures, including six related structures that specialized in a specific health issue or a specific region or promoted the engagement of the business sector. However, the complexity in structure caused confusion. An interviewee explained: "We have to wear two hats… as [Vital-P]

is funding and implementing [the spin-off] projects, and […] their staff is sitting in our offices"
(Q2). This situation used the resources of Vital-P in terms of staff time and attention, which could not be invested in working with the external and internal complexities that the ordinary MSP activities implied.
(3) Diverting employee attention. Third, Cata-P's and Vital-P's issue-oriented scope did not provide clear boundaries of what the MSP was about and, hence, where its priorities should be. "It's a bit schizophrenic in terms of how we shift from one thing to the other," one interviewee said (X5). The search for a unifying sense and missing reference points for priorities led the MSP staff and partners to focus their attention on how the MSP functioned rather than improving how they coped with the MSP's complex environment. In addition, given the lack of clear boundaries, many of the health problem's dimensions appeared relevant for them to address, which, in turn, required additional funding, support, and expert capacities. An interviewee from Vital-P echoed: "I think we need to narrow [but we] keep stepping out a little bit because there is an opportunity" (Q6). The constant search for unity in the MSP's diverse work and underlying confusion made it difficult to fully focus on harnessing the partner and employee diversity and responding to the health problems' complexities.
Based on these three implications, Vital-P and Cata-P's issue-oriented scope magnified the MSP complexities and caused internal inefficiencies and unwieldiness. Thus, instead of keeping the focus on managing the manifold sources of complexity, the MSPs kept changing their structures, mission, and strategies to better match the complexities. An interviewee (Q2) highlighted: "I suppose we [the MSPs] can be very internally focused, and that is where we've been, very internal, we should look outward. We should be asking what the world thinks, not what we think about how we are doing our job." An interviewee from Cata-P echoed: "We're so focused on these kinds of internal processes that sometimes we lose sight of what is important, what is urgent, and what isn't… We get obsessed with what I call bellyaching" (X5). A strategy review of Cata-P in 2017 underlined the need for prioritizing activities in view of the "many activities [scheduled for] the short term [and] given the secretariat's limited capacity and ongoing work-load… there is a concern that the new strategy implementation plan is adding a layer of complexity to this multifaceted environment." Several of our interviewees described this situation as "creating anxiety" (X3), "changing a lot" (Q4), "reinventing selves" (Q2), and "having burnouts" (X4). They noted that they were not happy about the situation as it prevented them from working properly and led to constant change. Furthermore, the MSPs faced challenges in securing continued donor support. For example, the donations for Cata-P decreased by 25% from 2016 to 2019, and those for Vital-P decreased by 21% in the same period (see Table 4 for data excerpts).

Discussion
In this research, we sought to examine how the partnership scope influences the way in which MSPs cope with external and internal complexities. Our case MSPs all addressed global health challenges in deprived areas, especially in the Global South, and harnessed resources and expertise from different sectors (Kolk, 2014). Our research provides a detailed picture of the complexities these MSPs had to navigate, including multiple external (i.e., the health problems' complex and global nature, increasing donor pressures, and the shift from the MDGs to SDGs) and internal complexities (i.e., complex MSP governance and operations and the MSP employees' diverse sectoral backgrounds).
Building on our comparative study, we found the MSP scope to center either on a specific health intervention, which could be used for one or several health issues, or on a specific or a set of specific health issue(s) on the basis of diverse interventions. Based on this distinction, our findings suggest that the scope influenced coping with complexity in three ways: allowing for clear messages vs. risking mixed messages for diverse stakeholder groups, providing clarity for vs. obscuring MSP structures, and helping focus attention vs. diverting attention. We integrate these insights into a framework describing the role of the MSP scope in managing complexity (see Fig. 2) and, in the following, develop the underlying propositions. We then discuss the broader theoretical and practical implications of our research for the management of MSPs, as well as for implementing the United Nations SDGs more generally.

Sources of Complexity
The strategic alliance literature uses the concept of alliance scope as a governance mechanism to manage the risk of opportunistic partner behavior (Oxley, 1997;Oxley & Sampson, 2004;Rufin & Rivera-Santos, 2012). However, the MSP setting offers an opportunity to consider the MSP scope in relation to managing multiple complexities. In this respect, our study suggests that it influences the management of complexities in three ways.
First, the scope of a MSP tends to shape its relationships with diverse stakeholder groups. The MSPs in our study represented a new way of organizing (Gray & Purdy, 2018), thus, it was challenging for internal and (in particular) external stakeholders to understand what the MSP was about. This risked creating confusion, such as for donors, implementation partners, the local community, and beneficiaries-all actors on whom the MSPs depended for resources and support for operations. Prior studies have shown, for example, that when an organization defines itself as a "bank" We have a lot of very good people who I think are motivated and experts in their fields. Our failing is that we are not harnessing it in an institutional approach.

