Rethinking relationships: clarity, contingency, and capabilities

Abstract Governments around the world have developed complex portfolios of relationships with other parties in order to carry out the various activities needed to govern. Reliance on the efforts of other parties has long been a feature of government activity, but in recent times governments worldwide have adopted contracting models to deliver outcomes to citizens and communities. This has especially been the case in their relationships with for-profit and nonprofit providers of public services. Despite the widespread use of contracting by government, how governments think about contracting is very narrow. Indeed, most government contracting models rely very heavily on a specific tool – outsourcing. This means that the practice of contracting is often very limited and reliant on a narrow set of tools and relationships. Such a narrow mindset shapes and frames relationships in such ways that governments will often fail to fulfill their aspirations in working with other parties. Adopting broader notions of relationships is important in moving beyond a fixation on tools. Such a shift in mindset, however, demands new ways of thinking about the various parties, relationships, and modes of working together. In turn, this would require very different sets of capabilities to effectively work together. In this article, I set out a much broader notion of relationships and explain what this means for the capabilities that governments need to manage much more complex portfolios. These capabilities extend beyond the skills of public servants, encompassing organizational and environmental factors. The end result being the need for governments to move toward broader, more strategic notions of relationships in order to deliver on the needs and aspirations of citizens and communities.


Introduction
Governments engage in a broad range of relationships with many different actors. In the delivery of public services, the scale and scope of both activity and actors has increased over time as governments have moved away from direct delivery to engaging others. There are many ways in which government's work with other parties; some common ways that this is done is through public-private partnerships, through joining up with other levels of government, or through outsourcing of service delivery. In doing so governments now seek to achieve outcomes for citizens and communities through complex systems of organizations, rules, and relationships. So much so that, in practice, very rarely "does a public sector agency have control over the whole process of producing public services for which it is responsible" (Alford and O'Flynn 2012, 3). Engaging other parties, however, is not without controversy. Not only do ideological differences and preferences shape how we view tools such as outsourcing, client coproduction or public private partnerships, but also we have seen very mixed results in the adoption of these tools around the world. In recent times spectacular failures such as the collapse of Carillion in the United Kingdom, a major private sector provider of services for and to government, has bought the complexity of public services back front-and-centre to public debate. 1 A key part of this debate relates to how effective government organizations are, and whether they have the requisite capabilities, to manage this increasing complexity.
It is widely agreed that the practice of governing has become increasingly messy, even disarticulated, with blurred boundaries, and increasingly interconnected and interwoven relationships (Frederickson 1999;Kettl 2016). Our thinking about these relationships in practice, however, has remained steadfastly narrow. In many cases, this means governments have been fixated on designing and implementing procurement and/or outsourcing processes. But the actual stewardship of these complex systems requires more than just managing procurement processes and ensuring contractual compliance. How these systems are designed, nurtured and managed is central to the achievement of government and societal outcomes.
The practice of building relationships between government organizations and external parties, however, does not appear to have developed past quite narrow, transactional approaches. Many governments seem to be locked into mindsets that draw down heavily on particular views of human behavior and push us towards designing specific types of contracts and governance structures. In my view, this is limiting the ability of government to delivery on the needs and aspirations of citizens and communities. The obsession with a particular tooloutsourcingrather than a much broader toolkit of relational forms means that we are not seeing the sort of strategic stewarding of complex systems that is needed in practice.
To help us in shifting this mindset it is important to think about why we have become so narrow in the practice of relationships, and also to lay out a much broader framework to guide decisions and action. This helps us to think more deeply about the capabilities that might be needed in practice. In exploring these issues, I make some opening comments about the scale of reliance on other parties and the nature of the relationships underpinning this activity. From here, I will set out the importance of clarity of terms and develop a much broader notion of relationships. This includes expanding the set of external parties that we tend to consider and the various ways in which we might join together. From here, I will discuss the demands that this places on governments to develop the broader capabilities needed to manage diversified portfolios of relationships and steward more complex systems. Capability extends far beyond developing the individual competencies of public servants, towards thinking about organizational capabilities and designing enabling environments.

Getting a handle on the scale and nature of relationships
Governments rely heavily on the efforts of other parties in pursuing outcomes for communities and citizens. They do this with a wide variety of external parties via various types of relationships. In talking about this, however, we tend to focus on only a very narrow set of actors and relationships. Once we get a handle on the complexity of government activity through other actors, especially as it relates to the delivery of government services, we see how together these activities create incredibly complex systems. How governments go about designing and stewarding these complex systems matters. And it matters a lot. Venturing into two key areas helps us to start to get a handle on this complexity: (i) the scale of government activity through other parties; and (ii) the nature of these relationships and how this impacts systems and outcomes.

