What Makes Health Care Special?An Argument for Health Care Insurance
While citizens in a liberal democracy are generally expected to see to their basic needs out of their own income shares, health care is treated differently. Most rich liberal democracies provide their citizens with health care or health care insurance in kind. Is this "special" treatment justified? The predominant liberal account of justice in health care holds that the moral importance of health justifies treating health care as special in this way. I reject this approach and offer an alternative account. Health needs are not more important than other basic needs, but they are more unpredictable. I argue that citizens are owed access to insurance against health risks to provide stability in their future expectations and thus to protect their capacities for self-determination.
Citizens in wealthy liberal democracies are typically expected to see to basic needs like food, clothing, and shelter out of their own income, and those without the means to do so usually receive assistance in the form of cash transfers. Things are different with health care. Most liberal societies provide their citizens with health care or health care insurance in kind, either directly from the state or through private insurance companies that are regulated like public utilities. Except perhaps for small co-pays or deductibles, citizens are not expected to see to their health needs out-of-pocket, at least for some basic level of care. Instead, when a citizen gets sick, the state or her insurance company foots the bill, and this is so regardless of how well off that citizen might otherwise be.
Shlomi Segall captures what is "special" about health care here by observing that, on this arrangement, health care is distributed in isolation from other social goods, particularly income and wealth (2007, 344). Ordinarily, when a citizen consumes some resource, she is expected to [End Page 561] pay for it, and thus to forego consumption of something else of equivalent value. But with health care, the amount of care a person receives does not affect, nor is it affected by, the contents of the rest of her bundle of social goods. (Again, perhaps with the exception of a small co-pay, and perhaps only for some basic level of care; hereafter I leave these qualifications aside.) In this way, health care is distributed more equally than it might be if it were treated continuously with mere consumer goods, at least in the sense that no one is denied needed care due to inability to pay.
The predominant liberal approach to justice in health care holds that the moral importance of health justifies treating health care as special in this way.1 Different theorists articulate the moral importance of health differently, but they agree that health's moral importance is what makes health care distributively special. I argue that any account of the moral importance of health will be at once too narrow and too broad. It will be too narrow in the sense that not all health care services will promote the morally salient feature of health, whatever that is taken to be, and it will be too broad in the sense that not all of the things that promote the morally salient feature of health will be health care services. No account of the moral importance of health will capture all and only health care services, and thus no account will justify treating health care (and health care alone) as special.
In this paper, I put forward a different justification for treating health care as special from within a liberal view. Liberalism begins from the idea that respect for persons requires that we protect each person's capacity to form, revise, and pursue a rational plan of life. This entails, among other things, that each person is owed a fair share of the rights, opportunities, and other social resources necessary for the effective exercise of that capacity. When it comes to fair shares, though, what matters is not only the resources that people can count on, but also how reliably they can count on them. At some margins, people's capacities for self-determination are better served by a more secure bundle of resources rather than a larger bundle. I suggest that this is why liberal democracies should provide their citizens with various forms of social insurance: health insurance, of course, but also insurance against sudden unemployment, against outliving one's savings (i.e., old-age pensions), against disability, and so on. They should provide these insurance policies because of the threat that risk poses to the capacity for self-determination, and because of the weaknesses of private markets in delivering these kinds of insurance products efficiently. My focus here is health care insurance, but the arguments I raise could be extended to other forms of social insurance as well.2 [End Page 562]
Once we have a case for providing people with insurance against health risks, the case for treating health care as special in Segall's sense falls naturally into place. The very purpose of health insurance is to prevent unexpected health care needs from overwhelming a person's budget; its purpose, in other words, is to isolate a person's health care consumption from the rest of that person's basket of social resources. My view, then, is that health care should be distributed in isolation from income and wealth because citizens should have insurance against health risks, and citizens should have insurance against health risks in order to provide stability to their future expectations and thus to protect their capacity to form, revise, and pursue a rational plan of life.
I begin in the next section by rejecting the view that the moral importance of health justifies treating health care as special. The following section sets out my view that health care should be treated as special because of the importance of providing citizens with security against health risks, and because of the failures of private markets in delivering health care insurance.
THE MORAL IMPORTANCE OF HEALTH VIEW
Perhaps the strongest motivation for treating health care as special comes from the problem of expensive and unequal health needs. Health care costs in the developed world exhibit a pattern that economists refer to as the "eighty–twenty rule:" in any given year, twenty percent of patients typically account for about eighty percent of costs.3 A theory of justice that treats health care continuously with other basic needs, trusting that citizens will purchase the care they need out of their fair income shares, would leave many citizens unable to afford necessary care. This seems intuitively unjust, and is perhaps the strongest motivation behind the desire for a distinct account of justice in health care—one that justifies isolating the distribution of health care from that of other social goods.
The problem of expensive and unequal medical needs does not by itself provide a justification for treating health care as special on a liberal view, however. This is because of liberalism's strong presumption against paternalism in the distribution of social resources. If the problem is simply that some people have very expensive health needs, the liberal solution should be to give those people money; it is a further step to the claim that the sick are owed health care services or health care insurance provided in-kind. [End Page 563]
One reason for liberals' anti-paternalism has to do with efficiency: providing goods in kind will often leave people worse off than they could be with a more fungible form of transfer which they could use to purchase something they might prefer. But there is also an argument from equality, in that providing goods in kind will inevitably privilege certain conceptions of the good over others (Heath 2008, 500; Dworkin 2000, 147ff.). A hedonist committed to the principles "live fast" and "die young" might be better off with the cash equivalent of his health care needs to put towards a sports car or an exotic vacation. A social order that provides him with only health care will leave him disadvantaged in the pursuit of his conception of the good relative to others whose life plans favor more staid pursuits. This is not to say that the hedonist is owed a check instead of health care; it is only to point out that, if he is not owed a check, he is owed an argument that explains why.
