The Principle of Restraint: Public Reason and the Reform of Public Administration

Public Administration, Public Reason and Restraint in Reform

Contrary to public perception, public administration agencies are in charge of much more than rigidly implementing decisions made by politicians. At all levels, public administrators are left considerable discretion in carrying out their jobs, not only because they have factual knowledge, coming from experience, of the sorts of cases typically dealt with by their agencies and how best to handle them to pursue pre-determined goals. The discretion exercised by public administrators crucially involves important decisions about values, and it could not be otherwise. When handed down from above, the values a public administration agency is instructed to pursue often need to be left vague. The same value might need to be specified differently depending on context, and it is extremely difficult, if not impossible, to pre-determine how values should be specified across the whole variety of concrete cases public administrators might face. For similar reasons, public administrators are left considerable leeway in arbitrating the conflicts that often arise in practice among the multiple values that their agency is instructed to pursue.

Public administrators’ discretion is well known to political scientists (Evans, 2010; Friedrich, 1940; Gailmard and Patty, 2007; Lipsky, 2010; Sowa and Selde, 2003) but has received little attention within normative political theory. Still, as stressed by a recent and growing literature, large spaces of discretionary power exercised by unelected officials open extremely important normative questions about how such spaces should be regulated and how administrators should exercise discretion so as to satisfy important political ideals. For example, Joseph Heath (2014) explores the requirements of accountability that should be met by public administrators, stressing in particular accountability to other administrators. John Boswell and Jack Corbett (2018) argue that public administration agencies should become more inclusive, transparent and justificatory, to create positive feedback loops that might motivate members of the general public to behave more deliberatively. Bernardo Zacka (2017) lays down a set of standards (efficiency, fairness, responsiveness and respect) that should guide street-level bureaucrats, calling on them to develop appropriate dispositions and an appropriate character.

My particular interest in public administration is linked to several ideals discussed by these theorists. Specifically, I am interested in the way the acknowledgement of the discretion inevitably left to public administrators should work as a call to arms for the authors working from within the framework of public reason. Public reason theorists stress that law and policy decisions, backed by the coercive sanctions of the state, impose a specific burden on the actors making them: a duty to make such decisions so that they can be justified to the public. Therefore, the fact that public administrators make crucial discretionary decisions about policy should leave a particularly strong impression on them. Is it possible to reason publicly at the level of public administration, and if it is, what are the conditions that enable public reasoning at that level? Until reassuring answers are provided to these questions, one might fear that many policy decisions in our administrative-heavy states are completely beyond the reach of public reason.

Public reason theorists are attentive to the differences among the choice situations faced by decision makers at different levels, like the judiciary, representative institutions and so forth, proposing different accounts of what it means to reason publicly at each level (Greenawalt, 1995; Rawls, 1997: 783–787). Although, in line with the rest of political theory, comparatively little attention has so far been paid to public administrators, Henry Richardson convincingly puts forward his theory of ‘specification’ as the shape that public reason should take specifically at the level of public administration.

Elected politicians have no choice but to leave considerable room for administrators to creatively add content to the abstract directives provided to them, so as to bring such directives to bear on the concrete tasks each administrative agency is in charge of and, crucially, to manage the tensions among values that emerge during the process. This now-familiar feature of public administration is connected to the ramification of political decision-making in administrative-heavy states, where difficult questions are broken down into progressively smaller pieces in the passage from parliaments and government cabinets to different government departments and then to administrative agencies (Richardson, 2002: 227). Specification is apt for public administration agencies in that it aims to explain how decision makers can still be guided by loyalty to the directives, and therefore, the framework of values handed down to them from above, either through legislation or otherwise, even in this context (Richardson, 2002: 215–219).

As a public reason theorist, Richardson stresses that administrators are supposed to make creative decisions in their public capacity. Consequently, they have a duty of justification towards the public. At a minimum, this means that the reasoning behind decisions must be such that it can be opened up and transparently explained to the public, therefore excluding outright appeals to the opaque intuitive sense that, for example, a certain course of action is just wrong. This is a taller order than it might seem because, as explained by Richardson, the balancing of conflicting values in a purely intuitive manner is widely seen as the only way to give answers to concrete practical problems once we recognise that there often is a plurality of irreducible and conflicting values relevant to the issues at hand. Crucially, intuitive balancing of plural values is by nature impossible to make transparent to others. ²

How can these desiderata for the public reasoning of public administrators be kept together? The search for coherence among plural values is crucial to doing so. According to Richardson, administrative agencies should feel free to amend the parameters set for them from above by narrowing such parameters, specifically by adding clauses that determine, for example, in what specific circumstances and by what specific means a certain policy end should be pursued. However, loyalty to the original parameters is safeguarded because this process of narrowing down should follow a very specific train of thought. Administrators should reflect upon the point behind the various directives they are handed down and see whether it is possible to revise them through narrowing them down in such a way that it would bring them all closer together, as in a more coherent big picture. In Richardson’s (1990: 300) words, this process of amending the ends of policy aims to ‘explain some of them in terms of others’, not just to establish logical consistency by removing possible tensions among them. Here, he follows John Rawls’s influential idea of a wide reflective equilibrium, proposed as a model of justification in political philosophy, where many forceful considerations often pull in different directions. In this context, justification is a matter ‘of everything fitting together into one coherent view’ after a process of revision through which some considerations are amended or even rejected to progressively create a greater overall equilibrium (Rawls, 1971: 19).