Understanding scale
Getting a clear handle on the scale of government activity through other parties is actually quite difficult. Attempts to do so, however, indicate that the scale of externalization is significant. Studies of the phenomenon of "shadow government" (Klein 2007;Reich 2007;Williams 2013) and of "government by proxy" (Kettl 2010) indicate that there are many more actors working on behalf of government (and funded by it) than many might expect. In the US, for example, by the early 2000s there was six government-byproxy employees carrying out the work of government under contract (Dilulio 2003). Investigations into the use of outsourcing by the US military found that outsourcing in the Iraq and Afghanistan wars had become the default option for activities once considered "core function, mission-critical, and organic capabilities" Commission into Wartime Contracting, 2011, 13). In Britain, government expenditure on outsourcing alone was estimated at 11% covering everything from school meals to running prisons (The Economist 2018). The emergence of this shadow government and the expansion of government by proxy is challenging to measure, and it has not made the job of governing in practice easier: "contracting out poses the most fundamental, ongoing governance and public management puzzles" (Kettl 2010, 239).
The decision to undertake activity through and with other parties is not inherently good or bad (although both positive and negative effects are possible). We should care about scale for different reasons; because it has broad ranging effects on factors such as government capacity, provider supply chains, markets and ecosystems. Scale matters also because of the nature of relationships that exist; an issue that will be discussed in more detail later.
We can try to grasp scale in various ways. The two most obvious, although not necessarily straightforward, routes should be through analysis of the extent of government expenditure and the number of contracts. Whilst these are the most obvious, they are not necessarily straightforward due to how governments in different parts of the world collect, classify, code and report data about these various arrangements. This can also be especially challenging within countries, where different jurisdictions adopt different practices. Governments also make different types of data more or less accessible. Looking at particular types of data, however, can give us some baseline notion of scale even if we cannot cover the full terrain. For example, O'Flynn and Sturgess (2019) found that the Australian national government expenditure on goods and services from for-profit and nonprofit providers was A$90-100bn in 2017-2018; a level three times what the national government spent on direct employment (O'Flynn and Sturgess 2019). 2 Whether this makes Australia a high or low compared to other nations matters to some, although it matters perhaps more to think about what outcomes governments are seeking, rather than looking just at spending. Using the most recently available OECD data we found that Australia is near to the average as measured by proportion of GDPspending 13.1% compared to the OECD average of 13.2%. But is higher in terms of proportion of government expenditurespending 35.4% compared to the OECD average of 30.3% (OECD 2017).
Looking at the number and relative size of procurement contracts can also help us to get a handle on scale. In recent analysis, we found that the Australian national government entered into 73,458 new procurement contracts for goods and services in 2017-2018. This figure captures spending on everything from military vehicles, to management consulting, immigration detention centers, healthcare services, travel for public servants, and information technology. 3 The value of which we estimated at around A$71bn (O'Flynn and Sturgess 2019). Of course, this does not cover all government expenditure to other parties, just those classified as procurement of goods and services. Nor can we say whether this number is too high or too low. What we can say is that over the past decade, the average number of procurement contracts entered into was almost 74,000 per annum and that average annual expenditure was A$47.3bn. Over time, the annual value of procurement contracts in Australia is increasing (Department of Finance 2019; O'Flynn and Sturgess 2019).
These issuesexpenditure and numbercome together when we look at the relative size of contracts. The intersection of these will differ across nations, however our analysis in the Australian context showed that most procurement arrangements are quite small: 62% have a value of less than A$80,000; 95% have a value of less than A$1m (Department of Finance 2019; O'Flynn and Sturgess 2019). Perhaps more important was the finding that substantial expenditures are locked into a small number of complex arrangements. Our analysis shows that almost three-quarters of the total value of expenditure was tied to just 0.5% (n ¼ 333) of the total number of contracts (Department of Finance 2019; O'Flynn and Sturgess 2019).
Even though this analysis was not comprehensive it does provide insight into the considerable activity that is undertaken via other parties. Looking only at procurement data, we can see that the amount of expenditure and the number of arrangements entered is extensive. It is also important to realize beyond gauging scale, that this has important implications for what sorts of capabilities government need to ensure that it manages relationships effectively. The complexity of procurements is mixed. For example, some of the very high value arrangements such as offshore immigration detention centers are highly complex, whereas the purchase of travel services can be considered more straightforward. 4 This makes managing portfolios of relationships challenging because of the difficulties in specifying, measuring and monitoring in a range of activities. 5 And this complexity needs a more contingent approach to management, not a one size fits all solution. For example, a large number of similar type of contracts might be effectively managed using routinized approaches, whereas more complex, uncertain and difficult to specify in advance activity, may need quite different relational architecture.

