1. THE PUBLIC JUSTIFICATION REQUIREMENT

The PJR holds that public officials ought to pass laws and policies that are justifiable to the public that they serve. A law or policy is justifiable to the public if and only if each member of the public has sufficient reason(s) to endorse it.Footnote ⁴ A member of the public has sufficient reason(s) to endorse a particular law or policy if she is rationally committed to endorsing it. Whether a law or policy is justifiable to members of the public does not therefore depend on whether they would actually endorse it.

Scholars generally motivate the PJR by arguing that public officials must comply with it if they are to respect members of the public qua free and equal persons. Some argue that public officials’ use of coercion is respectful of members of the public if and only if this power is used in ways that are rationally acceptable to them (Nagel Reference Nagel1991: 36–37; Rawls Reference Rawls1996: 136–137; Gaus Reference Gaus, Christiano and Christman2009: 206–208; Quong Reference Quong2011: 1–2). Others argue that public officials’ acting on behalf of members of the public is respectful of them qua free and equal persons if and only if officials do so in ways that members of the public could endorse as free and equal legislative co-authors (Bird Reference Bird2014: 201–204).

Scholars committed to the PJR disagree on important matters of interpretation and scope. First, for a law or policy to satisfy the PJR, must members of the public accept it on the basis of public reasons – i.e. reasons they can reasonably expect other citizens to share (Rawls Reference Rawls1996: 213–227; Quong Reference Quong2011: 261–273); or, can they accept it on the basis of private reasons – i.e. reasons they cannot reasonably expect other citizens to share (Vallier Reference Vallier2011)? Second, are public officials only state officials, or can they also be officials of international agencies such as the various arms of the United Nations? Similarly, must members of the public be co-citizens, or does the international community at large constitute a public in some important respects? Third, what is the scope of the PJR? Does it only apply to questions of constitutional rights and basic questions of justice (Rawls Reference Rawls1996: 227–230), or, does it apply to all spheres of public policy (Quong Reference Quong2011: 273–287)? Finally, for a law or policy to satisfy the PJR, is it enough that members of the public have sufficient reason to accept it when it is formulated in very general terms – i.e. as a set of principles (Nagel Reference Nagel1991: 33; Rawls Reference Rawls1996: 136–137)? Or, must all members of the public have sufficient reason to accept it in all its detail?

Fortunately, I need not resolve all of these questions to determine the implications of the PJR for the evaluation of health states. Regarding the issue of scope, my analysis presupposes that the PJR applies to questions of priority setting in health; but, this presupposition is a reasonable one. First, although Rawls limits the PJR to questions of ‘constitutional essentials’ and ‘basic questions of justice’, the question of health care resource allocation would seem to qualify as a basic question of justice. Although Rawls (Reference Rawls1996: 21, 223–230; Reference Rawls and Kelly2001: 174) abstracts away from the problem of health care in Political Liberalism, he understands basic questions of justice to include those of social and economic equality, of which health care allocation is clearly an example, and also makes clear in Justice as Fairness that the question of health care falls within the scope of a political conception of justice.

Regarding the question of which institutional officials count as public officials and whether the international community constitutes a public, my analysis has important implications even if these questions are answered in the narrowest way possible – i.e. that public officials are only officials of a domestic state and that the public they serve are state residents. As I note above, state PHDMs face important questions of resource allocation in health care and so even if the PJR only applies to state officials, determining its implications for the evaluation of health states would still be a significant contribution. As well, I need not adjudicate the question of public vs. private reasons since it is public reasons that are typically offered in favour of different health states metrics.

My analysis does depend on the question of whether the PJR only requires that policies or laws be acceptable to members of the public when formulated as general principles, or whether it requires that citizens have sufficient reason to endorse all aspects of a particular law or policy. As I discuss in Section 4, Hausman's metric satisfies the PJR on the former interpretation, but not the latter.

Are there cases where there is no law or policy that citizens have sufficient reason to endorse? Yes. In these cases, citizens reasonably disagree about a law or policy. By a reasonable disagreement here, I follow Rawls (Reference Rawls1996: 54–58) in meaning a disagreement by two or more persons that is not the result of errors in reasoning, bias, or stupidity, but that is consistent with each person exercising her rational capacities in a conscientious way. Where citizens reasonably disagree about a law or policy, public reasoning about a policy or law is inconclusive: it does not imply a conclusion that all citizens have sufficient reason to endorse (Gaus Reference Gaus2003: 215–216).