A non-paternalistic rationale for the in-kind provision of health care services might appeal to the fact that meeting health care needs is somehow more important or more urgent, morally speaking, than merely satisfying people's preferences. T. M. Scanlon expresses a view like this in "Preference and Urgency:"
The fact that someone would be willing to forego a decent diet in order to build a monument to his god does not mean that his claim on others for aid in his project has the same strength as a claim for aid in obtaining enough to eat (even assuming that the sacrifices required of others would be the same).(1975, 659–60)
On a view like this, the objective importance of meeting certain needs underlies society's obligation to provide necessary goods like health care, independent of the subjective importance that people may attach to having those needs met.
Considerations like this motivate what I will call the moral importance of health view, or "the MIH view" for short.4 Roughly, the MIH view holds that health's moral importance explains health care's distributive specialness. On the MIH view, health care is important because health is important, and health in turn is important either intrinsically or because of the contribution that health makes to some further value or values.
Many liberals who have defended some version of the specialness thesis have done so by appeal to the moral importance of health.5 My aim in this section is to show that the MIH view cannot give a satisfactory justification for treating health care as special. I pay particular attention to the work of Norman Daniels, but I also present my argument in a general form. [End Page 564]
One problem with the MIH view is that it is not easily reconciled with the liberal commitment to neutrality among conceptions of the good. It is difficult to give an account of "basic" or "urgent" needs that will not be bound up with some particular conception of the good, one that some citizens may reasonably reject (Engelhardt, Jr. 1996, 382–84). If we want a conception of justice to be "political" in Rawls's sense—formulated independently of any particular comprehensive moral view and capable of serving as the object of an overlapping consensus (Rawls 1993, Chapter 4)—then we may think that the MIH view is doomed from the start. But that is not my objection to the MIH view.
My objection is that, for any value (or combination of values) put forward to explain the moral importance of health, not all health care services will promote that value, and not all of the things that promote that value will be health care services. Each of these creates problems of its own. Most obviously, because the MIH view treats health care as instrumental to some further value, it cannot justify the provision of health care that does not actually contribute to the value or values in question; thus many important health care services will inevitably be excluded from health care justice on the MIH view. Just as important, though, the MIH view also cannot fail to justify the provision of other goods, non-health-care goods, when those goods actually do contribute to the value in question. This is not always a problem, of course; health care may not be the only good that should be treated as special. But the logic of the MIH view pushes its defenders toward treating lots of goods as special, ranging far beyond what most liberals think should be socially provided.
Norman Daniels's Just Health
We can see these problems clearly in Norman Daniels's landmark account of justice in health care, which is by far the most sophisticated and influential exemplar of the MIH view.6 Daniels sets out his view as follows:
(1´) Since meeting health needs promotes health (or normal functioning), and since health helps to protect opportunity, then meeting health needs protects opportunity. (2´) Since Rawls's justice as fairness requires protecting opportunity, as do other important approaches to distributive justice, then several recent accounts of justice give special importance to meeting health needs.
Here we see the basic outline of the MIH view: meeting health needs is important because it promotes health, and health is important because [End Page 565] it protects opportunity. It is the value of opportunity that ultimately explains the importance we attach to meeting people's health needs. Because liberal theories of justice like John Rawls's require that we protect citizens' opportunities, these theories can be understood to attach "special importance" to meeting health needs.
Two brief clarifications of Daniels's view are in order. First, by opportunity, Daniels means opportunities to pursue life plans. Health is of special moral importance because pathology, the absence of health, reduces the range of opportunities available to individuals to construct and pursue their life plans or conceptions of the good (Daniels 2008, 35). While Daniels does not deny that health is also important for other reasons, for him it is people's fundamental interest in maintaining a normal range of opportunities to form and pursue life plans that explains why we attach special importance to meeting health needs. This understanding of opportunity is an important shift from Rawls, who understands opportunity in a more limited sense. Rawls's opportunity principle requires that we protect fair equality of opportunity to compete for jobs and offices, not equality of opportunity to pursue any of our life plans (Rawls 1999, 63). I return to this point later.
Second, Daniels understands health needs to encompass more than just health care services as those are traditionally understood. Daniels has long emphasized that the determinants of health are many, including adequate nutrition, sanitation, and other public health measures (1985, 32; 2008, 42–43). In his more recent work, he has also emphasized the impact of the social determinants of health on population health, including such things as socioeconomic status, workplace hierarchies, and social integration. Daniels argues that these, too, should be included among health needs and so should also be distributed more equally than other goods (2008, 79–102). Here I am concerned with Daniels's argument for treating personal health care services as special, and I leave aside any complications that may arise from the broader scope of Daniels's project.7
Daniels's account exemplifies both of the problems I identified with the MIH view: not all health care services promote opportunity, and not all things that promote opportunity are health care services. Thus Daniels's view fails to provide an adequate justification for treating health care as special. I take up each of these problems in turn.
First, many important health care services cannot plausibly be said to promote opportunity. Many simply relieve pain, prolong life, or facilitate a dignified death (Buchanan 1984, 63; Segall 2007, 347–49; Stern 1983, [End Page 566] 346–49). This is especially important given that around 40% of all health care spending in the developed world goes toward the elderly, and a large chunk of that goes toward those in the last few months of their lives.8 Whatever else we may wish to say about these folks, their major opportunities, their major life plans, are all behind them. Providing health care in these kinds of cases will look mysterious on a view like Daniels's (Segall 2007, 347–49; 2010, 33).