Understanding the nature of relationships
Governments work with other parties for many reasons and, as will be discussed later, these relationships can take many different forms. In work that John Alford and I did on public service delivery, we argued that most decisions about externalization are based on the potential to increase value and/or reduce costs for example, through economies of scale or scope, specialization, flexibility, innovation and learning, or complementarity (Alford and O'Flynn 2012;see also O'Flynn and Sturgess 2019). It is also true that political decisions, shaped by the ideology of different parties and their views on who is best to undertake particular tasks, drive who does what in practice. In other words, decisions to outsource (or its opposite) can often be political. In some nations the taste for outsourcing expands when more conservative governments are elected; the election of President Trump in the United States was accompanied by a push to shrink the state and expand outsourcing. In other nations, various tools of privatization have been bipartisan, for example in Australian where governments of the "left" and "right" have deployed these tools equally (O'Flynn 2018).
As will be discussed later, there is a broad spectrum of relational forms available. However, it is fair to say that in many countries the dominant way of thinking about relationships between government and external parties has been narrowly constructed and anchored in transactional approaches. This is manifest in the widespread use of outsourcing models, and in some nations, the development of entrenched outsourcing mindsets (O'Flynn and Sturgess 2019). This is also reflected in the ways in which the government relies heavily on competition, competitive tendering, heavily specified services, incentive-based contract design, and monitoring schemes that signal low trust.
Whilst our focus was on the Australian Public Service, this mindset is not unique to Australia; indeed, this is a characteristic of many systems (O'Flynn and Sturgess 2019; Sturgess 2017). Whilst outsourcing can be suitable, even optimal, in some situations, developing an outsourcing mindset effectively boxes governments into very narrow ways of framing and managing relationships with other parties. This means that attempts to broaden out ways of working together often fall flat. For example, there is considerable evidence that the rhetoric of more collaborative or joined up approaches is more powerful than actual practice. O'Flynn et al. (2011) andHalligan (2014) found government organizations are often stymied by transactional approaches to their relations with other parties; similar experiences have been identified in a range of studies. Paun and Blatchford (2014), for example, showed this in the case of public service agreements in the UK; and Head and O'Flynn (2015) in relation to Indigenous disadvantage and environmental management in Australia.
On the nature of relationships then, when governments lock into narrow transactional approaches they limit how they can go about achieving outcomes for citizens and communities. Thinking much more broadly about who government works with, how it might structure these relationships in different ways, and what this means for the capabilities that governments need to develop, will be the focus on the remainder of the discussion.

Getting clear on terms: contracting is not outsourcing
It is fairly common to conflate and confuse a variety of terms. One area that has already been flagged in the discussion is that often outsourcing and contracting are used as if they are were the same thing. However, this masks the difference between a specific tool (outsourcing) and a broader relationship toolkit (contracting). To address this confusion O'Flynn and Sturgess (2019; see Appendix A) developed a summary of the vast array of terms and definitions that relate to government engagement with other parties (see Table 1).
It is important to differentiate between various terms and to be clear about the nature of tools and relationships (Alford and O'Flynn 2012). In many contexts various terms are used interchangeably; this is especially the case for terms such as outsourcing and contracting. And, in recent years, commissioning. However, the terms outsourcing, contracting and commissioning reflect very different ideas and practices. One way to differentiate them is as follows: "commissioning is a strategic framework for determining needs and aspirations and then developing ways to achieve these; contracting involves a range of tools that join two or more parties together; and outsourcing is a specific tool that is used when government moves specific in-house services to external providers" (O'Flynn and Sturgess 2019, 14).
We might also think of this as a nested set of concepts. Outsourcing occurs where government moves specific in-house services to external providers. Here, we see the whole of the production of a service, for example, being granted to external parties (Alford and O'Flynn 2012). Contracting encompasses a much broader set of relationships where two (or more) parties are joined together and contracts can exist in many forms from transactional to relational. Thus, a very transactional form of outsourcing might be one form of contracting, but it not the only form. Another way to think about the difference is to say that outsourcing is a tool and contracting is the toolkit (O'Flynn and Sturgess 2019). The contracting toolkit therefore might range from very legalistic through to very norms-based approaches and can be on the spectrum from transactional to relational (Williamson 1979).
There are different ideas underpinning outsourcing and contracting and they have manifested in unique ways in practice. For our purposes, it is important to recognize that outsourcing is a tool that draws down heavily on agency theory which, when interpreted narrowly, can reinforce more transactional ways of engaging with other parties. Contracting is a broader toolkit, not a specific tool, which when underpinned by transaction cost economics opens us a much broader set of tools that can be used to connect parties together. Both have been important in informing and framing practice in public sectors across the world, but they are different. The challenge in practice is that many governments appear to have locked heavily into a narrow view of relationships and rely heavily on outsourcing effectively sidelining a range of other approaches that may be much more effective and efficient. There are several reasons for this tendency that have been identified in the literature: accountability systems, budgetary models, bureaucratic organizing models, silo cultures, path dependence, performance regimes and power being part of the long list used to explain this tendency (see O'Flynn 2014 for a summary). Such mismatches between task and relational form are often at the root of dysfunction in practice (Alford and O'Flynn 2012).
Opening up the toolkit and adopting a much broader notion of contracting, however, requires us to think more broadly about other parties, relational forms, and capabilities. In the next section, I explain in more detail different approaches to understanding relationships and how they influence practice before moving on to rethinking building blocks.

