Duke Law School
Duke Law Journal

| PDF Version | Search

   
Cited: 69 Law & Contemp. Probs. 103
[*pg 103]

OF HEAD TAXES, INCOME TAXES, AND DISTRIBUTIVE JUSTICE IN AMERICAN HEALTH CARE

LAWRENCE ZELENAK*


I. INTRODUCTION

II. THE DISTRIBUTIONAL EFFECTS OF THE HEALTH-INSURANCE HEAD TAX

III. THE ROLE OF THE INCOME-TAX TREATMENT OF EMPLOYER-PROVIDED HEALTH INSURANCE
   A. The Treatment of Health Insurance Under an Ideal Income Tax
   B. Health Insurance and the Actual Income Tax
   C. Moving the Actual Toward the Ideal

IV. CONCLUSION

FOOTNOTES


   

I. INTRODUCTION

This response to Clark Havighurst's and Barak Richman's powerful indictment of the distributional effects of the American health care financing system1 addresses two respects in which they have somewhat overstated their claims. First, they argue that the current system imposes an implicit head tax on all persons with employer-provided health insurance, the proceeds of which are used largely to provide health care for the uninsured, and that the result of this implicit tax-and-transfer system is a regressive redistribution of income. However, this response explains that once one considers the distribution of the benefits financed by the tax, as well as the distribution of the tax burden, the overall system is not necessarily regressive. Indeed, it may well serve to decrease income inequality. Second, they criticize the income-tax exclusion for employer-provided health insurance as "aid[ing] the moral-hazard enemy."2 This response, however, argues that the core of the current income-tax treatment of health insurance -- the exclusion from the tax base of the value of basic employer-provided health insurance -- makes good sense from a tax-internal policy perspective and is not a health-policy culprit. Other aspects of the current income-tax treatment of health care expenditures -- in particular, the exclusion of the value of non-basic employer-provided health insurance, the tax bias against non-employer-provided health insurance, and the tax bias against health-insurance cost sharing -- are both bad tax policy and bad health policy, and should be reformed. The core of the exclusion, however, should be retained.

Both of these comments are in the nature of friendly amendments. Even with the view advanced here that the implicit head tax is not necessarily regressive in its overall effect, the unmasking of the head tax suggests the transfers it makes possible should be financed by a tax keyed to ability to pay. Similarly, [*pg 104] even with the view that the core of the income tax exclusion is justified, much of the income-tax treatment of health care costs is in dire need of reform.


   

II. THE DISTRIBUTIONAL EFFECTS OF THE HEALTH-INSURANCE HEAD TAX

According to Havighurst and Richman, hospitals take advantage of their insurance-enhanced market power by charging insurers substantially more than the cost of caring for insured patients, and the insurers pass on the extra costs to their insureds as "a 'head tax,' which falls on individuals [with employer-provided health insurance] without appreciable correlation to wealth, income, or ability to pay."3 The hospitals use these excess payments to finance a range of unprofitable activities, most notably, research and development, and providing uncompensated care to the uninsured.4 Because low- and high-income insureds pay (more or less) the same amount of this quasi-tax, rather than paying in proportion to income or wealth, Havighurst and Richman conclude that this system of financing health care for the uninsured is regressive.5

Much of this analysis is correct. The identification of the disguised head tax is persuasive, and it is almost certainly true that lawmakers designing an explicit tax to finance health care for the uninsured would not have chosen a head tax.6 However, the implicit tax-and-transfer system they describe is not necessarily regressive. It is, of course, common usage to describe an income tax as regressive if taxpayers with higher incomes pay a lower percentage of their income in tax than do taxpayers with lower incomes. It is also common to describe a head tax as regressive with respect to an income base, because a head tax expressed as a percentage of income obviously declines as income increases. It is a doubtful practice, however, to apply a distributional label to a tax system without regard to the distribution of the benefits financed by the tax. A tax that would be labeled regressive under the usual tax-viewed-in-isolation approach may look very different when considered together with the use of the tax's proceeds. Even a head tax may be part of a progressive tax-and-transfer program. Suppose, for example, a society of just two persons. Rich has annual income of $100,000, and Poor has annual income of $20,000. Each person is subject to a $10,000 head tax, the entire $20,000 proceeds of which are transferred (as a cash transfer, as in-kind benefits, or both) to Poor. The overall effect of the system is progressively redistributive, increasing Poor's after-tax-and-transfer income [*pg 105] to $30,000, and decreasing Rich's to $90,000. Viewed in isolation, the head tax may be labeled regressive, but it is part of an undeniably progressive system.

Havighurst and Richman acknowledge that a complete distributional analysis requires consideration of the benefits financed by the tax, but they summarily conclude that "the manner in which the proceeds are used [does not] appear to rectify the apparent regressivity" of the tax itself.7 It is not clear, however, that this conclusion is correct. Of course, the distributional effects of the system described by Havighurst and Richman -- in which an implicit head tax on those with employer-provided health insurance finances health care for the uninsured, as well as research and development (R&D) -- are considerably more complicated than in the hypothetical world of Rich and Poor. For the moment, set aside the portion of the quasi-head tax used to finance R&D and consider only the portion of the head tax used to finance health care for the uninsured. Viewed in isolation, imposing a tax in the same dollar amount on low- and high-income workers with employer-sponsored health insurance is indeed regressive. On the other hand, transferring money from a higher-income group (those with employer-provided health insurance) to a lower-income group (those without health insurance) is progressive.8

How might one determine which effect predominates? The most common measure of the degree of inequality of a distribution of income is the Gini coefficient.9 The Gini coefficient is most easily explained graphically.

[*pg 106]

The horizontal (x) axis in Figure 1 represents percentages of the population, while the vertical (y) axis represents percentages of the total income of all members of the population. Any possible income distribution within a society can be indicated by a Lorenz curve, showing for the lowest x% of the population the y% of total income in the society belonging to that lowest x%.10 The straight line, running from the lower left to the upper right, is the Lorenz curve of a perfectly equal distribution of income, under which the lowest x% of the population always possesses exactly x% of the total income within the society. Any inequality of income distribution will produce a convex Lorenz curve, touching the line of equality at 0% and 100% of the population and below the line of equality at all other points. The Gini coefficient is the area between a society's Lorenz curve and the line of equality, expressed as a percentage of the total area underneath the line of equality. Thus, if the area between the line of equality and the Lorenz curve is A, and the area underneath the Lorenz curve is B, the Gini coefficient is A/(A+B). The higher the coefficient, the greater the inequality of the income distribution. At the extremes, the Gini coefficient for a society with a perfectly equal distribution of income would be zero, and the coefficient for a society in which all the income is possessed by the single richest member would be one. Whether a tax-and-transfer system is progressive or regressive can be measured by comparing the Gini coefficient of the society in the absence of the tax-and-transfer system with the Gini coefficient taking the tax-and-transfer system into account.