Daniels does nibble at this bullet, but he does not bite it quite as unreservedly as the logic of the MIH view requires. Daniels argues that we have reason to limit health care spending on those over the age of 75. On his "Prudential Lifespan" account, the problem of health care for the aged is to be viewed, not in terms of redistribution from young to old, but in terms of how a single prudent person would allocate her own consumption of health care over a complete life. A prudent person would reserve some of her health care resources for later in life, Daniels argues, but she would do so in the knowledge that life-years after a certain point become less valuable to the agent than life-years before that point, due to the normal ageing process (2008, 171–81).
What is significant about the Prudential Lifespan account is that, if it solves the problem of health care provision for the elderly, it does so only because prudent planners would not allocate their access to health care resources solely to protect their shares of opportunity (Brock 1989, 305). In other words, to the extent that the Prudential Lifespan account solves the problem, it does so by introducing non-opportunity considerations into the distribution of health care, thus moving away from the MIH view. As Segall observes, if Daniels's view is that the aim of just health care is to protect fair equality of opportunity, that would seem to imply, not that we should devote fewer public health care resources to the elderly, but that we should devote none (2007, 349).
Moreover, I would emphasize that although the problem that not all health care services promote opportunity is particularly acute when it comes to health care for the elderly, it is a perfectly general problem for Daniels's view. There are plenty of examples of basic health care services for the non-elderly that do not promote opportunity, such as treatment to relieve a nagging cough or a minor pain that does not interfere with a person's ability to pursue her life plans. All such treatments would be unjustified on Daniels's view, at least as a matter of basic justice.9
From the other side, not all of the things that promote opportunity are health care services. Daniels's argument would seem to commit him to the [End Page 567] view that all goods that promote opportunity, or at least all goods that promote opportunity to the degree that health care services do, ought to be treated as special.
Daniels welcomes this implication to an extent. We have already seen that his account of health needs puts the social determinants of health alongside personal health care services, thus extending special treatment to them (2008, 79–102). Daniels also favors treating certain non-health needs as special, such as education, on account of their great significance for opportunity (2008, 60–61). But in fact to justify treating most health care services as special, including those with a relatively insignificant impact on opportunity, consistency will force Daniels to extend special treatment to a whole host of other goods that have a comparable effect, going far beyond what most liberals would endorse.
This is especially true on Daniels's broader understanding of opportunity. If a liberal society must protect not just equality of opportunity to compete for jobs and offices, as Rawls would have it, but equality of opportunity to pursue any reasonable life plan whatsoever, then it would seem that we have a case for treating practically any good as special: education and health care, to be sure, but also housing, cars, smartphones, computers, and indeed anything a person might need for any possible purpose.
Daniels considers something like this problem in Just Health:
Suppose that supplying a computer to everyone who cannot afford one would do more to remove individual impairments to the normal opportunity range than supplying certain health-care services to those who need them. Does the fair equality of opportunity approach commit us to supply computers instead of or in addition to medical treatments?(2008, 59)
After noting that the problem of connecting equality of opportunity to any specific goods is a perfectly general one, Daniels responds that his approach:
rests on the specific calculation that institutions meeting health needs quite generally have a central impact on individual shares of the normal opportunity range and should therefore be governed directly by [Rawls's] opportunity principle.(2008, 59–60)
This response is puzzling, to say the least. The original question was not whether health care generally has a central impact on opportunity. The original question was what we should do when certain health care interventions prove to have a lesser impact on opportunity than alternative uses of those same resources, like providing computers. Daniels's reply [End Page 568] seems to dismiss the need for this kind of marginal thinking, but it is not clear on what grounds. If providing computers would contribute more to opportunity than certain health care interventions, and we are obliged to provide those health care interventions, shear consistency seems to suggest that we are obliged to provide the computers as well.
Importantly, although Daniels frames the computer question in terms of whether computers should be provided "to everyone who cannot afford one" (Daniels 2008, 59), that is not exactly what is at stake here. On Daniels's view, the contribution of health care to opportunity means that health care should be treated as special, which is to say that health care should be provided to everyone, regardless of how well off they may otherwise be; health care is not meant to be a targeted benefit for the disadvantaged, according to Daniels. If computers have an impact on opportunity that is comparable to certain health care services, then presumably they should be treated in the same way. And of course the same point could be made for all sorts of technologies, to say nothing of the other basic needs. Daniels argument implies that we should treat all determinants of opportunity as special, and the determinants of opportunity extend far beyond health care.
Daniels sometimes claims that what distinguishes health and education needs from other needs, and thus why the former should be treated as special while the latter may be left to citizens to purchase from their fair income shares, is that only the former are unequally distributed in the population (2008, 60–61).10 The idea seems to be that since everyone has the same needs in these other areas, no one is disadvantaged if their distribution is left to the market. The problem with this reply is that even if it is true that needs for scarce resources like computers are evenly distributed throughout the population, it seems clear enough that access to these resources is not. Arguably this would still be true—although to a lesser extent—even in a society that was fully just by Rawlsian lights; such a society would still include some inequalities in income and wealth, which would translate into differential access to those resources. It would seem that equality of opportunity—understood not just in terms of opportunities to compete for jobs and offices, but opportunities to form and pursue any reasonable life plan—should condemn such unequal access, giving us a reason to treat those goods as special, too.