Understanding relationships: building a more comprehensive approach
To better understand how governments might work with other parties we need to develop more comprehensive approaches. There are substantial differences between what we might call transactional and relational approaches, and these have, and do, influence the practice of how government works with external parties. However, this creates a very binary way of understanding arrangements between parties and is not reflective of the complexity inherent in practice. To better comprehend that complexity, we need to develop more integrated approaches, capable of recognizing not just these, but other approaches as well. To move towards that, I set out an overview of these approaches.

Being transactional
Two important conceptual approaches to understanding relationshipsagency theory and transaction cost economicshave had important influence on the practice of contracting and outsourcing. Transactions are the foundational building blocks of both approaches and they share specific assumptions about human behavior; for example, that individuals are self-interested and will act opportunistically.
Agency relationships are created when a principal (i.e. a government organization) engages another party (i.e. a provider) to act on their behalf. In doing so, the principal must craft an appropriate incentive contract to govern this relationship.
Agency theory assumes that the interests of the parties diverge, that the agent will act opportunistically because they hold more information than the principal, and that the principal will often have considerable trouble in monitoring the agent (Alchian and Demsetz 1972;Jensen and Meckling 1976;Moe 1984;Eisenhardt 1989;Donaldson 1990). Agency costs are the by-product of these challenges. Important ideas from agency theory have framed the practice of public administration including, for example aspects of the US Reinventing Government movement, United Kingdom Next Steps, and the separation of purchasers and providers in New Zealand and Australia. Despite the potential for broader types of relationships, in public management practice the adoption of agency theory inspired reforms has often led to highly specified activities, competitive processes, tight monitoring and the use of financial incentives to drive performance.
Transaction cost economics focuses more on the boundaries of organizations and how to weigh up alternative governance arrangements (i.e. markets, hierarchies, or something in between). Decisions about the optimal way to organize are the driving force as organizations seek to economize on the costs necessary for "developing, maintaining, and protecting the institutional structure" (Pejovich 1998, 9) or "the costs that arise when individuals exchange ownership rights to economic assets and enforce their exclusive rights" (Eggertsson 1990, 14). These costs are the result of bounded rationality, potential opportunism and asset specificity (Williamson 1979). 6 Making these decisions, however, is not easy and rest more on judgement than precision given the difficulty of calculating transaction costs, the high probability that the full set of costs is not known prior to deciding, and the challenge of comparing like-with-like.
Both agency theory and transaction cost economics rest on transactional foundations, share assumptions of human behavior, and build arguments about the optimal design of incentive contracts (agency theory) and governance structures (transaction costs) to manage transactions. Often these decisions are boiled down to questions of make-or-buy in the private sector or make-or-procure in the public sector (Williamson 1999;Alford and O'Flynn 2012). Both perspectives have a much broader view of contracting; contracts here are the mechanisms that bind parties together, not legal documents.
Transaction cost economics in particular, sketches out a range of contractual types and governance structures. Contracts come in many forms, from classical to relational. Different forms of contracting match with different governance structures and are more or less appropriate (in an efficiency sense) in different circumstances. 7 In general, the higher the uncertainty and the higher the level of specific investment that needs to be made in the exchange, the better (in an efficiency sense) it is for transactions to be governed using norms of trust and relational contracts. Some writers have argued that in practice, government contracting models have tended to be much more classical/ transactional than relational (O'Flynn 2004;Alford and O'Flynn 2012).
The former underpinning the experience of considerable amounts of governmentexternal party relations in many countries, for example outsourcing and public-private partnerships. It has also been argued that such mismatches can often be the root cause of problems that emerge in relationships between government organizations and other parties (O'Flynn 2004;O'Flynn 2007). For example, adopting classical contracting where more adaptive, iterative approaches to working together are needed. Research has suggested that optimal matching can be constrained by a range of factors, including decisions made by political actors (O'Flynn 2004), and the system-wide rules and regulations (Alford and O'Flynn 2012).

