Health, Luck and Moral Fallacies of the Second Best

Abstract

Individuals who become ill as a result of personal lifestyle choices often shift the monetary costs of their healthcare needs to the taxpaying public or to fellow members of a private insurance pool. Some argue that policies permitting such cost shifting are unfair. Arguments for this view may seem to draw support from luck egalitarian accounts of distributive justice. This essay argues that the luck egalitarian framework provides no such support. To allocate healthcare costs on the basis of personal responsibility would arbitrarily and publicly burden socially detectable risk-takers while undetectable risk-takers continue to get a free ride. That problem is unavoidable even on the assumption that distributive institutions outside the healthcare sector are fully just. In actual, farfrom-just societies, imposing personal liability for the costs of voluntary risk taking would be wrong for an additional reason. Doing so would tend to magnify existing distributive injustices. These conclusions draw attention to two common ‘moral fallacies of the second best’ that can arise when applying ideal normative theory to matters of institutional design and in real-world policy contexts.

Introduction

Bad eating habits, lack of exercise, smoking, excessive (or insufficient) alcohol consumption, and other so-called “lifestyle choices” contribute substantially to the burden of disease in many societies. Much of the cost of the resulting healthcare needs is borne by public programs, spread across private insurance pools, or absorbed by hospitals that pass the costs on to other patients and to the taxpaying public.¹ Thus, those who voluntarily put their health at risk often shift the monetary costs of their choices to others. Some normative theorists maintain that polices permitting such cost shifting are unfair. Those who choose to engage in risky behavior should bear the costs of any resulting medical treatment needs.² Other theorists may resist that conclusion on grounds of humanitarian concern or of solidarity, but maintain nonetheless that society has no obligation of justice to provide care for those who bring illness on themselves.³ US opinion polls suggest that the public supports greater personal accountability in matters of health, and government healthcare plans in both Europe and the United States have recently begun to implement policies that hold individuals accountable, in limited ways, for the outcomes of their lifestyle choices.⁴ I will argue that a policy of holding individuals financially liable, or otherwise publicly accountable for their health, can never be justified on grounds of fairness.⁵ This does, however, not preclude forward-looking justifications—holding individuals accountable in limited ways may have health-promoting incentive effects, and may be justified on those grounds.

In the following, I consider and reject several fairness-based arguments for personal accountability before focusing on arguments drawn from luck egalitarian accounts of distributive justice. While I do not endorse these accounts, they seem to offer the firmest foundation on which a fairness-based case for personal accountability could be made. My conclusion is that the luck egalitarian case fails on its own terms. When we take proper account of the limited social capacity to detect voluntary risk taking, considerations internal to the luck egalitarian framework suggest that personal responsibility should not be a factor in healthcare cost allocation decisions. This conclusion holds even in the realm of ideal theory—that is, even on the assumption that society’s other distributive institutions are fully just by luck egalitarian standards. The conclusion is even more compelling in the real-world context. Imposing personal liability for health in any of our actual, far-from-just societies would only magnify existing distributive injustices, adding to the burden of those who already are unjustly disadvantaged. The implications of socially undetectable risk taking and of unjust background institutions call attention to two common errors of moral reasoning, which I call moral fallacies of the second best. When not all of the optimal conditions presupposed in ideal theory are satisfied, it is a fallacy to suppose that satisfying more, but not all, of those conditions moves a real society closer to the ideal.

Blame, Fault or Option Luck?

Over the past century, the industrialized world has undergone an ‘epidemiological transition.’ The acute infectious and deficiency illnesses that once accounted for most of the burden of disease have been supplanted by chronic, non-communicable illnesses such as heart disease, cancer, diabetes, liver cirrhosis and others.⁶ Many of the behavioral risk factors for these illnesses are under the voluntary control of the individuals affected. Over the same time period, the emergence of vast medical and pharmaceutical industries has generated unprecedented costs for the treatment of these chronic illnesses, while the introduction of private health insurance and public healthcare systems has spread those costs across the general population. The current debate, with its emphasis on the burden that individual lifestyle choices can place on the public, has intensified in the decades since the long-term economic implications of these developments began to become apparent.⁷