Applying this sort of analysis to the implicit head tax identified by Havighurst and Richman would be a massive undertaking. It would require de[*pg 107] tailed information on the income distributions of persons with and without employer-provided health insurance, on the amount of the implicit head tax, and on the distribution of the health care benefits funded by the head tax. However, an extremely simplified model, not completely divorced from reality, can give some sense of the possible distributional effects of the quasi-head tax and the benefits it finances.

Suppose society consists of eight people with employer-provided health insurance and four without. In the absence of the quasi-head tax, the incomes of the people with health insurance would be $200,000, $150,000, $100,000, $75,000, $75,000, $50,000, $50,000, and $25,000. In the absence of benefits funded by the head tax, the incomes of the people without health insurance would be $50,000, $25,000, $25,000, and $10,000.11 Taking all twelve people into account, the society has a Gini coefficient of 0.405689.12 Now suppose a $4000 benefit for each of the four uninsured persons is financed by a $2000 head tax imposed on each of the eight insured persons. After the tax-and-transfer system has its effect, the insured persons have incomes of $198,000, $148,000, $98,000, $73,000, $73,000, $48,000, $48,000, and $23,000, while the uninsureds have incomes of $54,000, $29,000, $29,000, and $14,000. The Gini coefficient of this new income distribution is 0.392515. The lower coefficient, after tax and transfer, indicates that the head tax and the benefits it funds have decreased the inequality in the overall distribution of income. On these hypothetical numbers, the effect of a system of the type criticized as regressive by Havighurst and Richman is mildly progressive. To be sure, the head tax in isolation would be regressive. If the government took $2000 from each of the eight insured persons and tossed the money into the ocean, the post-tax Gini coefficient for the population consisting of just the eight insureds would show an increase in the inequality of the income distribution.13 This effect is more than offset, however, by the equality-promoting effect of transferring income from the more-affluent group to the less-affluent group.

The implicit head tax-and-transfer system described and critiqued by Havighurst and Richman does not necessarily reduce income inequality. With some effort, it is possible to construct an income distribution of insured and uninsured persons for which a particular head tax-and-transfer system would in[*pg 108] crease inequality.14 It is also possible that the portion of the head tax used to finance R&D increases inequality.15 However, Havighurst and Richman too quickly assume that the quasi-head tax they decry is regressive. When the head tax is considered together with the benefits it funds, it is quite plausible that the head tax is part of a system that reduces income inequality.

The above analysis assumes that the proper comparison to the current system of implicit head taxes and health care transfers to the uninsured is a world in which there are no head taxes and in which the uninsured receive no transfers, either going without health care or somehow paying for it themselves. Of course, starting from a different baseline would change the analysis. If the alternative to the current system were financing the same health care transfers to the uninsured by means of an income tax (with either a flat rate or graduated rates) on persons with health insurance, then obviously head-tax financing is regressive compared with that alternative. Havighurst and Richman adopt this perspective when they claim that high-income insureds benefit from the head tax "by having the needs of the uninsured . . . met by means other than equitable taxes."16 It makes more sense, however, to perform the distributional analysis of the current tax-and-transfer system by comparing it with the absence of any such system, rather than by comparing it with an alternative system somehow selected from among an infinite number of possible alternative systems. Therefore, Havighurst and Richman's conclusion that the current system is necessarily regressive is unwarranted.

At a different level, however, their view of the public-policy implications of their head-tax analysis is correct. Suppose the uninsured currently receive no health care, but society has just now decided that persons with health insurance are obligated somehow to finance health care for the uninsured. If the financing took the form of an explicit tax, it seems overwhelmingly likely that the public and Congress would think the burdens of the tax should be distributed among insureds not simply on a per capita basis, but with some sensitivity to their differing abilities to pay.17 Based on how Congress has decided to distribute other burdens, the explicit tax might take the form of a mildly progressive income tax, or perhaps something less progressive than that but still far from a head tax.

[*pg 109]

It is almost inconceivable that Congress would come any closer to a head tax than the flat rate (2.9%) wage tax currently used to finance Medicare.18 The Medicare tax can itself be described, in isolation, as regressive with respect to income. Although its rate is flat, and although it applies to all wages (rather than only to wages up to the Social Security wage tax ceiling19), the exclusion of investment income from the tax base means the rate of tax as a percentage of all income tends to fall as income rises. Nevertheless, income inequality clearly would be reduced (that is, the Gini coefficient would fall) if the benefits currently financed by the disguised head tax were financed instead by a tax modeled on the Medicare tax -- as might happen if the existence of the head tax became common knowledge.

Havighurst and Richman have performed an important service by unmasking the implicit head tax. The consequence of their exposé might be -- and should be -- the replacement of the implicit head tax with an explicit tax falling more heavily on those with more ability to pay. The accurate claim, however, is not that the current tax-and-transfer system is clearly regressive in its overall effects, but that even if it is progressive it is not progressive enough.


   

III. THE ROLE OF THE INCOME-TAX TREATMENT OF EMPLOYER-PROVIDED HEALTH INSURANCE

A. The Treatment of Health Insurance Under an Ideal Income Tax

Section 106 of the Internal Revenue Code excludes from an employee's gross income the value of "employer-provided coverage under an accident or health plan." Havighurst and Richman are critical of this provision for "aid[ing] the moral-hazard enemy."20 Although it is true that some important aspects of the current income-tax treatment of health insurance are difficult or impossible to defend, excluding the value of basic health insurance from the tax base constitutes sound tax policy and sound health care policy. Under an ideal (that is, normative) income tax, the treatment of health insurance would differ from current law in several significant ways, but it would retain the core of the exclusion for employer-provided health insurance. Furthermore, a revised version of the income-tax exclusion would not be vulnerable to the criticisms Havighurst and Richman level at current law.