One of Richardson’s examples will help better clarify specification. This example responds to the criticism that different government agencies devote different levels of effort to preventing fatalities. According to Richardson, to attack the US National Park Service for spending less than other agencies on safety in their management of the Grand Canyon is to miss the ramified nature of public administration, where different agencies are handed down different mixes of ends. Unlike, say, nuclear power regulators, the mission of the National Park Service in managing the Grand Canyon is centred on enabling the aesthetic enjoyment of sublime untamed nature. The safety of visitors is certainly part of the mix, but the only way to translate this end into practical measures that sit well with the overall mission of the National Park Service is to specify it in terms of safety measures of a restrained kind, such as discreet warning signs, not widespread fencing (Richardson, 2002: 237–241). Similar points can be made about other values. For example, the National Park Service rightly cares about accessibility to visitors, pursued through the construction, among other things, of roads and hospitality centres. Specification requires that decision makers be focused on figuring out how to design new constructions so that they can fit most neatly within their agency’s mission. Given that it is the enjoyment of untamed nature that should be enabled for the visitors to the Grand Canyon, new constructions will have to be unobtrusive, interfering as little as possible with the main observation points. Importantly, as exemplified by these cases, it seems perfectly feasible to transparently explain to others how a certain possible specification of an agency’s values leads to the most coherent picture of its mission, making specification into a full-blown model of public reasoning.

Richardson’s account of specification provides solid foundations for a theory of public reason at the level of public administration. It offers an appealing synthesis between (1) the recognition that internal tensions are unavoidable and any administrative agency must find a way to handle conflicts among its plural values and (2) the idea that the solution is for administrators to search for coherence in the sense of closest fit within their agency’s mission. Consequently, I will employ specification as the shape that public administrators’ public reasoning should take. Still, Richardson leaves many open questions. In his necessarily schematic representation of specification, the set of values that specifiers must handle is closed in that while the values from the set can be continuously revised through specification, no brand-new general value will be added from the outside later on in the process. ³ This framing, however, neglects a key part of the dynamic nature of the values of administrative agencies, whose missions are routinely questioned from above, with politicians frequently considering whether they should be reformed.

Such reforms can foster or hamper the ability of administrative agencies to reason publicly through specification. Unless we analyse such reforms, our understanding of public reason at the level of public administration will be incomplete in that we will be ignoring an important enabler of administrative public reason. Accordingly, this article aims to start a discussion of how to reform existing administrative agencies to protect their ability to satisfy public reason. Specifically, I intend to focus on the way this ability can be hindered by introducing new ends into an ongoing process of specification.

This article’s main contribution is the introduction of the PR. PR lays down that elected politicians or, more generally, the actors considering whether to revise an administrative agency’s mission should refrain from any reform that involves a switch to a considerably less tightly interconnected system of values underlying the work of the agency in question. PR excludes reforms that would turn the clock way back on the search for coherence that Richardson depicts as crucial to public administrators’ public reasoning. My discussion of PR will highlight that specification only works within certain limits; if there is too much distance between an agency’s different values, administrators become unable to reason publicly through specification.

We have seen that for Richardson, a mission whose constituent elements can be closely connected to and partially explained in terms of one another is necessary to bring general directives to bear on concrete cases in a discursive way that therefore can be transparently explained. With the National Park Service, for example, the search for coherence around the central notion that the Grand Canyon National Park is about the sublime enjoyment of untamed nature enabled reasoned distinctions between different candidates as concrete safety and accessibility measures that would have otherwise been impossible to draw.

What if PR was violated and the logical distance between the different values guiding the National Park Service grew considerably? What if the National Park Service was instructed to consider itself as being as much about enabling the sublime enjoyment of untamed nature as about subsidising the hospitality industry? Faced with this new and more extreme clash of ends, the agency would stop making transparent sense to its public when deciding whether massive hotels and restaurants should be allowed in the most breathtaking spots of the Grand Canyon or when answering other questions where the two ends pull in opposite directions. This is because the possibility would be gone of arbitrating such value conflicts by reducing them to the question of what solutions would further, with the greatest integrity, the sublime enjoyment of untamed nature. Decision makers would have no choice but to pick sides through intuitive balancing of the relative importance of the values involved. However, intuitions are by nature impossible to fully explain to others, violating a basic requirement of public reasoning.

