¹ In popular parlance, “welfare” refers to means-tested programs, such as Aid to Families with Dependent Children in the U.S. However, means-tested programs make up a small portion of the budget of contemporary welfare states.

² Lomasky, Loren E., Persons, Rights, and the Moral Community (New York: Oxford University Press, 1987), p. 13.Google Scholar

³ I am at work on a book, tentatively called Justice, Efficiency, and Community: Can the Welfare State Be Justified?, where I aim to do just that. Elsewhere I have argued that government provision and management of retirement pensions, or what I call old-age social insurance, is inferior to compulsory private pensions (e.g., the system in Chile) on the basis of virtually all principles or values predominant in contemporary political philosophy. See my essay “Can Old-Age Social Insurance Be Justified?” Social Philosophy and Policy, vol. 14, no. 2 (Summer 1997), pp. 116–44Google Scholar. A system of compulsory private pensions would not, in my view, be the most preferred institution, since I believe that voluntary private pensions would be more desirable. Nevertheless, adopting the former would mean the end of a central welfare-state institution.

⁴ My information comes from Glaser, William, Health Insurance in Practice: International Variations in Financing, Benefits, and Problems (San Francisco: Jossey-Bass Publishers, 1991), chs. 1–4 and appendix AGoogle Scholar; and White, Joseph, Competing Solutions: American Health Care Proposals and International Experience (Washington, DC: The Brookings Institution, 1995), ch. 2Google Scholar. Glaser's focus is on Germany, Switzerland, Holland, Belgium, and France; White's is on Germany, France, Canada, the UK, Japan, and Australia.

⁵ Switzerland is a special case. Only a few cantons require that one purchase health insurance. Universal coverage is achieved by subsidies to the sickness funds. Glaser, , Health Insurance in Practice, pp. 4, 61, and 561.Google Scholar

⁶ In Germany, around 10 percent of the population—made up of those above a certain income—is not required to purchase health insurance. Virtually all of them choose to purchase private health insurance. Ibid., pp. 6 and 504.

⁷ In some countries, e.g., Canada, private insurance is only allowed for benefits not covered by NHI. White, , Competing Solutions, p. 66.Google Scholar

⁸ For a clear discussion of the differences between these, see Goodman, John C. and Musgrave, Gerald L., Patient Power: Solving America's Health Care Crisis (Washington, DC: The Cato Institute, 1992), pp. 178–82Google Scholar; and Feingenbaum, Susan, “Body Shop Economics: What's Good for Our Cars May Be Good for Our Health,” Regulation, vol. 15, no. 4 (Fall 1992), pp. 25–31.Google Scholar

⁹ I combine casualty and catastrophic insurance in this hyphenated expression to emphasize that casualty insurance, since it insures against adverse events and does not reimburse one for services consumed, is usually just insurance against catastrophes. However, it is important to notice that the connection between casualty insurance and catastrophic insurance does not hold with equal strength in both directions. That is, while casualty insurance is almost always catastrophic-only insurance, catastrophic-only insurance need not be based on the principles of casualty insurance; that is, it need not be limited to reimbursement for identifiable losses or illnesses. For example, one could take out a high-deductible policy which covers all of one's medical expenses above a certain amount, regardless of whether those expenses were incurred due to some loss or illness. This would be catastrophic-only insurance in the sense that a high level of medical expenditures often indicates the occurrence of a medical catastrophe; however, reimbursement for a high level of medical expenditures is not casualty insurance, since the reimbursement is for the consumption of medical services, not for some identifiable loss or illness.

¹⁰ Sometimes casualty insurers will place limits on the insured's discretion. Auto insurers, for example, sometimes place limits on how and when you fix your car after a wreck.

¹¹ Sometimes, as in Australia, the government sets no fee, but then the government forbids private insurance from reimbursing above what Medicare (Australia's ambulatory NHI program) does. Since Medicare pays 75 percent of doctor fees, this amounts to de facto government fee regulation. See White, , Competing Solutions, p. 97.Google Scholar

¹² Because some countries do not keep any records of waits, and because there is rarely a national waiting list (as opposed to a regional or provincial list), it is often hard to get reliable information about the extent of waits. For an unusually detailed study, see Ramsay, Cynthia and Walker, Michael, Waiting Your Turn: Hospital Waiting Lists in Canada (Vancouver: The Fraser Institute, 1995)Google Scholar. Countries with a national health service tend to have the worst problems. As Robert Baker notes, the British National Health Service is sometimes dubbed the “British National Health Shortages,” given its extensive waits for many operations and procedures. See Baker, , “The Inevitability of Health Care Rationing: A Case Study of Rationing in the British National Health Service,” in Rationing America's Medical Care: The Oregon Plan and Beyond, ed. Strosberg, Martin, Wiener, Joshua M., Baker, Robert, with Fein, I. Alan (Washington, DC: The Brookings Institution, 1992), p. 208Google Scholar. Some countries with national health services have requirements for maximum waiting time; that these requirements fail to work is noted by Lenaghan, Jo, “Health Care Rights in Europe,” in Hard Choices in Health Care, ed. Lenaghan, Jo (London: BMJ Publishing Group, 1997), pp. 188–89Google Scholar. Glaser, , Health Insurance in Practice, p. 250Google Scholar, claims that waits are not a problem in countries that have private insurance companies and nonprofit sickness funds. But this claim is belied by Choices in Health Care, a report by the Netherlands′ Government Committee on Choices in Health Care (Rijswijk: Ministry of Welfare, Health, and Cultural Affairs, 1992), pp. 97–98Google Scholar, and by an admission of the German government in 1993 that waiting lists were about to become more common. See Harper, Timothy, “Rationing: What We Can Learn from Europe,” Medical Economics, 09 13, 1993, p. 140.Google ScholarPubMed