Most tax-policy analysts agree that differences in individuals' tax liabilities (and thus in their shares of the costs of government) should reflect those indi[*pg 110] viduals' differing abilities to pay.21 That is, of course, the beginning rather than the end of tax-policy arguments, because it is often debatable how differences in individuals' situations affect their abilities to pay tax. On one point, however, there is widespread agreement: no one has any ability to pay tax on the amount of income necessary to support life at a level of basic decency. Ability to pay is generated only by "clear income" -- the amount of income a person has in excess of subsistence needs.22 If the poverty threshold is, say, $15,000, then no one -- not even Bill Gates or Warren Buffett -- should be required to pay tax on his first $15,000 of income. Historically, Congress has agreed with this view. The combination of the standard deduction and the personal and dependency exemptions is designed to approximate the official poverty threshold,23 thus excluding the cost of subsistence from the tax bases of taxpayers at all income levels. Although in recent years Congress has denied the benefit of personal and dependency exemptions to taxpayers near the top of the income distribution, this appears to have been a matter of political expediency -- a technique for disguising a marginal tax-rate increase for affluent taxpayers -- rather than of changed principles.24

Clear-income analysis provides a non-subsidy justification for excluding the individualized cost of basic medical care from each person's tax base. The costs of basic medical care -- both basic health-insurance coverage and necessary care not covered by insurance -- obviously are costs of subsistence. Thus, just as the costs of basic food, clothing, shelter, and transportation should be excluded from the tax base, so should the cost of basic medical care. In the same way [*pg 111] that personal and dependency exemptions make the tax system sensitive to differences in the cost of subsistence attributable to differences in family size, a tax allowance for the cost of basic health insurance should be sensitive to age- and sex-based differences in the cost of such insurance.25 Such differences in the cost of basic health insurance coverage are significant: the cost of insurance for persons in some age-and-sex categories can be several times the cost for persons in other categories.26 If the cost of basic health insurance for a twenty-five-year-old man is a small fraction of the cost of basic coverage for a woman of childbearing age, or for a sixty-five-year-old of either sex, then a universal allowance of a fixed dollar amount, akin to the standard deduction, must be too generous to the young man or inadequate for the others.27 To the extent that the income-tax exclusion of employer-provided health insurance removes the age- and sex-[*pg 112] adjusted cost of basic medical care from the tax base, it is -- like the standard deduction and the personal and dependency exemptions -- part of the definition of the ideal income tax base.28 Thus, if a "subsidy" is a provision included in the Internal Revenue Code to accomplish some goal other than adjusting tax liability to reflect ability to pay (such as encouraging particular types of consumption behavior), then because the health insurance exclusion is justified by ability-to-pay concerns internal to the tax system, it should not be understood as a "subsidy." 29

B. Health Insurance and the Actual Income Tax

Although the "clear income" approach provides a non-subsidy justification for the exclusion of basic employer-provided health insurance, in some respects current tax treatment of health insurance and other medical costs differs significantly from the dictates of clear-income analysis. The implications of clear-income analysis are (1) that all costs of basic health care should be excluded from the tax base and (2) that no costs of more-than-basic care should be excluded. Current tax treatment is insufficiently generous on the first count and overly generous on the second.

On the first count, current law is insufficiently generous because most costs of basic health care not covered by employer-provided insurance are not excluded from the tax base. People who purchase health insurance outside of the employment context may, in theory, deduct their premiums as medical expenses under section 213, but the deduction is allowed only to the extent total medical expenses exceed 7.5% of adjusted gross income (AGI), and even that excess is deductible only if the taxpayer itemizes rather than claiming the standard deduction.30 In addition, the 7.5%-of-AGI floor applies to basic medical expenses not covered by insurance (co-payments, deductibles, expenses for treatments [*pg 113] not covered by a taxpayer's insurance, and all expenses of uninsured taxpayers). Under an ideal income tax implementing the clear-income concept, however, all such expenses should be deductible in full.31 Although clear-income analysis suggests that the tax favoritism for employer-provided health insurance (that is, favoritism relative to other health insurance and health costs not covered by insurance) is inappropriate, it does not follow that the exclusion is a subsidy. Rather, the failure to provide equivalent treatment for other health care costs is a sort of penalty, or anti-subsidy.

On the second count, however, current law is overly generous because it extends the exclusion for employer-provided health insurance beyond the value of basic insurance.32 Under current law, the entire value of employer-provided health insurance is excluded from income, even if the insurance covers, for example, the excess cost of a single-bed hospital room over a two-bed hospital room, longer-than-medically-necessary hospital stays, the cost of experimental treatments, or the cost of having a "concierge" primary-care physician.33 Although drawing the line between basic and more-than-basic health care would not be easy, the failure to make any attempt to do so under current law means the existing exclusion for employer-provided health insurance cannot be fully justified by clear-income analysis.34

C. Moving the Actual Toward the Ideal

The ideal income tax treatment of health insurance and other health care costs can be briefly stated: The costs of basic health care, and only the costs of basic health care, should be excluded from the tax base. A person who receives employer-sponsored health insurance should be able to exclude the value of basic coverage (however high or low that value happens to be in light of the person's age, sex, and any other relevant attributes35), and should be able to deduct any costs of basic health care not covered by insurance (including deductibles [*pg 114] and co-payments). To the extent the employer-sponsored insurance goes beyond basic coverage, that value should be included in the employee's tax base. A person who purchases health insurance outside of the employment context should be allowed to deduct the cost of basic coverage and the cost of any basic health care not covered by insurance. Although a move to this regime certainly would constitute a significant change from current law -- enlarging the tax base in one respect while shrinking it in another -- it would retain the exclusion for basic employer-provided health insurance at the core of current law.

Would this revised tax treatment avoid the powerful criticisms aimed by Havighurst and Richman at current law? One complaint -- which, as Havighurst and Richman note, has been made by many commentators -- concerns the "greater apparent value [of this tax subsidy] to higher-bracket taxpayers and those with the costliest coverage -- frequently one and the same."36 There are two parts to this complaint. First, when a high-bracket taxpayer and a lower-bracket taxpayer receive identical health insurance and exclude the same amount from tax, the tax reduction -- and hence the subsidy -- is greater for the high-bracket taxpayer. Second, affluent taxpayers frequently receive better and costlier employer-provided coverage, and thus are able to exclude more value from the tax base. Havighurst and Richman do not place much emphasis on these criticisms,37 but the criticisms merit attention here because of their popularity with other commentators.

