Strange magic: what can the emergence of ‘magic concepts’ tell us about policy implementation?

Abstract Pollitt and Hupe have identified the emergence of “magic concepts” in public administration and policy. These concepts are seductive but do not solve – and often render invisible – important policy challenges. In highlighting the role of magic concepts, Pollitt and Hupe demonstrate the importance of linguistic battles within the bureaucracies. In this paper we examine the emergence of magic concepts in the context of the Australian National Disability Insurance Scheme. We analyze semi-structured interviews with policymakers to elucidate their use of magic concepts – for example, the new NDIS regulatory body – for addressing implementation challenges in the NDIS. This paper examines implementation of the NDIS through the application of Pollit and Hupe’s “magic concepts” to a context-specific case. We argue that paying attention to where magic concepts emerge is critical – they can alert us to areas that require further inquiry and where problems are emerging in governance, practice, and policy implementation.


Introduction
"Never trust anything that can think for itself if you can't see where it keeps its brain." JK Rowling, Harry Potter and the Chamber of Secrets Language can give shape to the way that a public problem is perceived, and the solutions suggested for addressing it. Thus the study of language in public administration has strong foundations (Fairclough 2000;Pollitt and Hupe 2011;Fairclough 2013). We draw on these foundations in our examination of the language used to describe implementation challenges and potential regulatory solutions in the newly implemented Australian National Disability Insurance Scheme (NDIS).
This article uses the case of the Australian National Disability Insurance Scheme (NDIS) and the establishment of a new regulatory body for oversight of the scheme to explore the ways in which structural solutions are sought to complex problems.
More specifically, we show that there is a tendency to pose "magic concepts" (i.e. structural solutions) as a means by which to solve a range of governance and implementation issues. While it is increasingly recognized in the scholarly literature that structural solutions to organizational performance and governance challenges are often inadequate or problematic, they can still frequently be found in practice .
We argue that the creation of the new NDIS regulator equates with Pollitt and Hupe's (2011) notion of a "magic concept". Magic concepts in public administration are: "very broad, normatively charged and lay claim to universal or near universalapplication" (Pollitt and Hupe, 2011 p.643). As concepts, they are defensible and allow little room for opposition, but are simultaneously so vague that they do not necessarily solve underlying problems. Magic concepts have rhetorical value in that they can "facilitate new orientations and frameworks, initiate the launch of new research strategies, stimulate campaigns for additional resources, enthuse staff, and assist a number of other developments." (Pollitt and Hupe 2011: p652). In this paper, we argue that magic concepts can be deployed on context-specific levels to avoid tackling challenging problems and further muddy already complex governance arrangements and policy challenges (Pollitt and Hupe 2011). Where and when magic concepts are deployed can give insight into the perennial problems that plague public administration, and identifying them can help to support clarity in structural decision making. Pollitt and Hupe (2011) outline four characteristics of a "magic concept" in public administration: 1. Broadness: they are widely applicable, have a wide scope and high valency. 2. Normative attractiveness: overwhelmingly positive connotations, it is hard to argue against them. 3. The implication of consensus: they obscure conflicting interests and logics. 4. Marketability: the concept is known and used by the practitioners and academics, frequently appearing in communications material and referred to as solutions.
Other magic concepts identified in the literature include, "performance", "participation", and "innovation", all of which can be argued to meet the four criteria above (Pollitt 2003;Kettl and Kelman 2007;Pollitt and Hupe 2011). Pollitt and Hupe (2011) provide a closer examination of three other magic concepts which are widely used in policy: "governance", "accountability" and "networks", Through an analysis of the scholarly and grey literature they found that these terms are ubiquitous but deployed in normative ways whereby they are presented as wholly positive, but provide little if any specific direction. Pollitt and Hupe (2011) discuss magic concepts at a high level, identifying concepts that are ubiquitous across all areas of public administration, however, the notion is also useful on a context-specific scale to identify magic concepts within a given program, reform or set of practices. We think that many practitioners will be able to image such "magic concepts" within their own spheres of influence. Here, we have used Pollitt and Hupe's (2011) criteria to identify the use of a context-specific "magic concept" within the context of the implementation of the Australian NDIS.