In his classic 1977 broadside, “The Responsibility of the Individual,” John H. Knowles—then president of the Rockefeller foundation—wrote:

The cost of sloth, gluttony, alcoholic intemperance, reckless driving, sexual frenzy, and smoking is now a national, not an individual responsibility. This is justified as individual freedom—but one man’s freedom in health is another man’s shackle in taxes and insurance premiums.⁸

While he does not propose that individuals be held financially liable for the costs of self-caused illnesses, Knowles gives clear voice to the view that it is unfair for those who make imprudent choices to expect others to foot their medical bills. That claim might be interpreted in two ways. I will call them the ‘depravity-subsidizing’ and the ‘cost-shifting’ interpretations. Knowles’s markedly moralizing language could suggest the former. On this interpretation, society should not have to bear the monetary costs of “sloth,” “gluttony,” “sexual frenzy,” and the rest because such behavior is inherently immoral. The public should not have to subsidize personal depravity. Understood thus, the call for personal accountability for health has come under fire from critics who see it as offensively moralizing and as reinforcing an ideology of ‘victim-blaming’.⁹ By contrast, on the cost-shifting interpretation, the choice to engage in imprudent but lawful behavior is regarded as strictly a private matter on which no public moral judgment is made. What is immoral, on this interpretation, is not the behavior as such but simply the fact that the individual shifts its costs to others.

There is an air of common sense to the view that, if someone becomes ill because of his own imprudent choices, then it is unfair for others to have to bear the costs of treatment. Yet as Daniel Wikler points out, even this view can harbor a covertly moralizing element. Both the smoker who develops lung cancer and the mountain climber who suffers some comparably serious injury while climbing brings his misfortune on himself. Yet (outside of the philosophical literature) we hear few complaints about the social costs of dangerous recreational activities. The difference, Wikler observes, lies in our conception of fault.¹⁰ We do not fault the injured mountain climber (provided that he took the precautions that a reasonable mountain climber would take) because we regard mountain climbing as a valuable activity that develops and exhibits forms of human excellence. By contrast, if we fault the smoker for imprudence, that is because we regard smoking as a relatively worthless habit that typically reflects weakness of will. Our views about whether or not an individual has acted imprudently rest on substantive moral judgments about the relative worth of the various activities that individuals may lawfully choose to pursue. Yet the project of publicly sorting the freely chosen, lawful activities of citizens into the valuable activities and the worthless ones seems out of place in a liberal society. As Wikler puts it, “we should not expect that the only ones made to shoulder the costs are those who behave in ways that offend their neighbours.”¹¹

If there is any compelling argument for holding individuals liable for their self-caused illness, it must avoid these kinds of moralizing judgments. One principle that might serve as a basis for such an argument is that individuals should bear—not the costs of their immoral or imprudent choices specifically—but simply the costs of their free choices in general. If it is unfair for smokers to shift their healthcare costs to others, then it is also unfair for mountain climbers to do so. This view reflects what I will call an option-luck conception of healthcare cost allocation. The idea of option luck is owed to Ronald Dworkin, who famously distinguishes it from brute luck:

Option luck is a matter of how deliberate and calculated gambles turn out—whether someone gains or loses through accepting an isolated risk he or she should have anticipated and might have declined. Brute luck is a matter of how risks fall out that are not in that sense deliberate gambles.¹²

For Dworkin, mountain-climbing injuries and reasonable business investments that go south are paradigm examples of option luck. Being struck by lightning is a paradigm example of brute luck. While there is no bright line between the two, the distinction seems to mark a sort of datum of common sense: there is an important moral difference between things that just happen to us and foreseeable outcomes of things that we do. Analysis of this commonsense distinction forms the heart of so-called ‘luck-egalitarian’ accounts of distributive justice.

