In deciding how best to use its limited resources, the National Health Service needs to attend to health care needs fairly and efficiently. On the assumption that the constraints of fairness have been met, the NHS should try to use its resources to provide the greatest health benefits possible. To do that, it needs a quantitative measure of the health benefits provided by different possible actions. The measure that the National Institute for Health and Clinical Excellence mainly relies on is the QALY, the quality-adjusted life year. Although the institutional details differ, every health system faces fairness constraints and the need to compare the benefits of alternative policies.

Measures of health such as the QALY usually suppose that the value of people's health can be measured by integrating the value of their instantaneous ‘health states’ over time. No one believes that this is quite right: five minutes of severe migraine is not one-twelfth as bad as a severe migraine for an hour. But, like Paul Dolan, I shall accept this convenient simplification. The problem Dolan is concerned with lies in determining the value of health states. With death assigned a value of 0 and full health assigned a value of 1, the problem is to assign numbers to each of the health states in some health classification system. The system Dolan mentions, the EQ-5D has 243 distinguishable health states (plus death). The Health Utilities Index, developed in Canada, has 972,000.

Weights or values are assigned to a few of these health states by eliciting preferences from respondents, and weights are assigned to the rest by imputing to respondents a multi-attribute utility function whose parameters are estimated from the elicited preferences. Health economists have provided few arguments in defense of their reliance on preferences. Some simply equate value and preferences.Footnote ¹ Others assume that the value of health states consists in individual welfare and take preferences to be the only tractable measure of welfare.

As Dolan points out, relying on preferences is problematic. Since the weights attached to states such as the one Dolan describes (healthy apart from problems walking) differ systematically and dramatically between those who experience those problems and those who do not, the weights that derive from the preferences of one or the other of these groups (if not both) are mistaken. Dolan's diagnosis and solution are in outline as follows: ‘The focus of my critique is that the preferences of the general public are not reliable because the public are not good at assessing what it would be like to experience different states of health’ (p. 2). ‘To more accurately reflect the effect of different health states on people’s well-being, I propose that policy-makers in health and elsewhere should shift their attention from the measurement of preferences towards the measurement of experiences’ (p. 3). I agree that the preferences of the general public are not reliable, but I disagree about the reason why (Hausman, Reference Hausman2006). Dolan's diagnosis assumes that what matters about health states is their bearing on well-being, and he argues that the best measure of this bearing is the quality of the experience associated with the health state. These claims are dubious. For example, some of those with advanced dementia due to Alzheimer's disease appear to be genuinely quite happy. But obviously they are nevertheless severely disabled. Either experience is a faulty guide to well-being or well-being is the wrong way to assess health states or both. The cure for the inaccuracies of relying on preferences that Dolan proposes is worse than the disease. Although Dolan is right to criticize the reliance on preferences, he is wrong to propose that measures of health states be based instead on experiences.

Let us step back and to ask what a ‘correct’ valuation of health states would be, given that the objective is to guide the allocation of health-related resources (subject, of course, to the constraints of fairness). For other purposes, one might want different things from a system of valuing health states. In the quotation above, Dolan assumes that a correct assessment of the value of health states should ‘reflect the effect of different health states on people's well being’. Whether this is correct depends on what the objectives of the agencies distributing health-related resources should be.

There are two main objectives defended by economists, political theorists, and philosophers, which I will call the liberal and the welfarist. In principle, they are quite different, though in practice the differences may be small. The liberal view is that social policies should aim to provide a framework in which individuals are best able to pursue their own projects. Exactly what this requires is subject to debate, with libertarians holding that the protection of individual rights, narrowly conceived, exhausts the legitimate scope of social policies, while other liberals (more reasonably in my view) take a more expansive view of rights and also hold that social policies should aim to provide individuals with insurance against misfortune and resources and opportunities to pursue what they value. Liberals can regard the state as a protector, insurer, and arbitrator, but not as a big brother or an active partner in individual pursuits.

The welfarist view, in contrast, holds that social policies should advance individual welfare. It might consequently appear that a welfarist would endorse Dolan's view that the value of health states depends on their effect on individual welfare. But matters are more complicated. The welfarist thinks of the state as a bit like everybody's mother. Advancing individual welfare might require intrusions into individual life. After all, mother knows best. Nothing in the welfarist view automatically implies that social policy should aim at protecting individual rights or expanding the range of projects that individuals are able to pursue. Welfarists in principle are, however, typically, liberals in practice, because the constraints on social policy that the liberal favors on principle turn out to be excellent ways to advance individual welfare. Such is John Stuart Mill's classic argument in On Liberty. Although it is obviously possible for the government to coerce individuals for their own good, the risks of doing so, the likelihood that the government will be misinformed or badly motivated, and the costs of encouraging individual passivity make it wise to insist on the policies that liberals defend.

