Advice that resonates: explaining the variability in consultants’ policy influence

The increased presence of management consultants in public policy has resulted in concerns from practitioners and policymakers that consultants are too influential. At the same time, there are genuine reasons to downplay consultant influence on policy processes and to instead see consultants as ‘servants of power’. This tension raises questions about how much influence consultants have and the ways in which their influence might vary. Previous studies have shown how two concepts—openness and trust—are useful in understanding variability of consultant influence. As part of this, public decision-makers must be open to the involvement of consultants in a reform process and must also trust the specific consultants involved. At the same time, limited consideration has, to date, been given to discursive explanations of variability. Focusing its analysis on the influence of coalitions on education policy, this paper’s main contribution is to show that consultants will be more influential when their discursive repertoire demonstrates resonance with the narratives of the coalitions they are trying to influence. As part of this, consultants should deploy concepts and language that are recognisable to those coalitions, but which also align with the coalitions’ underlying narratives about the problem being faced and appropriate solutions.


Introduction
From humble beginnings (Kipping, 1999;Wren, 1987), management consulting has grown to become one of the world's most widely used services. Though figures vary, the value of the industry is generally seen as being in the billions (Chia, 2015;Momani, 2013;Saint-Martin, 2017), and many of the large consultancies have experienced significant financial growth over the last decade, even during the financial crisis (The Economist, 2013). This growth has involved increases in both scale and scope, and the consulting industry is now a particularly heterogenous one, with a wide variety of different service offerings in almost every industry (Hurl & Vogelpohl, 2021;Wargent et al., 2020). As part of this, management consultants have focused on working in the public sector (Howlett & Migone, 1 3 2013b, p. 242; Lapsley, 2009). This has resulted in consultants being seen as 'change agents' within complex processes of administrative reform (Hamilton-Hart, 2006;Lapsley & Oldfield, 2001;Leys, 1999;Shaoul et al., 2007) and as central actors in the new public management (NPM) (Beveridge, 2012;Hood, 1991;Saint-Martin, 2000). As part of this, consultants are brought into the public sector to provide ideas to policymakers in running government agencies more like businesses (Henkel, 1991).
The presence of consultants in systems of public policy and administration has become a key point of contention for practitioners and policymakers (Beveridge, 2012;Joint Committee of Public Accounts & Audit, 2018). Concerns have been raised regarding the degree to which consultants undermine fundamental notions of democratic accountability (Morton, 2021). Representations in the media and popular books have argued that consultants are 'witch doctors' (Micklethwait & Wooldridge, 1996) or a 'shadow government' (Guttman et al., 1976) who are 'plundering the public sector' (Craig & Brooks, 2006). Controversy has also emerged regarding consulting firms supporting authoritarian regimes or being complicit in alleged human rights abuses (Bogdanich & Forsythe, 2018). Embedded in these arguments are concerns that consultants are far too influential on the decisionmaking processes of governments. However, the cases supporting these concerns only represent a small proportion of the billions of dollars that governments spend on consultants each year (Lapsley & Oldfield, 2001;Momani, 2013). Likewise, consultants are still subject to a range of restrictions on their influence, including public scrutiny, professional disclosure rules, market competition, and client contracts.
Clearly, then, there is a question about the conditions under which consultants are more (or less) influential. For some, consultants are seen as being 'masters of the universe' and wielding enormous influence over democratic institutions. Yet there are legitimate reasons to be sceptical of this view as an overarching explanation. At the same time, limited work has been done to explore the specific factors that we can use to explain this variability. Considering this, the purpose of this paper is to develop a richer understanding of variability of consultant influence on government systems, structures, and processes.
Overall, the paper argues that there are three factors that shape consultant influence: openness, trust, and resonance. 'Openness' refers to the willingness or receptiveness of policy actors to the presence of consultants in a particular reform process. 'Trust' refers to the extent to which policy actors have confidence that they can rely on the advice of individual consultants or firms. This is related to the ways in which consultants and/or consulting firms demonstrate relevant skills, experience, or knowledge within a specific policy domain. Finally, resonance refers to the ways in which consultants deploy a discursive repertoire-concepts, language, and narratives-so that they demonstrate a degree of discursive affinity (Hajer, 2005) with the repertoire of other policy actors. It also refers to the extent to which consultants can usefully demonstrate to other policy actors how specific policy objectives might be met. Put simply, consultants will be more influential when state institutions are open to consultant interventions; when policymakers trust the consultants; and when there is resonance between the consultants' discursive repertoire and that of the policymakers.
To support these arguments, the paper proceeds as follows: The next section reviews what has been referred to as the 'consultocracy' literature to generate a conceptual understanding of the variability in consultant influence. This burgeoning literature supports two elements of the variability of consultant influence-openness of state institutions to the presence of consultants, and the extent to which public decision-makers trust consultants. At the same time, less consideration has been given to the role of discourse in mediating this variability. Discourse is particularly necessary to consider given that a long tradition in the business sciences has pointed to the fundamentally discursive nature of consultant influence (Clark & Fincham, 2002;Merilainen et al., 2004;Newton, 1996;Whittle, 2008). The empirical component of the paper, therefore, illustrates the concepts of openness and trust but also adds to current knowledge by exploring the discursive components of consultant influence. It does this by examining two separate but related processes in the Australian state of New South Wales (NSW). The first was known as the 'Better Services and Value Plan', while the second was called 'Local Schools, Local Decisions'. This case is particularly relevant given the variety of different roles that consultants played in the ultimate policy design. Following the empirical discussion, the paper describes the theoretical implications and areas for further work.