The first part of the complaint is the "upside-down subsidy" critique of health care tax policy, a critique closely associated with Stanley Surrey.38 According to this critique, it is unfair that a $1000 exclusion or deduction reduces the tax bill of a wealthy taxpayer in the thirty-five-percent bracket by $350, while a $1000 exclusion or deduction reduces the tax bill of a middle-class taxpayer in the fifteen-percent bracket by only $150. This is a powerful objection when the deduction or exclusion in question is a subsidy, but it has no application to deductions and exclusions that are an integral part of the definition of the normative "clear income" tax base. Like the standard deduction and personal and dependency exemptions, the exclusion of the value of basic health-insurance coverage is not a subsidy at all, and so cannot be an upside-down subsidy.39 If one accepts the argument that basic health care is not part of the nor[*pg 115] mative tax base because it generates no ability to pay tax, then it makes no more sense to say that the exclusion of such health care from the base of a tax with progressive marginal rates disproportionately benefits the rich than it would to say that the failure to tax people on nonexistent income disproportionately benefits the rich. In a footnote, Havighurst and Richman seem to agree with this view: "[C]haracterizing the exclusion . . . as a 'subsidy' might not be appropriate at all, since a taxing authority might simply find it fairer to tax individuals' income only after certain basic necessities were provided for. But the subsidy here is decidedly not limited to . . . only basic coverage."40 Their point seems to be that any exclusion for health-insurance coverage in excess of basic coverage is a subsidy and so is vulnerable to the upside-down critique of tax subsidies embodied in exclusions and deductions in an income tax with progressive marginal rates. This is correct, but the point would be moot under the suggested revision of the health-insurance exclusion. As for the second part of the first complaint -- that affluent taxpayers are able to exclude from income the entire value of the luxury-style health insurance they receive from their employers, while others are able to exclude only the lower value of their bare-bones insurance -- this is also true.41 Again, however, the point would be moot under the proposed limitation of the exclusion to basic health-insurance coverage.

Havighurst and Richman emphasize a different criticism of current law: that it "aids the moral-hazard enemy"42 by discouraging cost sharing (that is, deductibles and co-payments) which could combat moral hazard. The problem is that employer-provided health insurance is excluded from the tax base no matter how large its value, whereas cost-sharing payments generally must be made with after-tax dollars. As noted earlier, cost-sharing payments are potentially eligible for deduction as medical expenses under I.R.C. § 213, but medical expenses are deductible only to the extent total expenses exceed 7.5% of a taxpayer's AGI, and even that excess is deductible only if the taxpayer itemizes rather than claim the standard deduction. In practice, the vast majority of cost-sharing payments do not qualify for deduction under I.R.C. § 213. So, for example, if an employer offers its employees health insurance worth $5000 per year, with no cost sharing, each employee will exclude the entire $5000 from gross income; but if the employer offered employees health insurance worth $4000 per year, with expected annual cost-sharing payments of $1000, only $4000 of an employee's health care would be excluded from the tax base. The [*pg 116] results are a tax-induced preference for health insurance with little cost sharing, and increased moral hazard with all its attendant woes.

As Havighurst and Richman note, the tax bias against cost sharing was substantially reduced by the introduction in 2003 of favorable tax treatment for "health savings accounts" (HSAs).43 Under the new rules, a taxpayer with a "high deductible health plan" (HDHP) may make tax-deductible contributions to an HSA, and the contributions may be used to pay the taxpayer's cost-sharing expenses.44 Limiting these rules to taxpayers with HDHPs is significant because an HDHP must feature an annual deductible of at least $1000 in the case of self-only coverage.45 With the introduction of HSAs and HDHPs, equally favorable tax treatment is available for employer-provided health insurance with no cost sharing and for employer-provided insurance with high cost sharing,46 but insurance with moderate cost sharing remains disfavored. Of course, even with the tax bias against HDHPs eliminated, non-tax resistance to high levels of cost sharing may persist. In fact, employers and employees have not responded to the 2003 legislation by rushing to adopt HDHPs and HSAs, and no rush is expected in the next few years.47

Although not mentioned by Havighurst and Richman, there is another way in which taxpayers may pay their cost-sharing expenses with before-tax dollars. At the beginning of a year, an employee may agree to a salary reduction of a specified number of dollars in exchange for the employer's agreement to contribute an equal amount to the employee's "health flexible spending arrangement" (health FSA). Cost-sharing expenses incurred by the employee during the year can then be paid out of the health FSA, with before-tax dollars.48 The [*pg 117] health FSA is subject to a use-it-or-lose-it rule, however. If the taxpayer is lucky (or unlucky) enough to incur lower cost-sharing expenses than the amount of the salary reduction, the unused amount of the FSA is lost to the taxpayer forever. Although hard data are not readily available, it is generally assumed that utilization of health FSAs is not particularly high, in part because taxpayers are deterred by the use-it-or-lose-it rule,49 in part because of the difficulty of jumping through all the required administrative hoops, and in part because many employers (concerned about their own administrative burden) do not offer the option.50

The bottom line is, although HSAs and FSAs significantly decrease the traditional tax bias against cost sharing, they certainly do not eliminate it -- in particular with respect to insurance with moderate levels of cost sharing, and employees who are put off by the risk and burdens associated with FSAs. The simple and tidy solution, of course, would be to exclude from the tax base cost-sharing payments incurred in connection with basic health-insurance coverage. This would mean the exclusion of employer-paid cost sharing and, more significantly, the deductibility -- by itemizers and nonitemizers alike, and without the application of any percentage-of-AGI floor -- of all cost-sharing expenses. That is what is called for by clear-income analysis, and it would fully eliminate tax discrimination among various levels of cost sharing.

Although eliminating the tax bias against cost sharing would lessen moral hazard, Havighurst and Richman note that significant cost sharing may have unattractive distributional effects. When rank-and-file employees and higher-income employees are included in the same pool of employer-provided insurance and provided with nominally the same insurance coverage, conditioning eligibility for insurance benefits on a willingness to incur cost-sharing expenses [*pg 118] is likely to disproportionately discourage rank-and-file employees from seeking medical care, with the result that they receive fewer dollars of benefits than do higher-income employees.51 This suggests a damned-if-you-do-damned-if-you-don't cost-sharing dilemma for employer-provided health insurance. With low cost sharing there is severe moral hazard, but with higher cost sharing there is inequity in the distribution of insurance benefits. Havighurst and Richman offer a way out, however: combine significant cost sharing with the separation of employees into different insurance pools by income groups.52

Another route to the same goal would be to combine significant cost sharing with income-homogenous health-insurance pools outside of the employment context. That approach is not currently practical because of the tax favoritism for employer-provided health insurance over health insurance from other sources, but that favoritism would be removed under the normative approach to the taxation of health insurance proposed here (that is, creating a deduction, available to both itemizers and nonitemizers, and not subject to a percentage-of-AGI floor, for the cost of any purchased basic health insurance). Adoption of this proposal would be sufficient to rescue the tax system from Havighurst and Richman's indictment. The tax system would no longer create a bias in favor of employer-provided insurance, and thus would provide no particular encouragement to the creation of income-heterogeneous, employment-related insurance groups.