In cases where public reasoning is inconclusive, should the PJR be interpreted to require that governments not act to address the policy problem in question? Not necessarily. It is important to distinguish cases where citizens reasonably disagree about whether government should regulate a specific sphere of activity, from cases where all citizens have sufficient reason to regulate a specific sphere of activity but reasonably disagree about which particular regulations should be implemented (Nagel Reference Nagel1991: 164–165; Gaus Reference Gaus2003: 216). In cases of the former type, it is reasonable to think that government should not act since not all citizens have sufficient reason to think that government action is warranted (Nagel Reference Nagel1991: 164–165; Gaus Reference Gaus2003: 216).

In cases of the latter type however, government action is sufficiently respectful of citizens qua free and equal persons. First, all citizens have sufficient reason to endorse government regulation of some kind. Second, governments can choose policies in these types of cases in a way that is respectful of citizens, namely, through deliberative democratic procedures. Such procedures demonstrate respect for citizens qua free and equal persons since they provide them with the opportunity to reason publicly about the possible policy options, and grant them an equal say regarding the policy that will be chosen (Nagel Reference Nagel1991: 163; Gaus Reference Gaus2003: 218–221). In these types of cases therefore, although citizens do not have a sufficient reason to endorse one particular policy or law, they do have sufficient reason to endorse a deliberative democratic procedure as a way of choosing a policy or law (Nagel Reference Nagel1991: 163; Gaus Reference Gaus2003: 218–221). My analysis proceeds on the assumption that citizens have sufficient reason to accept at least some health-related government regulation, making the valuation of health states a problem for PHDMs.

In sum, according to the PJR, public officials ought to pass laws and policies that are justifiable to the public. In cases where no policy satisfies the PJR, but citizens have sufficient reason to accept that some government action is required, the policy should be chosen through deliberative democratic procedures. In what follows, I investigate whether citizens have sufficient reason to accept any of the three health states metrics identified above.

2. EXPERIENCED UTILITY METRICS AND DECISION UTILITY METRICS

Experienced utility metrics measure the value of health states by determining the hedonic quality of experience of persons living in those states (Dolan and Kahneman Reference Dolan and Kahneman2008: 215). The quality of this experience is measured by compiling reports of the experienced utility or subjective satisfaction of persons in particular health states. This can be done by having such individuals rate the quality of their feelings at random times of day – experience sampling – or complete diary entries detailing the quality of the previous day's experiences – daily reconstruction (Dolan and Kahneman Reference Dolan and Kahneman2008: 228–229).

Proponents of the experienced utility approach rightly understand it to be grounded in sensory hedonism (Dolan and Kahneman Reference Dolan and Kahneman2008: 215), the view that the good life is the pleasant life, where pleasures are understood to be certain feelings or sensations – i.e. experiences – having positive affect (Feldman Reference Feldman2004: 55). This understanding of the experienced utility approach provides it with its strongest justification since health is normatively significant primarily because of the way in which it affects people's wellbeing. For this approach therefore, the value of health states is determined by the quality of experience that they promise.

Because the experienced utility approach presupposes sensory hedonism however, some citizens do not have sufficient reason to endorse it. Citizens can reasonably disagree about whether sensory hedonism is the correct account of wellbeing. This is not to deny that sensory pleasures contribute to the goodness of many lives, and, as I shall argue below, pain caused by disease or disability does diminish the quality of people's lives. However, it is reasonable to think that the value of many projects is not solely determined by the quality of experience that they promise. For many, marathon running, backcountry camping, completing a PhD in philosophy, or raising children are deeply valuable projects, despite the fact that they often promise little sensory pleasure and a good deal of pain – e.g. stress, anxiety and physical discomfort.Footnote ⁵ These citizens are not therefore rationally committed to endorsing the experienced utility approach since they can reasonably object that wellbeing depends on factors other than the quality of one's experience.Footnote ⁶ Experienced utility metrics do not therefore satisfy the PJR.

