Gabriele Badano
Many disciplines from the humanities and social sciences pay focused attention to the role of numbers in relation to political institutions. For example, historians have looked at how the rise of quantitative disciplines such as statistics and numerical tools like cost-benefit analysis has been inextricably linked to (and effectively driven by) the evolution of public administration (Desrosières 1998; Porter 1995). Anthropologists have analyzed the way in which practices including commensuration and the proliferation of numerical indicators constitute specific modes of power (Espeland and Stevens 1998; Merry 2016). Political scientists have investigated how numerical targets and other quantitative measures crucially contribute to governance in many political systems (Bevan and Hood 2006; Boswell 2018). This list could be much longer. However, there is at least one discipline that does not come easily to mind and whose inclusion would seem like a stretch—political philosophy.
If we pause for a moment to think about the fact that quantification is not a prominent subject of analysis for political philosophers, we will likely be puzzled by it. Here I am talking about what is normally referred to as “mainstream” or “normative” political philosophy. Dominating English-speaking academic debates in the field since the 1970s, this broad approach is interested in the evaluation of institutional arrangements from a philosophical perspective and in the provision of positive recommendations when applicable. Due to this normative focus, political philosophy seems highly relevant to discussions about the role of quantitative tools in political decision-making and possible comparisons with more qualitative approaches. Indeed, political philosophy seems very well placed to contribute, among other things, to the critical analysis of the ills created by too heavy a use of numbers in law- and policymaking and how best to balance qualitative and quantitative tools in order to foster important values institutions should pursue.
The overarching goal of this chapter is to finally make political philosophy part of the conversation. In order to do that, I concentrate on a debate that, although not explicitly focused on numbers, has interesting implications for the normative evaluation of quantitative tools in law- and policymaking. In tracing these implications, I will describe how the loudest voice from political philosophy turns out to be one that warns against the use of numbers in political decision-making. However, an important part of what I aim to achieve in this chapter is to argue that we should move beyond that voice and acknowledge that quantification might also play functions that are very important from a normative perspective, complicating the picture of its evaluation.
Section 7.1 focuses on the limits of precision as explored by the supporters of the influential “capability approach” to justice or, more specifically, to its evaluative space. It reconstructs two main arguments that capability theorists advance to support their conclusion that we should be wary of excessive precision and, therefore, too heavy a use of numbers in political decision-making. Section 7.2 digs somewhat deeper into political philosophy, demonstrating that, although less obviously connected to quantification, a prominent family of critiques of the capability approach is also relevant to the question of whether institutions should choose more qualitative over more quantitative tools to make decisions. These critiques, centered on the requirement that institutions should pursue justice in a publicizable way, can be developed so as to strike a far more positive note about numbers than the capability theorists’ arguments do. Quantitative tools emerge as uniquely well suited to reduce complexity, potentially making justifications for laws and policies more manageable by the public. Finally, section 7.3 introduces a case study to reinforce the theoretical arguments advanced in the foregoing section. Specifically, it examines the National Institute for Health and Care Excellence (NICE), a British administrative body in charge of appraising drugs and other health technologies, to illustrate the idea that quantitative tools can foster public justification and even pose an obstacle to market forces in their dealings with institutions.
My argument is connected to several other contributions to this volume, especially, although not exclusively, those that are most philosophical. Like Greg Lusk (chapter 9 in this volume), I aim to show that, despite all the skepticism surrounding our discipline, academic work in philosophy can help further discussions about important social problems, including the ones concerning quantification. The connection to Stephen John (chapter 6) is even closer and extends to the substance of our analyses. A point that I wish to make is that political decision makers should at times use numerical measures to make an argument simpler in order to render it publicly available. This point has many similarities with John’s notion of “spurious precision,” which—as exemplified by the policy to encourage the consumption of five fruit and vegetables a day—consists in picking a single number while available evidence points to a broader range of values as the correct one. John explores the conditions under which it is ethical to resort to spuriously precise numbers in a way that is potentially very relevant to a research direction that I sketch but leave open in this chapter, focused on the ethics of using numbers for the purpose of simplification in the context of public justification.
