This paper examines leadership in practice, specifically the interaction of leaders and followers, taking account of context (Spillane, J.P. (2006). Distributed Leadership. San Francisco, CA: Jossey Bass). Employing Gronn's dimensions of concertive action and conjoint agency (Gronn, P. (2002). Distributed leadership as a unit of analysis. Leadership Quarterly, 13, pp. 423–451), different conceptualizations of distributed leadership (DL) are examined, and the influence of a health and social care context on attempts to enact DL is analysed. In so doing, boundary conditions or the limits to distributing leadership in health and social care are identified. The analysis suggests that the collective leadership DL model presented by Denis et al. (Denis, J.-L., Lamothe, L. and Langley, A. (2001). The dynamics of collective leadership and strategic change in pluralistic organizations. Academy of Management Journal, 44, pp. 809–837) is most likely to be enacted in the face of policy and professional pressures towards more concentrated leadership. However, where DL does not encompass conjoint agency, it will tend towards more towards ‘nobody in charge’ (Buchanan, D.A., Addicott, R., Fitzgerald, L., Ferlie, E. and Baeza, J.I. (2007). Nobody in charge: distributed change agency in healthcare. Human Relations, 60, pp. 1065–1090) or collaborative leadership (Huxham, C. and Vangen, S. (2000). Leadership in the shaping and implementation of collaboration agendas: how things happen in a (not quite) joined-up world. Academy of Management Journal, 43, pp. 1159–1175). Following the analysis, the authors argue that researchers need to move beyond a reified concept of DL, and ask a more straightforward question of how power is distributed. As Gosling et al. suggest (Gosling, J., Bolden, R. and Petrov, G. (2009). Distributed leadership in higher education: what does it accomplish? Leadership, 5, pp. 299–310), DL evokes an aspiration for the way leadership is configured, and draws attention to iterative relations between leadership, followership and context, but it is a conception of leadership that requires unpacking. This conceptual analysis, applied to health and social care, is offered in pursuit of this aim.
This paper integrates literature in the areas of distributed leadership (DL), sociology of professions and public administration, to examine the interaction of DL with institutional context, as DL is enacted in practice. The interaction of leadership with context represents a research gap, which an emergent body of research is seeking to address (see the special issue of Human Relations edited by Liden et al. 2009). Specifically, we focus upon the enactment of DL in a health and social care context, and in so doing attend to how the institutional context influences the distribution of leadership. We focus on health and social care for two main reasons. First, health and social care is an exemplar of how contextual influences linked to professional hierarchy and policy impact on attempts to distribute leadership (Currie et al. 2009a). Second, health and social care in England,1 as a fast mover in policy reform, represents a case from which international lessons may be drawn (Martin et al. 2009). We frame our review around answering three research questions.
What is distributed leadership?
We address this question by examining conceptual variation surrounding the concept of DL, which has led to ambiguity as to what it is, and what it is not. Everyone seems to (think they) know what it means, but there is no evidence of an underlying ideal type for DL (Bolden et al. 2008; Gosling et al. 2009). Within a range of theories about DL, it is not entirely clear what is being distributed, by whom, or whether it is being actively distributed, or merely taken (Peck and Dickinson 2008). Applying the two dimensions of DL (concertive action and conjoint agency) identified by Gronn (2002), we synthesize the body of work that relates to DL (for recent reviews of DL, see: Bennett et al. 2003; Leithwood et al. 2008; Woods 2004). In doing so, we examine how different authors have defined DL, and how they relate to one another, with a focus upon identifying some contingent features of DL.
Why and how have governments promoted DL in the public sector?
The emergence of DL reflects disillusionment with heroic models of leadership (Heifetz 1994). For Heifetz, the prevalence of ‘wicked issues’ (Rittel and Webber 1973), which health and social care organizations (HSCOs) are required to address, means that leaders leave many problems unsolved and hence often have to disappoint the expectations of followers. Distributed leadership offers a potential solution to this problem by seeking to reduce follower dependence by enabling followers to take on leadership, i.e. enact ‘adaptive’ leadership (Heifetz 1994). Distributed leadership is also viewed as desirable in public services because it is inclusive and aligns with recent organizational restructuring towards the flatter organization. It may foster collaborative and ethical practice, and avoid alienation associated with lack of power by those positioned as followers. Finally, DL leverages skills and strengths across the organization to enhance organizational resourcefulness (Gronn 2002; Hodgkinson 1991); and is considered to be particularly appropriate for complex, contemporary organizations, where knowledge is distributed. In this light, Currie et al. (2009a,b,c) highlights that government has increasingly viewed DL as a panacea for both poor organizational performance and the democratic deficit within public services. However, government's vision of DL as a policy panacea for poor organizational performance has been difficult to enact.
Does the health and social care context promote DL?