Decision utility metrics would seem to avoid this problem with experienced utility metrics since they allow health states to be evaluated by considerations other than quality of experience. Instead, these metrics measure the value of health states by determining the strength of people's preferences to avoid occupying them. These preferences can be elicited by means of a number of methods that enable the assignment of cardinal values, including rating scales, standard gambles and time trade-offs (Torrance Reference Torrance1986: 18–20).Footnote ⁷

The most promising justification for decision utility metrics is that wellbeing consists in the satisfaction of preferences. On this view, a person's life goes well to the extent that her preferences for different outcomes or states of affairs are satisfied (Griffin Reference Griffin1986: 10). By eliciting and measuring the strength of people's preferences to avoid particular health states, decision utility metrics determine the extent to which different health states can be expected to impact people's wellbeing (Dolan and Kahneman Reference Dolan and Kahneman2008: 215–216; Hausman Reference Hausman2006: 259). The value of different health states is determined by the strength of people's preferences to avoid occupying them.

Unfortunately, citizens can reasonably disagree about whether wellbeing consists in the satisfaction of preferences. As such, some citizens are not rationally committed to accepting decision utility metrics. The problem is that it is reasonable to think that the satisfaction of preferences is neither sufficient nor necessary for wellbeing.

First, it is reasonable to think that the satisfaction of preferences is not sufficient for wellbeing since people are not always made better off by the satisfaction of their preferences. The chief problem is that people may have false beliefs about the state of affairs or outcomes for which they form preferences. Suppose I grow up in a small town and, seduced by the movies of Woody Allen, formulate a strong preference to live in New York City. Upon arriving however, I am shocked by the cost of living, the constant noise, and the sheer busyness that life in New York City offers, facts that I was either previously unaware of, or failed to appreciate fully. As a consequence, I quickly return home to my small town. It seems counter-intuitive to claim that satisfying my preference to live in New York City improved my wellbeing.Footnote ⁸ This is a problem for decision utility metrics moreover since many people are likely to have false beliefs about what it is like to occupy certain health states (Hausman Reference Hausman2015: 85–90).

It is also reasonable to think that the satisfaction of preferences is not necessary for wellbeing. My wellbeing can be improved by states of affairs for which I have formulated no antecedent preference. Consider pleasant surprises:

It is reasonable to think therefore that the satisfaction of preferences is neither necessary nor sufficient for wellbeing. If the argument in support of decision utility metrics is that preference satisfaction constitutes wellbeing, and so decision utility metrics, by measuring the strength of people's preferences to avoid occupying particular health states, determine the impact of health states on people's wellbeing, some citizens do not have sufficient reason to accept these metrics.

Proponents of preference satisfaction accounts respond to these lines of argument by modifying their view, claiming that wellbeing consists in the satisfaction of informed preferences. On this view, one's wellbeing is determined by the satisfaction of the preferences one would have if one occupied a privileged epistemic standpoint – i.e. had unconstrained access to all of the relevant information concerning the goods and lives one must choose amongst (Griffin Reference Griffin1986: 11; Railton Reference Railton1986: 16; Sobel Reference Sobel1994: 790; Rawls Reference Rawls1999: 358–372). By occupying such a privileged epistemic standpoint, one can consider what it would be like to live one's life in different ways (Sobel Reference Sobel2009: 344). Moreover, it is arguable that decision utility metrics can be modified to accord with this view – to some extent – if respondents are provided with information regarding the health states they are asked to evaluate (Hausman Reference Hausman2015: 89–90).

Although I am sympathetic to the view that wellbeing is constituted by the satisfaction of informed preferences, scholars have raised a number of plausible objections to it. The problem with this view, according to these critics, is that the informed preference account is purely procedural. It does not therefore rule out cases where people's informed preferences are satisfied, but their lives seem intuitively poor in wellbeing; and it does not rule out cases where people's informed preferences are not satisfied, but their lives seem intuitively rich in wellbeing. Consider John Rawls's (Reference Rawls1999: 379–380) case of the grass-counter as an example of the former type of case:

Some scholars find it deeply counter-intuitive to claim that the grass-counter is living a life rich in wellbeing and this skepticism strikes me as reasonable (Arneson Reference Arneson1999; Crisp Reference Crisp and Zalta2015; Parfit Reference Parfit and Scheffler2011: 75–82).