Getting relational
Whilst transaction cost economics includes notions of relational contracting, this is quite different to proposals for a relational state. Here the relationship, not a transaction, is the foundational idea (Cooke and Muir 2012). More specifically, human relationships rather than transactions. Part of the explanation for this development is in response to the limitations of more transactional ideas that have permeated government. This has been most acutely so alongside the recent collapse of large for-profit firms that have been central to the delivery of services for government, such as Carillion in the UK (See Tizard and Mathias 2019). However, the relational turn has been building for some timethe focus on collaboration, more joined up working and partnerships being obvious parts of this development (O'Flynn 2018).
Relational approaches have especially been linked to complex problems and here it is argued that more complexity demands higher levels of integration and connection at the local level and deeper relationships with community (Muir 2013). Transactional approaches do not seem to tell the full story of what is going on here and writers argue that we need to move beyond the "tired debate of public versus private" (Muir 2013). Getting more relational in practice would require us to rethink who it is that government engages with, as much as how, and Muir made the case in his work for a "revolution on the supply side" so that waves of autonomous providers to emerge. 8 This relational revolution would, in this view, be the catalyst for new organizational forms and would require very different ways of thinking about relationships. Such ideas have been developing in policy practice for some timefrom the Big Society initiatives in the UK to an Australian third way. 9 An outsourcing mindset would be a poor fit, for example, with a supply side populated by a broader variety of actors and formsencouraging the expansion of cooperatives and mutuals being front and center for many nations. More relational approaches have also been linked to broader notions of stewardship, most recently by Tizard and Mathias (2019) who are championing a turn away from narrowly constructed outsourcing approaches, towards long-term, strategic stewarding of systems.

Moving toward integrated approaches
Transactional and relational approaches to understanding how government engages with external parties provide part of the picture, but neither comprehensively can explain what happens in practice. It seems, in fact, that we need to be able to account for both, and also what exists in between. Indeed, some recent research has looked at how combinations of transactional (i.e. sanctions) and relational (i.e. trust) factors shape performance in partnerships (Warsen, Klijn, and Koppenjan 2019). Not only must government be able to manage many different types of relationships, it must also have to manage the dynamics of relational and transactional within relationships. Such hybrid forms add further complexity to the mix.
In practice we need to think more about integrated approaches rather than either/or so we can better understand the full set of approaches that can be deployed, and ensure we have the capabilities to do so. Alford and O'Flynn (2012) began to develop the ideas of a portfolio approach which seek to offer more comprehensive and integrated ways of thinking about complex systems of service delivery and which move beyond narrow views of outsourcing. In this they looked at how to accommodate a broader range of ways of parties working together, and also the need to think system-wide. Their portfolio approach, from the outset, recognizes that public sector organizations rarely have control over the whole process of deciding and producing what needs to be done to achieve desired outcomes; "invariably [they have] to call on effort, information or compliance from other parties" (Alford and O'Flynn 2012, 3).
This more integrated approach is built around several important aspects: a broader understanding of external parties; understanding roles; connecting up; and motivations and motivators.

A broader set of external providers
For the most part narrow approaches focus on transactional relationships with forprofit and nonprofit providers; a classic outsourcing mindset. Alford and O'Flynn (2012) looked more broadly than outsourcing and bring a more diverse set of providers into the mix. This is possible by having a more inclusive definition -externalizationwhich refers to a situation where one or more external providers produce all or some of a service (see Table 1). This broader definition encompasses any entity outside the government organization that produces all or some of a service; these are what Alford and O'Flynn call external providers. Working with this definition means we encompass a much broader set of actors that are involved in producing public services: other government organizations; private firms; nonprofit organizations; individual volunteers; clients; and regulatees. This takes us beyond the typical, yet narrow, view of parties that government works withfor-profit and nonprofit providersto acknowledging that various individual and organizations are engaged in efforts that produce services, and therefore the outcomes and value that governments seek to produce.

Understanding roles
Broadening out these ideas also means thinking differently about how these parties join together, but also the roles they take on. This allows us to move beyond narrow notions of principals and agents (although not dismissing them) to capture quite different role combinations. Alford and O'Flynn (2012) noted that a relatively straightforward, but very helpful, way of thinking about this is think about two roles that are played in service delivery systems. The first is the deciding roledetermining what is to be produced, arranging for someone to do to, and providing the resources to get it done. And the second is the producing rolewhere the activities to turn resources into results are undertaken. They explained how different combinations of these roles produce very different relationships; for example, government might make decisions itself or do so jointly with others. Or it may decide and produce jointly (e.g. joined up collaboration with other government organization). A typical outsourcing arrangement, for instance, will involve government taking on the deciding role and external parties taking on the producing role.