Of course, eliminating the income tax as part of the problem would not necessarily make it part of the solution. Merely removing the income-tax bias in favor of employment-related groups might not cause a flowering of income-homogenous insurance groups, especially considering that inertia would favor the existing income-heterogeneous, employment-related groups. Some additional governmental push might be needed if income-homogenous groups are to flourish. One possibility would be to deny the exclusion of basic health insurance (employer-provided or otherwise) from the tax base in the case of income-heterogeneous insurance groups. Various non-tax regulatory solutions are also imaginable.

The remaining count in Havighurst and Richman's indictment of the health insurance exclusion is that it encourages rank-and-file workers to devote too much of their overall consumption opportunities to health care.53 Because of the tax laws, health insurance is received as a fringe benefit of employment, implicitly paid for by a reduction in cash wages. Because the wage reduction is almost invisible to the typical worker, the worker does not realize how much he is really paying for health insurance, and so he purchases (so to speak) more insurance than he would choose if he received all his compensation in cash and had to buy his health insurance with cash. As Havighurst and Richman explain, [*pg 119] "Precisely because their costs are hidden from them, employees are more likely to demand and expect expensive health care even when their true interest would be served by economizing."54 The culprit here is not the exclusion of employer-provided health insurance from the tax base; rather, it is the inconsistency between the tax treatment of employer-provided insurance and insurance obtained from other sources. By providing equivalent treatment for all basic health insurance regardless of source, the proposed revision of the tax laws would eliminate the problem identified by Havighurst and Richman -- and it would do so without changing the current tax treatment of employer-provided basic health insurance.

To sum up, Havighurst and Richman offer a powerful critique of the current income-tax treatment of health insurance as the cause of many of the worst problems with American health care today. It does not follow from their critique, however, that the income-tax exclusion for employer-provided health insurance should be repealed. An ideal income tax, designed to align tax liability with ability to pay, would exclude the cost of basic health care from the tax base. Implementing this ideal would broaden the tax base in one way (by eliminating the exclusion for more-than-basic employer-provided insurance) and narrow it in another (by allowing deductions for purchased insurance and for cost-sharing expenses), but it would retain the core of the current exclusion. This revised version of the exclusion also would eliminate all the features of current law responsible for the effects to which Havighurst and Richman object. Interestingly, the recent report of the President's Advisory Panel on Federal Tax Reform would reform the income-tax treatment of health insurance very much along the lines of the ideal income-tax model.55 In order to eliminate the favorable tax treatment of more-than-basic employer-provided health insurance, the Panel would limit the exclusion to $5000 of insurance value in the case of self-only coverage and to $11,500 in the case of family coverage.56 And to eliminate the tax bias in favor of employer-provided health insurance over [*pg 120] health insurance purchased in the individual market, the Panel also would allow a deduction (not subject to any percentage-of-AGI floor) of the cost of such insurance (up to $5000 or $11,500).57


   

IV. CONCLUSION

Havighurst and Richman have made an important contribution by uncovering hidden ways in which the current system of health care financing, including the income-tax treatment of employer-provided health insurance, has disturbing distributional effects. Their analysis deserves to play a leading role in the national health care financing debate. As stated at the outset, the two caveats offered here are in the nature of friendly amendments. Viewed together with the health care for the uninsured that it finances, the implicit head tax on those with employer-provided insurance may not be regressive, but it is certainly not progressive enough. And although the exclusion for employer-provided basic health insurance would not be objectionable under a new-and-improved version of the income tax, the treatment of health insurance under the current income tax is indeed in need of reform.


   