As stressed in the introduction, political philosophy is somewhat exceptional in that the contrast between quantitative and qualitative tools to make political decisions has never become a topic of focused interest. This, however, does not mean that existing theories have no implication for the evaluation of quantitative tools in law- and policymaking. In particular, the limits of precision, which have obvious relevance for the role that numbers should play in decision-making, are an important theme in the capability approach to political philosophy. Given how prominent the capability approach is in the discipline, and how recurring this theme is, the loudest voice coming from political philosophy is therefore one of warning against an excessive use of numbers. But let us now take a step back and briefly reconstruct what the capability approach is.
The main idea behind the capability approach is that we should look at a specific evaluative space when discussing justice or, in other words, when working out what should be redistributed among citizens or otherwise guaranteed to them. The capability approach is distinctively focused on what is sometimes called “midfare,” falling in between welfare economics’ traditional focus on the utility generated by different states and the focus on the external resources—like money and access to education—that are needed for persons to enter those states (Cohen 1993, 18). According to capability theorists, the focus must be on the state of persons itself, keeping in mind what flourishing human beings do and are. Capability theorists generally stress that human flourishing is made up of many different elements, including good bodily health, the satisfaction of sexual desires, being imaginative and playful, having close friends, and being politically active. As a result, they also underline that the capabilities to do and be what is authentically valuable, which should be provided to citizens by political institutions, are plural.
The two most important capability theorists, Martha Nussbaum and Amartya Sen, are often at pains to point out that precision is frequently beyond our reach when we reflect about the ends of politics and make decisions about them. These authors echo a tradition of thought that reaches as far back as Aristotle, who famously said, while reflecting on the method of his Nicomachean Ethics, that we should be only as precise as our subject matter allows us to be; when it comes to political matters, we can draw conclusions only “sketchily and in outline” (Aristotle 2000, 1094b–95a). According to Nussbaum and Sen, being vaguely right in this area is definitely preferable to being precisely wrong, which puts into question the precision afforded by the use of quantitative measures in law- and policymaking.
We can reconstruct at least two arguments supporting Nussbaum’s and Sen’s claims that precision is often doomed to be misplaced and should therefore be avoided—an argument from ambiguity and one from incommensurability. Starting with ambiguity, Nussbaum maintains that philosophers as well as politicians and administrators are allowed to make generalizations in talking about elements constitutive of the good life and making political decisions aimed at fostering them. However, such generalizations must remain vague, leaving room for multiple specifications of each component of human flourishing. Not only individual societies but also individual groups in society and single individuals retain the authority to specify for themselves what it means, say, to build strong relationships with others or to be politically active. This is motivated by obvious liberal concerns with the oppression of minority groups. Also, Nussbaum appears to believe that the issue of specification of human flourishing simply admits of multiple answers; drawing again on the Aristotelian tradition, she suggests that the best way to tackle that issue is to let different groups in society tap into the rich intellectual resources provided by their specific religious and cultural traditions.1
The incommensurability argument does not target precision or, for that matter, the use of numbers as such. However, it does target a way of simplifying complex matters and bringing to the table a heavily numerical approach that is highly influential in academic disciplines like normative economics as well as in political decision-making, namely, commensuration. By commensuration I mean the identification of a single metric that will be used to evaluate states of affairs and choose between different courses of action. As I have already mentioned, the capability approach is committed to the idea that a flourishing human life is made up of plural components. Moreover, such components are thought to be qualitatively different from each other. Therefore, they are not reducible to one another and even less so to utility or any other generalist metric of value. Therefore, when it comes to evaluating a state of affairs or comparing alternative courses of action, institutions will have to be attentive to the provision of each basic capability in its own right, without any attempt at commensuration (Sen 1987, 61–65).
Both these arguments, and the one from incommensurability in particular, are closely connected to Nussbaum’s and Sen’s classic attacks on gross domestic product, especially when used to compare quality of life in different countries for the purposes of development policy, and on cost-benefit analysis as a tool for allocating scarce resources in global as well as domestic policy. To attach a monetary value to human well-being is thought to create a problem of commodification, making key elements of human flourishing alienable for a price. More generally, Nussbaum’s and Sen’s work often gives the impression that to resist precision (and numbers) in political decision-making is to resist the march of market forces in society. For example, Nussbaum enlists Karl Marx in her critical analysis of precision, accepting his claim that alienation comes not only from attaching monetary value to well-being but also from commensuration: “Assume the human being to be the human being and its relation to the world to be a human one, then you can exchange love only for love, trust for trust, etc.”2
Given that the use of numbers and, therefore, quantitative tools to make law and policy decisions involves precision, the capability approach’s attack on precision clearly cautions against too great a use of numbers in political decision-making. The next step in my argument is to dig somewhat deeper into the intellectual resources provided by political philosophy, demonstrating that, although less obviously connected to numbers, an important objection to the capability approach also has significant implications for the role of quantitative tools in law- and policymaking. In tracing these implications, I aim to strike a more positive note about quantification than my argument has done so far. This is not to deny that Nussbaum’s and Sen’s arguments have merit. However, I believe that there is at least one more side to the story of quantitative tools in political decision-making than that of the alliance between numbers and market forces.