A central concern for all these varied contributions is consultants' ability to influence public policy (Lapsley & Oldfield, 2001;O'Mahoney & Sturdy, 2016;Saint-Martin, 2000). Across all contributions, there is general agreement that consultants' influence is based on their ability to provide ideas to public decision-makers, specifically when those ideas assist in processes of reform (Beveridge, 2012;Howlett & Migone, 2013a;Saint-Martin, 2000). At the same time, the literature provides several examples of situations in which consultant ideas have not been accepted, leading to a situation in which consultants have struggled to be impactful. For instance, Ashraf and Uddin (2013) show how a consultant intervention 'failed' leading to a 'disgruntled client'. Likewise, Saint-Martin (1998a) has compared the uptake of consultant ideas in Britain and Canada. As part of this, he has demonstrated how consultants have been particularly influential in Britain but less so in Canada. Given this, some academic authors see consultants as being highly influential, describing reforms as 'consultant dominated' that reflect 'private control over public policy' (Horrocks, 2009;Leys, 1999;Shaoul et al., 2007). Others, however, down play the influence of consultants, arguing that they occupy a more functional or 'process-driven' role (Howlett & Migone, 2013b) or that notions of overarching consultant influence are 'simplistic' (Hodge & Bowman, 2006). Taking this analysis further, several papers point to the contingent nature of consultant influence, arguing that consultants will be more influential when they serve the knowledge needs of government officials (Saint-Martin, 1998b). In this way, we can understand consultant influence as highly variable (see also Hamilton-Hart, 2006;Howlett & Migone, 2013a).
Given this variability, it is important to understand the factors that might make consultant ideas more or less likely to be adopted by policymakers. The first factor that is suggested by the literature is what we might refer to as openness. Here, the 'permeability' of state institutions to consultant interventions are key mediators of influence. Indeed, one of the more significant trends in public administration over recent decades has been the shift from government to governance (Bellamy & Palumbo, 2017). As part of this, the boundaries between different agencies have become blurred such that there is a high degree of permeability between organisations (Beveridge, 2012;Saint-Martin, 1998b). Such relations between the public and private sectors are, therefore, not given but are instead subject to contestation (Kooiman, 2004;Stoker, 2018). This can take several different forms, including secondments, 'instrument constituencies' (Sturdy, 2018), or public-private partnerships. As part of this, theorists have argued for a 'revolving door' between public agencies and consultancies (Cook, 2009). These arguments reflect developments elsewhere in the political science literature, including work on social movements (McAdam et al., 1996) and relational public administration (Bartels & Turnbull, 2020). Ultimately, then, these arguments point to the importance of consultants having access to the halls of power as an important part of their influence (Saint-Martin, 1998a, 2000. The second factor that is suggested by the literature is the notion of trust. This concept is embedded in explanations that see consultants as exercising influence through networks and inter-personal relationships with other actors in a particular policy sub-system (Cameron, 2010a(Cameron, , 2010bCzarniawska & Mazza, 2003;O'Mahoney & Sturdy, 2016;Shaoul et al., 2007). For instance, O'Mahoney and Sturdy (2016) show how, in the context of healthcare privatisation, consultants are able to draw on networks and trusting relationships with senior decision-makers to exert influence. The networks themselves are seen as a resource of consultant influence that can legitimise consultant ideas. As part of this, consultants gain access to decision-makers through cultivating ongoing, long-term, and reciprocal relationships with key decision-makers in government (Corcoran & McLean, 1998). This builds trust and so over time consultants fortify their influence by becoming 'trusted advisors' (Maister et al., 2002). This is, of course, antithetical to ideals of public sector procurement, which aim to achieve supposedly objective, arms-length relationships founded on merit, and ability to meet specified criteria (Raudla, 2013).
In this sense, the current literature on consultocracy reveals two explanations for the variability of consultant influence. Institutional arguments focus on the porosity of state institutions to consultants. Here, consultants' position within networked forms of governance gives them access to key decision-makers in government. Likewise, relational arguments see consultants as already embedded in social networks. Crucial to their level of influence here is their ability to gain the trust of senior officials. This goes beyond just openness-it is a close and reciprocal relationship between the different actors involved in, and which may continue beyond (Ylönen & Kuusela, 2019), a specific reform process.
At the same time, what needs further exploration are more discursive understandings of the variability of consultant influence on policy processes. Indeed, lessons from the 'critical consulting' literature (Clark & Fincham, 2002) demonstrate quite clearly the importance of consultants' discursive repertoire in their ability to influence the private sector (Bloomfield & Vurdubakis, 1994;Clark & Salaman, 1996;Newton, 1996;Whittle, 2008).

3
In this body of work, consultants are seen to discursively shape reality through the presentation and transmission of narratives and metaphors to their clients. This is generally operationalised at the organisational or actor level. This literature has, however, paid less attention to the ability of consultants to wield discursive influence over entire processes of public policy. At the same time, discursive studies of consultant influence on public policy remain nascent. The small number of discursive studies that do exist have explored the ways in which consultants use 'demand inflation' as a way of securing more work (Sturdy et al., 2020); the role of consultants in co-constructing problems with their clients through instrument constituencies (Sturdy, 2018); or how consultants enrol other policy actors to establish networks of support for a particular policy prescription (i.e. privatisation) (Jupe & Funnell, 2015). There is, thus, room for additional exploration of the ways in which consultants draw on language to shape the nature of policy formulation and implementation. More specifically, discursive understanding of why consultants might be more (or less) influential remains under-explored. As a result, and after a brief methodological discussion, the empirical component of this paper will apply the concepts of openness and trust, but also generate new insights by exploring more discursive understandings in variability of consultant influence.