¹³ See Goodman, and Musgrave, , Patient Power, ch. 11.Google Scholar

¹⁴ See Langdon, Steve, “Health Insurance Law,” Congressional Quarterly Weekly Report, vol. 54, no. 37 (09 14, 1996), pp. 2619–23.Google Scholar

¹⁵ For example, such laws have been passed in New York, New Jersey, and Florida. For a discussion of these laws, see Epstein, Richard A., Antidiscrimination in Health Care: Community Rating and Preexisting Conditions (Oakland: The Independent Institute, 1996), pp. 5–7.Google Scholar

¹⁶ See Wasley, Terree, What Has Government Done to Our Health Care? (Washington, DC: The Cato Institute, 1992), pp. 47–57Google Scholar; Goodman, and Musgrave, , Patient Power, pp. 156–61Google Scholar; and Califano, Joseph A. Jr., America's Health Care Revolution (New York: Simon and Schuster, 1986), pp. 40–47.Google ScholarPubMed

¹⁷ The reason hospitals began offering Blue Cross was that the Depression was taking a toll on the hospitals′ business. Service plans, with their “first dollar coverage” (i.e., no deductibles, but with limits on total expenses covered) were a way of stimulating business. See Wasley, , What Has Government Done to Our Health Care?, pp. 47–50Google Scholar; and Califano, , America's Health Care Revolution, pp. 41–42Google Scholar. Blue Shield was introduced after commercial insurers introduced plans to compete with Blue Cross.

¹⁸ Furthermore, prior to the 1930s the AMA took action to squash consumer-based service plans. Starting in the early part of the twentieth century, a number of industries, mutual aid societies, and consumer cooperatives also provided service plans for low monthly or annual fees which covered both hospital and physician services, but these were different from the Blues in a crucial respect. The cooperatives, mutual aid societies, or industries provided the care without the insurance companies: they hired company doctors who were paid a fixed salary, and/or they owned the hospitals or clinics themselves. The AMA eventually eliminated all of these by lobbying successfully for state laws which effectively required all medical service plans to obtain AMA approval. See Goldberg, Lawrence C. and Greenberg, Warren, “The Emergence of Physician-Sponsored Health Insurance: A Historical Perspective,” in Competition in the Health Care Sector, ed. Greenberg, Warren (Germantown: Aspen Systems Corporation, 1978), pp. 288–321Google Scholar; and Goldberg, and Greenberg, , “The Effect of Physician-Controlled Health Insurance: U.S. v. Oregon State Medical Society,” Journal of Health, Politics, Policy, and the Law, vol. 2, no. 1 (Spring 1977), pp. 48–78.CrossRefGoogle Scholar

¹⁹ However, hospital insurance was still confined to a small group, around 9 percent of the population. Around 12 million had hospital insurance, while the total population was around 132 million. The Blues never held less than around 45 percent of the entire private health insurance market until the early 1980s. As of 1993, the latest year for which I have been able to obtain figures, the Blues held around 36 percent of the market. See Health Insurance Association of America, Sourcebook of Health Insurance Data, 1988 (Washington, DC: Health Insurance Association of America, 1988), tables 1.2 and 1.3Google Scholar; and Health Insurance Association of America, Sourcebook of Health Insurance Data, 1995 (Washington, DC: Health Insurance Association of America, 1996), table 2.5.Google Scholar

²⁰ The cost-plus reimbursement system was reined in during the 1980s, as I discuss below.

²¹ Some of this pressure occurred when large firms began self-insuring. In this way, companies established their own rules and procedures; the insurance company's role, if any, was simply limited to processing the claims. Once a significant segment of large employers began to self-insure and engage in cost-control techniques, these techniques rapidly spread through the U.S. health-care market. See Goodman, and Musgrave, , Patient Power, pp. 195–201.Google Scholar

²² See ibid., pp. 201–8; White, , Competing Solutions, pp. 180–83Google Scholar; and Epstein, Richard A., Mortal Peril: Our Inalienable Right to Health Care? (Reading, MA: Addison-Wesley Publishing Company, 1997), pp. 420–30.Google Scholar

²³ On the DRG system, see White, , Competing Solutions, pp. 44–45Google Scholar; and Goodman, and Musgrave, , Patient Power, pp. 55, 60–62, and 302–11.Google Scholar

²⁴ Goodman, and Musgrave, , Patient Power, pp. 60–62 and 303–16.Google Scholar

²⁵ Ibid., pp. 163–67.

²⁶ Why does the same practice exist even after the demise of cost-plus reimbursement? I am not entirely clear on the answer, but some of the reasons seem to be the following. First, a significant percentage of hospitals′ revenues are from Medicaid and Medicare, and, as is often the case with government financed and managed programs, they do not have the same incentives as insurance companies in a genuine market to press for comprehensible prices. Second, the DRG system used by Medicare and Medicaid, and increasingly by private insurers, is still in a sense cost-plus, but with a time lag: DRGs reimburse hospitals not on the basis of present costs, but on the basis of last year's average costs (that is how the figure for reimbursement is arrived at). This still gives hospitals an incentive to inflate costs rather than to compete over price. Third, even cost-conscious employers are not as good at bargaining as patients would be themselves if they paid for hospital expenses out of their medical savings accounts or insurance plans. (Medical savings accounts are tax-free accounts designed to pay for deductibles and small medical bills. See the discussion in Section IIC.) On the first two points, see Goodman, and Musgrave, , Patient Power, pp. 306, 311.Google Scholar

²⁷ Ibid., pp. 53–55.