The guiding moral intuition of these accounts is that other things equal, social institutions should be arranged to limit or to eliminate distributive disadvantages that an individual suffers owing to circumstances outside of her control, or that result from risks that it would be unreasonable to expect her to avoid.¹³ Society should therefore compensate individuals for bad brute-luck outcomes, but not for bad option-luck outcomes. Luck egalitarians differ in their views about what we ought to distribute according to that rule (e.g., welfare, resources, capabilities), about where precisely to draw the line between choice and circumstance (e.g., in which category we should place expensive tastes that an individual acquires through no voluntary act of choice), and about the target of compensation (e.g., to nullify brute luck fully, or only to the extent that rational and fully informed individuals would insure against, under conditions fair to all). A specification along these three dimensions distinguishes any given luck-egalitarian theory.¹⁴ My argument addresses the luck-egalitarian framework in abstraction from these matters; nothing in what follows turns on how they are specified.

The luck-egalitarian framework seems to offer a systematic, non-moralizing and appropriately liberal rationale for a policy of holding individuals accountable for their option-luck healthcare needs. Indeed, many prominent luck egalitarians seem to regard such a policy as a natural consequence of the theory. Eric Rakowski maintains that “people who leap from airplanes, scale cliffs, or whirl around racetracks…cannot expect others to foot their hospital bills or aid their dependents if fortune is uncharitable,” and that under fair arrangements, “those who smoked or ate too much would rightly have to pay more for certain types of health insurance.”¹⁵ Richard Arneson questions whether public funds should be used to compensate an individual who is blinded through “deliberate and fully informed participation in a dangerous sport that often gives rise to injuries that result in blindness.”¹⁶ John Roemer argues that individuals should be held liable for their health-care needs in proportion to the degree to which they have acted as responsible agents.¹⁷ Cappelen and Norheim propose targeted excise taxes to make individuals pay for the health risks they freely take (though not fully for the outcomes of those risks).¹⁸ In the most extensive treatment of the issue to date, Shlomi Segall argues that while society should not withhold healthcare from the imprudent, considerations of justice do allow us to impose on these individuals at least part of the cost of their treatment.¹⁹ And while Dworkin takes no definitive stand, he suggests that “it seems reasonable” that insurance companies should charge mountain climbers and cigarette smokers higher premiums.²⁰

Most of these policy prescriptions are intended—in the first instance at least—as propositions of ‘ideal theory.’ In other words, they are intended as policies that would be appropriate for a society in which background distributive institutions (i.e., those outside of the healthcare sector) are fully just. In the next three sections, I consider the issue from the perspective of ideal theory. Even in this case, I argue, it would not be fair by luck-egalitarian standards to hold individuals financially liable for their option-luck healthcare needs. In subsequent sections, I consider the special challenges of applying the luck-egalitarian framework to this issue in our actual, far-from-just societies.

Option-Luck Healthcare Cost Allocation in Ideal Theory

How would a society effectively regulated by luck-egalitarian principles allocate healthcare costs? Presumably, treatment for brute-luck health outcomes, along with routine preventive care and screening, would be paid for either by a publicly funded healthcare system or by community-rated private health insurance. Community-rated premiums would not vary with personal risk classifiers, such as an individual’s age, sex, genetic endowment, family history, income level or pre-existing conditions. As we have just seen, however, most luck egalitarians seem to believe that to cover option-luck outcomes in this way would be unfair. On their view, society could justifiably require individuals to pay for their option luck health care either out of pocket or through supplemental option-luck health insurance available at regulated market rates. Option-luck insurance premiums would not be permitted to reflect brute-luck risk classifiers, but could reflect an otherwise actuarially fair assessment of the health risks of voluntarily chosen behaviors. Higher levels of voluntary risk taking would incur higher premiums. I will designate this sort of arrangement an option-luck add-on system of healthcare cost allocation, and will argue in what follows that reasons internal to the luck-egalitarian framework should lead us to reject such a system on grounds of fairness.