Methodology
The research supporting this paper focuses on two separate but related processes of reform in the Australian state of New South Wales (NSW). The first was called the 'Better Services and Value Plan', while the second was called 'Local Schools, Local Decisions'. The 'Better Services and Value Plan' was introduced at the end of the 2009-10 financial year in response to budget forecasts from NSW Treasury. These forecasts indicated that, because of the global recession, the NSW Government was facing challenging financial circumstances over the forward estimates period. The plan aimed to improve service delivery and contain government expenses to save the state approximately $500 million. As part of this, in 2009, NSW Treasury engaged two consulting firms-the Boston Consulting Group (BCG) and PricewaterhouseCoopers (PwC)-to conduct separate audits of the NSW Education Department. BCG reviewed what Treasury referred to as 'non-school-based costs', which included school maintenance, land, and procurement. PwC explored what was referred to as 'school-based costs'. This review examined sick leave, payroll, and long service leave (amongst other things) (PricewaterhouseCoopers, 2010, p. 2).
At a similar time, the NSW Education Department began developing a suite of reforms to the NSW public education system. This reform package ultimately become 'Local Schools, Local Decisions'. The central aim of 'Local Schools, Local Decisions' was to reallocate decision-making powers within the public education system, and thus give 'principals and their school communities a greater say over how they allocate and use their available resources to best meet the needs of their students' (NSW Department of Education, 2009). Importantly, after the BCG and PwC reports were finalised, Treasury officials provided them to the Education Department to implement. However, and as I will discuss below, these instructions were largely ignored, and officials in the Education Department embarked on their own process to understand potential solutions to the problems facing the NSW education system. During this time, draft versions of the consultant reports were leaked. After, several key actors in the NSW public education system began advocating against introduction of what they perceived as a problematic policy reform. These advocates argued that 'Local Schools, Local Decisions' was entirely a creation of management consultants, insofar as it (in their view) emerged directly from the BCG and PwC reports (Kaye, 2012;Sciffer, 2013).
This case was chosen for two main reasons. First, it represented an 'instrumental case' (Stake, 2003) in that there was significant contestation about the role and influence of consultants in shaping the ultimate policy that was adopted by the NSW Government. For education activists, consultants were the ultimate drivers of the reform process, while for others, consultants were just one actor amongst many. The second reason it was chosen was due to availability of data. Here, the fact that the reports were leaked was a particularly useful aspect of this process. More specifically, consultants are notoriously secretive (Baaij, 2013, p. xiv). As such, availability of data became a key consideration in my research design. The availability of reports that were otherwise Cabinet-in-confidence meant that I could delve into the ideas, discourses, and narratives by both BCG and PwC and thus had access to a primary source for consultants' discursive mechanisms.
For this case, two primary data collection methods were used to support my arguments and findings-document review and interviews. First, policy and practice documents were reviewed to understand the temporal sequencing of events and to, therefore, establish a timeline. In addition, the documents provided a general understanding of the key points of contestation within this process of reform. These documents included annual reports from government departments, Hansard transcripts, media articles, and consultant reports. These were identified through a Google search including relevant key search terms (i.e. 'local schools local decisions' or 'better services and value plan') or a search on the relevant website (i.e. Departmental website or Hansard). Documents were initially retrieved between October 2014 and March 2015.
Insights from the document review were then explored further through interviews. People were included on the initial interviewee list based on their seniority and degree of proximity to the case under examination. Ultimately, I explored the retrieved documents to identify senior decision-makers or advisors who worked directly on either the 'Better Services and Value Plan' or 'Local Schools, Local Decisions'. Once a person had been identified, they were contacted directly via email. A follow-up call was conducted shortly thereafter if a reply was not received promptly. Additional interviewees were added to the initial list in a 'snowball' approach and were continued until saturation had been reached. As part of this, the final question for each of my interviewees was 'is there anyone else you could recommend I speak to?' Ultimately, I interviewed fifteen people. This included three people from the NSW Department of Treasury, six (former and current) employees of the Department of Education, one person from the Teacher's Union, two people from the consulting firms, one person from the NSW Secondary Principals' Council, and two NSW politicians. Interviews were conducted from August to December 2015.
The interviews were conducted as narrative interviews. This approach was highly instructive in allowing for deeper understanding of the discursive components of the case. As part of this, I asked interviewees to focus on the first time they had either come across the idea of 'Local Schools, Local Decisions', or else the 'Better Services and Value Plan'. From there, I asked interviewees to explain to me their involvement either initiative. As part of this, I asked interviewees to relay their experience temporally, that is, in a sequence of events such that A led to B which led to C (Ayres, 2012). Thus, a crucial question in each interview was 'And then what happened?' (Feldman et al., 2004). In line with accepted techniques for narrative interviews, I provided neutral 'probes' to elicit further information about connections between events. Indeed, '[r]evealing those connections is a primary goal of narrative interviews' (Ayres, 2012).
Having completed these two data collection methods, my analysis followed a grounded theory approach (Glaser & Strauss, 1967). There were several crucial elements of this approach that was replicated in my methodology. First, the data collection and analysis stages were not separate. Rather, I reflected on and analysed the data 'in field', inductively drawing out inferences as the data came in. These inferences were then applied to new data as it was collected to refine emerging concepts or develop new ones. This meant that analytic codes and categories emerged from the data itself and not from pre-conceived hypotheses. As part of this, the 'constant comparison' method was adopted (Lewis-Beck et al., 2003). This involved making comparisons between theory and data at regular intervals throughout the project. Here, data and theory were compared abductively (Agar, 2010;Kurowska & Bliesemann de Guevara, 2021;Yanow, 2009;Srivastava & Hopwood, 2009). Ultimately, the aim was to construct theory rather than establish representativeness (Bryman, 2017).