²⁸ Singapore has certain features of market health insurance, and thus is a possible exception, if one considers it a democracy. See Massaro, Thomas A. and Wong, Yu-Nig, Medical Savings Accounts: The Singapore Experience (Dallas: National Center for Policy Analysis, 1996).Google Scholar

²⁹ Goodman, and Musgrave, , Patient Power: Solving America's Health Care Crisis (supra note 8)Google Scholar. Herzlinger, Regina, Market-Driven Health Care (Reading, MA: Addison-Wesley Publishing Company, 1997)Google Scholar, provides a useful elaboration (and, to some degree, alteration) of the Goodman-Musgrave proposal.

³⁰ But see pp. 99–100 for why I think the first and third conditions must be modified.

³¹ Present tax policies encourage people to choose low-deductible rather than high-deductible insurance, even when the cost of the more expensive premiums for the former outweighs the savings achieved by choosing the lower deductible. Since the premiums paid receive the double tax deduction described in the text, but the money paid out-of-pocket or put toward savings for the bills not covered by the deductible is paid in after-tax dollars, there is a strong incentive to choose the lower deductible. For example, suppose a $1,000 deductible policy costs $800 a year less in premiums than a $250 deductible policy (which may be the case, if one is middle-aged and lives in a high-cost area). Assuming a standard insurance policy that pays 80 percent of medical bills covered by the policy, the higher-deductible policy yields $600 less of health insurance coverage. This is far less than the $800 extra one has to pay in higher premiums; yet since the premiums are excluded from the employee's gross income, while the money paid for bills not covered by the deductible is fully taxed, for many people choosing the low-deductible policy is more rational (particularly if they are in a high tax bracket). See Goodman, and Musgrave, , Patient Power, pp. 44–46 and ch. 8.Google Scholar

³² Goodman and Musgrave leave this matter somewhat open, suggesting at one point that some groups (the young, those living in a low-medical-cost area) might be allowed to make larger annual deposits to their MSAs. See Patient Power, p. 258.Google Scholar

³³ Ibid., pp. 68–69.

³⁴ Or, more precisely, the nonindigent who are not wealthy. The wealthy can generally afford to pay their health-care bills whether or not they purchase health insurance.

³⁵ The reasons stem from the economics of insurance. Markets for insurance exist because there are gains from trade from pooling risks. On the demand side, consumers who have some degree of risk aversion gain from paying a certain amount each year to reduce the chance of a loss in any particular year. On the supply side, companies gain because they are reasonably confident that their payouts to insureds who suffer losses in any given year will be more than made up by income received from insureds who suffer no loss. However, there are no gains from trade for consumers who will suffer certain losses. Since the losses are certain, insurance companies cannot make money by using income obtained from policy-holders in a similar risk class who will not suffer a loss—for there are none. Thus, the company will have to offer “insurance” priced at a rate equal to the loss plus the company's profit and overhead, and no rational consumer would have a demand for such insurance (why pay, for example, $1,500 a year to “insure” against a certain loss of $1,200?). Of course, insurance companies could charge consumers who face certain losses the same premium as those with a low probability of facing losses, and use the income from the latter to subsidize the former; but given the strong tendency in a competitive market for insurers to charge different premiums for different degrees of risk, this is unlikely to occur.

³⁶ Goodman, and Musgrave, , Patient Power, pp. 98–99Google Scholar. They also note (pp. 96–97) that even as late as the 1950s in the U.S. consumers could purchase guaranteed renewable health insurance. Since this was at a time when employer service plans were not quite as completely entrenched as they are today, this may provide some evidence that, in a free market, the demand for such insurance would be considerable, and that insurers would meet the demand.

³⁷ A tax credit for health insurance premiums means that one can reduce one's tax bill by the amount of the credit; for example, if one would otherwise pay $3,500 a year in taxes, and one's health insurance premiums are $3,000 a year, a 30 percent tax credit would cut one's tax bill from $3,500 to $2,600. A tax credit is not regressive; if two people pay the same amount of health insurance premiums, then they get the same reduction in their tax bill. A tax deduction, on the other hand, means that a certain percentage of one's gross income is shielded from taxes; for example, if one is taxed on gross income of $30,000, but one can deduct 30 percent of one's $3,000 health insurance premiums, then one would be taxed on an income of $29,100. Tax deductions for health insurance premiums are regressive; that is, the value of the exclusion of the premiums from taxable income is more valuable for those in the higher income brackets. Nine hundred dollars excluded from taxation is less valuable for someone who is not in a very high marginal tax bracket than for someone who is in such a bracket. For more details, see ibid., pp. 41–43.