To give a little background, the objection to the capability approach that I aim to concentrate on comes from supporters of so-called resourcist approaches to justice’s evaluative space, which I mentioned very briefly in section 7.1. Like capability theorists, resourcists reject the idea that, as a matter of justice, institutions should focus on the level of utility experienced by citizens. In contrast to capability theorists, however, they also disagree that institutions should try to work out, democratically or otherwise, a “thick” theory of well-being for their population that includes the essential components of human flourishing. Resourcists believe that institutions should abstain from setting the normative ends of human life and focus strictly on the provision of all-purpose means that can be useful to citizens with life plans oriented toward widely different ends. These all-purpose resources—which typically include civil and political liberties, access to health care and education, and income—make up shorter lists of goods than Nussbaum’s and Sen’s capabilities.
John Rawls is a very prominent resourcist thinker. When explaining why resources should be preferred over capabilities as the currency of justice, resourcists working in Rawls’s tradition often make the point that institutions are not required simply to pursue justice, but to do so in a publicizable way. In other words, the goods populating justice’s evaluative space must work as part of an authentically public criterion of justice. Why is this important? Thomas Pogge focuses on the fact that whenever institutions allocate funds to meet a citizen’s needs, the public at large will bear the cost of this choice, both by paying the necessary taxes and by missing out on the benefits that could have accrued to them if funds had been allocated otherwise. According to him, widespread costs create an obligation on the part of institutions to be able to widely justify their choices (Pogge 2002, 208–9). An alternative and potentially complementary justification goes back to Rawls, who argues that public justification is necessary for reconciling the democratic idea that the members of the public hold ultimate authority over laws and policies with the fact that, in complex societies, they routinely have to live with laws and policies designed by others (Rawls 1993, 135–37).
Now, in Rawls’s words, “simplifications” are necessary in order for any criterion of justice to become public (1999, 79).3 To publicly justify the arrangements aimed at raising the prospects of the worst off in society, or society’s institutional scheme more generally, is to choose arrangements that the public can broadly agree on. However, determining what are the components constitutive of human flourishing or, more broadly, what are the ends of life, is extremely controversial, as is the question of how to arbitrate the conflicts between such ends when all feasible courses of action can advance only some of them at the expense of others (Pogge 2002, 208–13). Therefore, according to resourcists, institutions cannot concern themselves with the high talk of ends and should rather settle for the provision of mere all-purpose means that virtually everyone can be expected to want for themselves whatever their life plans. Unlike capabilities, all-purpose resources have the potential to become publicly recognized objectives of government.
Thus far in this subsection, I have simply reconstructed a classic argument against the capability approach. This argument, however, is not normally brought to bear specifically on Nussbaum’s and Sen’s attacks on precision. I will now make this connection, suggesting that my argument allays many of the concerns animating Nussbaum’s and Sen’s objections to precision and, therefore, quantitative measures in political decision-making. Their claim that precision sweeps under the rug crucial ambiguities surrounding human flourishing loses its force if we accept that the job of institutions is simply to provide mundane resources such as money, access to education, and health care, leaving thicker accounts of well-being for individual citizens to figure out. Similarly, the capability theorists’ worries about incommensurability are centered on the incommensurability of the different constitutive components of human flourishing, which, however, should not be the business of institutions.
In sum, on a resourcist account, the numerical tools that institutions are (at least in principle) supposed to use simply do not measure the thick version of well-being that is integral to Nussbaum’s and Sen’s ambiguity and incommensurability arguments against precision. Another way of putting this point is to say that Nussbaum and Sen do not really warn us against too wide a use of numbers in general; they only warn against too heavy a reliance on numbers provided that we accept their view of justice’s evaluative space, namely that the aims of law- and policymaking (and, in turn, the typical constructs to be measured by the numerical tools employed in a political context) should essentially have to do with human flourishing.