Crucial to the constant comparative method was the coding process. First, I established a set of descriptive codes. This allowed me to identify general topics of interest. After this, pattern codes were established. These were more abstract but had a higher degree of explanatory and theoretical power. As part of this, I grouped the coded data into themes and constructs. This allowed me to test the extent to which I could link any patterns in the data to the explanatory constructs I had devised (Andrade, 2009). It also allowed me to refine the constructs and then reapply them to both old and new data to generate a more fulsome picture of the object of inquiry.
I completed this work as a single coder. Within a grounded theory approach, several researchers have argued that working as a single coder is preferable as it allows the researcher to bring their background assumptions, intuition, and interpretations to add insight and depth to the analysis (Bradley et al. 2007). This approach has been contrasted with the more deductive approaches of quantitative research (Xu & Storr, 2012), which understands evidence as objective and value-free. Rather, the constructivist underpinnings of grounded theory recognise that all data are interpreted from a particular vantage point. As a result, the researcher is expected to engage in a process of sense-making drawing on the entirety of the research experience and ultimately acting as the research instrument themselves (Xu & Storr, 2012).
Considering these assumptions, instead of coming to a particular answer or 'objective truth', grounded theory studies seek to unpack the richness of a particular case with a view to generating 'thick description' (Denzin, 1989;Geertz, 1973). Though this term has a degree of controversy (Ponterotto, 2006), it is often understood as being the application of verisimilitude to the write up of results. This means that processes, people, and practices are described in detail and as a way of abductively analysing the particular object of inquiry. As part of this, the underlying context and 'webs of meaning' are interrogated to understand how they frame the actors and institutions involved in a particular case. In this way, thick description leading to theoretical inference is a hallmark of grounded theory methodology. Finally, the site or level of analysis was the interactions between consultants and coalitions. Indeed, one of the oldest and most entrenched concepts within general theories of policy influence is that of the coalition (McCool, 1998). As part of this, it is generally accepted right across the 'ontological divide' that coalitions are fundamental in shaping the design and delivery of public policy (Béland & Cox, 2016;Hajer, 1997;Sabatier, 1988). This made the consultant-coalition interaction fertile ground for interrogating the concerns of the study. Moreover, coalitions are not explored in much depth in the 'consultocracy' literature (but see Sturdy, 2018). As a result, unpacking the consultant-coalition interaction had the potential to add additional novelty to this burgeoning body of work.
In line with a grounded theory approach, I drew on an interpretivist-constructivist understandings of a coalition. More specifically, I adopted Hajer's (1997) notion of a 'discourse coalitions'. A discourse coalition is 'the ensemble of a set of story lines, the actors that utters these story lines, and the practices that conform to these story lines, all organized around a discourse' (Hajer, 1997, p. 47). This means that actors in a policy sub-system will coalesce around collective expressions of the interpretations they give to social and political phenomena. It is these shared narratives that mean discourse coalitions form in the first place, remain stable over time, and work collaboratively for collective action.

Three discourse coalitions
Three discourse coalitions (Hajer, 1997) formed part of the process under investigationthe 'efficiency' coalition, the 'system reform' coalition, and the 'Local Blame' coalition. The efficiency coalition was the first coalition to emerge in the case. It was a central part of the NSW Government's efforts to implement its 'Better Services and Value Plan'. This coalition was charged with trying to understand how to best make efficiency gains within the NSW public education system. As part of this, the coalition engaged consultants from BCG and PwC to conduct an expenditure review of the NSW Department of Education and Training. The 'system reform' coalition was then charged with implementing the findings and recommendations arising from the BCG and PwC reports. However, this coalition generally rejected the consultants' work. Instead, the coalition drew on the report of another consulting firm-McKinsey & Co. -in its sense-making processes. Finally, the 'Local Blame' coalition emerged in direct response to some of the findings and recommendations from the BCG and PwC reports. Key members of this third coalition began advocating against the reform work being completed by the Education Department. In this way, the consultant reports became crucial aspects of the coalition's advocacy efforts.
For each coalition, the expressed concern was how best to improve the performance of the NSW public education system. However, what this meant in practice differed between each of the coalitions. For the efficiency coalition, this was to be achieved by a more efficient system, with efficiency defined in highly managerialist terms. Alternatively, the 'system reform' coalition wanted a suite of reforms that were driven by educationalists and not managerialists. This reflected their training and experience either as former teachers, school principals, or long-term employees of the Education Department. Finally, the 'Local Blame' coalition generated a counter-narrative that argued for maintenance of the status quo. As part of this, the coalition argued that 'Local Schools, Local Decisions' would threaten teacher pay and tenure and, as such, should be opposed. Thus, each coalition had a very different understanding of what 'ought to be done' about public education in NSW and the best route to get there. In other words, each coalition generated its own story line as a way of interpreting the problems facing the NSW education system and the most appropriate ways to address those problems.