An initial question regarding an option-luck add-on system is: What are we to do with the uninsured who cannot afford necessary care for their option-luck health outcomes? Luck egalitarians are sometimes criticized on the grounds that in their view, these individuals should be left to die in the streets.²¹ Few luck egalitarians embrace this consequence. Most allow that such individuals should be treated at public expense—not as a matter of justice, but on humanitarian grounds or on grounds of social solidarity.²² In order to portray the luck-egalitarian framework in its most favorable light, l assume that a society regulated by luck-egalitarian principles will include a healthcare safety net for uninsured option-luck healthcare needs and—for similar reasons—an adequate social minimum of income.²³

A publicly funded safety net for the uninsured risk-taker does not, however, fully socialize option-luck healthcare costs nor eliminate all personal accountability for health. Individuals who fail to purchase the add-on would become eligible for safety-net resources only after exhausting their surplus assets (i.e., those deemed to be above the social minimum). The risk of asset forfeiture provides an incentive to purchase the add-on and thus insures a degree of personal accountability for lifestyle choices. Nonetheless, those already at the social minimum could have no such incentive, nor could those living close enough to the social minimum that the cost of add-on premiums would exceed their surplus assets discounted by the probability of having to forfeit them. Moreover, since the worst-case outcome for the uninsured is forfeiture of surplus assets followed by subsistence with free healthcare at the social minimum, even those with significant surplus assets might opt to forgo purchasing the add-on. Finally, even among those for whom the only rational choice would be to purchase the add-on, some will fail to do so (after all, these individuals are by definition inclined to tolerate unusually high and perhaps irrational levels of risk). Thus, an option-luck add-on system would provide only a limited way of holding individuals accountable for their choices.

It might seem as though these limitations could be overcome by making option luck insurance mandatory for all risk takers. To implement such a mandate, however, society would have to detect risk taking, and to do so ex ante—that is, in the healthy population and not simply at the point of care. Some risky behaviors—smoking and motorcycle riding, for example—can be detected at the point of purchase for those products, and individuals made to insure through targeted excise taxes.²⁴ Yet society cannot—without unacceptably intrusive measures—detect ex ante many or even most risky behaviors. Socially undetectable risky behaviors include bad eating habits, excessive (or insufficient) alcohol consumption, inadequate exercise, overwork, poor stress management, driving while intoxicated or while drowsy, illegal drug use, unsafe sexual activity, and many others. Empirical evidence suggests that the medical costs of socially undetectable risk taking in the United States greatly exceed the costs of the relatively few risky behaviors that can reliably be detected ex ante—a pattern that may be general in developed societies.²⁵ Thus, even if option-luck health insurance is mandatory for risk takers, much and perhaps most of the cost of option-luck healthcare will continue to be borne by the general population.

No human institution can perfectly embody any principle of morality. Luck egalitarians recognize that some cost shifting must be tolerated even under the best-designed institutions. Defenders of an option-luck add-on may assume, however, that if we can at least hold some people responsible for some of their option-luck health outcomes, that would be a step in the right direction—better than not holding anyone responsible for any of their option-luck health outcomes. This assumption fails. Given the unavoidable limits to socially available information about personal risk-taking, it is not a step in the right direction—from a luck-egalitarian perspective—to hold only socially detectable risk-takers accountable. Or so I will argue.

Vertical Versus Horizontal Equity in Healthcare Cost Allocation

Suppose our town aspires to adhere, insofar as possible, to a luck-egalitarian conception of distributive justice. Among other things, we want to hold individuals (or households) accountable for their water consumption choices. Those who choose to consume more water should pay more; those who chose to consume less should pay less; and those who make the same consumption choices should pay the same. Let us say that half of the households consume 1,000 gallons per day each (‘the big users’), while the other half consume only 100 gallons per day each (‘the moderate users’). At those consumption rates, the cost to our town of satisfying the total demand for water is one cent per gallon.²⁶ Thus, if costs are divided equally, each household pays $5.50 per day for its water usage. As it happens, homes in the east side of town are served by pipes to which reliable and accurate meters can easily be added on. It is not, however, technically feasible to add meters onto the pipes serving the west-side homes. The question is whether we ought to install meters in the west side, and set water utility rates accordingly, on the assumption that doing so would at least be a step in the right direction.