FOOTNOTES


Copyright © 2006 by Lawrence Zelenak.
* Pamela B. Gann Professor of Law, Duke University School of Law.
1. Clark C. Havighurst & Barak Richman, Distributive Injustice(s) in American Health Care, 69 LAW & CONTEMP. PROBS. 7, 7-82 (Autumn 2006).
2. Id. at 36.
3. Id. at 28.
4. Id. at 22-25. Other important instances of subsidized activities include providing services under Medicare and Medicaid (to the extent the costs of rendering services exceed allowable payments) and complying with the unfunded federal mandate imposed on any hospital that maintains an emergency room and accepts Medicare payments to treat emergency room patients without regard to ability to pay. 42 U.S.C. § 1395dd(a) (2000).
5. Havighurst & Richman, supra note 1, at 26-27.
6. Id. at 28-29.
7. Id. at 28.
8. On the higher income levels of those with health insurance, as a group, than those without health insurance, see U.S. CENSUS BUREAU, CURRENT POPULATION REPORTS, P60-229, INCOME, POVERTY, AND HEALTH INSURANCE COVERAGE IN THE UNITED STATES: 2004, at 25 tbl.7 (2005) (reporting that of persons living in households with less than $25,000 of income, 24.3% lacked health insurance coverage in 2004; that percentage fell to 20.0% for persons in households with income of $25,000 to $49,999, to 13.3% for persons in households with income of $50,000 to $74,999, and to 8.4% for persons in households with income of $75,000 or more).
9. Corrado Gini, Measurement of Inequality of Incomes, 31 ECON. J. 124 (1921). For a description of alternative measures of inequality (including the Theil index), with a discussion of the advantages and disadvantages of the various measures, see JULIE A. LITCHFIELD, WORLD BANK POVERTYNET, INEQUALITY: METHODS AND TOOLS (1999), www1.worldbank.org/prem/poverty/inequal/methods/ litchfie.pdf. The example in the text employs the Gini coefficient because it is both the most commonly used and the most intuitive of the inequality measures.
10. See Max Lorenz, Methods of Measuring the Concentration of Wealth, 9 PUB. OF THE AM. STAT. ASS'N 209, 217-18 (1905).
11. Although those without health insurance are poorer as a group, as in the real world there is some overlap in the income ranges of those with and without insurance. See CENSUS BUREAU, supra note 8, at 25.
12. All Gini coefficient calculations in this paper were done by the author, with the use of the Resa Corporation's Gini coefficient calculator. P. Wessa, Office for Research Development and Education, Free Statistics Software Version 1.1.18 (2006), http://www.wessa.net.
13. The Gini coefficient for an eight-person population with incomes of $200,000, $150,000, $100,000, $75,000, $75,000, $50,000, $50,000, and $25,000, is 0.323276. The Gini coefficient for an eight-person population with incomes of $198,000, $148,000, $98,000, $73,000, $73,000, $48,000, $48,000, and $23,000 is 0.330571.
14. For example, suppose a society has three insured persons with initial incomes of $200,000, $25,000, and $20,000, and one uninsured person with an initial income of $30,000. The Gini coefficient of the four-person society is 0.495455. If each insured person is taxed $2000 in order to finance a $6000 benefit for the uninsured person, the resulting Gini coefficient (for incomes of $198,000, $23,000, $18,000, and $36,000) is 0.502727. The example is a bit rigged, however, because $4000 of the $6000 in the tax-and-transfer system is taken from people who are poorer-even before the transfer-than the person to whom that $4000 is transferred.
15. This depends, of course, on the assumptions one makes about the distribution of the benefits of R&D[0].
16. Havighurst & Richman, supra note 1, at 29.
17. Havighurst and Richman are of this view: "[A]ny projects that public lawmakers would be willing to support would almost certainly be financed in less regressive ways [than a head tax]." Id.
18. The 2.9% rate is the sum of the 1.45% tax imposed on employees, I.R.C. § 3101(b)(6) (2005), and the 1.45% tax imposed on employers, I.R.C. § 3111(b)(6) (2005).
19. See I.R.C. § 3121(a)(1) (2005) (limiting the imposition of the employer and employee taxes to finance Social Security benefits to the "contribution and benefit base").
20. Havighurst & Richman, supra note 1, at 36.
21. See, e.g., JOEL SLEMROD & JON BAKIJA, TAXING OURSELVES 62-64 (3d ed. 2004). But see LIAM MURPHY & THOMAS NAGEL, THE MYTH OF OWNERSHIP (2002) (arguing that distributive justice should be concerned with the distribution of after-tax income, rather than with the distribution of tax burdens relative to pre-tax income).
22. See, e.g., HAROLD M. GROVES, FEDERAL TAX TREATMENT OF THE FAMILY 10 (1963); Gerard M. Brannon & Elliott R. Morss, The Tax Allowance for Dependents: Deductions Versus Credits, 26 NAT'L TAX J. 599 (1973); Theodore P. Seto & Sande L. Buhai, Tax and Disability: Ability to Pay and the Taxation of Difference, 154 U. PA. L. REV. 1053, 1083 (May 2006) ("[T]here appears to be widespread consensus that . . . some initial portion of every taxpayer's income should not be subject to income tax."); Lawrence Zelenak, Children and the Income Tax, 49 TAX L. REV. 359, 361-72 (1994) (discussing the history and implications of the clear-income concept).
23. See S. REP. NO. 99-313, at 31-33 (1986), as reprinted in 1986-3 C.B. 31-33 (indicating that one goal of the Tax Reform Act of 1986 was to prevent the imposition of the income tax on families at or below the poverty level). Since 1986, the dependency exemption amount has been nearly identical with the increase in the Department of Health and Human Services (HHS) poverty threshold caused by the addition of one family member. For 2006, for example, the $3300 exemption amount differs by only $100 from the $3400 amount in the HHS guidelines. Rev. Proc. 2005-70, 2005-2 C.B. 979, § 3.17; Annual Update of the HHS Poverty Guidelines, 71 Fed. Reg. 3848 (Jan. 24, 2006).
24. The phaseout of personal and dependency exemptions for high-income taxpayers is contained in I.R.C. § 151(d)(3) (2005). In 2006, the phaseout begins for married couples filing joint returns at an adjusted gross income (AGI) of $225,750, and is completed at an AGI of $348,250. Rev. Proc. 2005-70, 2005-2 C.B. 979, § 3.17. The denial of exemptions to high-income taxpayers is scheduled to be gradually phased out, with the phaseout beginning in 2006 and to be completed in 2010. I.R.C. § 151(d)(3)(E), (F) (2005). The complete repeal of the phaseout is effective for only one year, however. As with all the provisions of the 2001 tax legislation, the repeal of the phaseout is subject to sunset at the end of 2010. Economic Growth and Tax Reconciliation Act of 2001 § 901, Pub. L. No. 107-16, 115 Stat. 38, 150.
25. The exclusion of employer-provided health insurance does not extend to insurance provided to an unmarried domestic partner of an employee unless the domestic partner qualifies as a dependent of the employee. Treas. Reg. § 1.106-1 (2004).

Because many employers now provide such non-excludable coverage, the IRS has been required to explain, in private letter rulings, how taxable employer-provided insurance is to be valued. In its first ruling in this area, the IRS stated that the taxable value of a non-excluded fringe benefit is "the amount that an individual would have to pay for the particular fringe benefit in an arm's length transaction" and that, in the health-insurance context, it is "the amount that an individual would have to pay for the particular coverage in an arm's-length transaction (i.e., at individual policy rates)." I.R.S. Priv. Ltr. Rul. 90-34-048 (May 29, 1990) (citing and applying Treas. Reg. § 1.61-21(b)(2)). The IRS soon changed its position, however. In Private Letter Ruling 91-09-060, the IRS stated that in the case of group health insurance "the amount includible . . . is the fair market value of the group medical coverage, notwithstanding that the fair market value of the group coverage may be substantially less than the fair market value of individual coverage or the subjective value of the coverage to the employee." I.R.S. Priv. Ltr. Rul. 91-09-060 (Dec. 6, 1990). The IRS has adhered to this position ever since. See, e.g., I.R.S. Priv. Ltr. Rul. 92-31-062 (May 7, 1992); I.R.S. Priv. Ltr. Rul. 97-17-018 (Jan. 22, 1997); I.R.S. Priv. Ltr. Rul. 2003-39-001 (June 13, 2003).