The efficiency coalition
As noted above, the Better Services and Value Plan emerged in the 2009-10 state budget as a response to projections from NSW Treasury that the state was facing significant shortfalls over the forward estimates period. Given this, Treasury was the logical agency to take on a lead role. Over time, officials from Treasury and the NSW Education Department coalesced around a storyline of efficiency. Within this storyline, the coalition argued that, to achieve better student outcomes, the NSW public sector needs to be made more efficient. The particular understanding of efficiency here was an input-output model-what Downs and Larkey (1986) have called 'managerial efficiency'. This aimed to provide the same level of service delivery for less money. As one member of the coalition described it: … in this case, we were looking to improve efficiency with no impact on services. TREASURY OFFICIAL One of the Treasury's central tasks was to deliver 'line-by-line expenditure audits' of government agencies-commonly referred to as expenditure reviews. However, Treasury did not have the in-house skills to deliver this task. As a result, the coalition was forced to turn to external consultants. These were sourced from a pre-qualified panel that contained the 'usual suspects', which included many of the large, multi-national firms such as BCG, PwC, or KPMG. In addition to the members of its own department and the consultants, officials from NSW Treasury engaged representatives from the NSW Department of Education and Training. All actors worked together in what was described to me as 'collaborative teams' that: … would brainstorm. We would put ideas on a whiteboard, we would capture those on an Excel spreadsheet, we would then go through consultations, talking to all those different kinds of stakeholders, saying 'well this is our initial 10, 20, 30 ideas, which of these might have merit' … INTERVIEWEE 6 Due to its lack of in-house skills, Treasury required the consultants to 'up skill' Treasury staff in some of the methodologies that were needed: We didn't have the skills or resources … part of it was genuinely about building our own skills sets, and making sure that our people were becoming competent in the methodologies and approaches and the analytics required … TREASURY OFFI-CIAL To a degree, then, Treasury was reliant on the expertise that the consultants could provide. This became crucial in giving consultants access to the process in the first place: … we were very clear that we were trying to source expertise and subject matter knowledge, that the people whose CVs were in the back of the pitch were really important to us and we would weight heavily towards the skills and capabilities of those people … TREASURY OFFICIAL But it was not just their expertise. In addition, it was their identity as 'outsider' that was particularly beneficial in shaping the consultants' ability to influence the process. In a document titled Introduction to Expenditure Review, Treasury states that consultants can bring in 'an independent perspective to complex issues'.
Moreover, this 'outsider' status meant that Treasury could use the consultants to move around roadblocks from the Education Department. These roadblocks related specifically to access to Education Department data. One of the central issues that Treasury had in conducting the expenditure review without outside assistance was accessing the relevant data. These data were necessary in allowing Treasury officials to analyse any reform proposals; to determine the feasibility of such proposals; and to ascertain whether the proposals would be effective. As one Treasury official told me: … part of the [expenditure review] process was to make sure we were getting the data we needed out of the agencies. Actually getting data from the agencies in a useable format is quite difficult … TREASURY OFFICIAL Moreover: … Treasury always feels like things are being hidden from them. Now, whether that's true or not, it's a perception on Treasury's side because they never feel that they have the intellectual horsepower to match the department ... EDUCATION OFFICIAL Hiring the consultants was one method through which Treasury could circumvent these issues to get access to the kinds of data that they needed: … we ran into some resistance from a number of education staff about access to data … but I think [PricewaterhouseCoopers] were able to help drive getting that access … TREASURY OFFICIAL Ultimately, the reports produced by both PwC and BCG understood the NSW public education system in highly managerialist terms. As part of this, schools were likened to small businesses, with principals acting as CEOs. For instance, on page 34 of the BCG report, the consultants recommend that principals be given extra training to effectively manage their 'business'. Likewise, on page 18 of their report, PwC argues that employing principals has been too heavily focused on educational leadership, at the expense of other management functions. Because of this, PwC recommends that principals be provided with training to be able to better manage their schools. Thus, both consultancies attempted to reframe the function and identity of the NSW school system away from a focus on education and towards a focus on management.
This approach aligned strongly with the vision that Treasury had for the future of the education system. For instance, for many years before the expenditure review, Treasury had established what it referred to as 'game plans'. Part of these 'game plans' was a policy reform known as 'school devolution': … in Treasury, we had game plans -how would you see us in five years' time? It was high-level strategic stuff … we definitely had devolution as one of the strategies … and that was before we'd even done the Better Services … TREASURY OFFICIAL The basic idea underlying school devolution (a.k.a. school autonomy, site-based management, or principal autonomy) is that certain decisions are made by the school instead of a central bureaucracy like Treasury or the Education Department. Devolution here can take one of two forms. The first relates to decisions over teaching and learning (i.e. curriculum, assessment, etc.) while the second relates to management (i.e. resource allocation). Evidence suggests that the former has positive impacts on student outcomes, while the latter does not (OECD, 2009). The form favoured by BCG and ultimately endorsed by Treasury very much related to a more managerial understanding of, and approach to, education. In this way, the concepts and language used by the consultants very much aligned with Treasury's managerialist understanding of how best to reform the NSW public education system. Moreover, the consultants' identity as outsider not only gave them access in the first place, but also meant that Treasury was more willing to trust their advice and recommendations.