So let us assume that whether policy is liberal or welfarist in principle, it is liberal in practice. That means that what matters about health states from a public perspective is their effect on the range of projects that individuals can pursue rather than their effect on individual welfare. Setting aside again issues concerning fairness, agencies with responsibility to the public should distribute health-related resources so as to protect individual rights and to secure for individuals the resources and opportunities to pursue what is most important to them. The information that the policy-maker needs concerns the relevance of health states to the range of activities that are open to individuals and the capacities individuals have to succeed and excel in those activities.

Neither the bearing of health states on well-being nor the extent to which individuals prefer health states, nor the subjective experience of the health states will coincide, even approximately, with a measure of their bearing on the range of activities that are open to individuals. Each of these four possible ways of ranking health states is distinct, and the four rankings will often conflict. Consider, for example, from these four perspectives a sensory deficit such as deafness. From the first three perspectives the quality adjustment called for by deafness is uncertain and probably rather small. The extent to which deafness limits well-being is uncertain (though of course the experience of losing one's hearing is a wrenching one). Many people who cannot hear live very well all the same. With respect to preferences, it is noteworthy that, when faced with the possibility of some restoration of their hearing, many of those who are deaf prefer to remain deaf. I know of no studies on the subjective experience of those who are deaf, but clearly many are very happy. Only the fourth ranking in terms of activity limitations explains why deafness is a significant disability. Among the projects that remain open to those who are deaf are, of course, many of the greatest value, but what is relevant from a practical liberal perspective is what range of projects are open to people, not how well people's lives turn out. Although a successful life requires a benign environment, how well people live is largely their own responsibility.

In contrast to Dolan, who writes, ‘In this context, I suggest that ‘relevant information’ should include an understanding of the likely future experiences associated with the different choices’ (p. 3), I suggest that subjective experience is not a good guide to well-being, and well-being is not what matters to public evaluation. Even if it were true that ‘an individual’s assessment of her life offers a more defensible measure of well-being than the satisfaction of her actual preferences’ (p. 7), health states should not be evaluated by their bearing on well-being. How people feel when in a particular state of health or how they expect to feel has no clear relation either to their well-being or (more importantly) to the range of projects they can successfully pursue.

Dolan has been influenced by recent work by psychologists, including especially Daniel Kahneman (Dolan and Kahneman, Reference Dolan and Kahneman2008). This work provides compelling evidence that both people's anticipation of their future feelings and their memories of their past experiences may conflict with the sum of the momentary pleasures or pains they experience (Kahneman, Reference Kahenman, Kahneman and Tversky2000). For example, individuals who are subjected to an uncomfortable procedure such as a sigmoidoscopy, which is prolonged by an additional period of milder discomfort at the end, typically judge the episode as a whole to be less painful than a sigmoidoscopy that is not similarly prolonged. If one ‘adds up’ current experiences to determine the experiential quality of whole episodes, one avoids such conflicts. Moreover, current feelings can be linked to neurological indicators, and correlate with factors such as facial expressions. Consequently, Kahneman and his co-workers maintain that momentary feelings rather than anticipations or recollections of experiences are what matters, and they have devised methods, such as the day reconstruction method that Dolan mentions to measure momentary feelings.

In my view, Kahneman and Dolan draw the wrong conclusions from the experimental results. Rather than supporting a concern with momentary feelings, the conflict between the sum of such feelings on the one hand and people's anticipations and recollections on the other undercuts the significance of feelings. Suppose that policy P would lead to a change in health that individuals would recall as a greater improvement than the changes due to policy Q, but policy Q would lead to a larger sum of net momentary good feeling. From a whole-life perspective, which makes people better off, P or Q? The fact that one faces such a choice gives one reason to look elsewhere to appraise policies.

The choice is not, as Dolan puts it, between ‘think[ing] of public interventions as benefiting people if their preferences are satisfied or if their experiences are enhanced’ (p. 7). To decide whether a health state described as ‘some pain or discomfort’ is worse than another described as ‘moderately anxious or depressed’, liberal public policy needs to know instead which state poses the more serious limitations on what people can do. To focus as Dolan argues on the values of the momentary experiences of health states is, I think, a step backward from relying on preferences. We ought instead to be taking a step forward away from welfare, preferences or experiences toward measures of the bearing on health states on what people are able to do.

Rather than joining Dolan ‘in a research endeavour that seeks to ration health care in ways that improve the real experiences of how people think and feel about their lives rather than in ways that satisfy their hypothetical preferences over how they imagine thinking and feeling’ (p. 8), I would urge readers to explore ways to ration health care in ways that are both fair and that expand and protect the range and value of the projects that individuals can pursue.