Leadership researchers have established the importance of situation to the enactment of leadership (e.g. see contingency theories of leadership: Fiedler 1973; Lawrence and Lorsch 1986). More recently, researchers have focused on the interaction of leadership with context (see the special issue of Human Relations, edited by Liden et al. 2009). As Gosling et al. (2009, p. 300) argue, ‘to appreciate the function of a distributed perspective upon leadership requires recognition of the social, political and power relations within organisations’. However, while DL is regarded as important in health and social care, particularly when change and improvement are required, beyond a limited number of studies (e.g. Denis et al. 2001), there is little consideration of how DL is enacted on the ground (Hartley and Benington 2010). We address this question by drawing on research in public administration and the sociology of professions. In doing so, we explain the complex institutional context in which HSCOs operate. Furthermore, we suggest that the complexity of the professional and policy institutions may render any attempts to enact DL, as conceived by Gronn (2002), difficult. In essence, the health and social care context creates a paradox for DL.
Drawing on our answers to the three questions above, in the discussion section we examine how DL may be enacted in a health and social care context. In doing so, we view our contribution as twofold. First, we seek to provide better conceptual clarification by synthesizing the body of work surrounding DL, using Gronn's (2002) dimensions of concertive action and conjoint agency. Second, our analysis provides insights into how organizational context influences the enactment of DL, specifically through professional and policy influences in health and social care. In short, we construct an account of DL that is grounded in everyday practice (Spillane et al. 2004).
What is distributed leadership?
Policy-makers internationally have converged upon effective leadership as a panacea to improve the performance of HSCOs (Hennessey 1998; Kakabadse et al. 2003). In particular, two possible models of leadership are favoured by policy-makers. The first model privileges transformational leadership (Behn 1998; Bellone and Goerl 1992; Eggers and O'Leary 1995; Hennessey 1998). Transformational leadership encompasses charisma, inspiration, individualized consideration and intellectual stimulation, with the leader maintaining a continuous challenge to followers by espousing new ideas and approaches (Bass 1985; Bryman 1992; Storey 2004). In its academic conception, it does not necessarily prevent distribution, although in its translation into organizational practice, it has tended towards a ‘heroic’, individualistic approach, including the effects of government policy intended to turn around HSCOs (Currie and Lockett 2007). Hence, we position the policy variant of transformational leadership in the bottom right quadrant of Figure 1, since concertive action and conjoint agency appear very limited (Gronn 2002), and a great deal of leadership agency is assumed, which fails to account for the effect of context upon DL (Spillane 2005, 2006; Spillane and Diamond 2007; Spillane et al. 2004).
The second model represents DL. Relative to individual leadership, with its long history in theory and practice, DL is a much more recent idea. It was first suggested by Gibb (1954), but lay dormant until its rediscovery by Brown and Hosking (1986). Since then, the descriptive and prescriptive literature on DL has blossomed. The boundaries of the concept, however, have been somewhat blurred by the range of different terms employed to describe leadership that extends beyond the individual located within the upper echelons of an organization. As Spillane (2005, p. 149) suggests: ‘DL often is cast as some sort of monolithic construct when, in fact, it is merely an emerging set of ideas that frequently diverge from one another’.
For the purposes of our analysis (and parsimony), we employ Gronn's (2002) concertive action and conjoint agency dimensions of DL. We do so because Spillane's (2005) notions of ‘collaborative’ distribution (where individuals play off one another, creating a reciprocal interdependence between their actions), ‘co-ordinated’ distribution (where leadership practice is stretched over different activities that must be performed in a particular sequence for leadership practice), and ‘collective’ distribution (where leadership practice is stretched over the practice of two or more leaders who work separately), map closely onto Gronn's (2002) DL dimensions. Furthermore, we return to Spillane's work in the discussion, when considering how leadership is ‘stretched over the social and situational contexts’ (Spillane et al. 2004, p. 5).
Concertive action encompasses three patterns of activity (Gronn 2002). First, concertive action relates to the spontaneous collaboration within an organization, where leadership actors with different skills, expertise and different occupations or organizations ‘coalesce’ to pool expertise, work jointly on a task and regularize conduct for the duration of the task. Second, concertive action involves a shared role, which emerges from two or more people who intuitively develop close working in an emergent fashion, within an implicit framework of mutual understanding. Third, working together becomes institutionalized over time into concertive mechanisms or structures, which may be grafted onto existing governance arrangements. For example, within a ‘professional bureaucracy’ (Mintzberg 1979), traditional norms that a ‘leader is first among equals’ may engender an institutionalized leadership structure, such as a committee.
Conjoint agency is more about the direction of leadership influence and relates to the alignment and synchronization of leadership action across different individuals, which results, first, in interpersonal synergy, and second, reciprocal influence (Gronn 2002). Leadership actors synchronize their individual leadership acts in real time through orientating not just towards their own plans, but also towards those of their peers. In so doing, actors draw upon their sense of group membership and open themselves up to, influencing, and being influenced by, their colleagues. There is reciprocity regarding the influence of two or more parties upon one another, which occurs in a manner akin to a ‘zigzagging spiral’, with each person in the leadership arrangements bearing the accumulated effects of successive phases of influence, as they begin to influence one another again (Gronn 2002).