The system reform coalition
The efficiency coalition had instructed the Education Department to implement the findings and recommendations of the BCG and PwC reports. To do so, the Department established a Better Services Program Office (BSPO). The BSPO's membership included experienced educators, senior departmental staff, and others who were bright 'up and comers' within the Education Department. None of the members of this office were part of the efficiency coalition, nor did they have any direct contact with consultants from either BCG or PwC. However, after careful review, members of the BSPO decided that the reports were largely unimplementable. This was because the ideas and narratives represented in those reports were far too embedded in the managerialist thinking that had been represented by BCG and PwC. As one interviewee explained: … [the BCG report] was a report that could only work in a board room of some sort of corporation that had no external stakeholders other than its shareholders … EDU-CATION OFFICIAL Instead, the coalition embarked on its own line of inquiry to interpret and understand problems and potential solutions within the sector. As part of this, BSPO members recognised that they would need to interpret the education sector in ways that were more aligned with how educators-and not the efficiency coalition-understood the education system. This was described to me by a key member of the BSPO as follows: … we had to have a model of solving the problem that was internally generated and bought into … [so that we could] recognise and address the things that had prevented some of these things from happening in the past … EDUCATION OFFICIAL As part of this, the BSPO engaged widely with experts right across the education portfolio to 'think deeply' about the future of the education system in NSW. This meant that members of the coalition: … realised that the only way to get to [a better education system] was to have an educational reform program that was driven by educators, and not by Treasury analysts, so we conceived a new reform that had some connections to the old ideas but was an entirely new creature … EDUCATION OFFICIAL This meant that actors in the BSPO felt that they needed a new understanding of the problems facing the education sector, and how those problems might be resolved-in other words, they needed a new narrative. This narrative was no longer about efficiency through cutting costs, but rather about how to create system-wide improvements in NSW public schools. As part of this, actors from right across the Education Department met to discuss the future of NSW public education: … [there] were people from across the organisation in various roles, people in central roles, people in front-line roles, we had regional directors, a group of really smart people, who had different perspectives and wanted to make the system a better place … EDUCATION OFFICIAL As part of this, members of the BSPO developed a narrative that was more about system reform than efficiency per se. Through constructing its own narrative, a new discourse coalition emerged-what I refer to as the system reform coalition. This coalition recognised that it would need to understand the 'root causes' of any problems, and any barriers to addressing those causes. This new understanding was not encapsulated in issues related to reduced revenue because of the global recession, but rather through the phrase 'authority, funding, and accountability'. Through their investigations into the underlying causes of the problems facing NSW public education, key members of the coalition began to interpret the root causes as a misalignment between these three key concepts. Indeed, from the system reform coalition's point of view, significant improvements could be made to the NSW public education system by addressing this misalignment. The main issue here was that the person who was accountable for a decision was not the same person who had authority to make that decision, nor were they the same person who had control over the funds to implement that decision.
This new narrative ran contrary to that of efficiency. As a result, the ideas from BCG and PwC were further marginalised. Despite this, the ideas and narratives of another consultancy proved quite influential on this group. This was McKinsey & Co.'s 'Good to Great' report. First published in 2010, the 'Good to Great' report analysed 20 public education systems from around the world, all with different levels of 'performance'. It did this to understand how school systems can improve their overall performance from 'poor' at one end to 'excellent' at the other. To do this, the report examines the kinds of interventions that individual school systems have undertaken to register 'significant, sustained, and widespread student outcome gains' (Mourshed et al., 2010). In this way, the main substance of the report explores similar issues to those that were being considered by the system reform coalition. The report frames the interventions within an 'improvement journey', which links different initiatives and programmes to different stages of system reform and performance-from 'poor' to 'great'.
McKinsey's 'improvement journey' meant that, in the view of the system reform coalition, McKinsey had produced a clear set of actions that the coalition could incorporate into their understanding of how to reform the NSW public education system: McKinsey gave us a way to say 'ok, yeh, we do agree that the best place to be is highly skilled educators making great decisions on the ground, we all support that, but there is a pathway to get to that, and here is the groundwork that you have to do to create a system that can operate with that degree of freedom'. So that was real value from my perspective of that McKinsey report, it gave us a framework to talk about the reforms had gone before, to position that as part of a journey and a story about where the system was going, and to be clear that over multiple phases, that's where we were headed, but there was some interim work to be done … EDUCA-TION OFFICIAL In other words, in the eyes of the coalition, McKinsey generated a much clearer 'implementation story': … it was more a matter of having a new lens through which to look at that, and what it really helped us with was telling an implementation story, about how might we get from where we are to more being the system we want to be … EDUCATION OFFI-CIAL This meant that McKinsey was of much greater assistance to the coalition than either the BCG or PwC reports: … so [the Good to Great report] became more of a thread in the argument, much stronger in fact than anything that had come out of BCG or PwC … EDUCATION OFFICIAL But it was not just McKinsey's narrative that influenced the coalition. Rather, the identity of its lead author-Michael Barber-was of particular significance. Mr Barber had served under former British Prime Minister Tony Blair. During Blair's first term, Barber worked as chief advisor to the Secretary of State for Education. During Blair's second term, Barber served as Chief Advisor on Delivery. Barber's work had given him somewhat of a celebrity status in the NSW education portfolio-at least in the eyes of those in the coalition: Michael Barber was kind of a celebrity at the time in the education space, he was the education messiah, and having his name on [the Good to Great report] helped a lot … EDUCATION OFFICIAL Moreover, Mr Barber had developed a reputation as being an authority on policy implementation through the release of several books, including Deliverology 101 . As one education official told me: I think 'Deliverology' was making a bit of a splash at that point … we had the Premiers Delivery Unit stuff happening, there was a whole bunch of stuff going on that was Barber-adjacent in New South Wales, and so [the 'Good to Great' report] had a lot of currency. I think he'd been out recently, so there was a fair bit of noise around him … EDUCATION OFFICIAL This meant that the coalition was willing to read the McKinsey report in the first place and therefore bring McKinsey's narratives into how the coalition was framing and understanding the work that needed to be done to support systemic improvement.

The 'Local Blame' coalition
The final coalition is the 'Local Blame' coalition. This coalition was made up of officials from the New South Wales Teachers' Federation (NSWTF) (i.e. the teachers' union), as well as education activists, left-leaning academics, and politicians. As with the other two coalitions, this one was bound together through a shared interpretation of the problems facing the NSW education system. However, in the view of this coalition, the problem was 'Local Schools, Local Decisions' itself. The coalition's main objection to this reform was that it was perceived to be just another version of school devolution. In the coalition's perspective, moves towards devolution would threaten the jobs of many members of the teaching workforce. It would also shift political risk away from government and onto schools, 1 3 thus creating 'local blame' (Kaye, 2012). As a result, the coalition generated its own counter-narrative as a strategic response to the threat posed by 'Local Schools, Local Decisions'. This counter-narrative was generated through a series of meetings and presentations from key members of the coalition, thus creating the space for the coalition to form.