Some citizens can therefore reasonably object to the use of decision utility metrics when these metrics are grounded in the claim that preference satisfaction is the correct account of wellbeing. But, there is another justification for these metrics. As I argue above, in cases where public reasoning regarding a policy is inconclusive but where there is agreement that some policy is necessary, the PJR implies that government should choose a policy by means of a deliberative democratic procedure. It is arguable that suitably designed decision utility metrics can operate as such a procedure for the purposes of evaluating health states since they involve citizens directly evaluating health states. Indeed, decision utility metrics, insofar as they determine the value of health states by incorporating and averaging the health-related preferences of a representative sample of people, can be understood as a democratic procedure that gives equal weight to each person's conception of wellbeing. Decision utility metrics, after all, allow hedonists to evaluate health states by reference to the hedonic quality of experience of people in those states. They allow perfectionists or those holding objective list views of wellbeing to determine the value of health states by reference to the substantive goals and projects that they deem to be valuable. And, they allow people to determine the value of health states by reference to the values and preferences they take to be constitutive of a good life. If playing soccer is my favourite activity, I can rate health states that involve the loss of a foot very low. If I prefer intellectual activities to physical ones, I can rate health states involving loss of cognitive capacity lower than those involving loss of physical capacities.

However, before concluding that decision utility metrics – understood as democratic procedures – are best from the standpoint of the PJR, we need to first establish that public reasoning regarding the value of health states is indeed inconclusive. I argue below that this conclusion is not warranted.

3. HAUSMAN'S ACTIVITY LIMITATION/DISTRESS METRIC

In his recent book, Valuing Health: Well-Being, Freedom, and Suffering, Daniel Hausman defends what I shall refer to as the Activity Limitation/Distress (AL/D) metric. On Hausman's (Reference Hausman2015: 155) view, the value of health states should be determined by: (1) the extent to which these states limit the freedom or activities of people; and (2) the pathological suffering they involve.

Hausman's (Reference Hausman2015: 158–159) central justification for this metric is that the value of health that matters to health policy is not the ‘private value’ of health, that is, the value of health to people's individual wellbeing; but rather the ‘public value’ of health, that is, the value of health to the public. Since the purpose of the liberal state, Hausman (Reference Hausman2015: 159–160) argues, is not to promote private views of the nature of wellbeing or the good life, government policy ought not to be governed by such views. The central responsibility of the liberal state, Hausman (Reference Hausman2015: 160) argues, is instead to ‘protect rights, secure justice, to address need and suffering, to expand opportunities, and to insure individuals against catastrophes, even of their own making’. A state that successfully fulfils this responsibility, Hausman (Reference Hausman2015: 161) claims, is a ‘liberal facilitator state’.Footnote ⁹

Since the liberal facilitator state must be largely neutral amongst private views of the good life or wellbeing, focusing instead on expanding ‘the range of personal choices that are open to its citizens’, the public value of health states should depend on the extent to which particular diseases and disabilities limit the activity of individuals (Hausman Reference Hausman2015: 163). Since the chief responsibility of the liberal state is to facilitate the activities of its citizens, the public valuation of health states should be based on the extent to which particular health states limit these activities (Hausman Reference Hausman2015: 163). Additionally, Hausman (Reference Hausman2015: 163) claims, because the liberal state also has duties of care and compassion, it has a legitimate role in ‘reducing suffering’. For this reason, the public value of health states should also depend on the suffering that these states involve (Hausman Reference Hausman2015: 164). With respect to health policy therefore, Hausman (Reference Hausman2015: 166) defends the position he calls ‘benevolent liberal facilitation’:

Hausman's AL/D metric, partially grounded as it is in the idea of liberal neutrality, is promising from the standpoint of the PJR. As I note above, the idea of liberal neutrality is often understood to be grounded in the PJR. However, Hausman (Reference Hausman2015: 160) does not attempt to show that his approach satisfies the PJR, relying instead on the general idea that the purpose of the liberal state is to ‘preserve and enlarge the possibilities for the successful pursuit of a wide range of individual goals, rather than joining state power to the efforts of individuals to enable them to reach their goals’. Additionally, the suffering component of Hausman's metric is not justified by appeal to the idea of liberal neutrality, but instead by appeal to the state's duty of compassion. Thus, even if one were tempted to accept the claim that the activity component of Hausman's metric satisfies the PJR, Hausman provides no reason to think that the suffering component of his metric does so.