³⁸ This is suggested by Herzlinger, , Market-Driven Health Care, p. 258.Google Scholar

³⁹ On this conceptual or terminological matter, I have been influenced by Temkin, Larry S., Inequality (New York: Oxford University Press), pp. 7–8.Google Scholar

⁴⁰ Rawls, John, A Theory of Justice (Cambridge: Harvard University Press, 1971), p. 302Google Scholar; Rawls, , Political Liberalism (New York: Columbia University Press, 1993), p. 291.Google Scholar

⁴¹ Two helpful summaries of the “Equality of what?” literature are Cohen, G. A., “On the Currency of Egalitarian Justice,” Ethics, vol. 99, no. 4 (07 1989), pp. 906–44CrossRefGoogle Scholar; and Sen, Amartya, Inequality Reexamined (Cambridge: Harvard University Press, 1992).Google Scholar

⁴² Rawls's difference principle is the great exception here: his conception of the worst off does not incorporate a responsibility or choice condition. But post-Rawlsian egalitarianism invariably incorporates such a condition, for reasons I discuss in Section IV. In any event, though Rawls's difference principle does not incorporate a responsibility or choice condition, the distinction between those who are and are not responsible for their choices is recognized by Rawls in other parts of his theory and writing, particularly in his argument that justice does not require that one subsidize voluntarily acquired expensive tastes. For a valuable discussion of how Rawls does recognize a distinction between voluntary and involuntary inequalities, despite the wording of the difference principle, see Kymlicka, Will, Contemporary Political Philosophy: An Introduction (Oxford: Clarendon Press, 1990), pp. 73–76.Google Scholar

⁴³ Cohen, , “On the Currency of Egalitarian Justice,” p. 908Google Scholar. I added the modifier “bad” before brute luck, because it is not an essential part of egalitarianism that the effects of good brute luck must be extinguished or modified. See Vallentyne, Peter, “Self-Ownership and Equality: Brute Luck, Gifts, Universal Dominance, and Leximin,” Ethics, vol. 107, no. 2 (01 1997), pp. 330–31CrossRefGoogle Scholar. The distinction between brute and option luck originated with Ronald Dworkin, who played the crucial role in contemporary egalitarianism's incorporation of a responsibility or choice condition. See Dworkin, , “What Is Equality? Part 2: Equality of Resources,” Philosophy and Public Affairs, vol. 10, no. 4 (Fall 1981), p. 283Google Scholar. There is, unfortunately, no canonical definition of the brute luck/option luck distinction. Dworkin originally defined it so that option luck is luck that results from a deliberate or calculated gamble, but later egalitarians have modified this, probably because Dworkin's definition seems too restrictive—the key intuition behind the distinction is whether choices significantly influence one's outcomes, and choices can play a significant role even where one does not deliberate or calculate. My use of the distinction comes from Vallentyne's gloss on Dworkin's distinction.

⁴⁴ All rights imply that someone other than the right-holder has duties to respect the right. Negative rights imply duties to refrain from some action; positive rights imply duties to do something. Basic rights are rights which have a considerable degree of moral weight, so that they typically defeat claims of aggregate utility or individual well-being.

⁴⁵ Another possible explanation is that fixing inequalities in health care is more feasible than fixing inequalities in health.

⁴⁶ Rawls, , A Theory of Justice, pp. 60–61Google Scholar; Rawls, , Political Liberalism, p. 291.Google Scholar

⁴⁷ For a discussion of the main lines of argument many egalitarians use to defend basic rights, see Shapiro, Daniel, “Liberalism, Basic Rights, and Free Exchange,” Journal of Social Philosophy, vol. 26, no. 2 (Fall 1995), pp. 104–5CrossRefGoogle Scholar and my references therein.

⁴⁸ Rawls, , A Theory of Justice, p. 73.Google Scholar

⁴⁹ Daniels, Norman, Just Health Care (New York: Cambridge University Press, 1985), chs. 2 and 4CrossRefGoogle ScholarPubMed. Basically the same account is given, in a more compressed manner, in Daniels, Norman, Light, Donald W., and Caplan, Ronald L., Benchmarks of Fairness for Health Care Reform (New York: Oxford University Press, 1997), pp. 19–22.Google Scholar

⁵⁰ Daniels, et al. , Benchmarks of Fairness, pp. 27–28 and 43–44.Google Scholar

⁵¹ Ibid., p. 46.

⁵² As noted by Buchanan, Allen, “Health-Care Delivery and Resource Allocation,” in Medical Ethics, ed. Veatch, Robert (Sudsbury: Jones and Bartlett, 1997), pp. 347–48.Google Scholar

⁵³ Daniels, et al. , Benchmarks of Fairness, pp. 25–26.Google Scholar

⁵⁴ Ibid., pp. 57–58. Daniels's critical description of proposals in the U.S. Congress for NHI (none of which even mentioned the issue of rationing) as using “doublespeak,” shows that he would be critical of present-day NHI secretive criteria for rationing. See ibid., pp. 116–19.

⁵⁵ See Dworkin, Ronald, “Justice in the Distribution of Health Care,” McGill Law Review, vol. 38, no. 4 (1993), pp. 883–98Google ScholarPubMed; and Dworkin, , “Will Clinton's Plan Be Fair?” New York Review of Books, 01 13, 1994, pp. 1–8.Google ScholarPubMed

⁵⁶ The key pieces are Dworkin, , “What Is Equality? Part 1: Equality of Welfare,” Philosophy and Public Affairs, vol. 10, no. 3 (Summer 1981), pp. 185–246Google Scholar; Dworkin, , “What Is Equality? Part 2: Equality of Resources,” Philosophy and Public Affairs, vol. 10, no. 4 (Fall 1981), pp. 283–345Google Scholar; and Dworkin, , “What Is Equality? Part 3: The Place of Liberty,” Iowa Law Review, vol. 73, no. 1 (1987), pp. 1–54.Google Scholar

⁵⁷ Dworkin does acknowledge in a note that this is not true of all risks; he mentions that it would seem fair to charge smokers more than nonsmokers. As I shall argue in the next section, this concession helps to undermine Dworkin's defense of NHI. See Dworkin, , “Will Clinton's Plan Be Fair?” p. 4, note 10.Google Scholar

⁵⁸ It is not entirely clear why Dworkin introduces the concept of prudence at this point. Perhaps the reason is that he is working within the contractarian tradition in moral and political philosophy, where justice is determined by seeing what rational, prudent people would choose under certain conditions.