Remaining outward-looking
Becoming self-absorbed (the most common example), we know more or less what it is, and the concept is easy to apply for external and internal audiences (Dhalla & Oliver, 2013). Conversely, because of the lack of understanding of what it means to be an MSP, either nothing or many different things resonated with multiple audiences.
In this context, a function-oriented MSP scope made it possible to focus on the focal health intervention and to craft consistent and simple messages around it. As a result, it provided a tool for the multiple stakeholders to understand what the MSP was about, what it did and how, and how they could support it. What resonated was, for example, Shape-P's focus on vaccination and Innov-P's focus on R&D for a specific disease. In turn, an issue-oriented scope risked producing mixed messages around the MSP and confusing stakeholders even more. This added complexity in terms of constantly needing to clarify and justify what the MSP was about, making it difficult to align the diverse partners and employees, and attracting and sustaining stakeholder (e.g., donor) support. Thus, we suggest that:

Proposition 1 A function-oriented MSP scope allows for clear messages around the MSP to sustain support; an issueoriented scope risks producing mixed messages and confusion for the multiple stakeholders.
Second, the MSP scope tends to influence the required number of formal structures and, consequently, the tasks, plans, and procedures for each area of activity (Albers et al., 2016). In view of the multiple complexities that the studied MSPs faced by operating on social issues and across sectoral and geographical boundaries, they already had to stretch their limited resources and capacities as much as they could. In this respect, a function-oriented scope made it easier to develop concise MSP structures that where interlinked based on clear interdependencies. This internal clarity freed up the ability to handle the complexities related to the MSPs' social, cross-sector, and country-spanning activities. Internal clarity thereby prevented further uncertainty (Adobor & McMullen, 2002) and promoted the MSP's agility.
Conversely, an issue-oriented MSP scope tended to multiply internal structures. For example, different interventions required different activities and structural backing from an operational perspective and, by working in parallel, often also prompted the duplication of support structures. Without building on clear internal interdependencies, the structures became complex and drew attention, as expressed in our evidence on constant structural change. As a result, decisions took time, inefficiencies emerged, and time, energy, and attention were drawn away from managing the manifold social, cross-sector, and country-related complexities. On this basis, we suggest that: Third, the MSP scope has direct implications on attention (Oxley & Wada, 2009). Although complexity was both necessary and unavoidable for our sample MSPs, individuals' attention is finite (Oxley & Sampson, 2004), and they only have a limited capacity to work with complexity (Mintzberg, 1979;Simon, 1997). In this regard, our interviewees underlined that a function-oriented MSP scope provided them with a focal lens to simplify and comprehend the dynamics, actors, and features of the targeted health problem. The MSP scope worked as a natural filter to observe and identify the necessary sources of complexity (Greenwood et al., 2011). For instance, Shape-P's focus on a specific health intervention and Innov-P's focus on research helped them set boundaries and priorities and, this way, helped simplify the complexity towards a level they could operationally work with. A function-oriented scope further helped the MSP actors speak the same language (shaped around vaccines or research) and align different stakeholder interests through focused intervention.
By contrast, an issue-oriented MSP scope seems to divert attention as multiple sources of complexity appear to be relevant. Specifically, targeting a (set of) health problems with multiple interventions made it difficult to set clear boundaries and identify priorities, and it risked increasing the number of complexities around each intervention. Confusion about priorities and the MSP's core distracted the MSP actors-in particular, its employees-and curbed their capacity to cope with the complexities the MSP generated. These insights suggest that, somehow similar to a concept used in physics, an issue-oriented scope functions as a concave lens that diverts attention, while a function-oriented scope works as a convex lens that trains attention on the primary focus.