Spillane et al. (2004) provides examples of DL that ground our conception of DL, albeit from the context of education. Distributed leadership activities in an educational context, as described by Spillane, focused on teaching staff meetings, typically involving a number of leaders whose actions are cognizant of what other leaders are doing. For example, a formal leader (principal) may take a more strategic view, moving meetings along, synthesizing what has been said, and setting expectations of followers. Meanwhile, another of the leadership team (e.g. assistant principal) performs some of this role as well, but acts more as the ‘detail person’ in identifying problems and generating solutions. In another example, the principal develops an evaluation process for teaching with others in the leadership team, but acts as an ‘authority’ figure, and limits herself to summative evaluation. Meanwhile, another of the leadership team, with greater rapport with teaching staff, engages in formative evaluation with regular feedback to teachers about their practice. According to Spillane, such reciprocal interdependences involve individuals playing off one another, with the practice of person A enabling the practice of person B, and vice versa. In HSCOs, focusing upon psychiatrists, Gronn (2000) provides a further example of DL. He describes the role constellation formed by three senior psychiatric hospital administrators as representing more than the aggregated efforts of the individuals, since, ‘complementary specialization enabled each man to engage in actions and operations of his own choosing for which he was best fitted within a jointly agreed-upon framework of activities, in pursuit of the interests and well-being of the hospital’ (Gronn 2002, p. 332).
Employing Gronn's (2002) dimensions, we argue that DL requires both concertive action and conjoint agency. Concertive action without conjoint agency may result in a divergence of the direction of leadership, where leadership is not aligned and synchronized across individuals, nor is reciprocal influence of individuals upon each other evident. Similarly, conjoint agency without concertive action may represent an aggregation of leadership in a common direction, but may not generate the desired synergies or reciprocal influence that should emanate from a pure form of DL. Employing Gronn's (2000) example of psychiatry, one psychiatrist may clinically lead her team towards therapeutic interventions with a group of patients, while another may lead his team towards pharmacological intervention with the same group of patients, each team delivering their interventions in isolation from the other. In DL terms, each clinical leader may be aware of the actions of each other, but the interactions between the different strategies for mental health intervention are not considered, with resultant poor clinical outcomes. Such an approach deviates from the policy desire to enact active and co-ordinated approaches to intervention to enhance clinical effectiveness (Anderson et al. 2008; Greden 2001; NICE 2004).
We now map the different conceptions of DL in relation to one another, based on the extent to which they adhere to Gronn's (2002) dimensions of concertive action and conjoint agency (see: Figure 1). In doing so, we suggest that Gronn's (2002) conception of DL, encompassing conjoint and concertive action, is located in the upper left-hand quadrant of Figure 1. The approaches described immediately below –‘shared leadership’ (Pearce and Conger 2003); ‘SuperLeadership’ (Sims and Lorenzi 1992); ‘team leadership’ (Katzenbach and Smith 1993); and ‘collective leadership’ (Denis et al. 2001) – all assume a degree of organization for DL driven by formal leaders. In essence, for these authors, DL does not mean the absence of a leadership hierarchy, but recognizes that the enactment of DL requires top-down activity. In contrast, other authors emphasize DL as a relatively bottom-up phenomenon, characterizing DL as ‘nobody in charge’ (Buchanan et al. 2007) or ‘collaborative leadership’ (Huxham and Vangen 2000). In broad terms, those more top-down driven DL models are more likely to ensure that direction is aligned (conjoint agency), but less likely to engender the widespread synergy and ongoing reciprocal influence (concertive action) described by Gronn (2002), i.e. they tend towards the bottom-left quadrant of Figure 1. Meanwhile, those more bottom-up models of DL are likely to engender greater synergy and ongoing reciprocal influence, but may have less confidence that the direction of DL is aligned in a conjoint manner, i.e. they tend towards the top right quadrant of Figure 1.
‘SuperLeadership’ (Sims and Lorenzi 1992), ‘team leadership’ (Katzenbach and Smith 1993) and ‘collective leadership’ (Denis et al. 2001) lie within the broad confines of ‘shared leadership’ (Pearce and Conger 2003). Shared leadership, first, re-envisions the ‘who’ of leadership, with leadership representing a set of practices that can and should be enacted by actors at all levels, rather than a set of personal characteristics and attributes located in senior-level managers. Second, it re-envisions the ‘what’ of leadership through its emphasis upon social interactions around leadership influence, i.e. shared leadership is a group phenomenon, with followers playing a role in influencing and creating leadership. Finally, shared leadership re-envisions the ‘how’ of leadership by focusing on the skills and abilities required to create conditions in which collective learning can occur. The shared leadership approach is consistent with DL, as described by Spillane (2006), with a focus on the interaction of leadership and followers, taking account of context in which this occurs. In mapping it onto Gronn's (2002) dimensions of DL, we make the following observations.