Although the rulings are not explicit on the point, it is reasonably clear that the IRS contemplates that coverage within a particular group has the same value for each covered individual, regardless of an individual's age, sex, or other attributes. See William V. Vetter, Restrictions on Equal Treatment of Unmarried Domestic Partners, 5 B.U. PUB. INT. L.J. 1, 7 (1995) (reading the rulings in this way). If this is the correct way to value employer-provided health insurance, then I.R.C. § 106 does not actually feature sensitivity to age- and sex-based differences in the cost or value of health insurance excluded from gross income, because there are no such differences in the group-insurance context. The later private letter rulings, however, are clearly inconsistent with general tax principles of valuation and are best understood as an IRS retreat in the face of political and administrative objections to the application of general valuation principles in this context. It is reasonable for the valuation of taxable employer-provided insurance to reflect the lower cost of group coverage than individual coverage, but it is not reasonable to pretend that group coverage has the same value for a twenty-five-year-old male as for a sixty-five-year-old (of either sex).
26. Insurers use indices to express how the cost of health insurance coverage varies according to age and sex. An index number of one for a particular age and sex indicates that the cost of coverage for a person of that age and sex is the same as the average cost of coverage for the entire insured population. For example, the index for a thirty-year-old man is 0.574, so if the average cost of coverage for the entire population is $2000, the cost of his coverage would be $1148. The index for a fifty-year-old woman is 1.762, so the cost of her coverage would be $3524-more than triple the cost of coverage for the younger man. Lawrence Zelenak, A Health Insurance Tax Credit for Uninsured Workers, 38 INQUIRY 106, 108 (2001).
27. Similarly, a fixed allowance cannot distinguish between the person who remains healthy all year and so incurs no cost-sharing expenses (that is, co-payments and deductibles) and the person who incurs several thousand dollars of such expenses.
28. Of course, if the cost of basic health insurance were built into the standard deduction and the exemptions, also providing an exclusion for the actual value of health insurance received from one's employer would amount to two tax-free allowances for the same cost. It does not appear, however, that the cost of basic health insurance is built into the standard deduction and the exemptions because the standard deduction and the exemptions do not feature the age- and sex-based adjustments which would be required to reflect the cost of basic health insurance. But see I.R.C. § 63(f)(1) (2005) (allowing a larger standard deduction to taxpayers sixty-five or older).
29. To be sure, the Staff of the Joint Committee on Taxation (Staff) lists the exclusion for employer-provided health insurance, I.R.C. § 106 (2005), as an item in the tax-expenditure budget, thus indicating a view of the exclusion as a subsidy. STAFF OF THE J. COMM. ON TAXATION, JCS-1-05, ESTIMATES OF FEDERAL TAX EXPENDITURES FOR FISCAL YEARS 2005-2009, at 37 tbl.1, (Comm. Print 2005). This contrasts with the Staff's "clear income" view of personal and dependency exemptions, as well as the standard deduction, as part of "the normal structure of the individual income tax." Id. at 3. Although the Staff's treatment of the entire exclusion as a tax expenditure is dubious, two points are worth noting. First, under current law the exclusion is not limited to the cost of basic insurance, and the Staff is correct that the exclusion of the cost of more-than-basic insurance should be classified as a tax expenditure. Second, the Staff has an announced bias in favor of tax expenditure classification in debatable cases; the Staff categorizes an item as a tax expenditure whenever there is "a reasonable basis for such classification." Id. at 2.
30. I.R.C. § 213(a) (2005) (imposing 7.5% of AGI floor). Under I.R.C. § 162(l) (2005), however, self-employed taxpayers may deduct their health-insurance costs in full, whether or not they itemize, and without the application of any percentage-of-AGI floor.
31. As described later in this article, under some circumstances health savings accounts (HSAs) and health flexible spending arrangements (health FSAs) permit the exclusion from the tax base of cost-sharing expenses, without regard to the limits imposed on deductions under I.R.C. § 213. See infra notes 43-50 and accompanying text. As also described later in this article, however, in many cases HSAs and FSAs are not adequate substitutes for an unrestricted deduction for basic health care costs not covered by insurance.
32. Havighurst and Richman make this point. See Havighurst & Richman, supra note 1, at 36-37 n.86.
33. See Abigail Zuger, For a Retainer, Lavish Care by "Boutique" Doctors, N.Y. TIMES, Oct. 30, 2005, § 1, at 1 (describing the trend toward "concierge" personal physicians).
34. Although the exclusion of insurance coverage for more-than-basic health care is objectionable under clear-income analysis, the exclusion for insurance coverage featuring little or no cost sharing is not. As explained in the text, the entire cost of basic health care should be excluded from the tax base, whether that cost consists of (1) the higher cost of insurance with no cost sharing or (2) the sum of the lower cost of insurance with high cost sharing and the deductibles and co-payments paid by a person with such insurance.
35. But see I.R.S. Priv. Ltr. Rul. 91-09-060, supra note 25 (taking the unreasonable position that the value of taxable employer-provided health insurance does not depend on any attributes of the covered individual).
36. Havighurst & Richman, supra note 1, at 36.
37. Id. at 37 (describing them as "provid[ing], at best, only weak evidence of the pervasive injustice we observe in U.S. health care"). In downplaying the significance of these criticisms, they note that "the regressive consequences would be entirely offset if, as is arguably the case, the government replaces the revenue it loses through such tax expenditures by taxing other income at higher progressive rates." Id.
38. STANLEY S. SURREY, PATHWAYS TO TAX REFORM 134-38 (1973) (expressing this view of deductions and exclusions); see also Harvey E. Brazer, The Federal Income Tax and the Poor: Where Do We Go from Here?, 57 CAL. L. REV. 422, 441 (1969) (noting that the tax benefit from a dependency exemption of any given dollar amount is greatest for a taxpayer in the highest bracket).
39. For criticisms of other attempts to apply the upside-down subsidy critique to adjustments to the tax base designed to measure ability to pay, see Zelenak, supra note 22, at 363-65 (regarding dependency exemptions), and Seto & Buhai, supra note 22, at 1093 (explaining that a deduction or an exclusion, rather than a credit, is the proper tax mechanism if the goal is "to adjust for differences in ability to pay").
40. Havighurst & Richman, supra note 1, at 37 n.86.
41. Notice, incidentally, that this part of the complaint does not depend on the existence of progressive marginal rates; even under a flat tax it would be inappropriate to subsidize extravagant health-insurance benefits for the well-to-do.
42. Havighurst & Richman, supra note 1, at 36.
43. Id. at 38; see also I.R.C. § 223 (2005) (allowing a deduction for contributions by the taxpayer to the taxpayer's HSA).
44. If the taxpayer's employer makes contributions to the taxpayer's HSA to cover the taxpayer's cost-sharing expenses, including contributions made under a salary-reduction agreement, those contributions are excluded from the taxpayer's gross income. I.R.C. § 106(d) (2005). However, the sum of deductible and excludable annual contributions may not exceed an inflation-adjusted ceiling amount, which is $2700 (self-only coverage) or $5450 (family coverage) in 2006. I.R.C. §§ 223(b), 106(d) (2005); Rev. Proc. 2005-70, 2005-2 C.B. 979, § 3.22.
45. The minimum deductible increases to $2000 in the case of family coverage.
46. As Havighurst and Richman note, in one respect taxpayers with HDHPs and HSAs receive better tax treatment than taxpayers with employer-provided health insurance with little or no cost sharing-contributions may accumulate within an HSA, and no tax is imposed on the HSA's resulting investment income. They accurately characterize this as a "new tax shelter for the well-to-do." Havighurst & Richman, supra note 1, at 39 n.96.
47. In a recent survey by the Kaiser Family Foundation, only 2.3% of employers reported offering HSA-qualified HDHPs, and only 15% of employees offered that option selected it; the Foundation estimated that a total of 810,000 workers were enrolled in HSA-qualified HDHPs nationwide. KAISER FAMILY FOUNDATION, EMPLOYER HEALTH BENEFITS 2005 ANNUAL SURVEY § 8 (2005), available at http://www.kff.org/insurance/7315/upload/7315.pdf. In the same survey, 2% of employers not currently offering HSA-qualified HDHPs (including 7% of employers with 1,000 or more employees) indicated they were likely to do so next year. Id.; see also STAFF OF THE J. COMM. ON TAXATION, supra note 29, at 37 (estimating expected tax-expenditure amounts for the five-year period 2005-2009, and indicating only $2.7 billion for HSAs over the five years, compared with $493.7 billion for exclusions under I.R.C. § 106 and $44.1 billion for deductions under I.R.C. § 213).
48. See I.R.C. § 125 (2005); Prop. Treas. Reg. § 1.125-2, Q&A (7) (Mar. 7, 1989); I.R.S., DEP'T OF THE TREASURY, PUBL'N NO. 969, HEALTH SAVINGS ACCOUNTS AND OTHER TAX-FAVORED HEALTH PLANS 12-13 (2004), available at http://www.irs.gov/pub/irs-pdf/p969.pdf (describing health FSAs).