Proposition 3 A function-oriented scope allows the MSP stakeholders, and employees in particular, to focus attention, whereas an issue-oriented scope risks diverting attention.
Overall, based on these three facets, the MSP scope seems to have implications at different levels. First, by making it possible to send clear (or generating mixed) messages, it influenced perceptions, in particular at the macro level, in that it helped sharpen or blurred the MSP's position, needs, and expertise in the global health landscape. Second, by shaping the MSP structures, it influenced the MSP management at the meso level (i.e., MSP secretariat and partner interaction level). Third, it functioned at the micro (e.g., MSP employee) level by helping focus or diverting their attention. On this specification our study insights emphasize the role of scope rather than size per se: Among the MSPs struggling with complexity were the comparatively biggest MSP (Cata-P) and a relatively small MSP (Vital-P), whereas the second largest MSP (Shape-P) provided evidence of successfully coping with complexity.
Consequently, our research suggests that a functionoriented scope helps MSP actors filter overwhelming complexities and, in this way, stay focused, function efficiently, and remain attentive to the MSP environment. On this basis, our data insights for Innov-P and Shape-P depict a working environment in which MSP employees were happy and able to sustain the collaborative energy and donor support. Conversely, an issue-oriented scope emerges in our study as one that magnifies complexities. As a result, the respective MSPs became internally focused and continuously tried to adapt to new complexities with additional structures and internal change, and yet, they struggled to navigate these complexities. By blurring priorities and what the MSP stood for, the collaborative energy and the donor support risked facing a decline. Thus, we propose that: Proposition 4 A function-oriented scope helps manage complexities by working as a filter, while an issue-oriented scope challenges coping with complexities by working as a magnifier.
We now turn to the theoretical, practical, and ethical implications of our research.