To place my argument for PR on firmer ground, I now turn to a recent reform to NICE, a public body working at arm’s length from the British Department of Health. NICE is relevant here because, as discussed in the next two sections, it not only has traditionally enjoyed considerable discretion in filling large gaps within the abstract parameters set for it from above but also shares Richardson’s regard for transparency and justification. The reform I aim to analyse, which saw NICE take over the CDF, constitutes a clear violation of PR and will help me illustrate the serious problems associated with loss of coherence. Also, to look at the CDF promises to be fruitful in that PR provides non-obvious reasons why commentators should be critical of it.

NICE and CDF

Founded in 1999, NICE appraises drugs and other health technologies. It issues recommendations as to whether local NHS commissioners in England and Wales should purchase the health technologies that have been referred to NICE for evaluation. To issue a recommendation, NICE requires evidence about the financial costs and clinical effectiveness of the technology under appraisal, where effectiveness is expressed in terms of quality-adjusted life years (QALYs). Based on this evidence, NICE calculates whether the added cost to the NHS of funding the technology would fall below £20,000–£30,000 per QALY gained – NICE’s famous cost-effectiveness threshold.

NICE explains that in the context of finite healthcare budgets, commissioning a new technology that is more expensive than the one currently employed for the same purpose always has opportunity costs – benefits lost by divesting from other services in the NHS. Therefore, through the comparison with its cost-effectiveness threshold, NICE (2013: 14) checks whether funding the appraised technology would create more QALYs for its patients than it would displace somewhere else in the NHS.

Cost-per-QALY estimates, and therefore, the cost-effectiveness of technologies provide the centrepiece of NICE’s decision-making method but are not the only consideration driving its recommendations. Although NICE is extremely unlikely to reject any technology falling below £20,000/QALY, it can approve technologies that cost between £20,000 and £30,000 or, in exceptional circumstances, even beyond £30,000 per added QALY if they are supported by other considerations, including severity of the target disease, socio-economic disadvantage of potential recipients, a special premium placed on health benefits delivered to end-of-life patients and the provision of ‘innovative’ health gains that cannot be properly captured by the QALY. ⁴ Therefore, NICE decision makers reach well beyond cost-effectiveness, examining a variety of other considerations grounded in values like fairness and compassion. Creating a link with the previous section’s discussion of public reasoning, NICE (2008: 10) has a strong commitment to transparently justifying its decisions, including the way all relevant considerations combine to form supporting rationales.

Since July 2016, NICE’s process has been further complicated because NICE took over the CDF, which had been running independently of NICE (and suffering from great financial problems) for a few years. NICE’s revolutionised CDF is a ring-fenced fund with an annual budget of £340 million, to be used to resolve the uncertainty surrounding new cancer drugs that nonetheless have the potential to satisfy NICE’s cost-effectiveness threshold and the other criteria outlined in the previous paragraph. The most conspicuous change brought to NICE by the CDF is that in appraising cancer drugs, NICE can now issue a new kind of recommendation. In addition to either simply recommending a cancer drug for routine use under the NHS or rejecting it as clear example of bad value for money, NICE can now decide that it should instead be funded for up to 2 years ‘under the CDF’ if the evidence about costs and benefits shows ‘plausible potential for a drug to satisfy the criteria for routine commissioning, but … there is currently too much uncertainty surrounding the clinical data and consequently the cost-effectiveness estimates to make such a recommendation’ (NHS England, 2016: 11).

When recommending a drug under the CDF, NICE identifies the areas of uncertainty and develops a framework for data collection aimed at resolving it. The pharmaceutical producer must commit itself to collecting the data as instructed while the drug is funded under the CDF. At the end of the 2-year period, NICE will reconvene to decide, in light of the new data, whether the £/QALY ratio of the drug in question is actually favourable and, subsequently, whether it should be recommended for routine commissioning.

An important element of NICE’s data collection instructions is the primacy enjoyed by so-called observational data, obtained through registries established in the NHS to systematically collect data about cancer patients and their response to therapies. NICE (2016: 5) states that cancer registries are ‘the preferred option for data collection in the CDF’, ‘could be the sole source of outcome data’ and ‘will always accompany any other data source’. Because NICE (2016: 9) aims to fund drugs under the CDF for no longer than 2 years, the idea of initiating new randomised clinical trials (RCTs), which take considerable time, ‘would need careful consideration’.