At this point, some might object that agreement over political decisions and their justifications is simply too much to ask, casting doubt on the argument that I have outlined in this subsection and the way it rescues precision from the attacks of capability theorists. Rather than evaluate this objection, I will suggest that we could strike a positive note about numbers even if we decided to accept it. In the process of doing that, I will give a new spin to several keywords that have emerged during my reconstruction of the resourcist critique of the capability approach.
Let us then redefine public justification in a considerably less demanding way, as the ability on the part of law- and policymakers to place a justification for their choices “out there,” in front of the public. Moreover, for any justification to be truly addressed to the members of the public, transparency in the disclosure of relevant information cannot be enough, but must be coupled with an effort to make the provided justification manageable by them. Among other things, manageability excludes scenarios in which the public is simply flooded with and left bewildered by a huge amount of information, for instance due to the complex nature of the factual side of the law or policy issue at stake. The commitment to manageability I am introducing echoes classic accounts of public justification. For example, in an analysis by Jeremy Waldron (1987) that is widely discussed as a powerful description of what public justification liberalism amounts to, the mark of a well-ordered liberal society is that the principles governing its institutions are “available for public apprehension” (146) or “capable of explaining [themselves] at the tribunal of each person’s understanding” (149).
Repurposing another Rawlsian concept, there is room to argue that public justification as I have redefined it often requires simplifying devices that allow decision makers to provide the public with the skeleton of the reasoning behind a law or a policy. When complex issues are on the table, such devices are the only hope to produce justifications that members of the general public can reasonably be expected to manage, as opposed to perceive as spoken largely over their heads. Also, reducing justifications to their bare bones is often important for starting democratic deliberation across society. This is where we get back to the topic of quantification and quantitative measures, which by their nature reduce their objects to a few key dimensions while excluding others, and which are therefore very well placed to serve as the simplifying devices that public justification needs.
Now, a reader might ask, is this simplifying power really unique to numbers? The use of words in putting together a justification for a political decision already involves a degree of abstraction from the problem in question and the proposed solution as actually experienced, in its full complexity, by flesh-and-bone individuals. I do not mean to deny this. At the same time, however, it is hard to find any scholar of quantification who fails to stress how numbers are characterized by a uniquely strong tendency to leave behind some of the features of their intended objects of measurement. This, I believe, makes intuitive sense: if qualitative descriptions of social phenomena necessarily involve abstraction, it stands to reason that a particularly strong form of abstraction is involved in their quantification.
To quote but a few examples from the literature, Theodore Porter makes some points that go in this general direction when discussing the measurement of social as well as natural objects. Among many other cases, he describes how, before thermometers were perfected during the eighteenth century, the concept of temperature referred to a rich variety of features of the atmosphere as perceived by people. “To quantify was necessarily to ignore a rich array of meanings and connotations,” he writes, given that in general “quantification depends on the art of forgetting” (Porter 1994, 396). On her part, Sally Engle Merry focuses on the numerical indicators that are employed to monitor the implementation of human rights worldwide. Her very definition of numerical indicators depicts them as meant to turn complex data into a simple number (Merry 2011, S86–S87); for her, one of the overarching functions of numerical representation in public policy is to simplify complexity (Rottenburg and Merry 2015, 7). In fact, according to Merry, numbers tend to flatten individual differences, gloss over context, and lose information more generally.
Most of the authors discussing the link between quantification and simplification do it in explicitly critical terms. This holds true for Merry and even more so for other scholars. Drawing on Merry’s work, Jerry Muller (2018) complains about the tendency of metrics to restrict attention to a few aspects of their target social phenomenon, which often are the simplest and easiest ones to measure (23–27). Lorenzo Fioramonti (2014a) compares those who try to understand the economy through statistics to the blind characters in an Indian fable who try to learn what an elephant is by each touching one single part of it. The message that Fioramonti aims to convey is that gross domestic product and other available statistics focus only on small, isolated aspects of the economy, like the man who touches the trunk of the elephant and thinks he is faced with a snake, or the one who touches its leg and thinks the elephant is a tree (16–17). For Cris Shore and Susan Wright (2015), who look mostly at audits, “tunnel vision” is one of the main negative consequences of quantification (430–31).