Connecting up
Actually producing public services requires these roles and parties to be connected using different modes of coordination. A lot of attention has been placed on the narrow idea of outsourcing rather than broader notion of contracting. There are very many different modes of coordination that can connect parties together. One way to think about this is as a continuum (Alford and O'Flynn 2012): at one end is compulsion which relies on the threat of sanctions; at the other is collaboration which relies on trust and shared purpose. In between these two extremes we might use: supervision which relies on episodic instructions; classical contracting which relies on incentives and competition; and negotiation which relies on flexible agreements. These five examples are nodes on a continuum rather than definitive types. And relationships are often dynamic, rather than static; they might move between these types over time or in response to behavior. It is also important that we more fully recognize that the duration of the relationship matters as it helps to set expectations and shape behaviors of parties. In practice, the challenge is to adopt the optimal mode so that the nature of the relationship between the parties develops in the most appropriate way, including taking account of motivations and intended outcomes.

Motivations and motivators
How we design relationships matter and thinking about the motivations of various parties, and how to induce them to contribute to outcomes is important. Outsourcing mindsets rely on narrow views of human behavior and therefore tend to rely very much on carrots and sticks or reward and sanction models as a way to tap into extrinsic motivations. However human motivations are much more diverse than this, as are the motivators we can use to tap them. Matching motivations and motivators is central to effective relationship management and the achievement of outcomes by government organizations. Alford and O'Flynn (2012) develops a schema of motivations and motivators that captures a fuller picture than the typical "carrots and sticks" perspectives.
In their typology, Alford and O'Flynn (2012) identify four fundamental motivations or drivers, each of which partially (but never fully) explains behavior. Individuals, we know, are often complex mixes of motivations (Talbot 2005). We also know that motivations can be shaped and reordered through the use of different motivators or instruments through which we seek to tap these motivational drivers. In their schema, five motivators are important in transforming motivations into effort and action.
In terms of motivations, much attention is given to the first two listed below. And these are often contrasted to each other. Much less attention has been given to the third and fourth. Together the four provide a more fulsome account of the complex drivers of behavior.
Material self-interest: in which people value extrinsic material rewards or conversely the avoidance of sanctions. Intrinsic motivation: in which people gain satisfaction from the act of doing the work or carrying out the task itself.
Sociality: in which people gain satisfaction from a sense of social affiliation or approval from one's peers. Purposive values: in which people attach value to purposes beyond their own self-interest.
To turn these motivations into action we need to deploy motivators. Again, there is a lot of attention to some, and little to others. In the case of motivators, the first three tend to get most of the attention. Having a more fulsome set allows us to consider the importance of matching and the impact of mismatches.
Sanctions: the threat or actual application of punishments. Material rewards: tangible benefits such as money or other in-kind rewards. Intrinsic rewards: such as a sense of competence, self-determination, or the "warm glow" effect. Social motivators: extrinsic, nonmaterial such as conviviality, peer approval or disapproval. Perceived contribution to the public good: legitimacy of the task, mission alignment.
Building effective relationships relies in part on the effective matching of motivations and motivators. Extending these ideas allows us to plot some obvious pairing where we can expect to see better or worse outcomes (see Table 2).
Combining various parties, roles, and modes of coordination provides a much richer picture of how government organizations work with others. It also reflects more pragmatically the breadth of relationships that are in place. Matching the mode with the roles and parties is important and can have a substantive impact on results; these are the challenges faced most acutely in practice. Especially so where we are locked into narrow options. Broadening out our conception of contracting and adoption more integrated approaches means we might better avoid the mismatches that can create a range of unintended consequences and suboptimal outcomes for parties, users, and the public.

Getting a handle on costs and benefits
Working in a more integrated portfolio approach requires that we weigh up of various costs and benefits and that we move beyond either/or notions of transactional or relational approaches. Making the decision relies on sophisticated analyses of a range of (potential) benefits and costs. The decision principle being: government organizations should enlist external providers when the benefits outweigh the costs of doing so (Alford and O'Flynn 2012). Notions of costs and benefits have, however, been relatively narrow on focused almost exclusively on what we call the service costs and benefits or the "value for money" test. When adopting a portfolio approach, we also need to consider relationship and strategic benefits and costs. Assessing the full set of costs and benefits should drive decision making (Alford and O'Flynn 2012).

Service benefits and costs: net value of the service
Here we consider whether externalization will reduce the cost to government, increase value, or some combination of both. Additional value can be created in many ways. For example, through reducing costs because providers can generate economies of scale and pass on the savings or because external providers can deliver improved quality.
Other times value is created by accessing specialist skills that government organizations do not hold. Evidence on price varies widely, as does the evidence on quality; there is no definitive evidence that externalization reduces (or increase) the cost of services to government or that it increases (or reduces) quality. The evidence is mixed and depends on the specific circumstances and task. Weighing up service costs and benefits is not always straightforward. For example, when services are subjected to competitive tendering they may be specified differently. And decision criteria will differ; for example, if we can extract more value for the same price this is net gain even if we do not save money.