A New Framing for the Problems Raised by NICE’s CDF

The CDF lowers NICE’s evidentiary bar for issuing a positive recommendation for cancer drugs at two levels. First, the option to recommend a drug under the CDF is all about accepting, although temporarily, a greater amount of uncertainty about its cost-effectiveness than NICE tolerated in the past or will tolerate with other kinds of technologies. Second, consider the decision to have RCTs take a back seat to observational studies when it comes to collecting the data that will determine whether a cancer drug will be routinely commissioned after the CDF funding has ended. This decision makes positive recommendations on exit from the CDF more likely because compared to RCTs, observational studies are known to exaggerate the positive clinical effects of treatments and to be less sensitive to possible harmful effects, because of RCTs’ ability to better prevent biases (such as selection and performance biases) that tend to be skewed in the direction of positive results (Howick, 2011: 39–62). To be sure, when reconstructing NICE’s traditional approach to the assessment of clinical evidence, NICE decision makers explain that NICE has never given absolute priority to RCTs; observational studies are needed alongside RCTs because both have limitations that the other approach helps to compensate for (Littlejohns et al., 2010). However, NICE’s traditional approach never amounted to ‘a plea to abandon RCTs and substitute them with observational studies’ (Rawlins, 2008: 2159) or, as has happened with the CDF, to give priority to observational studies over RCTs.

These elements have already been criticised from an epistemic perspective. For example, the central role of observational studies has been described as ‘a major cause of concern’ because of their comparative openness to bias, exacerbated by the conflicts of interest of pharmaceutical companies in charge of collecting data (Grieve et al., 2016). What matters here, however, is that the choice to lower the evidentiary bar for positive recommendations is an issue of value, providing an instance of value reform. NICE (2008: 4) itself traces its need to make value judgements back to the fact that the best available evidence about costs and clinical effectiveness ‘is not always of good quality and hardly ever complete’. Given that scientific research never gives NICE absolute certainty as to whether a certain technology satisfies the £20,000–£30,000/QALY cost-effectiveness threshold and its other considerations, NICE must choose how much certainty is enough to recommend the technology. Any such choice to set the evidentiary bar somewhere (or to move it) is not scientific in nature, and the only non-arbitrary way of making that choice is to adjudicate between the values fostered by a stricter approach to evidentiary standards and those fostered by a laxer approach.

In short, underlying the changes brought by the CDF is an implicit restructuring of part of the framework of values served by NICE’s process. What values are promoted by NICE’s shift to a laxer approach to the evidence needed for the approval of cancer drugs and what values suffer from it? Regarding the values fostered by that shift, a natural answer points to onco-exceptionalism: the idea that cancer is a uniquely terrible disease, and therefore, higher priority should be placed on the treatment of cancer patients vis-à-vis patients affected by diseases that are equally serious in terms of QALYs lost. Indeed, there is no plan for, say, a Cardiovascular Disease Drugs Fund. Still, I intend to bracket onco-exceptionalism here because a closer look reveals that NICE’s CDF does not really put a premium on the fight against cancer at large, which helps us identify another value as centrally fostered by the CDF.

The CDF makes life easier for cancer pharmaceuticals alone. As discussed in the debate surrounding the transition of the CDF into NICE, this fact heavily short-changes cancer patients because greater health benefits would accrue to them if the CDF’s money was redirected towards cancer technologies like radiotherapy, surgery and diagnostics (e.g. Jack, 2014). Despite these objections, the remit of NICE’s CDF was not extended beyond drugs. Consequently, onco-exceptionalism is not the whole story, and there is at least another important value served in its own right by the CDF: facilitating drug coverage, or, in other words, facilitating the growth of the pharmaceutical industry.

It has already been noted that recent British governments have seen the subsidisation of pharma as one important end of health policy. The creation of the old CDF has been explained with reference to it (Maynard and Bloor, 2015: 221). Also, the protection of industry appeared to win the day once before in NICE’s recent history: in 2014, NICE refused to lower its cost-effectiveness threshold although long-awaited evidence had emerged about opportunity costs in the NHS, demonstrating that the threshold should be set as low as £13,000/QALY (Maynard and Bloor, 2015: 220). This reading of NICE’s refusal to lower its threshold is confirmed by the explanation for such refusal provided by Andrew Dillon, NICE’s chief executive. According to Dillon (2015), ‘reducing the threshold to £13,000 per QALY would mean the NHS closing the door on most new treatments’, therefore forgetting goals that government valued highly, including ‘encouraging an innovative UK research base’.

The protection of pharma might seem innocuous enough. However, against the background of finite NHS budgets, caution must be exercised before spending anything on a new treatment because the necessary resources could always be used on other beneficial interventions that will otherwise go unfunded. Therefore, it is extremely important that we are strongly confident that the new technology provides greater value for money than those that will have to be displaced or, more generally, those that could have been provided instead.

By accepting more of a promissory note than usual on entry into the CDF, and by privileging research methods at the end of the CDF period that are biased towards positive results, NICE greatly increases its risk of recommending funding for drugs that actually do not meet the £20,000–£30,000/QALY threshold and its other considerations. This increased risk hinders NICE’s ability to foster aggregated population health through its recommendations and also its ability to promote good value for NHS resources according to NICE’s set of criteria, bringing cost-per-QALY estimates together with other considerations (special attention paid to severity of disease, the end-of-life premium and so forth). This tension with population health and value for money is further compounded by the fact, noted above, that £20,000–£30,000/QALY is already an unrealistically generous estimate of the level beyond which new technologies displace more QALYs in the NHS than they create.