Implications for the Management of Multi-stakeholder Partnerships
The SDGs guide the agenda of social, economic, and environmental development activities. In this role, the SDGs are ambitious (Van Tulder & Keen, 2018) cannot be addressed by single actors (Van Tulder, 2018;Voegtlin & Scherer, 2017), and, therefore, openly call for partnerships (see SDG #17). This makes MSPs influential and important actors and, consequently, increases the need for more research to understand how MSPs can function more effectively and efficiently. As research on MSPs is still nascent, we investigate the role of the MSP's scope in dealing with external and internal complexities. To do so, we focus on MSPs grouped around SDG #3 that sought to address complex and interrelated health problems at a global scale.
Prior literature highlights that managing MSPs is a difficult endeavor since they need to cope with partner diversity (Gray & Purdy, 2018) and challenges associated with collaborative governance (Quélin et al., 2017). However, such internal complexity, in particular through the resulting creativity and resource pools, seems necessary to address the complexities underlying the social issues that the MSPs target (Schneider et al., 2017)-specifically, their dynamic, multifaceted, and interwoven nature (Dentoni et al., 2018;Van Tulder & Keen, 2018;Waddock et al., 2015).
Illustrating these internal and external complexities, our study first adds the acknowledgement of diverse employees working in the MSP secretariats to the internal challenges of partner diversity (Ashraf et al., 2017) and governance complexities (Provan & Kenis, 2008;Quélin et al., 2017). This seems to be an emerging feature as operational MSPs become more institutionalized with a separate secretariat. While the diversity of MSP employees forms a means to better cope with partner diversity (Saz-Carranza & Ospina, 2011), it also adds another layer of complexity. As the SDGs call for globally integrated approaches that may prompt larger MSPs, we posit that it is important to understand how MSP secretariats work. We encourage future research to explore the interdependencies between the MSP employee and partner diversity, for example, with a focus on the synergies and additional conflicts that arise in managing these two facets of complexity.
With respect to external sources of complexity, our study expands the focus on the magnitude of issues (i.e., multifacetedness, dynamic and interwoven nature; see Dentoni et al., 2018;Van Tulder & Keen, 2018;Waddock et al., 2015) to also consider specific challenges arising from global, multilayered problems (Gray & Purdy, 2018), as well as their implications for the MSPs' everyday work. Likewise, our study illustrates that when MSPs become more institutionalized, another stakeholder group gains power: the donors (Kolk & Lenfant, 2015), who may have conflicting expectations of the MSP. Together, these insights contribute to developing a more fine-grained understanding of MSP-type specific complexities (e.g., due to their global and operational nature) that underlines the interrelatedness of external and internal complexities. Second, where these internal and external sources of complexity meet, our research introduces the MSP scope as a mechanism that influences how MSPs handle complexity. Overall, the complexities tended to push the MSPs to extend their boundaries and created constant pressure on them. Therefore, we add to previous research on MSP complexity management, which has primarily addressed either relational (Le Ber & Branzei, 2010a, 2010bSloan & Oliver, 2013) or governance dynamics (Bryson et al., 2006;, by examining how the MSP scope functions in this context. We posit that a function-oriented (rather than an issue-oriented) scope works as a guardrail (Smith & Besharov, 2019) and helps sustain the successful management of MSPs. We suggest that studying the MSP scope is critical because it influences a partnership's relations with its stakeholders (e.g., donors and partners) and forms a mechanism shaping the MSP structures. On this basis, our study shows the interlinkages between the MSP's scope and previous literature on the structural dimensions of partnerships (Bryson et al., 2015;Provan & Kenis, 2008).
Third, at this intersection, our research reveals the limitations of the common assumption that external complexity should be matched with internal complexity (Schneider et al., 2017). In the MSP setting, scholars suggest that more comprehensive partnership designs are required to address the inherent complexities of and interlinkages between longexisting grand challenges (Dentoni et al., 2018;Van Tulder & Keen, 2018). However, our research brings to the surface the argument that organizations and individuals have a limited capacity to work with complexity (Mintzberg, 1979;Simon, 1997), as well as their ability to work with complex structures and grasp what the corresponding organization (here, the MSP) is about.
On the surface, MSPs may appear to be a promising means to address multiple external complexities with internal complexity, but when we zoom in, significant limitations emerge. After all, complexity also needs to be reduced to allow the different stakeholders to connect and align their actions (Mintzberg, 1979). We argue there is a delicate balance for MSPs to be comprehensive while working efficiently and effectively in the short and long term. However, it is important to note that our insights are derived from health MSPs and, thus, require replication in other social issue settings, such as education or poverty.
Besides the theoretical implications, we posit that these insights are also valuable for partnership practitioners. Specifically, our research suggests that while a function-oriented MSP scope facilitates the management of MSP complexity, it may fail to live up to the call for more integrated and encompassing approaches that surround the SDGs (UNDP & World Bank, 2016). As such, our study suggests limiting or 'right-sizing' the expectations of what single MSPs can achieve, while simultaneously raising the question of how issue and intervention interlinkages can be addressed and managed. On this basis, the traditional diversification literature (Ansoff, 1957) might provide a useful lens to study how coordination might be achieved when MSPs are growing multi-directionally. Some interesting future research questions arise: When do MSPs enter new issue or intervention fields? Is unrelated diversification a reliable strategy for MSPs? And when does it become a viable option for MSPs to open up to new countries?
Overall, our study resonates with practitioners' calls for 'multi-stakeholder platforms,' which coordinate and provide leverage between different MSPs (The Partnering Initiative & World Vision, 2016). Questions for future research in this respect arise regarding how such platforms secure finance and support. And while our study suggests that (operational) MSPs remain critical to achieving the SDGs, more support infrastructure is necessary to overcome their capacity limitations. For managers involved in individual MSPs, our study recommends regularly re-evaluating the MSP scope.
Finally, our research also contributes to a better understanding of how the shift from the MDGs to the SDGs influenced existing MSPs. Our research suggests that the shift toward the SDGs brought new development issues to the foreground while removing the emphasis on others. For MSPs explicitly tailored to the MDGs, this entailed change in global visibility and, as such, donor and other stakeholder support. In addition, with the SDGs calling for comprehensive, issue-spanning approaches, MDG-based MSPs likely need to develop new capabilities (Dentoni & Veldhuizen, 2012;MacDonald et al., 2019). Considering the limited resources available in the MSP field and MSPs working at full capacity already, developing new capabilities poses challenges. Overall, more research is needed to extend our findings and investigate the MDG-SDGs transition in the MSP field, including potential shifts in power, players, and resource endowments.
On this basis, our research has important ethical implications for managing MSPs. First, one of the key insights from our study relates to the MSP employees' well-being. Our study shows that extending the boundaries of existing MSPs risks overwhelming their employees and harming their well-being. Thus, it is the ethical responsibility of leaders on the MSP board and executive team to consider their employees' well-being (Chughtai et al., 2015;Mo & Shi, 2017). Our research suggests that having a clearly defined (e.g., function-based) MSP scope helps improve employee well-being. Second, when MSPs become too complex, control over implementation (e.g., to avoid fraud and the misuse of partnership resources) becomes a challenge. To enhance accountability and MSP efficiency (Bryson et al., 2015), our study presents and helps better understand the role of the partnership scope as a mechanism that has previously been neglected.

Conclusion
MSPs can be useful in addressing humanity's grand challenges by mobilizing different actors in different countries and by bridging both funding and expertise across multiple sectors. Although very promising, MSPs involve critical complexities and limitations. Understanding their nature and the associated challenges to better leverage their capacity can help achieve the United Nations SDGs more productively. Our study contributes to this promise by elucidating the role of the MSP scope and by suggesting new research avenues to support MSPs on their SDG journey.