Here is the breakdown of how water usage costs would be allocated before and after the proposed metering system is implemented (Table 1):

Table 1 Daily per household water usage charges

Is metering a step in the right direction, by the standards of a luck-egalitarian theory? On the classical Aristotelian analysis, fairness or equity consists in two elements: treating like cases alike (‘horizontal equity’) and treating different cases differently and in proportion to their difference (‘vertical equity’). Both components are implicit in the luck-egalitarian framework. If we apply the principle of vertical equity, then partial metering seems to be better than none at all. At least with partial metering, some who use more pay more, and some who use less pay less. That might seem to be a step in the right direction. On the other hand, if we apply the principle of horizontal equity (those who make the same usage choices should pay the same) then it seems better not to meter. For if we adopt the metering plan, western big users will fare better than eastern big users, and eastern moderates better than western moderates. Assuming that the two principles of equity are given equal weight, there appears to be no reason to favor either course of action.

Our judgments about such a case should probably be sensitive to the relative population sizes of west and east. For example, if there were only two households in the west side and two thousand in the east, then metering would arguably be a step in the right direction. Thus if we give equal weight to horizontal and vertical equity, we should perhaps conclude that metering is best if the majority of households are on the east side—and that otherwise it would be better not to meter. Yet there is reason to believe that horizontal equity should weigh more heavily than vertical equity in a case like this. For even if the allocation of water costs under the two schemes (partially metered and unmetered) is equally fair/unfair, there is an added element of publicity to the unfairness in the partial-metering case. Prior to metering, each moderate user can say, “I know that the big users are shifting the costs of their water usage choices to me and other moderate users, but I cannot expect everyone else to take my word for it that I am a moderate user, and as a community, we do not know who is a moderate and who a big user. I am arbitrarily disadvantaged by this arrangement, but it is unavoidable that some will be arbitrarily disadvantaged, and from the public standpoint, I have been given the same equal chance of being arbitrarily disadvantaged as anyone else has.” By contrast, after metering, there are publicly identifiable winners and losers. The moderate users on the east side are the winners, and everyone else can envy them for benefitting from the metering plan. The big users on the east side are the losers, and they can envy everyone else. This does not sound like equal justice.

The somewhat counterintuitive moral of this story is that less free-riding is not necessarily more fair. Let us apply this to the case of an option-luck add-on system. The analogy is not perfect, but it is close enough to make the point.²⁷ What we should conclude, it seems, is that—giving equal weight to horizontal and vertical equity—if the costs of ex-ante socially identifiable risk taking exceed the costs of socially undetectable risk taking, then an add-on system is a step in the right direction (fairness-wise). Otherwise it is not. As already noted, this condition does not seem to be met in the US—where the costs of socially undetectable risk taking well exceed those of detectable risk taking.²⁸ Moreover, if, as I claim, considerations of publicity give us reason to weigh horizontal equity more heavily than vertical equity, then an option-luck add-on system would be a step in the right direction only if the costs of socially detectable risk taking were significantly greater than the costs of undetectable risk taking—that is, sufficiently greater to justify adding publicity to arbitrary disadvantage.²⁹ I conclude that even on the assumptions of ideal theory, an option-luck add-on system is, in actual societies, unlikely to be a step in the right direction from the standpoint of the luck-egalitarian framework. If considerations of personal responsibility have a legitimate role to play in health-cost allocation decisions, it should be justified on grounds other than fairness (for example, on grounds of producing beneficial incentive effects).³⁰

Ideal Theory and Moral Fallacies of the Second Best

The foregoing argument suggests a general point related to what economists call “the problem of the second best.” If several variables in a market model deviate from their optimally efficient (“first-best”) values—and if some of the variables are “constrained” (cannot take their first-best values)—then moving the other variables to their first-best values does not necessarily, or even probably, yield the second-best solution (i.e., the most efficient one consistent with the constrained variables). In general terms, “it is not true that a situation in which more, but not all of the optimum conditions are fulfilled is necessarily, or even likely to be, superior to a situation in which fewer are fulfilled.”³¹ The present case illustrates the problem of the second best for normative political theory.³² Even if implementing luck-egalitarian principles would—given perfect information availability—yield the “first-best” (fairest) solution, attempting to implement such principles in a context of limited information availability does not necessarily, or even probably, yield the second-best (i.e., second fairest) outcome. We can now formulate the first of two moral fallacies of the second best:

• 1. Institution or policy P would be required by justice, assuming perfect availability of relevant information.