The Australian National Disability Insurance Scheme (NDIS)
The Australian National Disability Insurance Scheme (NDIS) is Australia's first largescale national policy utilizing personalized budgets, passed into legislation in 2013 (Needham and Dickinson 2018). Under the NDIS, approximately 460,000 individuals who have a significant and permanent disability will receive personalized funding to pay for their care (Productivity Commission 2011). The scheme is due to be fully implemented across Australia by 2019 (Productivity Commission 2011;Collings, Dew and Dowse 2016).
The creation of the NDIS has involved a significant shift of responsibilities that have traditionally been the concern of state governments and territories, up to the Commonwealth level. This includes shifts in funding, safeguards, and regulatory protections for people with a disability and service providers (Productivity Commission 2011). Specifically, two new entities have been created as the key sites of governance and regulation for the NDIS. These are the National Disability Insurance Agency (NDIA) and a new regulatory body, which is yet to be given a name. The NDIA is charged with the implementation of the NDIS, including the creation and administration of funding packages to individuals, and market stewardship (Productivity Commission 2011;Carey, Dickinson, et al. 2017). The establishment of these bodies is part of a process of creating national consistency regarding the regulation and oversight of disability services, consistent with the shift from a state-based model to a federal model.
The new regulatory body that is being established for the NDIS is envisaged to play a wide range of oversight roles, taking over responsibilities from the states and standardizing these across jurisdictions (Commonwealth Department of Family and Community Services 2017). This regulatory body was not envisaged in the design of the NDIS, but rather emerged in response the demand for greater regulation of the sector and a nationally consistent approach (Commonwealth Department of Family and Community Services 2017). At present, how this body will be structured, its precise roles and how they will be shared or demarcated with regard to the NDIA is under consideration.
Since its launch, the NDIS has been plagued by the implementation challenges (ANAO 2016;Carey, Kay, and Nevile 2017;Carey, Dickinson, et al. 2017). These have included significant IT issues, capacity issues with the NDIA and accountability concerns (ANAO 2016; Malbon 2017; Productivity Commission 2017b). The establishment of the regulator has been proposed in response to the implementation issues thus its examination offers insight into the implementation problems of the NDIS and their perceived solutions.

Methods
This paper draws on data from a longitudinal study of the implementation of the NDIS (author reference). The aim of the study is to examine the experiences of largescale policy design and implementation in real time over a period of five years. The study utilizes a case study research design, enabling in-depth investigation (Yin 2014), across time. Semi-structured interviews (N ¼ 32) were conducted with key policymakers in the Commonwealth government charged with the design and implementation of the NDIS working in the Department for Social Services (DSS). Interviewees were responsible for designing and implementing the governance and regulatory architectures for the NDIS. Participants were identified by the Deputy Head of DSS on the basis of their role and expertise and were at the executive and senior executive level. Interviews were semi-structured and featured a range of questions designed to explore experiences of policy design and implementation regarding the NDIS. These included broad questions about implementation, governance, current challenges the scheme is facing as well as more specific questions regarding the establishment roles and responsibilities of the new regulator, the roles, and responsibilities of the NDIA, accountability mechanisms and market management.
Interviews were recorded and transcribed verbatim. We used Pollitt and Hupe's (2011) framework to identify where magic concepts were being deployed within discussions of the NDIS's implementation. The framework identifies four characteristics of magic concepts: broadness, normative attractiveness, the implication of consensus and marketability. We would add to this frequency of use. Once the regulator emerged as a potential magic concept we undertook purposive sampling to more deeply analyze how it was being described and used. We found that participants discussed six overlapping functions that the new regulatory body would perform. These coincided with the major implementation challenges presented to date by the scheme (see ANAO 2016; Malbon 2017; Productivity Commission 2017b).

Findings
Our interviews with Commonwealth policymakers found the concept of "the regulator" has come to be a "catch-all" for the problems that the NDIS is currently experiencing. For example, some of the roles that the NDIS regular has been said to take responsibility for include: safeguards on the scheme workforce, market stewardship, auditing, complaints, and provider registration. Some of these roles, such as safeguarding the workforce, are presented as the sole responsibility of the (currently nonexistent) regulator. Other roles appeared to be shared between the NDIA and the regulatorsuch as "market stewardship". The identification of the roles that the NDIS regulator is said to solve provides insight into policymakers' representations of major implementation challenges for the NDIS and highlights the lure of structural solutions.
In total eight overlapping roles for the new regulator were discussed, which are explored in turn below.