⁵⁹ Dworkin, , “Will Clinton's Plan Be Fair?” p. 4.Google Scholar

⁶⁰ Ibid., p. 5.

⁶¹ Dworkin, , “Justice in the Distribution of Health Care,” p. 898.Google Scholar

⁶² I will not discuss the issue of compulsory coverage, since as I noted in Section II, the Goodman-Musgrave proposal for MHI may not fully eliminate compulsion for the provision of some very minimal set of health-care benefits.

⁶³ See McGinnis, J. Michael and Foege, William H., “Actual Causes of Death in the United States” Journal of the American Medical Association, vol. 270, no. 18 (11 1993), pp. 2207–12CrossRefGoogle ScholarPubMed. I thank Robert S. Sade, M.D. for directing me to this reference. This article surveys and synthesizes articles published between 1977 and 1993 that discussed the causes of death in the U.S. The authors estimate that around 40 percent of all deaths in the U.S. are due to tobacco use, alcohol consumption, sexual activity, motor vehicle accidents, and the use of illegal drugs. A longitudinal study of 6,928 adult males in Alameda County, California, from 1965 to 1974, is also suggestive: it shows that good health practices (not smoking, moderate or no use of alcohol, seven to eight hours regular sleep, regular physical activity, proper weight, eating breakfast, not eating between meals), and not the initial health status of the survey respondents, were responsible for significant differences in mortality. See Breslow, Lester and Enstrom, James E., “Persistence of Health Habits and Their Relationship to Mortality,” Preventive Medicine, vol. 9 (09 4, 1980), pp. 469–83CrossRefGoogle ScholarPubMed. I have not located any studies that attempt to quantify the effects of lifestyle choices on mortality in Europe, but since a greater percentage of the population smokes in most European countries than in the U.S., it is not unlikely that a higher proportion of deaths in those countries could be attributed to “lifestyle” choices. See U.S. Congress, Office of Technology Assessment, International Health Statistics: What the Numbers Mean for the United States (Washington, DC: U.S. Government Printing Office, 11 1993), pp. 70–71 and 83–84.Google Scholar

It is important to stress two limitations of the above articles. First their focus is on premature deaths, not morbidity, illness, or injury. That lifestyle risks cause a certain percentage of premature deaths does not mean that they cause the same percentage of morbidity, illness, or injury. For example, women generally have greater morbidity than men, but tend to live longer. See Blaxter, Mildred, “A Comparison of Measures of Inequality in Morbidity,” in Health Inequalities in European Countries, ed. Fox, John (Aldershot: Gower Publishing, 1989), p. 199Google Scholar. Second, the first article cited above does not discuss how much of the differences in premature mortality is due to lifestyle factors.

Studies that do attempt to determine how much of differential premature mortality was due to lifestyle factors, as opposed to other influences such as years of schooling, relative income, and occupational status, have found that lifestyle factors played a role, but not the most significant role. The most famous of these is probably Michael Marmot's longitudinal study of four occupational classes of British civil servants working in the same office, which showed that the civil servants' occupational status had a stronger correlation with mortality from coronary heart disease than did lifestyle factors such as smoking. See Marmot, Michael, “Social Inequalities in Mortality: The Social Environment,” in Class and Health, ed. Wilkinson, Richard G. (London: Tavistock Publishing, 1986), pp. 21–33Google Scholar. Other longitudinal studies that show a stronger correlation between health and income and/or education than between health and lifestyle factors are discussed by Haan, Mary N., Kaplan, George A., and Syme, S. Leonard, “Socioeconomic Status and Health: Old Observations and New Thoughts,” in Pathways to Health: The Role of Social Factors, ed. Bunker, John P., Gomby, Deanna S., and Kehrer, Barbara H. (Menlo Park: Henry J. Kaiser Family Foundation, 1989), pp. 83–86Google Scholar. For reasons I discuss in note 71, however, all of these studies must be viewed with great caution and have various methodological and conceptual problems.

⁶⁴ Shapiro, Daniel, “Smoking Tobacco: Irrationality, Addiction, and Paternalism,” Public Affairs Quarterly, vol. 8, no. 2 (04 1994), pp. 187–203.Google Scholar

⁶⁵ See my “Addiction, Responsibility, and Drug Policy,” unpublished, 05 1997Google Scholar. See also DeGrandpre, Richard J. and White, Ed, “Drugs: In Care of the Self,” Common Knowledge, vol. 5, no. 3 (Winter 1996), pp. 27–48Google Scholar and the references cited therein.