Through the link I have drawn between public justification, manageability, and the need for simplifying devices and numbers, my aim has been to push back, at least to some extent, against this overwhelmingly negative assessment of the simplifying powers of numbers in political decision-making. Quantification is very well placed to help decision makers meet the need for public justification. This fact demonstrates that the tunnel vision induced by numbers can play very important roles and is not to be discarded lightly. Obviously, this point is not enough to cancel out the possible dangers associated with numbers that I have hinted at in the previous two paragraphs. More than anything else, this point highlights the need for further research—research into the ethics of using quantification to reach the level of simplicity required for justifications to be truly manageable by the public.
Although very important, this line of research is extremely complicated and therefore falls beyond the scope of this chapter. In concluding this section, I have room for only a few preliminary remarks about what it might mean to use the simplifying powers of numbers ethically. An ethical theory of quantification in this area should probably start with a few clear statements about how numbers must not be used. Decision makers should not make a justification for a law or policy manageable by the public by simply zooming in on its simplest aspects, or the ones easiest to measure, regardless of whether they are central to their argument for that law or policy. Even more importantly, the introduction of numbers into a justification offered to the public should not be motivated by merely rhetorical purposes. Numbers have a well-known allure (Seife 2010), but using them to simply move the public to accept a proposal, without aiming to add anything to their understanding, is inconsistent with public justification properly understood. Decision makers should use quantification to narrow the focus of their justifications for laws or policies to the few aspects that represent the innermost core of their argument. Their goal should be to reduce complexity to the extent necessary in order for the resulting justification to become manageable to the majority of the general public in the sense of the term explored in this subsection. To reiterate, this is but a preliminary description of what an ethical use of numbers in public justification amounts to, subject to revision and expansion in future investigations.4
Section 7.2 suggested that when it comes to normatively appraising the role that quantification should play in political decision-making, things are more complicated than the capability theorists’ arguments against precision appear to indicate. The task of this section is to reinforce the theoretical conclusions reached in section 7.2 by looking at the use of numerical measures made by a real-world administrative body, the one in charge of evaluating pharmaceutical and other health technologies for use in the National Health Service (NHS) in England and Wales.
The administrative body I will discuss is the National Institute for Health and Care Excellence (NICE). Created in 1999, NICE provides guidance to the NHS in several areas, but here I will focus exclusively on its health technology appraisal (HTA) process. Through this process, NICE recommends whether or not local NHS commissioners should purchase the health technologies that have been referred to it for appraisal. And if NICE recommends a technology, commissioners are legally bound to fund it under the NHS. Given the crucial role played by its HTA process within the NHS, NICE is often in the spotlight, both in civil society and among academics. In part because NICE itself has a clear sense that its methods are grounded in moral or otherwise philosophical assumptions (National Institute for Health and Care Excellence 2008), it has also been widely discussed (and criticized) from a philosophical point of view (see, e.g., Harris 2005a, 2005b). Political philosophers have been part of this conversation, bringing theories of the so-called pattern of distributive justice and of fairness in deliberative democratic decision-making to bear on the work of NICE and other healthcare resource allocation bodies (Daniels 2008). However, no one has so far brought together NICE and topics like precision and simplicity from the debate over justice’s evaluative space.
When evaluating a drug or some other health technology, NICE calculates the added cost to the NHS of producing an additional unit of gain in health-related quality of life through that technology compared to the technology that the NHS currently uses for the same indication. In technical language, this is the appraised technology’s “incremental cost-effectiveness ratio.” NICE’s measure of health-related quality of life is the quality-adjusted life year (QALY), which integrates the increases in life expectancy and the gains in quality of life brought by healthcare treatments. In so doing, the QALY reduces to a single numerical scale a huge variety of benefits, which are as diverse as those produced by life-saving surgery, life-extending drugs for terminal cancer, medications that slow down dementia, palliative treatments, surgery capable of restoring mobility, and treatments for depression. The QALY represents a heroic attempt at commensurating very different objects. Consequently, it provides a first excellent example of a numerical tool displaying the traits of a simplifying device.