The key feature of SuperLeadership is the emphasis placed on ‘leading others to lead themselves’ (Sims and Lorenzi 1992, p. 296), i.e. SuperLeadership develops capacity in others, so they are independent of formal leaders to stimulate their talents and motivation. In a similar vein, Katzenbach and Smith (1993) view team leadership as a process, whereby team leaders play the role of coaches or facilitators who focus on developing leadership in others by building commitment and confidence, removing obstacles, creating opportunities and being part of the team. The SuperLeadership and team leadership variants of the shared leadership approach highlight that leadership emerges beyond formally designated leaders, but engendering this requires organizing skills and social influence on the part of a senior leaders (Hosking 1988, 1991).
The overall effect of shared leadership in its SuperLeadership or team leadership variant is the need for some external or internal top-down organizing for DL. Consequently, the shared leadership variant of DL diverges from Gronn's (2002) conception of its concertive and conjoint dimensions because the circulation of initiative associated with concertive action may be constrained by a more hierarchical ordering of the network of leadership actors. We recognize that DL may develop over time to encompass concertive action and conjoint agency more fully within shared leadership approaches (Denis et al. 2005). Nevertheless, in relation to Figure 1, we position shared leadership, SuperLeadership and team leadership towards the middle, but tending towards the bottom left to emphasize the relative absence of concertive action.
The ‘collective’ leadership variant of shared leadership (see Denis et al. 2001) recognizes the influence of external agencies, notably externally imposed professional work arrangements and government policy on the enactment of DL in HSCOs. In doing so, Denis et al. (2001) emphasize how context frames the interactions of leadership actors, thus aligning with DL as presented by Spillane (Spillane 2005, 2006; Spillane and Diamond 2007; Spillane et al. 2004). They suggest that driving change within complex organizations requires that strategic leadership roles are shared, with each member of a ‘leadership constellation’ playing a distinct role, and all members working together harmoniously. Taking account of the critique of the limit upon the potential for any individual driving transformational change through a complex organization, Denis et al. (2001) argue that this, more collective leadership approach, assembles the necessary variety of skills, expertise and sources of influence and legitimacy. The model of collective leadership appears more formalized than the ‘collaborative’ leadership or ‘nobody in charge’ models described below, with distribution of leadership among a strategic group, which Denis et al. represent as exhibiting entrepreneurial or brokering behaviour. Therefore, the success of the ‘collective’ approach to DL necessitates a concentration of concertive action and conjoint agency, i.e. it appears less widespread than Gronn (2002) suggests. Thus, we position collective leadership in the middle range of DL, a view confirmed by Currie et al. (2009c).
In contrast to top-down forms of DL, Buchanan et al. (2007) prescribe that leadership for change agency needs to be very widely dispersed, i.e. the impetus for DL is ‘bottom-up’. Their prescription was illustrated through one of their three empirical cases, where senior management encouraged an innovative culture and allowed significant autonomy to clinical teams within their organization. Drawing on the case, Buchanan et al. (2007) suggest that leadership needs to be numerous, transient, fluid, migratory, ambiguous and distributed, with many leadership actors engaging and disengaging over time. Consequently, anyone who wishes to assume responsibility has a chance to be involved. Such a conception of DL stands in direct opposition to individual leadership, with Buchanan et al.'s (2007) prescription that change agency is best distributed through having ‘nobody in charge’. Buchanan et al. (2007) argue that the ‘nobody in charge’ variant of DL encourages spontaneous collaboration, i.e. promotes concertive action. However, such an approach may actively work against conjoint action, because leadership influence may become fragmented among the myriad actors trying to enact change agency. Based on the above discussion, we locate Buchanan et al.'s (2007) model in the top-right quadrant of Figure 1, although tending towards the middle, since their model highlights considerable potential for conjoint agency over time. Interestingly, the two other cases conducted by Buchanan et al. (2007) attest to the difficulties in enacting DL in healthcare organizations. This emphasizes that researchers need to study leadership in practice to understand how the situation impacts upon interactions across leaders and between leaders and followers (Spillane 2006).
Finally, Huxham and Vangen (2000) present the idea of ‘collaborative leadership’ as being enacted not only through the behaviour of actors identified as leaders, but also through things that happen because of the structures and processes embedded within a collaboration (Spillane 2005, 2006; Spillane and Diamond 2007; Spillane et al. 2004). However, the structures and processes are often outside the immediate control of members of collaboration, e.g. controlled by government policy-makers or by professional bodies (we expand upon this in the later section focusing upon the health and social care context). The result may be similar to the effect we attribute to the ‘nobody in charge’ model above (Buchanan et al. 2007). Although leadership activities clearly affect the outcomes of the collaboration, those ‘leading’ are frequently thwarted by dilemmas and difficulties, so the outcomes are not what they intended. In such a context of collaborative leadership, conjoint agency may prove particularly difficult to enact. In such circumstances, Huxham and Vangen suggest collaboration may be characterized by ‘inertia’, since prospects for change are limited as leadership fragments. For this reason, we position collaborative leadership in the top right quadrant of Figure 1, but tending further away from the middle than Buchanan et al. (2007), on the basis that there may be a high degree of spontaneity in interactions, but the effect of these are less likely to be aligned in a conjoint manner, hence organizational inertia.