• 2. Therefore, implementing P under conditions of less than perfect information availability is at least a step in the right direction.³³

This fallacy has general and far-reaching implications for the application of luck-egalitarian accounts to matters of institutional design. For most of these accounts (Dworkin’s is the exception) an individual’s rightful distributive share depends on specific biographical circumstances that could not feasibly be incorporated into the design and administration of public institutions.³⁴ The assumption that, despite these limits, public institutions should be designed as though information availability were not limited and then implemented with as much information as is practically accessible is a moral fallacy of the second best.³⁵

From Ideal Theory to the Real World

The foregoing argument proceeds from the ideal-theoretic assumption that background distributive institutions—i.e., those outside of the health-care sector—are fully just. The distributive institutions of actual societies are, however, far from just by luck-egalitarian standards. Some individuals are worse off than others because, through no fault of their own, the market value of their developed abilities is lower. Others have suffered adverse brute-luck life events such as serious illness, disability, job layoffs, criminal victimization, natural disasters, and the like. Even the most comprehensive social welfare systems of actual societies fall short of fairly compensating these individuals from the standpoint of the luck-egalitarian framework. Moreover, even in the most egalitarian actual societies, the income and wealth levels of an individual’s parents are highly predictive of her income and wealth levels as an adult. This indicates that many individuals are disadvantaged by the brute-luck circumstances of their family socioeconomic background.³⁶ What are the implications of these background injustices for healthcare-cost allocation in actual societies? To isolate the relevance of unjust background institutions, let us assume that the problem of socially undetectable risk taking does not arise. Contrary to the conclusions above, we assume that society is capable, through acceptable means, of detecting every significant voluntary risk to health that individuals take. Should luck egalitarians in that case conclude that a policy of holding individuals liable for their option-luck healthcare costs is justifiable in far-from-just societies?

Some luck egalitarians explicitly warn against applying the theory directly to matters of policy in the actual world.³⁷ Others, however, emphasize the policy relevance of their accounts. John Roemer, for example, maintains that since the general implementation of luck-egalitarian principles in an actual society is not likely to be achieved all at once, we should seek to implement luck-egalitarian principles in one sphere of policy at a time, as the political opportunity to do so arises:

A distinction must be made between a general theory of distributive justice, which may entail equality of opportunity in various (but perhaps not all) spheres, and its practical implementation. I believe that progress toward distributive justice will advance at different rates in different spheres—health, employment, education, and income, for instance—and I shall endeavor to discuss equal opportunity as it might be advanced in each sphere.³⁸

In the sphere of healthcare, Roemer believes that a luck-egalitarian conception of equal opportunity would be advanced by holding individuals financially liable for their option-luck healthcare needs.³⁹ I will consider in a moment what sort of case might be made for this implementation strategy. It is clear, however, that the general theory of the second best bars any inference of the following sort:

• 1. Institution or policy P would be a part of any fully just society.

• 2. Therefore implementing P in a far-from-just society would be a step in the right direction.

This is a second (and different) moral fallacy of the second best. Here the constrained variables have to do not with limits to information availability, but with policy fragmentation. As Roemer correctly observes, policy reforms are by their nature piecemeal: they are limited to specific spheres (the healthcare system, the tax code, labor and employment law, the public education system etc.) What Roemer does not emphasize is that the net distributive effect of a reform in any given sphere will depend on interaction effects with the existing policies of other spheres. These must be taken as fixed or constrained variables. The general theory of the second best tells us that what would be the fairest policy for a given distributive sphere on the assumption that other spheres are also fairly regulated is not necessarily, or even probably, fairest—taking other spheres of policy as they are.