While the coalition's objections to 'Local Schools, Local Decisions' were often couched in the language of student outcomes, it was deeper concerns about changes to pay and entitlements that also played a significant part in their sense-making processes. In a submission to the consultation on 'Local Schools, Local Decision', the NSWTF discusses an earlier trial of devolution at 47 schools in NSW (known as the '47 Schools Trial'). Here, the NSWTF argues that: [The 47 Schools Trial] is part of a cost cutting strategy that aims to undermine and dismantle the existing formula-based approach to determining school staffing entitlements. The long-term aim is to have the size of the school's staffing budget determine the number and mix of staff, rather than a government guaranteed staffing formula that ensures equitable provision for all students (Diamond, 2011, p. 3) As part of this argument, the coalition again draws on the BCG report as a way of supporting their interpretation of government intentions: [The NSWTF's] opposition to the [47 Schools Trial] has been vindicated by Boston Consulting Group's 'Expenditure Review of the Department of Education and Training (DET) -Initial Scan' prepared for the previous NSW Government and leaked to the media a week prior to the state election on March 26 this year (Diamond, 2011, p. 3) Indeed, the coalition's interpretation of the 'Local Schools, Local Decisions' process was due to an interpretation of consultants as having a specific agenda of cutting costs within the public sector. As part of this, consultants were seen as being a central-if not the most important-actor. As a key figure in the 'Local Blame' coalition described it: … that's why [governments] employ companies like Boston Consulting, Pricewater-houseCoopers, because … this is their [i.e. the consultancies'] bread-and-butter … to destroy from within public services … they are hired guns, and their job is to establish the pre-conditions for privatisation, corporate takeover and the diminution of public provision, that's what they exist for … UNION OFFICIAL In this view, consultants represent some kind of 'power behind the throne'. That is, such an interpretation would suggest that policymakers have limited agency and power as compared to the consultants. Indeed, after I asked the same interviewee about the consultants' role in 'Local Schools, Local Decisions', he answered (in no uncertain terms) that the consultants designed the entire thing.
Essentially, then, the 'Local Blame' coalition was concerned about a shift in power. By providing principals with more authority, by giving them greater power over decisions related to hiring and firing of staff, budgets, and contract administration, there was a very real threat to the influence of teachers. While the coalition's counter-narrative was couched in normative language about student outcomes, shifts in the locus of power within NSW schools proved just as significant a concern. In the eyes of the 'Local Blame' coalition, power should remain with teachers and/or the department to make these kinds of decisions.
Crucial to this view was two things. First, an assumption that the public sector is always moving towards and is privileging private sector understandings of the world and, second, that consultants are the primary actor through which these changes are made possible. In this way, the coalition was using the identities of consultants as a political tool to support their counter-narrative. The consultants involved in this case clearly had an impact on various aspects of the process, but in different ways and to varying extents. BCG and PwC had the most significant impact on the efficiency coalition. As part of this, the coalition drew on the consultants' narratives and metaphors as a tool through which they could direct the actions of the Department of Education and Training. At the same time, BCG and PwC were unable to shape the actions of the system reform coalition. Instead, this second coalition drew on the work of McKinsey & Co. to understand how best to shape the institutional arrangements within the NSW public education system. Finally, neither BCG nor PwC were influential directly on the 'Local Blame' coalition. Instead, this coalition had already formed a counter-narrative about the purpose and effect of 'Local Schools, Local Decisions'. The leaked consultant reports were then used as a way of evidencing this counter-narrative. This case, therefore, demonstrates the varying ways in which consultant influence on public policy can vary-even within a specific reform process. The first factor that explains this variability was the openness or permeability of the different coalitions to the presence of consultants in this reform process. The efficiency coalition did not have the skills or experience to drive the expenditure reviews themselves. As a result, they were forced to rely on consultants to get the work done in the first place. This meant that the coalition was genuinely open to the presence of consultants-they had to actually get the work done. On the other hand, the system reform coalition was in the luxurious position of being able to choose the ways in which consultants were to be used in their reform efforts. As part of this, the coalition wanted a systemic reform that was 'driven by educators' not 'Treasury analysts'. In this way, the coalition was open to consultants generally, just not those from BCG or PwC. Finally, the 'Local Blame' coalition was arguing against reforms contemplated by 'Local Schools, Local Decisions'. This coalition developed a counter-narrative that positioned consultants as vehicles for Treasury to centralise control over the education system. In this way, the coalition was not open to acting on the ideas of BCG and PwC, nor did they consider that the NSW Government should be acting on the consultants' advice. Instead, these were used by the coalition as argumentative ammunition in support of their own counter-narrative.

Varieties of influence
The second factor that was relevant to the influence of consultants in this case was trust. For each coalition, the ways in and extent to which the various consulting firms were trusted was quite different. The efficiency coalition had no prior relationship with either BCG or PwC. However, the procurement process gave Treasury the opportunity to scrutinise the skills, background, and experience of the various consultants and ensure that they were the 'right people for the job'. Importantly, the consultants were also able to draw on the branding of their respective firms as a way of signalling to Treasury that they could be trusted to be involved in the process. For the system reform coalition, the identity of Michael Barber was a crucial factor in the coalition's willingness to consider McKinsey's ideas and narratives. In this way, he was a trusted personality and someone worth taking seriously. Finally, for the 'Local Blame' coalition, the consultants were not trusted at all. As a result, the firms did not influence this final coalition but instead the coalition sought to fight against whatever the consultants were arguing for. In this way, the consultants' identity became a major factor in the extent to which they were trusted.