Having set out the framework for conceptions of DL (variation in the extent to which DL encompasses concertive action and conjoint agency), we now develop a contingent analysis to consider which forms of DL are likely to be enacted within the context of HSCOs.
Why have governments promoted DL in the public sector?
Policy-makers across the world, concerned with improving the performance of public service organizations (PSOs), have looked to leadership as a ‘panacea’ for organizational ills, particularly in countries subject to Anglo-American influence (Hennessey 1998; Kakabadse et al. 2003). Leaders of PSOs, like CEO ‘superstars’ in the private sector, are cast in a heroic, transformational mould (Khurana 2002) and encouraged to exhibit characteristics such as charisma, inspiration; individualized consideration and intellectual stimulation, with the leader maintaining a continuous challenge to followers by espousing new ideas and approaches (Bass 1985; Bryman 1992; Storey 2004).
However, as individualistic models of transformational leadership have emerged in practice within HSCOs, so too has criticism of the application of the transformational leadership model. Heifetz (1994) is particularly critical of current leadership theory because it relies too much on the capability of a single leader, often situated at the top of an organization (Heifetz 1994). This reinforces the dependence of others on the leader for answers to a problem, but with the leader ill-equipped to provide the answer. Instead, Heifetz (1994) argues that solving problems through the delivery of public services lies beyond the capacity of any one person. Consequently, Heifetz (1994) argues that leadership must facilitate a process of ‘adaptive work’ in which those who are led undertake difficult work to identify what is wrong and to forge solutions.
Increasing recognition of the limits to individualistic leadership has led policy-makers to extend the reach of leadership influence through the promotion of DL beyond a single, ‘heroic’ individual (Currie et al. 2009a,b,c). In England, following the emphasis on leadership in the Modernising Government White Paper (Cabinet Office 1999, p. 57), virtually all public services felt compelled to respond with initiatives to promote leadership (Hartley and Allison 2000; Storey 2004). Reports which emphasize leadership have followed, focusing upon leadership in: central government (Cabinet Office, Performance and Innovation Unit 2001; Home Office 2001), local government (DETR 2003), the police service (NPLF 2002), defence (Modernizing Defence People Group 2000), education (DfES 2003), higher education (HEFCE 2004, p. 34), health and social care (NHS Institute for Innovation and Improvement 2007). We suggest that the UK's new Coalition Government (Conservative–Liberal Democrat) elected in 2010 is likely to continue the previous Labour Government's preoccupation with leadership. Hartley and Benington (2010), examining leadership in health and social care, argue that the financial crisis and reductions in public expenditure will lead to a sustained emphasis on the role of leadership. Reflecting this, with respect to government policy in health and social care, the Coalition Government has continued to roll out the ‘Darzi’ reforms outlined in his Next Stage Review (Department of Health 2008) which seek to distribute leadership to frontline clinicians. Further, evident in their first White Paper for the NHS (Department of Health 2010) is a continued push to distribute leadership to service users, as well as frontline clinicians, who are expected to lead in concert with managerial leaders. We also see DL in the policy aim to position frontline clinicians, specifically GPs (general practitioners of community physicians), as leaders of the planning and commissioning of health and social care provision in England.
Although there is some empirical evidence that leadership has a significant effect on organizational performance (Agle et al. 2006; Bycio et al. 1995; de Hoogh et al. 2004; Howell and Avolio 1993; Ogbonna and Harris 1999; Waldman and Yammarino 1999), critics note that, beyond a limited number of studies, much of the evidence has been circumstantial or anecdotal (Porter and McLaughlin 2006). Therefore, the link between leadership and performance might be best regarded as an act of faith, rather than an empirically proven fact, particularly in relation to PSOs (Van Wart 2005). Compounding this problem, conceptions of leadership in PSOs have been poorly defined (Fairholm 2004), so no one leadership approach can be linked to superior organizational performance. Nevertheless, DL is viewed by policy-makers in the UK as a means by which poorly performing PSOs can be ‘turned round’, to combat social exclusion, facilitate joined-up government, build relationships with multiple stakeholders and engender citizen and user participation (Currie et al. 2005). Having explained why governments have sought to promote DL in the public sector, we now turn to the question of how?
How have governments promoted DL in the public sector?
England represents a ‘fast mover’ regarding implementation of transformational leadership and DL within PSOs. Policy-makers have acted in two ways to encourage more effective leadership of PSOs. First, they have implemented structural reform away from markets and hierarchies (in rhetorical terms at least), towards network forms of service delivery. Second, they have implemented large-scale leadership education initiatives to orientate public services leaders towards DL. We deal with each in turn.