In what follows therefore, I first outline Hausman's AL/D metric. I then investigate whether it satisfies the PJR.

3.1. What is Hausman's AL/D Metric?

Hausman's AL/D metric claims that the value of health states should be determined by (1) the extent to which they limit people's activities; and (2) the pathological suffering they involve. By ‘pathological suffering’, Hausman (Reference Hausman2015: 171) means the physical discomfort – e.g. negative sensations such as pain or nausea – or emotional distress – e.g. negative mental states such as depression or panic – that are component parts of health states, not reactions to them. Hausman recognizes that it is likely impossible to construct a complete, fine-grained ranking of the possible forms and combinations of physical discomfort and emotional distress. But, he suggests that a complete ordering of health states with respect to the disvalue of pathological suffering is possible if these states are sorted into a small number of broad categories such as: ‘severe suffering, moderate suffering, mild suffering, or little or no distress’ (Hausman Reference Hausman2015: 172). Such a coarse ordering of health states, Hausman (Reference Hausman2015: 172) claims, is likely to be sufficient for the purposes of priority setting.

To measure the extent to which health states limit people's activity, Hausman (Reference Hausman2015: 176) proposes that health states be ordered in terms of the following classification of activity limitations:

1. 1. Not limited.

2. 2. Minor social or occupational limits.

3. 3. Major social or occupational limits.

4. 4. Physically limited in IADL.

5. 5. Physically limited in ADL or cognitively limited in IADL.

6. 6. Cognitively limited in ADL.

By social or occupational limits, Hausman (Reference Hausman2015: 176) means ‘limits on the range of occupations, recreations, and social relations that are available to the individual as a consequence of their health’. The social and occupational activities in question are those available to people generally in their society, and do not refer to specific activities or projects individuals have chosen to pursue (Hausman Reference Hausman2015: 176). ADL – activities of daily living – include: ‘eating, getting in/out of bed, inside mobility, dressing, bathing, and toileting’ (Hausman Reference Hausman2015: 174). IADL – instrumental activities of daily living – include: ‘light housework, laundry, meal preparation, grocery shopping, outside mobility, travel, money management, and telephoning’ (Hausman Reference Hausman2015: 174). Hausman (Reference Hausman2015: 176) classifies cognitive limits as more severe than physical limits since people who are physically incapable of performing ADL or IADL may still be cognitively capable of engaging in a variety of intellectual activities. Those who are severely cognitively disabled, by contrast, are unable to direct their lives.

Hausman does not commit himself to the specific categories he proposes – he recognizes that research may demonstrate that different classification systems may prove to be more accurate and intuitive. But, Hausman's system provides a framework for classifying different health states in terms of (1) the extent to which they limit people's activity; and (2) the pathological suffering they involve:

Particular health states can be understood as ‘limitation/distress (L/D) pairs’, with the above chart allowing decision-makers to map particular health states onto one of the 24 cells (Hausman Reference Hausman2015: 177). Each cell would then be assigned a QALY value from 0 to 1, with 1 indicating full health.

The central challenge for Hausman's (Reference Hausman2015: 179) approach is to assign these QALY values in a principled way, thus constructing a ‘cardinal scalar measure of the public value of limitation/discomfort pairs’. To construct such a measure, Hausman must not only determine the value, on scales of 0 to 1, of the different categories of activity limitation and distress he identifies above, but also combine these scales into a single metric, making it possible to compare the value of health states that limit activity but involve no distress and vice versa, and assign values to health states that limit activity and involve distress. As Hausman (Reference Hausman2015: 180) recognizes, this involves answering a number of difficult normative questions: should the worst possible health state be assigned a 0, or is such a state worse than death? Should increases in activity limitation always yield the same decline in QALY value, regardless of the level of distress and vice versa (Hausman Reference Hausman2015: 180–181)? If the answer to this question is affirmative, how should numbers be assigned to the gaps between different categories of activity limitation and different categories of distress (Hausman Reference Hausman2015: 181)? This last question is particularly difficult, requiring one to compare the disvalue of activity limitations to the disvalue of different levels of distress.