The point is illustrated by the following example. Suppose that a lot of our slaves are heavy smokers. The prevalence of smoking-related illness in the slave population is high, and the cost of the resulting medical treatment needs is substantial. Our long term goal is a society fully regulated by luck-egalitarian principles, but progress in some spheres—including, regrettably, the sphere of slave labor—is not politically achievable (slavery is a constrained variable). Nonetheless, we would like to take a step in the right direction. It is proposed that a good start might be to hold slaves who smoke accountable for their free choices. We calculate that if slaves who smoke were to work for one extra hour each day, the extra wealth that they produce would offset the costs of treating their smoking-related illnesses. Therefore we propose to implement a system that requires smoking slaves to work one additional hour per day.

While the example illustrates the fallacy, it is not immediately clear how illuminating the example is for the issue of healthcare cost allocation in any given actual societies. Contingent considerations will determine whether or not a policy of imposing option-luck healthcare costs on individuals in an actual society would be a step in the direction of greater fairness or not. Among the relevant contingencies are the likely impact of the policy on members of society who are unjustly disadvantaged, and the likely contribution of the policy to the long term goal of implementing luck-egalitarian principles across all appropriate distributive spheres. If, (unlike our slaves) the unjustly disadvantaged of an actual society are likely to be net beneficiaries of such a reform, then it could be a step in the right direction. Alternately, if such a reform in the sphere of healthcare would significantly advance the cause of achieving justice across all social institutions, then a short-term step backward, in terms of fairness in the overall social distribution, might be justified on grounds of intergenerational justice. Neither of these contingencies, however, can plausibly be said to obtain for actual societies.

The Worst-off and the Unjustly Disadvantaged in the Real World

To assess the likely impact of a proposed policy reform on the unjustly disadvantaged, it is first necessary to identify them. For most luck-egalitarian accounts, this is not a straightforward task. Since there is in principle no lower limit to how badly off one can become as a foreseeable result of her own free choices, the luck-egalitarian framework does not—as some egalitarian theories do—license a straightforward identification of the unjustly disadvantaged with the worst-off (or indeed with any salient social group).⁴⁰ Distributive injustices on most luck-egalitarian accounts are strongly path-dependent—we identify them by looking not only at an individual’s relative distributive share, but also at the biographical circumstances by which she came to have it. Much of this information is socially unavailable. Unless some salient group can serve as a reliable proxy for the unjustly disadvantaged, an impact assessment of the desired kind may be impossible.

It can be shown—perhaps not surprisingly—that the economically worst-off of actual societies are, from a luck egalitarian standpoint, a good proxy for the unjustly disadvantaged of those societies. There are three reasons for this. First, the economically worst-off are disproportionately affected by adverse brute-luck life events; second, they are disproportionately disadvantaged by unequal opportunity; and third, they are systematically deprived of the social bases of self esteem.

In the actual world, the worst off are disproportionately affected by adverse brute-luck life events—illness, job injuries and layoffs, criminal victimization and the like. Not only do they experience such misfortunes with greater frequency than the better off—the poor are also harder hit by any given misfortune. They are less likely than the better off to be adequately insured, less likely to have a well-resourced support network of family and friends, and generally less capable of absorbing losses without catastrophic financial outcomes like the loss of a home, of retirement savings or of college savings for the next generation.⁴¹ In actual societies, the worst off are thus the most likely to have been unjustly disadvantaged by uncompensated adverse brute-luck life events.

Moreover, as noted above, the income and wealth levels of an individual’s parents are highly predictive of her income and wealth levels as an adult. The data on social mobility in the United States, for example, indicate that about one-third of those in the lower two income quintiles would be in a higher quintile (i.e., the third, second or first) if family socioeconomic background conferred no advantage or disadvantage for an individual’s life prospects.⁴² The numbers for the United Kingdom are comparable, and the correlation between family background and socioeconomic status is robust even in the most egalitarian societies.⁴³ Any such correlation is unjust by luck-egalitarian standards (as indeed by the standards of any plausible egalitarian theory) as such a correlation indicates that significant numbers of the economically worst off occupy that position simply because of the brute-luck circumstance of family background.