Second, from 1997 to 2010 the Labour Government established a centralized and formalized leadership development system across all domains of public services. The leadership development systems are variable in their scope and effect, with school leadership most exposed to systematic development of DL, encompassing frontline, middle level and senior level staff, through the National College for Leadership for Schools and Children's Services (http://www.nationalcollege.org.uk). In contrast the systematic development of DL in health and social care has been rather slower, although its development has more recently gathered pace, reflected in the development of the ‘NHS Leadership Quality Framework’ as the basis for leadership programme offerings at different levels of the NHS (NHS Institute for Innovation and Improvement 2007), and the formation of the NHS National Leadership Council. The NHS National Leadership Council represents a strategic mechanism for infusing a culture of leadership throughout the NHS, with managers and clinicians working together, and leadership development distributed across all professionals and many levels of the NHS, including the appointment and development of ‘Darzi Fellows’ to provide clinical leadership to drive reforms.
Currie et al. (2009b,c) note, however, that while such leadership programmes are aimed at creating a new institution of DL, they remain largely abstracted from the professional and policy constraints upon leadership influence in public services settings. The leadership programmes, promoted by the centres outlined above, may assume more scope for concertive action and conjoint agency than is possible in HSCOs. Further, policy-makers assume public services settings are relatively homogeneous in the promotion of generic leadership models. Thus, those leading child protection agencies, with its myriad professions and organizations, are exposed to similar leadership models as head teachers, who lead one organization and one profession (Dudau 2009). The mosaic of professions arrayed in a hierarchical fashion within HSCOs, which is outlined in the next section, presents an additional significant challenge to DL. In short, public services contexts are heterogeneous, which may shape the way in which attempts to promote DL are enacted.
Does the health and social care context promote DL?
The implementation of DL in a health and social care context faces institutional challenges related to professions and policy. Many HSCOs conform to the professional bureaucracy archetype (Mintzberg 1979), which has implications for leadership. A powerful professional core of staff (e.g. doctors in a HSCO) may exercise significant autonomy over the means and ends of service delivery and self-regulate their activities, with limited scope for leadership intervention outside the ranks of this professional cadre (Friedson 1994; Hebdon and Kirkpatrick 2005; Wilding 1982). This powerful professional core of the organization is traditionally represented by a ‘leader’ drawn from their ranks, whom the others consider ‘first among equals’, with a notion of collegiality underpinning decision-making (Kirkpatrick 1999; Sheaff et al. 2004). However, we note that more managerial modes of organizing health and social care can disrupt these collegiate arrangements, as illustrated in Currie et al.'s (2010a) study of children's services networks, as leadership is more formalized and aligned with organizational accountability requirements.
Second, based upon horizontal and vertical distribution of knowledge and jurisdiction (Abbott 1988), a professional logic of hierarchy is dominant, which remains essentially paternalistic and authoritarian (Bate 2000). This may limit the distribution of leadership beyond the powerful professional group, owing to significant power disparities regarding who can lay claim to knowledge and jurisdiction over expert matters. Within the health care domain, power is likely to be concentrated with specialist doctors (Fitzgerald and Ferlie 2006). Reflecting this, academic commentators have also reported that others have struggled to assert themselves in influencing doctors, e.g. nurses (Currie et al. 2010b; Nancarrow and Borthwick 2005) and managers (Ackroyd 1996; Ferlie and Pettigrew 1996; Ferlie et al. 1996; Harrison et al. 1992). Where different HSCOs (e.g. healthcare providers and local government organizations) come together in networks, the distribution of power is less clear, particularly between doctors and social workers. Meanwhile, professionals from agencies outside health and social care, such as police or youth workers, find they are marginalized in leadership influence (Huxham and Vangen 2000). Finally, despite policy encouragement to allow for DL to service users, health and social care delivery is likely to remain professionally defined as a consequence of traditional professional hierarchy (Martin 2008).
The effect of professional hierarchy upon DL is twofold. On the one hand, collegiality that underpins decision-making might be more supportive of DL, as a professional college acts in a concertive and conjoint manner, represented by the lead professional. On the other hand, collegial decision-making is not free of politics, with power likely to be concentrated within elite groups, such as specialist doctors, and a degree of internecine conflict within and between professions. Consequently, the concertive and conjoint dimensions of DL are likely to be moderated by the dynamics of professional hierarchy. Figure 2 depicts the effect of such forces as moving DL downwards and across from its pure form (i.e. the upper left quadrant), towards the middle of the matrix.
Earlier, we highlighted how government policy supported DL through structural reform towards networks, and more systematic DL development. However, as with professional hierarchy, the effect of policy can be rather mixed. Specifically, ‘public managers (in England) are constrained by the fact that they work within a set of legal, regulatory and policy rules and demands, and are required to be accountable for their and their organization's actions’ (Ferlie et al. 2003, p. S9). Such ‘target-based leadership’ has been characterized as orienting leadership towards individualism, rather than collectivism, as accountability has been concentrated in the few, rather than many (Currie and Lockett 2007; Currie et al. 2009a,b,c). Its effects upon leadership ‘on the ground’ within English HSCOs has been highlighted by the way in which those at the apex of the management hierarchy have been castigated for failures in the delivery of health and social care. Recent examples include the ‘sacking’ of health and social care leaders, such as the Director of Children's Services, Haringey Local Safeguarding Children's Board, following the death of ‘Baby P’ (Laming 2009), and the resignations of CEOs of hospital trusts at Mid-Staffordshire, and Maidstone and Tunbridge Wells, following patient deaths attributed to poor quality service (see http://www.healthcarecommission.org.uk). Within this context, leadership is likely to be concentrated, rather than concertive and conjoint, because any leader situated at the top of the organization may be unwilling to distribute leadership to others, and others unwilling to take on leadership.