One aspect of this clash of values is particularly relevant to my argument. As I now proceed to demonstrate, while the commitment to facilitating drug approval is a recent addition to NICE’s mission, the values damaged by the new CDF arrangements have a long history within NICE. Of course, NICE has revised its process many times since its inception, finding room for values as diverse as cost-effectiveness, fairness and compassion. However, until recently NICE did that by integrating the inputs it received from government into a rather coherent big picture of a watchdog, making sure that technologies only receive a share of scarce taxpayers’ money if they guarantee good value for patients. The protection of industry found no space in NICE’s traditional mission – a fact that will take us back to PR.

Ever since an early amendment to its Establishment Order, NICE has been instructed to secure ‘the effective use of available resources in the health service’, and therefore, to pay attention to the opportunity costs of new technologies (Statutory Instrument 1999 n. 2219, 1999). The Directions from the Secretary of State for Health, published in 2005, legally require NICE to consider ‘the broad balance of clinical benefits and costs’, therefore endorsing an interpretation of relevant opportunity costs as value lost for patients (and not, say, for industry; Directions and Consolidating Directions to the National Institute for Health and Clinical Excellence, 2005, emphasis added). This interpretation is confirmed by Kalipso Chalkidou’s reconstruction of the objectives that the government laid out for NICE when it was created. The goal of translating taxpayers’ money into population health figured prominently, alongside the reduction of geographical variation in care quality (Chalkidou, 2009: 3). This interpretation of the value to be created by NICE can still be found in the recently published NICE Charter, whose introduction to NICE’s mission is punctuated by statements picturing NICE (2014) as aiming to ‘deliver the best possible care’ or ‘the best value for patients’ within available resources.

This is not to say that before NICE’s CDF existed, NICE had pursued a strict understanding of value for money as aggregated population health. To be sure, NICE’s creation and first few years were deeply influenced by an approach to health economics that is fond of pure cost-effectiveness analysis, developing NICE in the direction of a tool to maximise QALYs, and therefore, population health for the amount of money available (Williams, 2004). However, often because of external pressure coming from government, over the years, NICE incorporated into its methods several ‘other considerations’, which can outweigh cost-effectiveness and often embed very different values from it. For instance, as mentioned in the previous section, a few of them oppose to cost-effectiveness’s purely maximising logic a regard, grounded in fairness, for who gets healthcare benefits; based on such considerations, a QALY creates greater value if its beneficiaries are badly off, for example, in terms of severity of disease or socio-economic status. Still, while progressively expanding relevant sources of value beyond cost-effectiveness, for a long time, NICE managed to stay within a coherent picture of its mission, focused on securing all dimensions of value for money for the recipients of NHS interventions.

The end-of-life premium is a great case in point. NICE introduced it in 2009, in response to political pressure picking up on complaints that NICE had been rejecting too many terminal cancer drugs (Cookson, 2013: 1133–1134). However, when specifying this new criterion, NICE managed to identify general circumstances for its application that capture a minimally plausible account of added value to patients created by the technologies satisfying them, although this account strays far from cost-effectiveness. In deciding that unfavourable cost-per-QALY estimates (effectively up to £50,000/QALY) would be tolerated if patients had less than 24 months to live without the technology under appraisal and such technology added at least 3 months to their life expectancy, NICE (2009) appeared to make room for a new mix of considerations, such as the extreme severity of the diseases in question, the magnified subjective importance that a few more weeks can have at the end of life and perhaps compassion for patients who might need those extra weeks to set their personal affairs in order. Note the contrast with the CDF, which does not make any effort to restrict attention to patients who lack alternative treatments or for whom the drug in question would otherwise create an exceptional amount of value.

In sum, until recently, industry growth was not part of NICE’s mission in its own right. The nature of NICE as a watchdog, in charge of keeping the industry in check, is confirmed by the very first principle NICE gives to itself in its publication Social Value Judgements; if there is ‘not enough evidence on which to reach a clear decision’, NICE (2008: 16) should never recommend in favour of a drug. To be sure, NICE has always been instructed to include the diffusion of innovation among its goals. However, while the promotion of bare innovation is connected to a concern for industry growth, the instructions handed down to NICE always urged it to promote ‘the benefits to the NHS’ that innovation might have (Directions and Consolidating Directions to the National Institute for Health and Clinical Excellence, 2005) or ‘high-value’ innovation (Chalkidou, 2009: 3). When innovation was formally integrated into NICE’s (2013: 19) method, it was in fact reduced to the ability to make ‘a significant and substantial impact on health-related benefits that are unlikely to be included in the QALY calculation’. In contrast with the CDF’s relaxing of evidentiary standards, and echoing Social Value Judgements’ first principle, those benefits must be ‘demonstrable’ (NICE, 2013: 73).