⁶⁶ For a good critique of such views, see Peele, Stanton, The Diseasing of America: Addiction Treatment Out of Control (Boston: Houghton Mifflin, 1991).Google Scholar

⁶⁷ On the relationship between education and mortality, see Valkonen, Tapani's longitudinal study of seven European countries, “Adult Mortality and Levels of Education,” in Health Inequalities in European Countries, ch. 7, pp. 142–72Google Scholar. In “Socioeconomic Status and Health: An Examination of Underlying Processes,” in Bunker, et al. , eds., Pathways to Health, p. 19Google Scholar, David Mechanic says that “[e]ducation is one of the most consistent predictors of measures of mortality, morbidity and health behavior.” In “Socioeconomic Status: A Personal Research Perspective,” in Ibid., p. 141, Michael Grossman and Theodore J. Joyce say that education has been shown to be a more important causal determinant than occupational status or income. Leonard A. Sagan also argues that education is a more important causal determinant of morbidity than wealth; see Sagan, , The Health of Nations: True Causes of Sickness and Well-Being (New York: Basic Books, 1987), pp. 176–78.Google Scholar

Of course, even if education is a better causal determinant or predictor of mortality and/or morbidity than income, this does not mean that income level is unimportant. For some studies that stress the effect of income on mortality, see Pappas, G., Queen, S., Adden, W., and Fisher, G., “The Increasing Disparity in Mortality between Socioeconomic Groups in the United States, 1960 and 1986,” in New England Journal of Medicine, vol. 329 (1993), pp. 103–9CrossRefGoogle ScholarPubMed, cited in Fein, Oliver, “The Influence of Social Class on Health Status: American and British Research on Health Inequalities,” in Journal of General Internal Medicine, vol. 10, no. 10 (10 1995), pp. 577–86CrossRefGoogle ScholarPubMed; and Haan, et al. , “Socioeconomic Status and Health,” p. 84.Google Scholar

⁶⁸ On self-efficacy, see Bandura, Albert, “Self-Efficacy Mechanism in Physiological Activation and Health-Promoting Behavior,” in Neurobiology of Learning, Emotion, and Affect, ed. Madden, John IV (New York: Raven Press, 1991), pp. 229–69Google Scholar; Herzlinger, , Market-Driven Health Care, pp. 60–62Google Scholar and the references cited therein; and Sagan, , The Health of Nations, pp. 187–94Google Scholar. Sagan also discusses the role of hope and coping skills on pp. 180–81 and 184.

⁶⁹ Cohen, , “On the Currency of Egalitarian Justice,” pp. 930–31Google Scholar, discusses the example of a cheerful person and says that, in conversation, Dworkin viewed cheerfulness as a borderline case.

⁷⁰ For discussion of the relationship between education and lifestyle choices, see Sagan, , The-Health of Nations, p. 179Google Scholar; Illsley, Raymond, “Comparative Review of Sources, Methodology, and Knowledge [of Health Inequalities],” Social Science and Medicine, vol. 31, no. 3 (1990), p. 230CrossRefGoogle Scholar; and Mechanic, , “Socioeconomic Status and Health” (supra note 67), p. 19Google Scholar. There are a number of studies showing a relationship between socioeconomic status and risky lifestyles, but for reasons discussed in note 71, such studies are of limited use, since socioeconomic status, as typically defined, combines education with other influences, such as occupational status and income.

⁷¹ Studies that show a correlation between one influence (e.g., lifestyle or income) and another (e.g., mortality or morbidity), or that show differing degrees of correlation between different influences and a certain measure(s) of health, are always vulnerable to the “third variable” problem: correlation is not proof of causality. Studies that show a correlation between socioeconomic status—defined as a combination of income and/or occupational status and/or education—and health outcomes are particularly problematic, if the correlation between the various factors making up socioeconomic status is not that high, as Victor Fuchs points out in “General Comments of Conference Participants,” in Bunker, et al. , eds., Pathways to Health, p. 226Google Scholar. Furthermore, some studies are not longitudinal, which raises the suspicion that the relationship found is a temporary one. Many studies are of only one country, which makes them of limited use. Morbidity studies are particularly tricky, since (a) authors often use different notions of morbidity (is it defined in terms of some objective signs of pathology, or in terms of self-reported symptoms, or by an inability to perform certain “normal” tasks?) and (b) there is a big gap between actual morbidity and reported morbidity. On different concepts of morbidity, see Blaxter, , “A Comparison of Measures of Inequality in Morbidity” (supra note 63), pp. 206–21Google Scholar. On the gap between reported and actual morbidity, see Illsley, , “Comparative Review of Sources, Methodology, and Knowledge,” p. 233.Google Scholar

Finally, the factors being measured are notoriously difficult to measure accurately (e.g., diet and physical activity). One reason “years of schooling” may show up as a better predictor of mortality or morbidity than other factors is simply that “years of schooling” is relatively easy to measure.

⁷² Perhaps an egalitarian might reply that where basic rights are at stake, egalitarian principles simply do not apply. Since there is a basic right to free speech, the egalitarian principle, whether it be correcting for unchosen inequalities or correcting for inequalities per se, would not apply. However, this reply begs the question. Egalitarians who justify a basic right to free speech do so by arguing that there are communicative choices vis-à-vis the exercise of one's conception of the good that need to be protected. If egalitarians acknowledge that communicative activities are influenced by a mixture of choices and unchosen circumstances—but they nevertheless, despite this mixture, defend a basic right to free speech and inequalities resulting from the exercise of this right—then the same kind of defense should apply to health insurance systems.