The result of policy emphasis upon performance management and accountability drives leadership towards concentration rather than distribution. The target-based element of policy may moderate both concertive and conjoint dimensions of DL, and counter the more supportive dimension of policy, which encourages DL through educational means and structural reform. In short, we should recognize that aspirations for DL coexist with a transition from collegial models of leadership towards more managerialized models which concentrate accountability (Bolden et al. 2008, 2009; Collinson 2006; Gosling et al. 2009).
In summary, we suggest that in HSCOs in England, professional hierarchy and traditional power relationships, combined with a strong centralized performance regime, will act to stymie policy-makers' aspirations for enacting DL. As such, the health and social care context creates a paradox for DL.
Returning to Spillane's (2005) work on DL, individuals play off one another, whether positioned as leaders or followers, creating a reciprocal interdependence between their actions. It is not just a function of one or more leaders' actions. Further, their interaction is framed by the context in which they find themselves. The HSCOs present particular challenges for DL, because structural power is all-pervasive (Denis et al. 2005; Peck and Dickinson 2008), so that one key aspect of leadership is the ‘ability to pull together a powerful alliance with diverse internal and external actors’ (Denis et al. 2005, p. 454). As Collinson (2006) notes, traditional dichotomous identities of leader and follower are increasingly ambiguous and blurred in the face of dynamic organizational contexts, as not only do leaders impact followers, followers also impact leaders. From this perspective, we seek to understand the influence of the health and social care context upon DL, drawing on insights from literatures on public administration and the sociology of professions.
Focusing on the dimensions of concertive action and conjoint agency, we now conceptualize how the health and social care context influences attempts to distribute leadership (see Figure 2). We suggest that the enactment of DL in HSCOs converges towards the middle of the diagram at the intersection of the axes in Figure 2, with a tendency towards the bottom right-hand quadrant. This is a result of target-based policy pushing DL towards concentration, even individualism, as represented by the thicker of the two lines. The policy pull upwards through network structures and DL education, represented by the thinner line in Figure 2, is counteracted by target-based performance management. Meanwhile, professional power is represented as a centrifugal force that, on the one hand, promotes DL through collegiality, but on the other, fragments or concentrates DL owing to its hierarchical arrangements.
Consequently, the enactment of DL in HSCOs in England appears closest to ‘collective leadership’ (Denis et al. 2001). It is perhaps not surprising that the form of DL enacted in the English health and social care context is close to collective leadership, because Denis et al.'s (2001) model of collective leadership was developed through studying large-scale change (hospital mergers) in Canada. However, our analysis of DL in the English health and social care context does diverge from this model, in that it is less concertive than characterized within the collective leadership model. An explanation for this is that the Canadian healthcare context described by Denis et al. is one where central targets appear relatively absent in comparison with England. This allows leadership to be more collective in Canada, promoting both concertive action and conjoint agency. Meanwhile, in the English health and social care system, centrally set targets for which leaders are accountable drive a more individualistic orientation, so that leadership is concentrated with an elite at the apex of the organization. As Bolden et al. (2008, 2009) note, DL can represent an idealized notion of equitable and inclusive working practices, which glosses over more managerial models of public administration.
That our analysis of DL diverges from the model presented by Buchanan et al. (2007) is more surprising, since the latter is firmly grounded in the English health and social care context. Buchanan et al.'s (2007) empirical study characterized DL as more concertive, and less conjoint. How do we explain the divergence of our characterization of DL from that of Buchanan et al. (2007)? First, we suggest that the title of their study, ‘nobody in charge’, may overstate the empirical case, since the presence of a leadership constellation (‘a partnership pair’ and ‘core of four’) is indicated as important at certain time points in the change initiative they follow. Other commentators confirm the temporal dimension to DL arrangements and highlight the requirement for some formal concentration of leadership in the early stages of any initiative towards DL (Currie et al. 2011; Denis et al. 2005). Second, we also note that the change initiative was fairly narrow (a specific clinical service improvement), compared with the large-scale mergers examined by Denis et al. (2001). In effect, the narrow focus of the study would minimize potential problems of a lack of conjoint agency and hence lead to a focus on the concertive action dimension. Finally, Buchanan et al.'s (2007) cross-case comparison of similar attempts at clinical service improvement in two other empirical sites revealed patterns of DL which were rather more concentrated than in their exemplary case of DL. Reflecting upon these differences, we suggest that patterns of DL on the ground, within HSCOs, are contingent on a range of factors, as we have outlined in the discussion above: the presence of central targets, the scale of the change initiative driven by leadership, different temporal stages of DL activity. To this we might add another crucial contingent factor: the extent to which a mosaic of professions characterizes the domain of change driven by DL. The basis for this lies with our contention that Gronn's (2002) conception of DL was generated within an educational context, which is characterized by a single profession (teachers), compared with health and social care, which encompasses a mosaic of professions. In the latter case, concertive action and conjoint agency proves more challenging, since the means and ends of leadership activity are likely to be more contested across professions, i.e. it may be more difficult to stretch leadership (Spillane et al. 2004).