The two objections I have raised against Daniels's view combine to present a dilemma. The more expansive Daniels makes his notion of opportunity in order to respond to the first objection, the more vulnerable [End Page 569] he becomes to the second, while the narrower he makes his conception of opportunity to respond to the second, the more vulnerable he becomes to the first. If he is worried that some important health care services are not justified on a narrow account of opportunity, he might move to a more expansive definition. In doing so, however, he is bound to capture more non-health care services in his net. On the other hand, if he narrows the definition so as to include fewer non-health care services, he is likely to find himself excluding more traditional health care services, too.
We have already seen this dynamic at work in Daniels's view. As mentioned earlier, Rawls's principle of fair equality of opportunity applies to opportunities in a very narrow sense: opportunities to compete for jobs and offices (1999, 63). Daniels finds that he needs to broaden Rawls's notion of opportunity to include opportunities to pursue all life plans, not just careers. This is because many important health care services do not affect our opportunities to compete for jobs, but they are important nonetheless: infertility treatment, for example, or any form of health care for those beyond working age (2008, 59–60). But broadening the notion of opportunity in this way also lets in many things that are not health care services. If we are ultimately concerned to protect equality of opportunity to do effectively anything, to pursue any life plan whatsoever, then there is little reason to think that health care will have any distinctive role to play; at best, it will be one of myriad factors that determine a person's share of opportunities.
I conclude that opportunity cannot explain why we distribute health care in isolation from other goods. At best, it can explain why we distribute the things that most contribute to opportunity in isolation from other goods. Some health care services will fit that bill, but many will not. And many of the things that fit that bill will be things other than health care services.
Other Variants of the MIH View
I have discussed Daniels's account at length because his is the most developed account of justice in health care on offer, in the liberal tradition or elsewhere. But I think that the objections I have raised against Daniels's view would apply with equal force to any attempt to justify treating health care as special by appeal to the moral importance of health. For any value or values put forward to explain the moral importance of health, there will always be some health care services that do not promote that value, and many things that promote that value that are not health care services. Thus we will not have a justification for treating health care (alone) as special. [End Page 570]
One exception might be health itself. Someone might suggest that health itself is of intrinsic moral importance, and this explains the importance we attach to meeting health needs. Even here, though, there is enough space between health and health care to run my argument. Not all important health care services contribute to health. Palliative care and other end-of-life services are prime examples. Just as important, not all of the things that contribute to health are health care services. Indeed, a health system that was designed primarily to promote health would probably not be one that focuses on personal health care services at all. A health care system designed to promote health would focus instead on public health measures and other social services, simply because the former have proven much more effective at promoting health than the latter. As Dan Brock explains:
But health care's impact on both health and health inequalities is quite limited; for example, medical care is estimated to account for only about one-fifth of the life expectancy gains in the twentieth century. More important, inequalities in health among individuals and groups that are within human and social control are not primarily the result of inequalities in access to or use of care. . . . The crucial point is that differences in the incidence of illness and injury from social causes swamp the effects on health of differences in access to and use of medical care to treat that illness and injury.(2000, 31, emphasis in original)
For reasons like these, even a view that locates the moral importance of health in health itself would not justify treating health care as special.
In general, to say that health care is distributively special because of the contribution that health care makes to some further value or values is to make the justification of health care's specialness in some sense teleological. There is then the question of whether a particular health care service helps to promote the value in question. There is also the question of whether health care services are the only things, or even the most significant things, that protect or promote the value in question. My claim is that there is no value or combination of values that happens to be most effectively promoted by treating health care as special, i.e., by distributing health care services in isolation from other social goods. If we want a justification for treating health care as special, we would do well to look elsewhere.
THE RISK POOLING VIEW
I suggested earlier that the problem of expensive and unequal health needs may be the strongest factor underlying the desire for an account of justice in health care that justifies treating health care as special. But [End Page 571] another way of describing the unequal distribution of health needs that we observe ex post is to say that they represent significant risks to an agent ex ante. If there is a reason for treating health care differently from other social goods, even other basic needs, I argue that it is because health needs are more uncertain for an agent. Food, clothing, shelter, and other needs are all quite predictable, and so it is reasonable to expect people to see to those needs out of their fair income shares. Health needs are more difficult to plan for, and in this respect they pose a "special" threat to a person's ability to form, revise, and pursue a rational plan of life.11 It is for this reason, I argue, that health care should be isolated from the distribution of other social goods through health care insurance.
I begin this section by arguing that failing to attend to risk per se is a significant oversight in Rawls's influential theory of justice. I then develop a simple model of an insurance scheme to show how risk pooling as a form of social cooperation can reduce people's exposure to risk and thus protect their capacity for self-determination. Next, I discuss why certain forms of insurance, including health insurance, should be mandatory for all citizens. I then return to the main thread of the argument to show how health care insurance justifies treating health care as special, that is, distributing health care in isolation from other social goods. I close by considering some objections.
Risk and Self-Determination
Liberals recognize that in order for a person to form and pursue a plan of life effectively, she must have means at her disposal (Rawls 1999, 79; Ripstein 2006, 1402–04). But the connection between means and ends is deeper than is often appreciated. It is not just that, in order to achieve any of her ends in particular, an agent must have the necessary means (although that is true). More than that, what ends it is reasonable for a person to set for herself in the first place depend upon the means she can count on. A person cannot craft a rational plan of life in a vacuum. What it makes sense for a person to aim to do or to be depends crucially on the resources she can expect.