The centerpiece of NICE’s HTA method is the comparison between the incremental cost-effectiveness ratio of the technology under appraisal and NICE’s soft incremental cost-effectiveness threshold, ranging from £20,000 to £30,000 per QALY. NICE’s threshold can be read as another example of a numerical measure simplifying an enormously complicated object. Indeed, it can be read as a measure of (or at least a first approximation of) good value for NHS money. If a technology costs less than £20,000 per QALY, NICE is extremely unlikely to reject it. NICE can recommend a technology whose incremental cost-effectiveness ratio falls in the £20,000–£30,000/QALY range, provided that support for that technology is offered by any item on a list of additional considerations—severity of disease; benefits delivered at the end of life, on which there is a premium; disadvantaged groups, to whom special attention is paid; benefits accrued to children, which are given extra weight; and several others. NICE can approve technologies above the £30,000/QALY mark if such considerations provide an exceptionally strong support to them (National Institute for Health and Care Excellence 2013).5
A paper that I coauthored with Stephen John and Trenholme Junghans (Badano et al. 2017) explores the process through which NICE arrived at the £20,000–£30,000/QALY figure for their cost-effectiveness threshold. The justification that health economists normally provide for NICE’s use of a threshold (and that NICE explicitly endorses) is that the threshold is a measure connected to the opportunity costs of newly funded technologies against the background of the NHS’s finite budget. Specifically, it is meant to measure the cost-per-QALY level beyond which a newly funded technology would displace greater QALYs across the NHS through disinvestment than it would produce. Going back to a comment I made in the previous paragraph, the threshold is supposed to measure the level beyond which the technology under appraisal counts as insufficient value for money in terms of net QALY output. To serve this function, the threshold should be set at the level of the cost of a QALY produced through the least cost-effective treatment currently funded under the NHS (McCabe et al. 2008).
However, John, Junghans, and I found that when, in 2001, NICE set its threshold at £20,000–£30,000/QALY, it was nowhere near having the necessary empirical information to identify how much it cost to produce a QALY through the least cost-effective treatments that the NHS was providing. In fact, NICE did not have a solid evidence base to set its threshold until much later, around 2013. What NICE did to identify the £20,000–£30,000/QALY figure was to look back at the HTA recommendations it had issued case by case until 2001, based on its decision makers’ best insights. The figure derives from the simple fact that NICE found that while rejections had been very rare below £20,000/QALY, the likelihood of a negative recommendation increased above it, and a positive recommendation had been extremely unlikely beyond £30,000/QALY (Badano et al. 2017, 155–58).
From NICE’s stated health-economic perspective, the choice of the £20,000–£30,000/QALY figure seems to make little sense. Still, we argue that once we switch to a political philosophical perspective, there are things to be said in favor of NICE’s choice to nonetheless pin down explicit numbers for its threshold, at least until the relevant evidence emerged in 2013. For the sake of my argument in this chapter, those things link back to points made in section 7.2.
The choice to place the £20,000–£30,000/QALY figure at the very center of NICE’s HTA process made it much easier for NICE to explain to the public how its decisions were made, and thus where this or that recommendation against new technologies came from (Badano et al. 2017, 159–63). This fact increased both public transparency and what this chapter has called “manageability by the public,” in turn contributing to NICE’s ability to publicly justify its decisions in the less demanding sense of the term outlined in section 7.2.2.
To better see how manageability was helped by NICE’s choice, just imagine an alternative scenario in which NICE abandoned its plan to place at the core of the HTA process its best estimate of any numerical measure of good value. We can imagine that NICE made this choice in 2001, either for lack of a solid evidence base or because they could not stand the simplifications involved in both commensurating hugely different benefits on the QALY scale and giving a place of honor to any single measure of value for money. In that scenario, NICE’s recommendations for or against any new technology would become largely a matter of judgment by those in charge of the decision, to be exercised in identifying all the relevant considerations for the case under appraisal, in solving their conflicts, in working out how the context of the specific decision changes relevant considerations and how to balance them, and the like.
It is extremely difficult, if not impossible, to imagine how decision makers could transparently explain their judgment calls to others.6 Even if we assumed that they could, such calls would deal with a complex plurality of considerations and be highly sensitive to context, creating the problem of flooding the public with overly complex information which I underlined in section 7.2.2. In other words, by giving up the numerical centerpiece of its decision-making method, NICE would have given up the solid backbone it has traditionally been able to go back to in order to explain its choices to the public in a way that is simple enough to be largely manageable by them.