⁷³ Economist Robert Sugden has raised an objection to this argument, noting that risk rating is based, to some degree, upon unalterable or virtually unalterable characteristics of an individual (e.g., upon age or sex). That one's premiums are based in part on such characteristics provides no incentive for more responsible behavior, since one cannot alter (or can only alter with great difficulty) such characteristics, and one is not acting responsibly or irresponsibly by simply having such characteristics. However, the arguments in this essay are comparative ones. That risk rating is based only in part on characteristics or behavior that one can alter is still an improvement (vis-à-vis acting responsibly in the future) over the absence of risk rating, since the former provides more information than the latter concerning the cost of one's (partially) alterable behavior.

⁷⁴ See note 70.

⁷⁵ Knowledge about some risky activities, such as smoking, is widespread. But it is one thing to know this in the abstract and another to have this information communicated in a direct and powerful way. Public health campaigns about the risks of smoking may not be as effective as health insurance that raises one's premiums based on one's behavior. For evidence that smokers, at least in the U.S., are not ill-informed about the risks of smoking, see Viscusi, W. Kip, Smoking: Making the Risky Decision (New York: Oxford University Press, 1992), pp. 77–78.Google Scholar

⁷⁶ Notice also that the combination of medical savings accounts plus risk rating not only provides a way for people to become informed about the effects of different lifestyle choices, but encourages them to budget over time to reap the benefits of healthy lifestyle choices or pay the costs of unhealthy ones. See Goodman, and Musgrave, , Patient Power, p. 251.Google Scholar

⁷⁷ Much of my argument in the text concerning the relationship between insurance and moral hazard is indebted to Abraham, Kenneth, Distributing Risk: Insurance, Legal Theory, and Public Policy (New Haven: Yale University Press, 1986), ch. 2.CrossRefGoogle Scholar

⁷⁸ Neither NHI nor MHI has an obvious advantage on the issue of combating intentional (fraudulent) moral hazard. Many policies in MHI will pay the insured directly, which will give some of them an incentive to commit fraud; on the other hand, companies which are lackadaisical in combating fraud are at a competitive disadvantage. NHI does not pay the insured directly, which removes that incentive for fraud; but payment to health-care providers can provide the same kind of incentives, particularly with fee-for-service medicine.

⁷⁹ By definition, my analysis in the text does not apply to emergencies. Even here, casualty insurance has a built-in incentive to limit moral hazard, namely, that it would not pay for a trip to an emergency room per se, but would only pay if something serious was uncovered.

⁸⁰ On the difficulties NHI systems have faced historically in getting doctors to accept expenditure targets, see Glaser, William A., “How Expenditure Caps and Expenditure Targets Really Work,” The Milbank Quarterly, vol. 71, no. 1 (1993), pp. 97–127.CrossRefGoogle ScholarPubMed

⁸¹ “Rationing” is an ambiguous concept. In any system where all wants or preferences cannot be satisfied—that is, any system in the real world—some method or procedure is used to determine which wants or preferences get satisfied. In that sense, both markets and governments ration. However, as Ellen Frankel Paul has pointed out to me, rationing also means the division of a fixed supply with fixed shares. In that sense, it is quite misleading to say that markets ration, since, over time, markets expand the resources or supply available (while government per se cannot expand resources).

⁸² I adopt this phrase from a description of Oregon's rationing system, which I discuss below. See Strosberg, Martin A.'s introduction to Rationing America's Medical Care: The Oregon Plan and Beyond, ed. Strosberg, Martin A., Wiener, Joshua M., Baker, Robert, with Fein, I. Alan (Washington, DC: The Brookings Institution, 1992), p. 3.Google Scholar

⁸³ A citizen jury is a “trial” whereby those responsible for health-care allocation decisions are “cross-examined” by experts, after which citizens on the jury deliberate and offer their assessment of those allocation decisions.

⁸⁴ Daniels, et al. , Benchmarks of Fairness, p. 57.Google Scholar

⁸⁵ See Lenaghan, Jo, “Health Care Rights in Europe: A Comparative Perspective,”Google Scholar in Lenaghan, , ed., Hard Choices in Health Care (supra note 12), p. 188.Google Scholar

⁸⁶ See Strosberg's introduction to Strosberg, et al. , eds., Rationing America's Medical Care, pp. 3–7Google Scholar; and Kitzhaber, J. and Kemmy, A. M., “On the Oregon Trail,” British Medical Bulletin, vol. 51, no. 4 (10 1995), pp. 813–17.CrossRefGoogle ScholarPubMed

⁸⁷ See Ham, C., “Synthesis: What We Can Learn from International Experience,” British Medical Bulletin, vol. 51, no. 4 (10 1995), pp. 821–28CrossRefGoogle ScholarPubMed. In Holland, a government commission proposed something similar to Oregon's rationing system, but so far this has had no effect on public policy. See van de Ven, P. M. M., “Choices in Health Care: A Contribution from the Netherlands,”Google ScholarIbid., pp. 785–88.

⁸⁸ Daniels, et al. , Benchmarks of Fairness, p. 58.Google Scholar

⁸⁹ Daniels, , Just Health Care, p. 35.Google Scholar

⁹⁰ See Emmanuel, Ezekiel J., The Ends of Human Life: Medical Ethics in a Liberal Polity (Cambridge: Harvard University Press, 1991), pp. 130–33Google Scholar, for some cogent arguments along this line.