This argument is nothing new. Liberals have used arguments in this spirit to justify entitlements to various all-purpose means, like basic rights, opportunities, and a fair share of income and wealth. Liberals go wrong, however, when they suppose that what matters is only the means an agent can expect. What matters is also how reliable or how secure those expectations are. [End Page 572]
On Rawls's theory of justice, for example, the justice of a society is a function of how that society distributes certain all-purpose means to citizens in various social positions. Rawls's difference principle directs us to maximize the lifetime expectations of income and wealth for members of the least-advantaged group (Rawls 1999, 81–86). On the most natural interpretation of the word "expectations," though, the lifetime expectations of members of a certain social position are a mean, an average, and a mean can conceal great variation. The mean does not necessarily tell us much at all about the shape of the actual distribution of income and wealth to actual people. The mean does not tell us how income and wealth are spread across the lives of different individuals within the least advantaged group, nor how the income and wealth of a single person within that group are spread over different times in the course of her whole life. The mean does not tell us how many members of the least advantaged group do worse (or better) than expected, or by how much. These sorts of facts about a person's life cannot be read off the mean income of her social group, yet they are extremely important to a person's ability to form, revise, and pursue a conception of the good.
For this reason, liberals should care not only about the expected value of a person's bundle of resources, the mean; they should also care about the variance. Variance is a measure of how "spread out" a set of numbers are, that is, how far numbers in the set lie from the mean. The farther numbers diverge from the mean, the greater the variance. For example, the set of numbers [0, 9, 18] has the same mean as the set [8, 9, 10], yet the variance of the first set is much greater because numbers in the first set are not so tightly clustered around the mean. If we think of these sets of numbers as representing the payoff structures of possible gambles, and we suppose that each outcome within the set is equally likely, then the expected value of the gamble is given by the mean, while we can think of the variance as a measure of how "risky" each gamble is.
Because the variance is independent of the mean, maximizing the expected value of a gamble does not necessarily do anything to reduce one's exposure to risk per se. For that reason, Rawls's difference principle is consistent with arbitrarily high levels of risk. Indeed, the difference principle is consistent with some members of the least-advantaged group winding up very badly off indeed, even as the expectations of members of that group are maximized. Some may do quite well, but others may find themselves suddenly unemployed, or sick and in need of expensive medical care, or temporarily or permanently disabled and unable to work at all. [End Page 573] This can be true no matter how fine-grained we make our procedure for identifying the least-advantaged group.
For these reasons, a liberal society, one concerned to protect each citizen's capacity to form, revise, and pursue a rational plan of life, should aim to make the expectations of the least advantaged—indeed, the expectations of all citizens—not only larger but also more secure. A liberal society can accomplish this by making sure all citizens have access to insurance policies.12
How Risk Pooling Works
Political philosophers sometimes represent insurance as a special case of gambling. On this picture, when you buy insurance, you are betting that an unlikely but disastrous event will occur; your insurance company is betting that it will not (Dworkin 2000, 95–96). Some kinds of insurance work like this, but chiefly those forms of insurance that deal with unusual and hard-to-quantify risks, as when Lloyd's of London underwrites a policy for Keith Richard's hands (Hacking 2003, 28–29). Most insurance schemes operate by pooling large numbers of similar risks together. In this way they reduce everyone's exposure to risk rather than simply transferring risk from one person to another.
This is possible because of a phenomenon that statisticians refer to as the "law of large numbers:" increasing the number of trials of an experiment will cause the average result to converge on the expected value (Hacking 2001, 189ff.). For example, a fair coin flip has a 50% probability of coming up heads, but we know that flipping it 10 times is quite unlikely to yield exactly 5 heads (p = .24). However, we also know that, as we increase the number of coin tosses, the frequency of heads we observe will tend to converge on 50%. (If you flip it 1,000 times, your chances of getting between 49–51% heads are about 2 in 3; to get between 48–52%, your chances are about 9 in 10.) In other words, increasing the number of trials induces statistical stability. This large numbers effect is the chief mechanism through which insurance schemes are able to produce benefits for their members (Heath 2006a, 322). It enables individuals to swap a gamble with an expected value of x for the certainty or near-certainty of x (less, of course, the costs of administering the scheme).
We can see how this is possible with the help of a simplified example.13 Imagine a merchant who owns a small ship that she uses to take her goods to market overseas. Suppose that on any given journey there is a 20% chance that the ship will sink, and suppose that a lost ship will [End Page 574] cost her $100,000. The expected cost of this risk is then $20,000 (20% of $100,000), but as things stand she will not actually lose $20,000. Instead, either the accident will occur and she will lose $100,000, or it will not occur and she will lose nothing. Let us assume for the sake of this example that, like most people, our merchant is risk-averse; she would gladly accept a guaranteed loss of $20,000 (or even a bit more) to escape the 20% chance of losing $100,000.
She cannot achieve this on her own. But suppose she meets ninety-nine other merchants who face identical (though, we shall suppose, independent) risks. The hundred of them might agree to pool their risks of shipwreck and hold any potential losses in common. Should they do this, each of them swaps her 20% chance of a $100,000 loss for the gamble represented in Figure 1 (above).14 As Figure 1 shows, the most likely outcome, represented by the tallest point on the curve, is that 20 of 100 ships will sink, for an average cost to each member of $20,000—exactly the expected loss with which they all began. Outcomes that diverge very far from the mean of $20,000 become increasingly less likely, to the point where losses smaller than $5,000 or more than $35,000 per person are very nearly mathematically impossible—and this with as few as 100 people pooling risks. With more members in the pool, the range of realistically possible losses would narrow even further (See Figure 2, below).