While thus far, this review of the literature has steered clear of post-structuralist conceptions of leadership, our critique of DL highlights that leadership power relations and identities are blurred, multiple, ambiguous and contradictory. Collinson (2005, 2006) argues that leadership practice might be considered as a process of working through a series of dialectical relationships arising from the juxtaposition of conflicting ideas, forces and differential power and resources. In the health and social care context, these relate to policy and professional institutions, so that followers are encouraged to step up to leadership, but, at the same time, there are limits to this owing to professional hierarchy and, notwithstanding this, leaders may be required to act as ‘calculative followers’ in the face of accountability pressures. In summary, within the context of HSCOs, when considering professional and policy forces together, their interaction is likely to result in a more concentrated form of DL than policy-makers intend (Currie and Lockett 2007; Currie et al. 2009a,b,c). Consequently, the DL enacted in HCSOs diverges from the form of DL described by Gronn (2002), which encompasses significant and widespread concertive action and conjoint agency.
In conclusion, we concur with Gosling et al. (2009) that researchers need to move beyond a reified concept of DL, and ask a more straightforward question of how power is distributed. As Gosling et al. (2009) suggest, DL evokes an aspiration for the way leadership is configured, and draws attention to iterative relations between leadership, followership and context, but it is a conception of leadership that requires unpacking. We offer our conceptual analysis, applied to health and social care, in pursuit of this aim.
To appreciate DL, and its application in HSCOs, requires recognition of the social, political and power relations within organizations (Gosling et al. 2009). Our analysis, in Grint's terms, acts as a ‘corrective’ (Gosling et al. 2009), to draw attention to the variety of constituents of leadership, and to the iterative relations between tasks, actors, roles and organizational context, all of which are constitutive of the practice and experience of DL (Grint 2000). Rather than a blueprint or prescription for leadership, DL should be viewed as a perspective, a conceptual or diagnostic tool for thinking about leadership (Spillane 2005) that engenders this corrective analysis. Our critique allows for greater reflexivity about DL, so that researchers come closer to understanding how it is enacted on the ground (Spillane 2006).
To date, DL ‘apologists’ have not adequately clarified the role and contribution of individuals as continuing sources of organizational influence within a distributed framework (Gronn 2009). The current high profile accorded DL runs a risk of replacing one longstanding pattern in leadership, focused on individualism, with a polar opposite. However, commentators argue that leadership configurations fuse different degrees of focused and distributed tendencies (Bolden et al. 2008, 2009; Gosling et al. 2009; Gronn 2009). Indeed Bolden et al.'s (2008, 2009) research, albeit carried out in higher education, may represent a more accurate description of situational leadership practice, including both individual leadership and collective leadership working in tandem. Bolden et al. (2008) note that, in a professional bureaucracy, of which HSCOs are exemplars, pressures for top-down influence in times of change exist in dynamic tension with traditional values associated with professional collegiality. Our conceptual analysis highlights that some sources of influence carry more weight than others, and are anchored in different sets of resources, i.e. managerial accountability upwards as the senior lead, or concentration of power resulting from professional hierarchy.
Regarding further research, we suggest that our model of DL, which is grounded in the health and social care context, should be subjected to rigorous empirical examination. We contend that there is unlikely to be a generic template for the practice of DL, even within the specific context of health and social care. In particular, there is a need to draw out the contingencies that frame DL. We note that a new ‘Coalition’ government elected in 2010 in the UK changes the policy context which impacts DL in England. While the new government continues an emphasis upon leadership in HSCOs, there seems more emphasis on market forms of organization, and less on networks, with a reduction in central control through performance targets, i.e. DL may be bounded differently. Second, although the majority of research into DL has focused on PSOs, the appeal of DL is equally strong in the private sector (e.g. Teece 2007; Nonaka and Toyama 2002). Further research should identify the boundary conditions framing the enactment of DL in the private sector. Finally, we recognize limits to our study, notably that the empirical context was set in England. Consequently, we encourage further research in different public services contexts, particularly outside England, where the effect of centralized performance management regimes may be less pronounced, as was revealed in the Canadian health and social care system by Denis et al. (2001).
The electorate in the UK elect a government which covers England, Scotland, Wales and Northern Ireland. At the same time, the electorate in all countries except England elect their own devolved government which develops policy in areas such as health and education, with the effect that the policy context can vary across the constituent countries of the UK. Our empirical case is set in England and we recognize that the policy context may impact differently in other constituent countries of the UK.