Ensuring workforce quality through worker screening
One of the most important roles envisaged for the regulator is worker screening and qualifications to protect against poor practices in the sector. There has been a great deal of concern over responsibility and accountability surrounding the workers within the NDIS scheme (Commonwealth Department of Family and Community Services 2017; Malbon 2017). As one policymaker explained: Worker screening is a component of the overall regulatory picture. The states and territories will be responsible for that. For some of them it will be an extension of their existing, like working with vulnerable children, vulnerable people check in the A.C.T … . And the regulator has a responsibility then to, I guess, essentially making a report in if there are concerns about workers … This is an industry where often the most vulnerable participants are supported in the most unseen private spaces, and their risk is higher. GP_30 For this participant, the introduction of the new regulator means that responsibility for worker screening will be lifted from the states and brought under the Commonwealth. It is worth noting however that state-based schemes, such as working with vulnerable people checks, will remain the responsibility of state governments (Commonwealth Department of Family and Community Services 2017). Hence coordination will continue to be needed between state governments and the new regulator. While the regulator may solve some issues regarding worker screening, greater complexity will persist in coordinating state and national practices, reducing the "magic" nature of the regulator.

Guard against risky providers and ensure provider standards
Earlier experiments in marketization in Australia demonstrated the potential for "risky" providers to enter the market (Considine 1999). That is providers who seek profit at the expense of participants/consumers. In the care sector, the risks are high, as experienced within the aged care sector where breaches of human rights and death have occurred (Reynolds 2000). This participant envisaged the regulator's role to extend from the individual employees through to the organizations in order to prevent these types of predatory behaviors: The bad thing happens. They drop the person, some terrible assault, whatever it is. Who's accountable? So, if they're employed by [an organisation] or whatever, then the government goes to [that organisation], and it goes, what the hell? Was the person supervised properly? Were they trained properly? What were your assurances around that? What went wrong? Was this just a completely random rogue individual? Or is something systemic happening with your organisation? … And we'll hold the firm to the greater extent somehow accountable for what happened … The job of a regulator is to check whether that's the case or not. GP_20 Similarly, the regulator was seen as necessary because of a lack of shared information between state governments about poor providers: The States do not share at the moment, they do not share at all information about what they're doing in relation to bad behavioural practice of any of the others. G_P26 Finally, one policymaker envisaged the role of the regulator to extend as far as ensuring providers functioned effectively in the market: "the regulator's responsibility is more around the individual providers and making sure they're functioning effectively" (G_P32). These varied roles of the regulator rely on the normative notion that more regulation of the sector is an inherently good thing and will circumvent problems emerging in the Scheme from individual harms through to organizational functioning. Here the regulator has considerable marketability in relieving community concerns about the risks of the scheme.

Market stewardship
Responsibility for market stewardship has been an area of growing concern regarding the NDIS (Carey, Dickinson, et al. 2017;Productivity Commission 2017b). As noted in a recent report by the Australian Productivity Commission (a premier policy advice body): To date, market stewards have not given market supply issues sufficient and timely attention. This may be partly due to ambiguity about who is responsible for market development, what market stewardship entails, why it is needed, and lack of capacity or impartiality of governments to undertake market stewardship functions. As such, 'the system for identifying and addressing emerging issues is currently informal and unfixed in the NDIS implementation. (Productivity Commission 2017b, p.394) Market stewardship involves protection against market gaps and thin markets through monitoring, steering, supplementation, and active diffusion of the best practice (Gash 2014; Carey, Dickinson, et al. 2017). Policymakers identified three different aspects of market stewardship the regulator would, or could, be involved in. At the broadest level, policymakers felt that the regulator has an important role in market stewardship: "the response to a lot of the things in the market will be what powers are available and their appropriate application by the regulator" G_P20. Moreover, that these would need to be carried out in coordination with the NDIA and could also extend to financial reporting by providers in the market: So, the functioning of the market is clearly a role for the regulator. I'm not even going to go into the detail, because it is what it is. It's also clearly a role for the NDIA who have an important stewardship role around doing that signalling of where there are gaps or opportunities, or spaces to fill. G_P30 The regulator will, I call it almost prudential oversight. So, there is a large provider here, and you're getting consistent complaints about quality. That indicates that provider may actually have some financial difficulties. So, I see the regulator having that role of, prudential oversight sounds too draconian, but they're going to have the market intelligence about where there is an oversupply of providers, where there is behaviour that is resulting in poor quality outcomes. Again, that's an exchange with the NDIA. G_P31.
As we will explore later in this paper, this proposed structure presents a major coordination issue for shared governance over the NDIS between the NDIA, the regulator, and Commonwealth government.
As implementation has progressed there has been growing concern over the risk of "thin markets" and market gaps, where an undersupply of service providers persists. Guarding against thin markets is increasingly seen as a major part of the market stewardship function of government (Carey, Dickinson, et al. 2017;Productivity Commission 2017b). Policymakers suggested the regulator would also play a role in identifying non-robust markets: keeping a very close eye where we know we have potential for market failure, and making sure that there's a level of robustness around the oversight there that doesn't hinder organisations, but encourages. When we've got those situations, which we will have, particularly in rural and remote regions where choice is very limited, that those services are doing the best they can. And striving to do better. Rather than potentially being a monopoly in the market. G_P20 One of its roles is to look at the market, and try and look at risks and when you've got a thin market or a failed market G_P26 Part of this role included providing advice on price setting, which is currently carried out by the scheme actuaries (without input from the Commonwealth) (Walsh and Johnson 2013; Treasury 2014): "the regulator will be signaling participants as consumers in the market will send providers. And about the regulator's role in really amplifying that. Both from an individual practice level, and I think that will come up particularly in the complaints space". G_P25 The need for government to act as a "provider of last resort" in the case of thin markets or market gaps has been raised in two recent Productivity Commission reports (Productivity Commission 2017a; 2017b). The Commission noted that "While the NDIA typically does not supply any supports … it has been suggested that, as the scheme scales up, the NDIA might need to act as a provider of last resort for communities where the market is thin." (Productivity Commission 2017a, p.19). In our interviews, policymakers suggested the regulator could determine when the NDIA or government needed to step in as a provider of last resort: The regulator will have information that will be kind of canary in the coal mine. Things look like they might be about to go badly. And I think particularly in those concepts around provider of last resort. G_P30 one of the things it would worry about is the extent to which there are actually solutions in place, such as provider of last resort G_P26 Here the regulator is being deployed as a magic concept to the problem of providers of last resort. In particular, it is obscuring logics regarding how government should deal with market failure through market stewardship. By placing these concerns on a yet-to-be-established regulator, capacity development for the highly vexed issue of market stewardship in the NDIA could be sidestepped.