Relationship benefits and costs: getting the relationship to work effectively
Relationships matter because they provide the architecture for getting things done. Parties must coordinate action and interests may differ; thus relationship costs emerge. These are the costs of managing the relationship and ensuring that an external party acts in such a way that what needs to get done, is done. These costs derive from four main relationship management tasks: defining the servicewhat is it is that has to be done; determining who is to produce the servicefinding and engaging the external party; ascertaining whether the service has been providedmonitoring, managing, oversight; and inducing good performancethrough the use of various motivators. However these tasks are undertaken, cost will be generated; whether that be the costs of trust building or the costs of close monitoring. These relationship costs must be considered against the positive relationship benefits that accrue. For example, well-specified contracts can bring clarity to the roles and responsibilities of parties; well-managed collaborations can build social connection.

Strategic benefits and costs: longer-term positioning and competencies
Engaging with external providers poses both strategic benefits and costs; here we are concerned with how externalization may impact on internal competences and strategic positioning, including legitimacy and trust. Time can play an important factor here as short-term gains on services and relational factors may impact strategic benefits and costs in the longer-term. Externalization can generate strategic benefits in various ways through a more strategic focus by government organization, through political and solidary benefits. In other words, by being associated with particular providers. Strategic costs are incurred when core competences are lost or the public standing of an organization is compromised to the extent that the long-term viability of its mission is compromised. On the former, the government organization may lose its ability to decide what it needs, or how to carry it out. On the latter, we are concerned most with risk of losing public trust or stakeholder support, based on what external providers do when acting on behalf of the government organization.

Making decisions: a contingency framework
The nature of these relationships changes over time and in response to a range of pressures. And these systems are dynamic, not static. Demands on government change alongside the shifting needs and aspirations of communities, political ideology and preferences recasts relationships, provider markets and supply chains transform over time, and technological change can offer new options to government (O'Flynn and Sturgess 2019). Decision makers need ways of figuring out how to respond to these dynamics, including determining who to work with and how. Making these decisions, however, relies on judgements about purposes, identification of who might be involved and the analysis of various costs and benefits. However, drawing these aspects together is complex. Alford and O'Flynn (2012, 226) develop a contingency framework to aid public managers and argue that: Instead of presuming on the basis of ideology, fashion, political convenience or transient budgetary circumstances that an activity should be externalized or alternatively conducted in-house, a contingency approach aims to choose the best option for the circumstances. Its touchstone is concern for what is likely to bring about the greatest value to the public.
The starting point for analysis of whether and how to externalize is purpose. To understand purpose, we ask two important questions: (i) What value are we seeking to produce? And (ii) What benefits to the public do we expect to derive from the performance of this function. The questions force us to consider purpose and open up the potential of identifying a broader range of ways of achieving outcomes. And focusing on purpose forces us to think why we are doing something not just high.
Once we have a handle on purpose and what we are trying to achieve, we can focus on two related questions: (i) What specific activities do or could lead to the achievement of desired outcomes? and (ii) Who is or might be involved in those activities? Some have argued that using techniques such as value chains can help public managers to map out the production processes involved in creating outcomes (see Alford and O'Flynn 2012 for a detailed discussion and applications). This technique is useful because it allows public managers to identify which external parties do (or might) contribute to outcomes and what relationships we have (or might have) with these parties. Considerable judgement is exercised in these steps and this requires the develop of very different sets of capabilities for those involved in deciding on the nature of relationships and those that will manage themoften quite different sets of actors. Figure 1 provides an overview of the contingency framework from Alford and O'Flynn (2012)

Capabilities
Looking to develop a more integrated framework requires us to rethink, to some extent, capabilities. Thinking at the systems level and stewarding across portfolios of relationships is critical to longer-term effectiveness ( 1. Competencies of the individual public servant: the skills, attributes, and characteristics of those who need to manage relationships and those that need to make externalization decisions. Different types of externalization require different competencies whether in relation to deciding or managing relationships. 2. Capabilities of public sector organizations: how structures and processes are arranged to facilitate the tasks of relationship managers and to mitigate the obstacles to their effectiveness. Particular attention is needed to more strategic human resource management, organizational structures, cultural mapping and matching, and portfolio management. 3. Enabling environment: how to configure government-wide structures and processes so that the other two levels can be more effective. This may include more flexibility in relation to budgeting or rule application and more contemporary approaches to risk management.
Governments need to give considerable attention at each level because: … in the end, public sector organizations must develop the capability to manage these complex portfolios and networks of external providers. And not simply as discrete providers engaged in a narrow part of the service delivery story, but as interconnecting webs of providers focused on the production of public value (Alford and O'Flynn 2012, 253).
Capabilities then, are much more complex that just looking at what specific skills public servants might need. Developing capabilities requires us to understand the nature of the task at hand (e.g. is it simple, complex, or somewhere in between?); the nature of the relationship that government engage in (e.g. are we coercing, collaborating, cooperating?); and how do we might choose between these very many and varied options (e.g. the contingency approach). Developing the capability to steward these complex systems of various relationships, working with (and through) a full range of external providers, means we need to rethink individual competencies, organizational capabilities, and how we might reshape the enabling environment. Neglecting any of these levels means that the practice of managing these relationships is often challenging.