Hausman (Reference Hausman2015: 180–186) provides an illustration of how one might answer these questions that appeals to public, not private, values; but, he admits that his account is inconclusive. He claims therefore that the assignment of values to limitation/distress pairs should ultimately be left to a public procedure, whether this takes the form of public deliberation, public surveys, or citizen juries (Hausman Reference Hausman2015: 186–187).

4. HAUSMAN'S METRIC: A CLEAR ALTERNATIVE TO DECISION UTILITY METRICS?

Although I think there is a principled difference between Hausman's metric and decision utility metrics from the standpoint of the PJR, this last objection raises an important question. Given that Hausman's metric relies on a procedure to evaluate health states, to what degree is it really superior to decision utility metrics from the standpoint of the PJR, where the latter metrics are understood to offer citizens a fair democratic procedure for adjudicating their reasonable disagreements about the nature of the good life? I take this question to not only be theoretically interesting, but also, as I explain below, of interest to policymakers.

Hausman would no doubt claim that although his metric relies on a deliberative procedure to evaluate health states, there is an important difference between the procedure his metric employs, and the procedure decision utility metrics employ. Decision utility metrics assign values to health states by (1) eliciting people's personal valuations of them; and (2) aggregating these valuations. In the face of reasonable disagreement regarding the nature of wellbeing, these metrics therefore aggregate individuals’ private views regarding the impact of various health states on their personal wellbeing, whatever their view of their own wellbeing happens to be. By contrast, Hausman envisions a different type of procedure for assigning QALY values to limitation/distress pairs. Rather than appeal to people's private preferences regarding these pairs, Hausman (Reference Hausman2015: 186) suggests that QALY values be assigned through a deliberative process in which participants reason together about the value of different limitation/distress pairs. As an example of such a process, Hausman (Reference Hausman2015: 187) suggests the use of deliberative groups such as citizen juries. As Hausman (Reference Hausman2015: 186) puts it:

As this quote suggests, Hausman (Reference Hausman2015: 179–186) offers an illustration of an evaluation procedure that he claims avoids reliance on the aggregation of individual preferences. This procedure is not intended to be the final word on this matter, but only intended to illustrate how such an evaluation might begin (Hausman Reference Hausman2015: 179). Most notably, Hausman (Reference Hausman2015: 182–183) suggests that the value of different activity limitations could be determined by (1) the (adjusted) median labour incomes, and (2) the reported ‘subjective experience’ of those with the limitations in question. Median incomes, Hausman (Reference Hausman2015: 182) suggests, can be understood to offer a public valuation of different limitations. Reported subjective experience, Hausman (Reference Hausman2015: 182–183) claims, can act as a proxy for the value of different activity limitations since people with greater limitations are likely to be more miserable.

Hausman would probably claim that his proposed evaluation procedure constitutes an important difference between his metric and decision utility metrics from the standpoint of the PJR. However, I am deeply sceptical that Hausman's proposal of tasking deliberative groups to reason together to assign QALY weights to limitation/distress pairs offers a clear alternative to the procedure employed by decision utility metrics. It is simply not clear how participants can rationally assess the disvalue of certain limitation/distress pairs except by asking how these pairs impact people's wellbeing, and it is not clear how participants can do this without employing – perhaps implicitly – particular conceptions of wellbeing. Hausman's own specific proposal for the evaluation of limitation/distress pairs illustrates this problem. Evaluating health states on the basis of median labour incomes seems, at base, no different from appealing to people's private preferences. What are labour incomes, after all, but the result of people's aggregated preferences for certain types of services? Additionally, the appeal to reported subjective experience seems to use a sensory hedonist metric to measure the disvalue of activity limitations, which seems highly questionable if the goal is to get away from evaluating health states on the basis of controversial conceptions of wellbeing.

Now, there may be a variety of different ways in which people's private views of wellbeing could be employed to evaluate limitation/distress pairs and Hausman may have reasons to favour one rather than another.Footnote ¹⁶ For example, Hausman might think that these pairs should be evaluated by deliberative groups, tasked with answering the question: to what extent are the particular limitation/distress pairs likely to impact the wellbeing of one's fellow citizens? Hausman might also think that certain illiberal conceptions of the good life should not be considered in the evaluation of health states for the purposes of informing government policy, and so think that there are constraints on the conceptions of wellbeing that may be appealed to.