Complaints
There has been a great deal of public scrutiny over the handling of participant complaint functions within the NDIS (Morton 2017). In part this has stemmed from capacity issues within the NDIA and the fact that, at present, the NDIA is both implementing the scheme and handling complaints from participants (often about the NDIA itself) (ANAO 2016). Handling complaints was the most commented on role for the regulator: The regulators role … will be strongly around those complaints, serious incidents. G_P30 [The regulator] specialises in quality audits and complaints. G_P20 Run the complaints, the critical incidents, the registration function and the restrictive practices kind of functions. G_P28 Regulator, when they come on-board, will also have some market insights because they'll be the register of providers and they will also be getting complaints etc. G_P13 As noted by the last quote, the complaint function is also viewed as giving the regulator market information that would support the market stewardship functions. However, the more crucial function for the NDIS regulator to play in the handling of complaints is to enforce the Quality and Safeguards framework (Commonwealth Department of Social Services 2015) that protects against abuses within the disability care system. Here, the regulator is being deployed to solve pressing market stewardship problems over and above more straightforward regulatory roles.

Registration of providers
All providers who want to enter the marketplace for the NDIS need to be registered and comply with quality and safeguarding registration standards (Commonwealth Department of Family and Community Services 2017). The other commonly discussed and fairly straightforward responsibility for the regulator was registration of providers, which is currently done by the NDIA: The arrangements that are in place now between the states and territories around the NDIA and the registration will change dramatically because what we envisage is that the regulator will have, will be operating all of that registration function. Rather than at the moment it's a bit the states, it's a bit the NDIA G_P28 The regulator will take responsibilities from the states and territories, but also from the NDIA in terms of that just very administrative job of signing a provider up. But states and territories still have responsibility under the framework. So it's not a just hand it over and go. It needs to continue to be quite a collaborative model, going forward. G_P30 The registration of providers is a very procedural and administrative based task, and assigning such a role to the NDIS regulator frames the regulator as taking up some burden of administration from the NDIA and states and territory governments. This is in contrast to the framing of the role of the regulator as providing market stewardship, where we can see it is more clearly deployed as a magic concept to alleviate concern over major implementation challenges.