Moreover, John, Junghans, and I suggest that the creation of a numerical threshold and the publicization of decisions made on its basis have served as a catalyst for a broad democratic debate in civil society over the allocation of NHS resources—a debate whose scaleis arguably unparalleled across the world and that is credited by some for having convinced large sectors of the British public of hard truths, including the need to ration healthcare resources (Badano et al. 2017, 159–63). Whatever we might want to say in favor of NICE’s choice in 2001 to pin down an explicit figure for its threshold while waiting for in-depth empirical research, things changed in 2013 when NICE decided to stick to its £20,000–£30,000/QALY figure even though the much-awaited evidence had just made it clear that the threshold should actually be lowered to around £13,000 per QALY. From that point on, the choice to use a £20,000–£30,000/QALY threshold has created a brand-new contradiction between the decisions made by NICE about new technologies and the justifications NICE has kept issuing for them—justifications that, still describing the threshold as meant to identify the cases in which new technologies would displace more QALYs than they would create, cannot consistently incorporate a threshold set as high as £20,000–30,000/QALY. As politically hard as it might be, public justification appears to require that NICE revise its threshold and make the difficult choice of lowering it.
Let us now turn to a more recent development in NICE’s history, which is especially relevant to the alleged alliance between numbers and market forces that emerged during my discussion of capability theorists in section 7.1. In 2016, NICE took over and completely restructured the Cancer Drugs Fund (CDF). The CDF is a ring-fenced fund with an initial annual budget of £340 million, to be used by NICE to temporarily fund cancer drugs that would have otherwise been rejected, therefore enabling the collection of new evidence during the process. Now that the CDF is in place, NICE can recommend a cancer drug for usage under the NHS, albeit under special arrangements, even if NICE decision makers believe that its cost-effectiveness is still very uncertain. These arrangements require that over the following two years, the company producing the drug will have to collect further evidence aimed at reducing the existing uncertainty. At the end of this period, NICE decision makers will reconvene to decide whether the new evidence is sufficient to warrant confidence in the cost-effectiveness of the drug under appraisal and, therefore, whether it should be discontinued or recommended for routine commissioning under the NHS (NHS England Cancer Drugs Fund Team 2016).
This reform to NICE’s HTA process is interesting in many respects. However, I wish to focus on the data collection arrangements between NICE and the producers in charge of dispelling the uncertainty surrounding their drugs. These arrangements break with NICE’s traditional method in giving priority to observational studies over randomized controlled trials (RCTs) when it comes to determining the QALYs produced by drugs and, in turn, whether or not such drugs meet NICE’s £20,000–£30,000/QALY threshold. RCTs, the cornerstone of evidence-based medicine, measure statistical associations, and thereby provide a highly quantitative method for investigating effectiveness. Given that pharmaceutical producers have only two years to collect evidence while their drugs are funded under the CDF, however, NICE states that “careful consideration” must be given to any plan to start new RCTs (National Institute for Health and Care Excellence 2016, 9). Therefore, observational data, obtained through NHS registries that systematically collect information about cancer patients and their responses to treatment, may well be the only evidence submitted to NICE at the end of the two-year period.
NICE’s choices about data collection arrangements did not happen in a vacuum. They were inspired by an increasingly influential model for the collection of medical evidence, which is critical of the rigidity of RCTs and advocates a switch to flexible “adaptive” pathways to evidence gathering and “real-world” observational studies. Many commentators note, however, that RCTs are rigid for a reason: the tight constraints imposed on evidence collection are necessary for minimizing manipulation by researchers with vested interests in the outputs of their studies, like the producers in charge of collecting observational data under the CDF—or, for that matter, the great many clinical scientists worldwide who are funded by the pharmaceutical industry (Grieve et al. 2016). Also, observational data are known to be worse than RCTs at countering biases toward positive results, increasing the likelihood that researchers will conclude that a drug is beneficial even if it is ineffective or even harmful. In turn, this increases the likelihood of NICE mandating that the NHS pay pharma for drugs on exit from the CDF that are either useless or outright harmful for patients (Davis et al. 2016; see also Howick 2011, 39–62).