Conclusion
This article has covered a lot of territory. The purpose has been to set out some foundational ideas that can help us to think more strategically about what it is that government does, with who, and how. It provides the foundational building blocks for thinking much more comprehensively about the nexus of relationships that governments rely on to deliver on outcomes for citizens and communities.
Shifting mindsets is hard work. Simply saying we want to rethink relationships and move beyond outsourcing or procurements mindsets to a much more diverse portfolio of ways of working, differentiated ways of operating, and stewarding complex systems requires concerted efforts in theory and practice. We need to continue our work in building these connections, in laying out this complexity and providing ways in which we can map out new ideas that help us in the practice of governing.
When we look at the work of government is actually done, we see this complexity clearly, but we continue to design systems, processes and relationships that seek uniformity, and which narrow our ability to delivery on the needs and aspirations of citizens and communities. We need new building blocks to guide us and this article is part of that effort.
Notes 1. For more on the collapse of Carillion in the United Kingdom see the report from the National Audit Office (2108) online at: https://www.nao.org.uk/report/investigation-intothe-governments-handling-of-the-collapse-of-carillion/ and also the report by the Public Administration and Constitutional Affairs Committee on lessons from the Carillion collapse online at: https://publications.parliament.uk/pa/cm201719/cmselect/cmpubadm/ 748/74802.htm 2. The figure captures annual expenditure on goods and services, not the total value of new contracts. 3. The Australian Government Department of Finance provides summary information online at: https://www.finance.gov.au/procurement/statistics-on-commonwealthpurchasing-contracts/ 4. On the complexity of immigration detention contracts see O'Flynn (2014) Manus Island takes Australia to the edge of outsourcing, The Conversation, March 6. Online at: https:// theconversation.com/manus-island-takes-australia-to-the-edge-of-outsourcing-23647 5. On some of the challenges of specific, measuring and monitoring see O'Flynn (2015) Big Bucks up for Grabs When Governments Outsource Unemployment, The Conversation, February 23. Online at: https://theconversation.com/big-bucks-up-for-grabs-whengovernments-outsource-unemployment-37903 See also Donaldson (2016) Perverse incentives at work in Nauru subcontracting The Mandarin, August 17. Online at: https:// www.themandarin.com.au/69040-nauru-under-reporting-shows-perverse-incentives-atwork/ and Donaldson (2017) Calls to rethink offshore detention after scathing audit, The Mandarin, January 18. Online at: https://www.themandarin.com.au/74257-call-to-rethinkoffshore-detention-management-after-scathing-audit/ 6. Opportunism goes beyond assumes of self-interest and assumes individuals will act with guile (Williamson 1979); bounded rationality is recognition that agents are cognitively limited, adaptive and goal oriented and therefore they are not perfectly rational beings as is assumed in the homo economicus model of agents (Simon 1957); asset specificity relates to the level and type of investment made by suppliers in relationships (Williamson 1979). 7. For example, classical contracts match best with market governance, and relational contracts match best with unified (hierarchy) or bilateral (long-term partnership) governance structures (Williamson 1979(Williamson , 1999. Classical contracts are highly formalised and transactional and often thought of as 'black letter law' agreements; relational contracts are governed by norms, reciprocity and trust. Theoretically, relational contracts deal better with complexity, uncertainty, high frequency interactions, and increasing levels of specific asset investment; whereas classical contracts deal better with short, selfliquidating transactions (Williamson 1979). 8. By autonomous institutions, Muir (2013) is referring to self-governing entities that are "bounded from the over-reach of bureaucracy and profit" and because of this are "most likely to foster strong relationships and service innovation". 9. Readers may be interested in the broad range identified in O'Flynn and Sturgess (2019: Appendix B) which covers: social enterprises, mutual, cooperatives, independent professional teams, prime contractor models, integrator models, agile sourcing, modular contracting,, social impact/benefit bonds, arms' length bodies, joint ventures, management insertion, and GOCOs. The Appendix provides links through to examples from across the word.