⁹¹ Daniels mentions both of these problems, along with some others, in “Four Unsolved Rationing Problems: A Challenge,” Hastings Center Report, vol. 24, no. 4 (07–08 1994), p. 28.Google Scholar

⁹² Furthermore, in some NHI systems with sickness funds, one has limited choice in picking the fund. In Germany, for example, members of the 90 percent of the population that must stay within NHI are often assigned sickness funds on the basis of their occupation and/or region. See Glaser, , Health Insurance in Practice, pp. 502–3.Google Scholar

⁹³ As noted in Maxwell, R. J., “Why Rationing Is on the Agenda” British Medical Bulletin, vol. 51, no. 4 (10 1995), p. 765CrossRefGoogle ScholarPubMed; and Harrison, S., “A Policy Agenda for Health Care Rationing,”Google ScholarIbid., p. 892.

⁹⁴ See Aaron, Henry J. and Schwartz, William B., The Painful Prescription: Rationing Hospital Care (Washington, DC: The Brookings Institution, 1984), pp. 101–2Google Scholar. That this still occurs in Britain is argued in Klein, R., Day, P., and Redmayne, S., “Rationing in the NHS: The Dance of the Seven Veils—In Reverse,” British Medical Bulletin, vol. 51, no. 4 (10 1995), pp. 769–70CrossRefGoogle ScholarPubMed. That most patients in Britain accept the doctor's views that no further treatment is warranted is argued in Harrison, , “A Policy Agenda for Health Care Rationing,” p. 892.Google Scholar

⁹⁵ Aaron, Henry J., “The Oregon Experiment,”Google Scholar in Strosberg, et al. , eds., Rationing America's Medical Care, p. 109.Google Scholar

⁹⁶ In what follows, I rely on Illsley, , “Comparative Review of Sources, Methodology, and Knowledge” (supra note 70), p. 233.Google Scholar

⁹⁷ Ibid., p. 231.

⁹⁸ The following paragraph is much indebted to Goodin, Robert E., Grand, Julian Le, and Gibson, D. M., “Distributional Biases in Social Service Delivery Systems,” in Not Only the Poor: The Middle Classes and the Welfare State, ed. Goodin, Robert E. and Grand, Julian Le (London: Allen and Unwin, 1987), pp. 131–32.Google Scholar

⁹⁹ Systems with a national health service tend to centralize care within a certain region, while, in a market, services are typically dispersed throughout a city or country. See Goodin, et al. , “Distributional Biases in Social Service Delivery Systems,” pp. 128–30 and 133–38.Google Scholar

¹⁰⁰ Julian Le Grand presents evidence that this occurred in the British National Health Service, i.e., that middle-class capture occurred; see Grand, Le, The Strategy of Equality: Redistribution and the Social Services (London: G. Allen and Unwin, 1982), ch. 3Google Scholar. For an opposing viewpoint, and an argument that Le Grand used inappropriate methodology to measure the extent to which the British National Health Service allocated resources equitably, see O'Donnell, Owen and Propper, Carol, “Equity and the Distribution of UK National Health Service Resources,” Journal of Health Economics, vol. 10 (1991), pp. 1–19.CrossRefGoogle ScholarPubMed

¹⁰¹ Menzel, Paul, Strong Medicine: The Ethical Rationing of Health Care (New York: Oxford University Press, 1990), p. 143.Google Scholar

¹⁰² Furthermore, the egalitarian could point out that the relative disadvantage for the poor in getting access to preventive care gets worse over time, since MSAs that are used to pay for such care are tax-free over one's lifetime, and are more valuable for those in a high tax bracket than for those in a low or zero bracket.

¹⁰³ It might be objected that from an egalitarian point of view, it is more important for the poor to have routine and noncatastrophic care rather than catastrophic care, since having the former prevents the need for much of the latter. Under MHI, the poor may very well not have access to routine and noncatastrophic care in two kinds of cases: illnesses which must be paid for out-of-pocket or out of one's savings under the terms of a high-deductible policy; and preventive care, which is not covered by casualty insurance when there is no adverse event or loss.

As for the former, it would be less inegalitarian to provide better access to such care than to catastrophic care only if the bulk of catastrophes resulted from such low-cost illnesses. However, this is not the case. Most catastrophic illnesses result from chronic degenerative conditions that start in middle and old age. Furthermore, it is worth reiterating a point already made in the text, that under MHI, the poor will get a (refundable) tax credit for a somewhat lower-deductible policy than would be available with a tax credit to the non-indigent.

The issues relating to preventive care are more complex, but the argument does not hold here either. First, some preventive care involves expensive diagnostic, high-tech equipment (CAT scans, mammograms, etc.); thus, NHI's rationing of high-tech equipment affects access to preventive care. Second, the higher rate of time preference of the poor suggests that they will not take advantage of preventive services, even if such services are available (I have found no data on whether the poor under NHI take good advantage of preventive care, so this remains a theoretical point, rather than an empirical one). Third, and most important, even if we were to suppose that the poor will take advantage of subsidized preventive services, and that these will not be rationed by NHI, the logic of the argument is suspect. For preventive care to be effective, it must cover most or all of the relevant population. Thus, for an egalitarian to plausibly argue that it is fairer for the poor to have better access to preventive care than catastrophic care, he must argue that it is better to treat all of the poor—most of whom are healthy now and only some of whom will later have the disease we are seeking to prevent—than it is to treat a smaller group, who will actually be sick later. There is little in egalitarianism that would support such a conclusion; on the contrary, focusing resources on people now who are in acute need seems to be more in keeping with the spirit of egalitarianism.