In practical terms, the impact of this arrangement is that each member swaps the small chance of a big loss for the certainty of a small loss. What they achieve in mathematical terms is a substantial reduction in the variance of their future expectations. The expected value of the risk does [End Page 575] not change: these risk pooling arrangements come with expected costs of $20,000, exactly like the 20% chance of a $100,000 shipwreck with which our merchants began. (In reality, a risk pooling arrangement like this will have an expected cost of somewhat more than $20,000 because of transaction costs.) But thanks to the law of large numbers, the subjective risk is reduced considerably. Everyone can be very confident of losses between, say, $16,000 and $24,000 (with 1,000 people pooling risks, they can be over 99% confident of it), and they can be utterly certain of avoiding anything like the $100,000 cost of bearing a single shipwreck alone.
Obviously a risk pooling arrangement like this does not eliminate or even reduce the risk of shipwreck; it simply redistributes the losses over the whole pool. Still, it would be misdescribing this arrangement to say that its primary purpose is redistributive. By its nature, an insurance arrangement involves ex post redistribution from lucky to unlucky, from those who do not suffer the insured loss to those who do. But people do not join an arrangement like this to help the unlucky through hard times. They join an arrangement like this to reduce their own exposure to risk (Heath 2006b).
The logic of risk pooling is therefore more appropriately described in terms of solidarity rather than equality.15 By committing to share each other's fate in certain ways, each member of the risk pool can advance her own interests more effectively. Risk pooling may have an egalitarian element if premiums are structured progressively, but this is distinct from the logic of risk pooling per se. For example, if the merchants in our shipwreck pool offered insurance to some of their poorer members on more favorable terms—if in other words the wealthier merchants cross-subsidized [End Page 576] the premiums of poorer ones—this would introduce a properly egalitarian dimension to the scheme.16 I return to this point later.
An Argument for Universal Coverage
Insurance is beneficial because it reduces people's exposure to risk, providing stability in their future expectations. This is of interest to liberal theorists because, among other things, it protects people's ability to make effective plans for the future. But to show that insurance is beneficial in this way does not yet explain why certain kinds of insurance should be mandatory; it does not yet provide an argument for social insurance.
In general, liberals should favor allowing individuals to decide for themselves the balance of risk and security that is appropriate for them by purchasing the insurance products they want on the market according to their own level of risk aversion. But for certain kinds of insurance markets, including health insurance markets, this is not always possible. Private markets in health care insurance consistently fail for reasons that have long been familiar to economists (Arrow 1971). The principal reason has to do with the unequal distribution of health risks, which generates problems of adverse selection.
Adverse selection occurs when individuals who face different levels of risk are able to obtain insurance at the same price. When that happens, those who face higher levels of risk will be more inclined to buy insurance than those who face lower levels of risk; at any given price, insurance is simply a more attractive deal for higher-risk individuals. Over time, this "adverse selection" of risks increases the per capita liabilities of the insurance pool, which in turn drives up the price of insurance. As the price rises, more good risks may begin to drop out of the pool, causing per capita liabilities and thus premiums to rise even further. In this way, the bad risks drive out the good (Akerloff 1970, 492–94). In the end, this may lead to the dreaded "insurance policy death spiral:" premiums drift ever upward, to the point where the pool finally prices itself out of existence.17 The pool may try to prevent this by engaging in more careful underwriting practices, but these too are costly and raise the price of insurance for everyone.18
Adverse selection is a perfectly general problem with insurance markets, but it is endemic in markets for health care insurance. This is because practical and legal barriers often prevent insurers from making accurate determinations of a person's risk level and then pricing insurance plans accordingly. It is not as easy to conceal your home's risk of fire from an [End Page 577] insurance underwriter as it might be to conceal your own risk of heart disease, and of course there are good privacy-related reasons for this. But the consequence is that adverse selection runs amok in the individual health insurance marketplace, leaving many people unable to obtain health insurance (Akerloff 1970, 492–93).
The most effective solution to the problem of adverse selection is to force the good risks back into the insurance pool, that is, to force everyone to carry health insurance. This can be achieved by having the government itself act as insurer, as in Canada, or by forcing all citizens to purchase health insurance on the private market, as in the U.S. under the Affordable Care Act. Having the state provide medical care directly, as under Britain's NHS, can also be seen as an indirect way of pooling health risks.
This offers a non-paternalistic rationale for universal coverage. The problem of adverse selection means that the alternative to mandatory health insurance is often that health insurance will not be available at all to large segments of the population, and where it is available it will be on very unfavorable terms. In other words, the alternative to paternalistically forcing all citizens to carry health insurance is leaving many citizens to go without health insurance.
Insurance and Specialness
Looking at the operation of the shipwreck insurance scheme described earlier, someone might suggest that its logic shows that the merchants in the pool believe that meeting people's shipping needs is special. They certainly treat shipping needs in accordance with the requirements of specialness. No one in the pool is expected to meet their shipwreck needs out-of-pocket. If one of the merchants loses her ship, the pool compensates her for the loss. This is true whether she happens to be a very wealthy merchant who could easily afford to replace the ship herself, or a poor trader who could never afford such an expense on her own. The pool distributes funds in isolation from (the rest of) members' income and wealth and, incidentally, on the basis of shipping need.