Placing Restraint on Firmer Ground

The creation of NICE’s CDF is part of a process that perfectly exemplifies a violation of PR. The CDF constitutes a sharp break with, if not a betrayal of, what NICE stood for until recently. For around 15 years, NICE was committed to the specification of a broadly coherent picture of itself as a watchdog, essentially pitted against industry, or at least narrowly focused on the interests of NHS patients in cautiously recommending how to use scarce taxpayer money. With the CDF and Dillon’s defence of NICE’s decision to stick to the £20,000–£30,000/QALY threshold, NICE has effectively transitioned from industry watchdog to industry watchdog and industry facilitator, entrusted to pursue an end – industry subsidisation – that comes at the expense of NHS patients whenever it is pursued in its own right.

The goal of furthering the interests of patients and that of furthering the interests of pharma could be reconciled in a revolutionised picture of NICE’s mission, but only by moving to a level where the different elements making up NICE’s mission are much more loosely connected than they traditionally were. NICE would have to accept that it is now in charge of a much broader understanding of the sorts of value that can be produced within available NHS resources, for patients, for industry and perhaps for society at large. As stated in Section 1, PR excludes precisely these sorts of great losses of coherence in an agency’s mission.

This reform of NICE well illustrates my point that violating PR leads to a drastically reduced ability, on the part of administrators, to transparently justify decisions to the public. Let us go back to the time when new empirical studies emerged that examined the cost of a QALY in the NHS, concluding that NICE’s threshold should be lowered. Assuming that NICE was at that time already committed to both pushing pharma to deliver value for money for patients and subsidising industry, it would have looked equally logical to respond to the empirical evidence by leaving the threshold at its current level or by lowering it.

As in Section 1’s hypothetical case in which the National Park Service is instructed to add the subsidisation of the hospitality industry to the core of its mission, NICE’s loss of coherence involves the loss of precious argumentative resources to transparently explain why it should go one way rather than the other in the face of difficult questions. Indeed, if Dillon had been pushed to provide a full justification for NICE’s choice to leave the threshold as it was, it is difficult to imagine him doing more than resorting to intuitive balancing, explaining that in the case at hand, the growth of pharma felt weightier than the health benefits lost through displacement across the NHS. In contrast, an alternate version of NICE still unambiguously built around the guardianship of patients’ interests would have had clear what it should choose: lower the threshold. Importantly, the fit between that choice and the pursuit of good value for patients within scarce resources could have easily provided NICE with a transparent justification for it, without any need to fall back on intuitions.

Turning to a hypothetical example that is more closely focused on cancer drugs, what if questions were raised – for example, through parliamentary oversight of NICE – about whether the evidentiary bar for positive recommendations regarding cancer drugs is set precisely at the right level? Specifically, it could be asked whether NICE should not further relax the requirements regulating the evidence to be provided in support of drugs on exit from the CDF, perhaps by admitting other forms of observational evidence that might be widely available but even more prone than currently used registries to exaggerate drugs’ clinical effectiveness.

Obviously, NICE could reply that it is unwilling to lower the evidentiary bar any further because any step in that direction would feel like pursuing one of NICE’s goals (industry subsidisation) at too high a cost for another (value for money across potential NHS patients). However, the question is, would NICE be able to dig deeper, providing also a discursive justification to transparently explain this reply to the public or would this act of intuitively balancing two disparate considerations be all they can offer? Again, it is extremely difficult to imagine how NICE could offer anything more than the brute and opaque act of intuitive weighting, now that PR has been violated, and therefore, key justificatory resources are gone that were provided by NICE when cultivating a broadly coherent picture of its mission as centred around pushing pharma to deliver value for money for patients. In contrast, a clearly articulated justification could be provided if NICE could unambiguously picture its mission as centred on the pursuit of the interests of NHS patients in the context of scarce resources. This pursuit would require NICE to closely monitor the opportunity costs that any newly recommended drug has for patients across the NHS and, in turn, to reject further reliance on observational evidence known to systematically exaggerate the positive effects of new drugs.

NICE’s CDF has strengthened my case for PR by illustrating the impediments to administrative agencies’ public reasoning that are brought by reforms that violate PR. It has also reinforced my call for a wide-ranging theory of public reason at the level of public administration, which does not stop with Richardson’s account of specification as the reasoning method that should guide public administrators. Given the impact of reforms like NICE’s CDF, such theory should also include a thorough analysis of how to reform administrative agencies to protect their ability to use that method to reason publicly.

Moreover, NICE and the CDF can help to clarify the status of PR as a pro tanto principle, subject to be overridden under certain circumstances. Part of the agenda of the latest democratically elected British governments has been to use healthcare resource allocation policy to promote pharma growth. It can be argued that this fact confers democratic legitimacy to the goal of industry subsidisation and therefore, provides a reason why the health secretary and his ministers should integrate it into NICE’s mission. This reason would have been particularly strong if the importance of integrating industry subsidisation and value for money for patients in healthcare resource allocation policy had been a focus of the winning party’s manifesto or otherwise of the latest electoral campaign of the prime minister. At any rate, there is at least a set of possible countervailing considerations to be carefully considered against PR, with the CDF and other cases.