Guarding against "disorderly exit"
A key policymaker working on the formation of the regulator discussed the notion of preventing disorderly exit from the market. Preventing disorderly exit refers to ensuring that providers exiting the market (for a range of reasons) do not leave market gaps or problems in their wake (Fisher 2016). This could involve the regulator ensuring organizations have market exit plans for organisations above a certain size.
Disorderly exit. So, a provider disorderly exiting for a whole range of reasons, it will be the regulator that sees that starting to happen. G_P30 While the regulator would be unable to monitor all organizations, nor would all departures from the market leave substantive gaps, it was suggested that they would have a role to play in certain situations. Particularly when a provider is of a certain size or the only type in a particular locality: "not all providers present a risk to participants of a disorderly exit. … So we've got to get that balance right." G_P30. Here we can see the regulator being applied again solve market issues, in this case, to guard against market gaps.

NDIA as co-regulator
Our participants were aware that many of the functions identified above by the policymakers for the regulator need to be done in conjunction with the main implementation agency for the NDIS, the National Disability Insurance Agency or NDIA. However, coordinating and sharing roles across the agencies presents boundary and coordination issues (Keast and Brown 2006), further suggesting the regulator is being used as a magic concept. Roles that were said to be shared between the two agencies included market stewardship and ensuring participants can exercise choice: Making sure that participants are active consumersthat's really a very strongly shared role, because the regulator won't have that much touch, except on interacting with participants, except where things are not going well in a kind of complaints, serious incidents way. Whereas the NDIA; that's their bread and butter. G_P30 The NDIA has a strong market stewardship role in being able to send signals to the market about where there are gaps or opportunities or those kind of things. And the regulator's responsibility is more around the individual providers and making sure they're functioning effectively. G_P31 As we move into a world where you've got six or seven or eight providers, and one's doing yoga and one's doing trumpet lessons, and one's doing personal support, the ability to say "no, it's a provider's responsibility to deliver some quality of life outcomes" starts to become a lot less of a robust concept. So it most definitely is a shared responsibility, and I don't want to be suggesting that the regulator has no interest in these things. However, there's a mechanism that's around the planning process. NDIA lacks their interaction in that. There's a mechanism that's around ensuring that providers are meeting a certain performance standard, and that will be clearly the regulators. G_P26 Precisely how market stewardship will be carried out or participant choice supported in the NDIA is still to be determined (Productivity Commission 2017a). Sharing these roles across two new agenciesthe NDIA and the regulatorespecially when the NDIA has been identified as having major capacity issues, is likely to add to confusion and complexity around these activities. In particular, there would need to be shared data and market monitoring activities: So, no one party owns all the market oversight but predominantly, the Agency [NDIA] will have an ongoing role because they'll know, it will be their data that says "Our support has been provided in particular localities … these are particular supports that can't be offered." But the Regulator will also be looking for trends in relation to the registration, complaints processes and may need to provide feedback back to the Agency like are there terms of business, for example, leading to providers doing some things that might not be appropriate in that space G_P21 However, at present, the NDIS Act (2013) does not permit the NDIA to share data with other agencies (Carey et al. 2018). For these functions to be shared, an amendment to the NDIS Act (2013) would be needed. Hence, coordination between these two new agencies is likely to present a range of new challenges to policymakers. Rather than solving the issues within the main implementation agency, the concept of the regulator was used to sidestep addressing known challenges. The problem of market stewardship faced by the NDIA could be shifted to the regulator or "magically" solved, even though in reality a highly complex networked governance arrangement would have to be negotiated between the two agencies. Further, without legislative change, the NDIA would be under no obligation to share the data that the regulator would need to make accurate assessments about market stewardship.