Finally, a narrow focus on the material deprivation of the worst-off group ignores what arguably is the greatest distributive injustice of developed societies. The economically worst off in these societies are deprived not only of second cars, beach vacations, and botox treatments, but also in many cases of college educations, the opportunity to develop their talents fully, and the chance to pursue careers and avocations that could have been sources of pride and fulfillment.⁴⁴ Thus, the magnitude of injustice suffered by the economically worst-off exceeds any narrowly economic reckoning. I conclude that the worst off of actual societies are a good socially identifiable proxy for the unjustly disadvantaged of those societies. In order to assess the impact of a policy on the unjustly disadvantaged we should look to the impact of the policy on the economically worst off.

Consequences for the Worst off and for Future Generations

A policy of imposing personal liability for option-luck healthcare costs will tend to be regressive in its effects, hitting the worst off the hardest and thus (if the argument of the previous section is sound) tending to aggravate the burdens of those who are already unjustly disadvantaged. The extent of the regressive effect will depend on the accountability mechanism—but to fix ideas, let us assume an option-luck add-on system like the one discussed above. In actual societies, such a system hits the worst off hardest for three reasons. First, the prevalence of unhealthy voluntary behaviors tends to be greater among lower SES groups.⁴⁵ Second, households at the bottom of the SES ladder are the most likely to be under-insured, and thus to incur out-of-pocket costs for lifestyle-related illnesses. Third, many of these households are, as already noted, in financially precarious circumstances, and the imposition of out-of-pocket costs could push them over the brink.

The only remaining question, is whether, on grounds of intergenerational justice, it might be justifiable to magnify existing injustices for the sake of advancing a long-term goal of implementing luck-egalitarian principles in all spheres of social policy. I think we should acknowledge that an affirmative answer must meet a high burden of proof. To justify imposing further costs on those who are already unjustly disadvantaged, we must have a high level of confidence that doing so will make some fairly significant contribution to achieving the final goal of a fully just society. Yet the political forces confronting a reform effort in a given sphere are likely to be idiosyncratic. There seems to be little reason to suppose that a luck-egalitarian reform in the sphere of healthcare would contribute even marginally to the implementation of reforms in other spheres. There are also opportunity costs (in terms of political capital, the mobilization of public opinion, and other advocacy efforts) of pursuing any given reform effort. It would seem preferable for luck egalitarians to push first for progressive reforms that are consistent with long-term luck egalitarian aims—for example, reforms aimed at achieving greater intergenerational social mobility—and to save the implementation of any regressive ones for last. While this is not, by itself, any reason to refrain from imposing liability for option-luck health costs on the better off, we have already seen in Sects. 3 and 4 above that the issues of vertical versus horizontal equity, and of publicity, provide reasons not to impose option luck health care costs on any sector of society. The conclusions of this section simply provide additional (and I believe weightier) reasons not to impose such costs on the worse off.

Conclusion

The luck-egalitarian framework does not—on its own terms—give us any reason to favor allocating healthcare costs on the basis of personal responsibility. Luck egalitarians mistake the implications of their own accounts when they endorse such policies. It is a moral fallacy to suppose that since a certain policy would be fairest given unlimited information availability, implementing it in the context of limited information is a step in the right direction. Even in an otherwise fully just society, imposing personal liability for health would, even while reducing the amount of free-riding, introduce a new and public dimension of unfairness by singling out a small minority of risk takers (the socially detectable ones) to bear added costs while the majority of risk takers (the undetectable ones) continue to impose costs on everyone. In the actual world, imposing personal liability for health would be an even bigger mistake. It is a moral fallacy to suppose that since a certain policy would be part of any fully just set of social institutions, implementing it against a background of unjust institutions would be a step in the right direction. Imposing accountability for health in actual societies would tend to magnify existing distributive injustices. These conclusions have general implications for the application of ideal normative theory to matters of institutional design. No matter how attractive they are in theory, principles that operate on information that is not socially available can offer little direct guidance for the design of social institutions. Moreover, even absent such limits to available information, the theory may offer little or no direct guidance in real-world policy contexts where the overall effect of reforms in any given sphere depends on interaction effects with other spheres.