A powerful distillation of these arguments is offered by some critics of this emerging model, who point out that accepting adaptive pathways effectively amounts to “adapting to industry” (Davis et al. 2016). Obviously, this point is controversial, and the supporters of adaptive and real-world approaches dispute the objections that are being raised against them.7 Still, their critics surely manage to make an appealing counterpoint to the view I have extrapolated from the capability theorists’ critique of precision, namely, that the march of numbers in political decision-making is one and the same as the march of market forces in society.
A common complaint about algorithmic or simply numerical approaches to political decisions is that they are often too inflexible. Still, we can now see how this rigidity might be an important asset in resisting market forces. In certain contexts, a heavily quantitative and therefore fairly rigid approach, which will capture nothing more than a simplified version of quality of life or any other value it is supposed to serve, appears to provide the only way to constrain industry in its dealings with policymakers, by forcing them to actually show that they can deliver at least some kind of value to the public. In contrast, due to industry’s economic and lobbying powers, a flexible and more qualitative approach risks leading to a scenario in which industry systematically has its way, to the detriment of other stakeholders like patients, taxpayers, and citizens at large. As far as NICE is concerned, a “numerical connection” linking attention to RCTs to the use of QALYs and finally to NICE’s threshold seems able to do good work and is, in fact, under attack by the pharmaceutical industry. This is despite the fact that, as I mentioned at the end of section 7.3.1, the threshold is already too generous with the new drugs that producers bring to the market.
Political philosophy seems highly relevant to existing debates over the role of numbers in political decision-making. However, quantification has not been the subject of focused attention within the discipline. The overarching goal of this chapter is to take a first step in the direction of filling this gap. Given the close link between numbers and precision, the critique of precision that recurs in the work of prominent capability theorists makes the loudest voice from political philosophy one of warning against quantification. However, by drawing on objections to the capability approach that stress the importance of public justification and, in turn, by simplifying devices in political decision-making, I argue that quantitative tools should not be dismissed quickly. The case of NICE has reinforced the impression that such tools can indeed perform functions of great importance within political institutions, strengthening the normative conclusion that in some circumstances they might well be the best choice for making law- and policy decisions.
If looked at from the perspective of political philosophy, the argument of this chapter is surprising. The main framework I employ to identify the positive functions that numbers can play is that of public justification, which owes much to the classic work of Immanuel Kant. As evident with authors like Rawls, many descriptions of public justification retain a distinctive connection to Kant’s first formulation of the categorical imperative, requiring that agents check that their maxims can be willed as universal laws in the sense that they must be acceptable from all perspectives (Vallier 2018). Broadly Kantian approaches to political philosophy, however, are normally pitted against, not classed as, number-heavy theories, which are most obviously exemplified by utilitarianism. The surprising bit of my argument is that there is a strong link between a Kant-inspired commitment to public justification and numerical tools in political decision-making, which, I suggest, are extremely well suited to simplify arguments to the extent necessary to make them public. This is not to say, of course, that the use of numerical tools is without dangers, from a Kantian or other perspective. I have highlighted the need for an ethical theory of how such tools are to be used—an urgent topic that deserves much focused attention.
1. Nussbaum (1992), 224–25. Out of respect for real-world democratic processes, Sen (2004, 333) goes a step further than Nussbaum, rejecting also the idea that scholars should try to draft a list of basic capabilities.
2. Marx and Engels (1978), quoted by Nussbaum (1992), 231. For another argument that the role of numbers in politics is to entrench the power of markets, see Fioramonti (2014a), 1–9.
3. Pogge (2002, e.g., at 208) sometimes speaks in terms of the “specificity” required by a public criterion of justice, bringing up another desideratum that is connected to the use of quantitative tools.
4. Interesting resources to continue this line of research could be provided by Stephen John and Anna Alexandrova, both writing in this volume, looking respectively at the ethics of spurious precision and at how to evaluate the effects on important values of political attempts to redefine them so as to build in quantifiability, comparability and so forth.
5. For a focused discussion of the considerations to be taken into account in addition to cost-effectiveness, see Rawlins et al. (2010).
6. For a classic account of the way in which judgment interferes with transparency, see Richardson (1990).
7. For example, see Guido Rasi and Hans-Georg Eichler’s letter (2016), written in response to the critics of adaptive pathways.