For reasons of space, I will not be able to determine exactly under what conditions these considerations of electoral democratic legitimacy can justifiably outweigh PR or which other considerations, if any, can override it. These issues concern how strong the pro tanto reasons provided by PR are. To solve them, I would need to get to the bottom of the intricate debate over the justification for, and therefore, the value we should assign to, public reason, choosing among the justifications I outlined in the introduction and many others that have been proposed. All I can do in this article is suggest that PR is strong enough not to be trumped by two further initially plausible aspiring countervailing principles.

Potential reformers might realise that the mission of an administrative agency is built upon a serious misunderstanding of its policy area. Here, the issue is not the platform on which potential reformers were elected – whether it included the reform of that agency or anything similar. They simply have solid reasons to believe that restraint on their part would be tantamount to cementing a big mistake. When big mistakes are on the table, should not a principle to do the right thing trump PR, mandating that all the necessary corrective elements be added to the existing mission of the administrative agency?

I believe it should not. In these sorts of cases, potential reformers should neither violate PR nor refrain from reform. They should be bold enough to discard the old mission of the agency in order to rethink it radically, with an eye on rebuilding it around a new mix of values that are close enough to one another, therefore protecting the ability of public administrators to reason publicly. This reaction seems preferable to violating PR also from the perspective of correcting the mistakes that triggered reform. Indeed, it is hard to imagine how it could be better to keep the old, misguided mission and blend it with new elements that are distant enough from it to counteract those mistakes. ⁵

Assuming, purely for the sake of argument, that he had identified a mistake at the heart of NICE’s understanding of the economics of the NHS, the then health secretary Andrew Lansley announced in 2010 a reform of NICE that, had it ever been pursued, would have exemplified the bold attitude I have sketched. Lansley rejected the very principle that clinically effective drugs should ever be turned down based on cost considerations, implicitly rejecting the need to look at the opportunity costs of new technologies and, in turn, at value for money across the population; frontline clinicians should be wholly in charge of prescribing drugs to their patients (Boseley, 2010). This proposal disregarded the basic fact that NHS budgets are finite. Consequently, it highlighted no big mistake at the heart of NICE. But, let us imagine for a moment that Lansley’s analysis effectively identified such a crucial mistake. Pursuing his proposal to the point of shutting down NICE’s health technology appraisal process would have clearly solved the mistake in question much more effectively than violating PR by switching to a loosely interconnected set of ends, to be served by NICE, still including value for money.

Readers might suggest that there is a different sense in which the mission of an administrative agency might be built on a mistake: whenever it ignores the interests of any of the main stakeholders affected by the agency. The job of administrative agencies should be to further the interests of all affected parties efficiently. Consequently, mechanisms of stakeholder involvement should be created so that every group can have a say in what agencies do. However, involving all stakeholders (in the case of NICE, including industry) in an agency’s decision-making comes at the expense of the coherence of its mission. Should we ever sacrifice PR for this principle of democratic accountability calling for stakeholder involvement?

My answer is again negative. This particular principle of democratic accountability would fit well within so-called aggregative theories of democracy, which are ultimately concerned with the mere interests of individuals, calling for the institutional arrangements that aggregate such interests in the best way, which generally means to advance them most efficiently across society. In fact, negotiation among stakeholders is typically proposed as one such arrangement. However, implicit in my choice to draw upon Richardson is a rejection of aggregative democracy that goes all the way down to public administration. Richardson and public reason theorists in general are deliberative democrats, who require that the interests of individuals and groups be justifiable in an appropriate way or else they should not drive political decision-making.

The deliberative camp emerged in opposition to aggregative democracy, raising numerous objections against the idea that institutions are meant to efficiently further the interests of stakeholders (Dryzek, 2000: 31–56; Young, 2000: 16–51). One such objection is that, given their disregard for public justifiability, aggregative conceptions are especially prone to reinforcing existing distributions of power, for example, by proposing decision-making methods that systematically go the majority’s way (Gutmann and Thompson, 2004: 16). This objection finds concrete confirmation in several critical explorations of stakeholder participation in healthcare resource allocation, including NICE. Different interest groups – for example, different patient-advocacy groups – are more or less successful in influencing decision-making depending on the amount of resources they can draw upon (Goddard et al., 2006: 83–85). Among them, industry emerges as uniquely influential, capable of even using patient groups as its ‘ground troops’ in steering policy (Ferner and McDowell, 2006).