The third factor is what I call 'resonance'. Put simply, resonance refers to the degrees of alignment between the narratives and metaphors put forward by a consulting firm and those of the client and/or coalition. The influence of BCG and PwC on the efficiency coalition was very much due to the underlying narrative of that coalition being couched in managerialist terms. Indeed, the very notion of efficiency that drove this coalition was embedded in discourses of scientific management. BCG and PwCs ability to couch their ideas in a similar language was very much a factor in whether the coalition took those ideas seriously or not. However, it also meant that the system reform coalition largely rejected the BCG and PwC reports and instead drew on the 'reform journey' put forward by McKinsey. Indeed, the language in the 'Good to Great' report draws on the discourse of education and system reform, rather than the language of management and accounting. As a result, it was far more influential on this coalition than the BCG or PwC reports. Finally, for the 'Local Blame' coalition, the existence of the managerialist language in the BCG and PwC reports again supported the counter-narrative-that 'Local Schools, Local Decisions' was about shifting the power dynamic away from teachers and towards principals and school management.
In this sense, we can identify resonance as an additional (and important) mediating factor between consultants and coalitions. As part of this, consultants are likely to be more influential when their ideas, narratives, and metaphors resonate with those of the client's coalition. This means that consultants need to be adaptable to the underlying worldviews and mental models that shape the ways in which coalitions understand the world. This involves understanding the fundamental philosophies that permeate through the coalition's sector or policy sub-system, and how those ideas are communicated by people in positions of power. It also means using the same key concepts that the coalition uses to explain and interpret their world, as well as understanding how those concepts are operationalised in practice. Finally, it requires demonstrating shared goals, shared values, and a coherent vision of some future state-as was clearly the case with Michael Barber and the system reform coalition.
Finally, the findings from this paper suggest that the three factors-openness, trust, and resonance-are interrelated. First, it demonstrated that coalitions may be more open to the presence of a trusted consultant (i.e. Michael Barber) and may be more willing to take the ideas of that consultant seriously. At the same time, just being trusted would not necessarily be enough for the consultant to be influential. Even if the consultant was trusted, their ideas may be terrible, which is likely to erode trust and thus make the coalition less open to engaging with that consultant in the future. Likewise, even if the consultant is trusted and their ideas are strong, it may be that the coalition is not open to engaging with the consultant, either due to lack of funding, institutional reluctance, or restrictive procurement rules that prevent consultants being hired in the first place. In this way, there is no single factor that is most important. Instead, the three factors should be seen as a 'reinforcing triangle'-all three must be present for consultants to be most influential.

Conclusion
This paper has sought to understand the ways in which consultant influence on public policy varies. Focusing its analysis at the level of the coalition, the paper has shown how openness, trust, and resonance are key mediating factors between consultants and coalitions. Its main contribution has been to introduce the concept of resonance and to show that consultant ideas, narratives, and metaphors are more likely to have an impact when they can demonstrate resonance with the underlying mental models of a coalition. This builds on previous studies in the newly emerging consultocracy literature. The previous studies have shown that openness and trust are crucial in explaining the variability of consultant influence on policy processes. As part of this analysis, organisations and actors have to be willing to incorporate a client into a policy process, and they also need to trust those consultants. These are necessary but not sufficient. For a consultant to truly shape the actions of a coalition, their advice also needs to resonate.
These findings contribute to other works in several ways. First, the findings speak to a large and well-established literature on management learning (Kantola & Seeck, 2011;Sahlin-Andersson & Engwall, 2002;Sturdy, 2004). Contributions in this field have a central focus on the ways in which management ideas permeate into and through organisations and how they contribute to fads or fashions (Butler, 1986;Grint, 1994;Sturdy, 2004). This literature speaks critically of consultants' use of 'rhetorical devices' as a way of reinforcing client anxieties (Sturdy, 2018, p. 80). In this way, there is a strong emphasis on consultants exercising discursive power over their clients. The concept of 'resonance' nuances the idea of rhetorical devices to make space for the ways in which client power and agency might affect the uptake of ideas. Indeed, and as my case material has shown, a client does not have to accept the negative narratives or ideas propagated by consultants. Likewise, the same literature describes 'cognitive distance' as an important concept through which we can understand the alignment between shared interpretive frames of different people and organisations (Nooteboom et al., 2007;Sturdy et al., 2009). Resonance is one way through which we can identify cognitive distance. Indeed, my case material demonstrates that it is more difficult for consultant narratives to resonate when there is high degrees of cognitive distance-as was the case with BCG and the system reform coalition. Finally, literature works on policy advisory systems have focused on notions of 'ideational compatibility' between advisors such as consultants and policymakers (Craft & Wilder, 2017;Laage-Thomsen, 2021). The findings in this paper align with those of the ideational compatibility model insofar as they also demonstrate openness and trust as being key mediators of influence. At the same time, this paper contributes to the policy advisory system literature by demonstrating how resonance might be another factor in creating cohesion within an advisory system. These findings also have implications for practitioners and researchers. Consultants can use the framework of openness, trust, and resonance to understand how they might be most impactful on public policy. More specifically, consultants might seek to build trusting relationships with organisations that regularly use consultants. Importantly, actors already enmeshed in a sector and with a pre-existing relationship would also be particularly valuable for consultants seeking to be influential. In addition, consultants need to deploy a discursive repertoire in a way that demonstrates discursive affinity with the concepts, language, and discourse of the coalition with whom they are working. For researchers, future research might focus on operationalising the framework developed in this paper. Quantitative metrics might be established that allow us to explore how much trust is needed, or ways to measure the extent of resonance. Likewise, the case study is from Australia and so case studies from other countries might be helpful, particularly those who have not pursued the NPM so vigorously.