Discussion
As noted at the outset of this paper, the NDIS has experienced a range of high profile implementation issues. In response to these issues and part of the overall development of the scheme, policymakers have proposed the establishment of a new regulator to take on a host of responsibilities relating to the scheme.
The sheer range of roles identified by policymakers through our interviews suggests that the regulator has become a "magic concept" by which to "solve" a host of implementation issues which are less straightforward functions for any regulator to undertake. In particular, these relate to market stewardship functions (monitoring and acting on thin markets or market gaps), decisions regarding where the government needs to step in as a provider of last resort, and supporting robust choice and control for scheme participants. There was broad agreement that the NDIS regulator will handle complaints from scheme participants and the registration of providers, perhaps because the complaint function in particular has received significant media attention and public scrutiny and is currently seen as a key implementation challenge (Morton 2017).
The establishment of an NDIS regulatory body within the context of the NDIS implementation meets, on a context-specific scale, Pollitt and Hupe's (2011) four criteria for magic concepts outlined in the introduction. In Table 1 we map the NDIS regulator against the criteria for a magic concept and include questions practitioners can ask in order to identify magic concepts.
We have demonstrated the way that Pollitt and Hupe's (2011) criteria of magic concepts can be applied on a context-specific scale to identify "magical" structural solutions. In addition, we argue that the way in which the concept of the NDIS regulator was deployed as a "fix all" by participants supports the claim that bureaucrats may look to establish new bodies and other structural solutions to solve difficult problems, rather than attempting to fix problems in existing agencies (Howlett and Rayner 2013). Rearranging the boundaries of public sector organizations is a wellrehearsed practice in public administration (Alford and O'Flynn 2012), but a tendency to turn to structural solutions before attempting other possible solutions should be avoided (Carey, Buick, et al. 2017).
Magic concepts rarely offer "reconciliation" of the problems that they are presented as addressing (Pollitt and Hupe 2011). In this sense, the NDIS regulator as a magic concept is slightly different to Pollitt and Hupe's original notion of a magic concept. The NDIS regulator, in the early stages of its conception, has been applied as though it is a magic concept, but at some point in the future the bureaucracy will pin down and define exactly what functions the NDIS regulator will and will not serve. At that point, the NDIS regulator will no longer be able to be used as a magic concept within the NDIS because its broadness, ubiquity, and applicability will be limited. It is uncertain what the final responsibilities of the regulator will be, but its remit is unlikely to provide solutions to all the problems that policymakers in our study have ascribed to it. Moreover, the fact that most of the identified roles need to be carried out in a coordinated fashion with the NDIA suggests that the establishment of the regulator will increase the complexity of governance arrangements for the scheme. Issues not solved by the NDIS regulator will remain as implementation challenges that must be addressed at some point by the bureaucracy. We argue that policymakers in the NDIS are looking to a "magic concept" to solve these problems rather than having a clear understanding of the roles and responsibilities of existing government agencies to manage those problems.
Many of the roles outlined for the new NDIS regulator originally sat solely with the NDIA, as outlined by the Productivity Commission blueprint (Productivity Commission 2011). However, as noted throughout this paper, major concerns have been raised about the capacity of the NDIA to deliver on its existing duties The range of possible roles for the regulator is noticeably broad and wide in scope, including: worker screening, provider quality and safeguards, complaints handling, market oversight, monitor thing markets, decisions about providers of last resort, registration of providers; supposedly addressing many current implementation challenges experienced by the scheme Normative attractiveness Is the concept hard to argue against because of its perceived progressive values?
No participants were arguing against the instigation regulator or suggesting that the NDIA capacity should be increased instead of putting resources into the new regulator solution Regulation is difficult to argue against, especially in the context of vulnerable populations and a scheme that is already experiencing implementation challenges Implication of consensus Does the concept dilute or deny conflicting interests, logics or arguments about the best solution to the context-specific problem?
All participants spoke with confidence about the role of the regulator, without details about how it would fulfil such a wide range of roles Marketability Is the concept known and used by many practitioners and feature in new policy documents, titles and new units within the context?
The NDIS regulator has marketability providing a narrative of a potential 'fix' for the many implementation challenges faced by the NDIS (ANAO 2016; Productivity Commission 2017b). Rather than address these capacity issues, a new entity has been proposed in the form of the NDIS regulator. Overall, our analysis suggests that it is important to pay attention to where magic concepts emerge during policy implementation to alert us to problem areas. As alluded to in the opening quote, it is risky to trust something when you cannot see "where it keeps its brain" or, in other words, it is risky to trust something when the details and intricacies of its functioning are not understood. NDIS bureaucrats would be wise not to trust that the instigation of the NDIS regulator will provide solutions to all the problems that they hope it will address. As Pollitt and Hupe (2011 p.654) argue practitioners "should not be seduced into thinking that these apparently unopposable ideas actually solve previous dilemmas or resolve awkward trade-offs". Given the NDIA will need to act as a co-regulator, and undertake many roles in conjunction with the new NDIS regulatory body, the capacity issues of the NDIA are likely to present on-going problems even after the NDIS regulator is established.

Conclusions
Using interviews with policymakers, we identified the use of the new NDIS regulator as a context-specific "magic concept" (Pollitt and Hupe's, 2011) used to manage expectations about the way that the Australian government will solve the implementation problems faced by the NDIS. We argue that when we find magic concepts in use we are likely to uncover problems and tensions which are being glossed over. As Pollitt and Hupe (2011 p.654) suggest, these magic concepts do not provide detailed recipes for action or solution: "Magic" is entertaining. It excites discussion, but when the show is over many hard choices remain". This means that while the identification of magic concepts can help us find gaps in problem-solving strategies, the concepts themselves should rarely be utilized as solutions.