The considerations advanced in the last two paragraphs only exclude involvement mechanisms that leave stakeholders free to work for their sectional interests – the sort of involvement detrimental to coherence and inconsistent with PR. I do not wish to deny the importance of stakeholder involvement in public administration, provided that stakeholders’ representatives accept that their role is to contribute to the coherent specification of the mission of the agency in question (Richardson, 2002: 219–222).

Two Objections

This section will reply to two possible objections. The first objection draws upon the framework of ‘sensemaking’, prominent in social psychology. Psychologists describe sensemaking as commonly used to arrive at decisions within organisations, especially in the face of unusual problems. To make decisions, members often ask what they are all about as an organisation, choosing a course of action accordingly. Thus far, this picture matches the normative account of specification of an agency’s mission proposed by Richardson for administrators. However, the scholars of sensemaking stress how organisational members can often make sense of their organisation’s identity even when the formal directives making up its mission are inconsistent. This is because members build sense out of formal directives as well as informal traditions and tacit understandings within the organisation (Weick et al., 2005: 410). Moreover, they are often found not to feel constrained in a strict sense by formal directives and the rest of their context; they bracket or actively reconstruct elements of such context during sensemaking (Weick, 1995: 6–16, 30–38). But if coherence can be created despite inconsistent directives, a critic might suggest, PR becomes redundant and should be rejected.

A closer look demonstrates how the literature on sensemaking does not actually make PR redundant. First, the scholars of sensemaking explain that it can fail, even to the point of leading a hospital to abysmal performance rates or a firefighting team to be decimated in the field. Importantly, they describe how excessively inconsistent stimuli, generated, for example, by too many conflicting principles for action or too many sub-cultures organisational members have to navigate, are among the main causes of such failures (Weick, 1993: 634–636; Weick and Sutcliffe, 2003: 78). Therefore, PR, which protects the coherence of the formal system of ends handed down to administrators, is helpful to keep failures of sensemaking at bay.

Second, researchers highlight how sensemaking ‘lies importantly in the hands of others’, especially of powerful outside actors. If members see that the view they have developed of their organisation’s identity is not accepted outside the organisation, this will destabilise their efforts at sensemaking, which they will restart, potentially in a recursive way (Weick et al., 2005: 416–417). Imagine that different reformers have over time revised the mission of an administrative agency with no respect for PR, adding more and more ends that point in radically different directions. Any coherent view of this agency’s identity now requires much of the above-mentioned effort, on the part of administrators, to bracket some parts of their formal mandate and radically reconstruct others. However, the greater this effort, the higher the risk that the identity administrators come up with will not be confirmed by elected politicians and other outside actors, who have the logical space to combine in markedly different ways the disparate elements making up the agency’s formal mandate. Rather than being able to rely upon a coherent view of their agency’s identity to reason publicly, administrators risk being thrown back, perhaps recursively, to the preliminary task of sensemaking. Under PR, the crucial ability of administrators to make coherent sense of their agency’s identity is therefore on much firmer ground. In addition, under PR, the way to a coherent view of one’s agency’s identity does not necessarily go through ignoring or revolutionising formal directives handed down by actors higher up in the democratic decision-making hierarchy, vindicating PR also from the perspective of democratic legitimacy.

The second objection is that PR displaces rather than solves the problems it should tackle. The growth of industry, including the pharmaceutical sector, should be a concern of the British government at large, if not perhaps of its Department of Health. Conflicts between the most disparate ends, banished by PR from administrative agencies, will always characterise the activities of government cabinets. Also, although the passage from a government cabinet to single government departments already provides the opportunity to restrict the attention of decision makers to substantially fewer values, considerably more closely connected to each other, each department will necessarily be in charge of more values than its individual administrative agencies. Does public reason gain anything from PR, or is PR just displacing incoherence to different levels?

The starting point of this article was that a great many important decisions are rightly made not only within each government department but also within specific agencies such as NICE. Therefore, PR improves the so-called determinacy of public reason, that is, the ability of decision makers to provide determinate answers to the questions they face without relying upon intuitions to arbitrate value conflicts or otherwise drawing upon resources external to public reason (Schwartzman, 2004). In fact, if PR is observed, at least decision makers from within administrative agencies become able to provide such answers. The question of how public reason’s indeterminacy can be reduced also at the level of government cabinets and government departments tackling inter-agency tensions is important and should be the subject of further analysis. However, it does not erase PR’s important contribution to public reason.

Conclusion

My argument has reinforced my initial call for normative political theorists, particularly if interested in public reason, to finally put administrative agencies firmly on their radar. Closer attention should be paid to both public administrators and their relationship with other levels of political decision-making if we want the requirements of public reason to be met in society. This article’s contribution has been to propose PR, which constitutes a first step towards a normative theory of the reform of administrative agencies. I have argued that PR preserves the ability of public administrators to transparently justify to the public the creative political decisions they routinely have to make. My justification for PR has used as a case study the CDF, a recent reform to NICE. At the same time, to bring such reform together with a discussion of PR has brought to light fresh reasons why commentators should be worried about the CDF.