But the reason that the insurance pool treats shipping needs as special in this way is simply that it is an insurance pool against shipwrecks. Compensating someone for an insured loss is just what insurance does. We do not need to appeal to the "moral importance of shipping" to explain it. The purpose of the arrangement is to reduce individuals' exposure to shipwreck risk. If the pool did not compensate individuals according to their shipwreck losses, it would not serve that function. [End Page 578]
Extending this to health care, the question of why health care should be treated as special, on my view, has a two-step answer. If the question is why health care should be distributed in isolation from other goods, the answer is because citizens should have health care insurance. If the question is why citizens should have health care insurance, the answer is to provide them with greater stability in their future expectations and thus to protect their ability to form, revise and pursue a rational plan of life. Once we grant that individuals should be insured against health risks, we have a justification for treating health care as special in the relevant sense, and this justification does not need to appeal to the moral importance of health.
Here the contrast with other basic needs is instructive. There is far less variance in people's food needs, for example. Food needs do not vary dramatically, either from person to person or at different times in a single person's life. Even people with special dietary needs—the vegetarians, the lactose-intolerant, sufferers from celiac disease, even those cursed with expensive tastes—have no trouble anticipating their future food needs and planning for them accordingly, given a fair share of income and wealth. If we worry that some people will not be able to afford enough food (or enough of the right kind of food), redistribution of wealth recommends itself as the optimal solution.
For these reasons, it is implausible to think that food should be treated as special. The state should not be in the business of doling out to each citizen a fair share of food, in isolation from other social goods. The only possible justification for a program like that would be paternalistic, grounded in the worry that citizens would not buy enough food or would not buy the right kind of food if left to their own devices.
The difference between health needs and food needs is not that health needs are more important, morally speaking, or even that they are important in a different way. The difference is that health needs are more unpredictable (PCSEP 1983, 23). It is because of this uncertainty that giving people cash in the amount of their expected future health costs is no guarantee that they will be able to meet their health needs going forward. For this reason, we do not need to appeal to the moral importance of health in order to explain why citizens should have health insurance, any more than we need to appeal to the "moral importance of employment" to explain unemployment insurance, or the "moral importance of being able-bodied" to explain disability insurance. Bearing the risk of losing your job unexpectedly, of being unable to work due to disability, or of losing your nest-egg due to unexpected medical bills are all real obstacles [End Page 579] to a person's capacity to form, revise, and pursue a certain plan of life effectively. Making sure that people can obtain insurance against these risks is an important part of protecting that capacity.
I have so far avoided questions about how health insurance programs should be funded. A variety of answers are possible, ranging from actuarial fairness at one end to funding out of general tax receipts at the other, with community rating lying somewhere in between. Which of these is most just will depend upon broader questions of distributive justice. As mentioned earlier, there is no necessary connection between risk pooling and equality, although risk pooling may incorporate cross-subsidization in order to respond to egalitarian concerns. From an egalitarian point of view, one virtue of my account is that it allows for the distribution of health care in isolation from income and wealth while leaving open the possibility of the progressive funding of health care insurance. On my view, questions about the financing of health care insurance are distinct from the question of what makes health care special.
In closing, I would like to point out that my account will not be open to the same objections that I raised against the moral importance of health view. The problem with that view, I argued, was that no account of the moral importance of health will capture all and only basic health care services. Some important health care services will not promote the value underlying health, whatever that is, and some things that are not health care services will promote that value. Thus the moral importance of health view cannot justify treating health care as special. My account might be thought to fall victim to a version of the very same objection. It is not true that every health need constitutes a risk in the relevant sense, and not all things that are risks in the relevant sense are health needs. I address these two problems in turn.
Regarding the first objection, it is true that not all health needs constitute risks properly understood. Routine health care, for example, certainly poses a problem for my account. There is no uncertainty surrounding whether you need a physical examination every few years; this is not a risk but a fact, and so insurance is not really appropriate for these kinds of needs. This is a bullet that I am prepared to bite. I do not think it is a grave injustice if people have to pay for routine health care out of pocket, or (what is essentially the same thing) if health insurance includes a small deductible. The vast majority of health spending is due to non-routine [End Page 580] care, and it is the expensive, bank-breaking treatments that raise the most troubling problems of justice. These are the expenses that most threaten a person's ability to plan for the future effectively.
It may be possible to cobble together efficiency-based arguments for why insurance should cover routine care. For example, large insurers are often able to bargain for better prices than individuals. Also, if check-ups are paid for by insurance, individuals may be more likely to take advantage of them, which may help keep overall health costs down by allowing doctors to catch serious conditions early.19 These will not necessarily be reasons of justice, but the end result will be the same. If those arguments fail, I do not see a non-paternalistic rationale for covering routine doctor visits. At that point, leaving citizens to pay for routine care themselves does not seem to me to raise serious problems of justice (even if it may cause some to decline care).20
From the other side, it is also true that not all risks are health risks, but I do not see this as a serious threat to my account, either. While there are many risks that both threaten a person's ability to make future plans and are difficult to insure against privately, there are also many forms of social insurance: unemployment, disability, old-age, and so on. My account could be extended to these cases as well. Insofar as each is a form of insurance, each will treat their respective risks as (so to speak) special in their own way—each risk will be addressed in isolation from the rest of people's income and wealth and on the basis of need. It is a virtue of the risk pooling account that it can explain why health care should be treated as distributively special without having to take on the dubious claim that health care is morally unique.
I have argued that an entitlement to health care should be grounded in the importance of providing security against health risks. Guaranteeing citizens access to health care insurance has the fortunate consequence of making sure that no one goes without needed medical care, but I have argued that this is not its primary purpose. Instead, its primary purpose is to provide greater stability in people's future expectations, thus enabling them to plan for the future more effectively.