However, if Hausman's deliberative participants can't avoid evaluating limitation/distress pairs by appeal to controversial conceptions of the good life, it seems far more democratic to employ something like a decision utility metric to assign these values. Policymakers could ask citizens directly to assign values to limitation/distress pairs, rather than doing so through the type of deliberative procedure Hausman has in mind which may exclude some voices or give greater weight to others. Importantly, on this proposal the decision utility metric in question would be nested within Hausman's AL/D metric. Respondents would be asked to assign QALY values to limitation/distress pairs where these pairs have already been ordered at a general level by Hausman's metric. The point would be to assign QALY values to each of the empty boxes in the above chart, with the chart imposing constraints on how respondents may assign these values: boxes to the right must receive lower QALY values than boxes to the left and boxes below must receive lower QALY values than boxes above. From here on, I shall refer to such a proposed metric as Hausman's AL/D metric*.

Hausman's AL/D metric*, though it contains a nested decision utility metric, is still different from a decision utility metric. Hausman's metric* places constraints on the ways in which respondents can assign QALY values: health states that involve greater activity limitation and greater pathological suffering must be ranked worse than health states that involve lesser activity limitation and lesser pathological suffering. Of course, this only matters if some respondents possess conceptions of the good life that would lead them to rank health states differently. While it is difficult to imagine anyone possessing a conception of the good life that would lead them to rank health states with mild pathological suffering as worse than health states with moderate pathological suffering (assuming the health states in question possess the same activity limitation); it is possible to imagine someone possessing a conception of the good life that would lead them to rank a health state that involves physical limitations in instrumental activities of daily living as better than a health state that involves only minor social or occupational limits (assuming these states possess the same degree of pathological suffering). If playing the violin is my passion and vocation, I might genuinely prefer being a paraplegic to having a condition that limits the dexterity of my hands.

There is thus a genuine difference at the level of principle between Hausman's AL/D metric* and decision utility metrics. Although both rely to some extent on people's conceptions of the good to evaluate health states, Hausman's metric* places constraints on the extent to which these private conceptions of the good can influence health state valuations. These constraints ensure that health states are ordered in terms of the degree to which they limit people's activity and the degree of pathological suffering that they involve. Given the interpretation of the PJR that I offer above, Hausman's AL/D metric* is still relevantly different from decision utility metrics, though not as superior as Hausman might claim.

Despite this difference at the level of principle however, it is important to recognize that Hausman's metric* may not imply health state valuations that different from those implied by decision utility metrics. Whether it does or not will depend on contingent empirical factors. For example, if both metrics employ the same elicitation tool – e.g. time-trade-off – and if there are very few respondents like the violinist above, both metrics should imply similar health state evaluations.

To sum up, decision utility metrics not only offer a sufficiently liberal procedure for evaluating health states should public reasoning prove inconclusive. If I am correct that Hausman's AL/D metric should incorporate a decision utility metric as a fair procedure for assigning values to limitation/distress pairs, decision utility metrics may not yield health state valuations that differ significantly from Hausman's metric so understood – i.e. Hausman's AL/D metric*. Why might this conclusion matter?

Suppose you are a policymaker of a liberal society and you are responsible for deciding how to assign QALY values to health states. You currently use a decision utility metric to do so but after reading Hausman's Valuing Health you are concerned that doing so does not constitute a liberal approach to health state valuation. Suppose too that switching from your current decision utility metric to Hausman's AL/D metric is quite costly, involving re-training of your staff, carrying out new evaluations etc. and that Hausman's metric faces a number of implementation challenges. Suppose finally that you agree with my claim above that the most democratic way to complete Hausman's AL/D metric is to nest a decision utility metric within it. Given this last conclusion of my analysis, it's not clear to me that you should make the switch. Unless your society has a substantial number of people disposed to rank health states in a way that would violate the constraints of Hausman's metric, switching is unlikely to change your health state valuations significantly and so is unlikely to change how resources are allocated. Additionally, my analysis also shows that it is possible to understand decision utility metrics as adequately liberal – i.e. not best from the standpoint of the PJR, but a close second. Preferring less just policies for reasons of cost or because of implementation challenges is always dangerous, but it strikes me that it's not clear what the best decision is on balance.