My proposal may seem somewhat counterintuitive given that the standard liberal view, the MIH view, holds that the order of explanation goes the other way. The MIH view holds that we have an obligation of justice to meet health needs. On the MIH view, it follows from our [End Page 581] obligation to meet health needs that individuals are relieved of the burden of bearing health risks alone, but health risks are socialized as a somewhat accidental consequence of socializing health losses. I argue that this position is exactly backwards: the reason we meet citizens' health needs is primarily to reduce individuals' exposure to risk. On my view, health losses are socialized primarily as a means to the socialization of health risks.
There is nothing in the nature of health needs to distinguish them, morally speaking, from other basic needs. Indeed, any attempt to articulate what makes health care morally important will fail to capture some health needs and will inevitably include many non-health needs. There is nothing in the moral nature of health needs to explain why we are required in justice to treat them differently from other basic needs. What distinguishes health needs from other needs is in their nature as risks. The risk of needing health care is very costly for a person to bear alone, and not just in the sense that, if a health risk should ripen into a health loss, a person would find herself very badly off. Bearing health risks alone is in itself costly to a person's ability to plan for the future, even in the happy event that no health risks ever become health losses. A person's ability to form, revise, and pursue a plan of life effectively is greatly enhanced by pooling these risks with others to provide stability in people's future expectations.
L. Chad Horne, PhD, is currently a visiting assistant professor of philosophy at Franklin & Marshall College. He received his PhD from the University of Toronto, and was recently a postdoctoral fellow at the Centre for Moral and Political Philosophy at the Hebrew University of Jerusalem. His research focuses on the theory of the welfare state, particularly theories of justice in health care.
A version of this paper was presented at the Canadian Philosophical Association's Annual Congress at Brock University in 2014. I am grateful to my audience there for their questions. I would also like to thank Donald Ainslie, Kelin Emmett, Joseph Heath, Shlomi Segall, and Sergio Tenenbaum for their many helpful criticisms and suggestions.
1. The moral importance of health view probably originated with Walzer's Spheres of Justice (1983, 86–91). The most prominent liberal defender of this view is Norman Daniels, whose position I discuss in some detail below. For other defenders of the moral importance of health view, see Brock (2001) (but see Brock  for some complications); Buchanan et al. (2000); Jacobs (2004); and Nielson (1989).
2. The contemporary discussion of social insurance and distributive justice begins from the work of Dworkin (2000) and Goodin (1985 and 1990). Recently there has been a resurgence of interest in the topic; see especially contributions [End Page 582] from Heath (2006a and 2011), Landes and Holtug (2015), Landes and Néron (2015), and Lehtonen and Liukko (2015).
3. In the US in 2010, 20% of patients accounted for 81.7% of total health spending, with the top 1% accounting for 21% of such spending (KFF 2013). The situation is even more dire in Ontario, Canada, where in 2007 just 10% of patients accounted for 79% of system-wide costs (Wodchis et al. 2012, 11).
5. See note one above.
6. Daniels's view has been elaborated over more than thirty years and has undergone significant changes. I focus on his view as presented in Just Health (2008), as I take this to represent his considered view, but see also Daniels (1985) and Daniels (2001).
8. For instance, people age 65 and over made up 13% of the US population in 2002, but they consumed 36% of total health care services (Stanton 2006, 3). One quarter of all Medicare spending in the US goes to patients in the last year of their lives (Hogan et al. 2001, 188).
10. To be fair, Daniels is speaking here about what distinguishes health care and education from other basic needs, not what distinguishes them from other opportunity needs. That said, this seems like an obvious reply he might raise to the line of argument I am pursuing here.
14. I assume here that the shipwrecks follow a normal, Gaussian distribution.
16. Ronald Dworkin's treatment of insurance tends to obscure this point because of his broader luck-egalitarian commitments. Dworkin holds that equality requires that people be compensated for bad brute luck, and because Dworkin supposes that the purpose of insurance is to compensate people for bad luck, he sometimes suggests that insurance per se promotes equality. In fact, insofar as Dworkin's insurance markets serve an egalitarian function, it is [End Page 583] only because Dworkin insists that the price of insurance should not reflect unchosen differences in risk. It is not actually insurance that promotes equality, in Dworkin's view, but rather the cross-subsidization of those with expensive, unchosen risks by those without. For Dworkin's view, see Dworkin (2000, Chapter 2). For more on the relationship between insurance and equality, see Heath (2011) and Landes and Néron (2015).
18. One benchmark estimates that marketing and underwriting practices account for nearly two-thirds of overhead costs in a typical private health insurance plan. In 1999, prior to the Patient Protection and Affordable Care Act, insurance overhead in the United States totalled $47 billion dollars, or nearly $260 per capita. Canada's single-payer insurance system, which effectively eliminates marketing and overhead, spent only $311 million on overhead that year, or only $47 per capita, in PPP-adjusted 1999 US dollars (Woolhandler, Campbell, and Himmelstein 2003).
19. A 2004 study examined the paradoxical fact that American states with higher levels of Medicare spending tend to enjoy lower quality of care. The authors claim that this is because states with higher levels of Medicare spending also have a higher ratio of specialists to general practitioners. Primary care is both cheaper and more effective than tertiary care (Baicker and Chandra 2004; see also Starfield, Shu, and Macinko 2005).
20. Perhaps the most influential study of this issue was conducted by Rand for 1974–1982. The Rand experiment randomly assigned individuals to insurance plans with different levels of cost sharing (deductibles). Not surprisingly, the study found that cost sharing reduced consumption of both unnecessary and necessary care. The study found no significant positive effect of free care on most measures of health, although those with certain specific health conditions did benefit from free care, with this effect being largest among the poor (Brook et al., 1984).