Arguably, Health Reimbursement Arrangements (HRAs) constitute a third means of paying cost-sharing medical expenses with before-tax dollars. If, pursuant to an HRA, an employer reimburses an employee (up to a maximum annual amount specified in the HRA plan) for cost-sharing medical expenses, the reimbursement is excluded from the employee's gross income. See generally I.R.S. Notice 2002-45, 2002-2 C.B. 93 (describing in detail the rules applicable to HRAs). HRA reimbursements must be financed solely by the employer; they may not be made pursuant to a salary-reduction agreement. (Any excess of the ceiling on annual reimbursements over actual reimbursements during the year may be carried forward for reimbursements in later years.) From the employee's perspective, having cost sharing covered by an HRA is the equivalent of having health insurance with no cost sharing at all. For this reason, HRAs are not discussed in the text as a means of obtaining favorable income tax treatment for cost-sharing expenses.
49. On the complexities of how a fully rational taxpayer should balance the tax savings from health FSAs against the use-it-or-lose-it risk, see Franklin Lowenthal and Phillip Storrer, Medical FSAs: An Expected Value Analysis, 100 TAX NOTES 521 (2003).
50. In a 2003 national survey by the Kaiser Family Foundation, 42% of employees surveyed indicated their employer or their spouse's employer offered a health FSA option, 49% indicated the option was not available, and 9% did not know. Of those with the option, 34% indicated they participated, 65% indicated they did not participate, and 1% did not know. KAISER FAMILY FOUNDATION, THE KAISER FAMILY FOUNDATION HEALTH INSURANCE SURVEY (2004), available at http://www.kff.org/insurance/upload/2003-health-insurance-survey-toplines.pdf.
51. Havighurst & Richman, supra note 1, at 43.
52. Id. at 45.
53. Id. at 55-57.
54. Id. at 38.
55. THE PRESIDENT'S ADVISORY PANEL ON FEDERAL TAX REFORM, SIMPLE, FAIR AND PRO-GROWTH: PROPOSALS TO FIX AMERICA'S TAX SYSTEM 78-82 (2005). The Panel's stated reasons for its proposal resonate with Havighurst's and Richman's critique of current law:
Because of the tax-preferred status of health insurance, people are more likely to buy health insurance that provides more coverage than they would in the absence of the incentive. Workers who purchase more health insurance may, in turn, use more health services, thereby increasing overall health spending. Estimates are imprecise, but removing subsidies for employer-provided health insurance could lower private spending on healthcare by 5 to 20 percent. . . .
. . . .
The Treasury Department estimates that the Panel's recommendation to cap the health-insurance amount at the average premium and provide an equal deduction to all taxpayers would reduce the number of uninsured Americans by 1 to 2 million people.
Id. at 80, 82.
56. Id. at 81. These are "the national average amount[s] projected to be spent on health insurance premiums in 2006."
57. Id. at 82. Given the purpose of equalizing the tax treatment of employer-provided and individually purchased health insurance, there appears to be a technical error in the proposal. In applying the $5000/$11,500 ceiling to the exclusion of employer-provided insurance, the proposal seems to contemplate valuing the coverage in the manner currently used by the IRS in valuing taxable coverage provided to the domestic partners of employees-by disregarding the effect of the age and sex of an insured person on the value of her coverage. See I.R.S. Prvt. Ltr. Rul. 91-09-060, supra note 25; see also supra text accompanying note 25. Thus, the proposal would exclude the entire value of employer-provided coverage for all members of an employer group as long as the average cost of coverage within the group was within the dollar ceilings, even if the value of coverage for some members of the group would exceed the ceilings if the valuation of the coverage reflected age- and sex-based differences. This is reasonable, as it approximates the results that would be achieved under the more theoretically proper approach of (1) valuing the coverage taking the age and sex of the insured into account and (2) adjusting the ceiling on the exclusion to take age and sex into account. The problem arises with respect to the treatment of individually purchased coverage, where a twenty-five-year-old-male might be able to deduct the entire cost of more-than-basic coverage under the $5000 ceiling, whereas a sixty-five-year-old of either sex subject to the same ceiling might not be able to deduct much of the cost of even bare-bones coverage. The solution, of course, would be to adjust the deduction ceilings to reflect differences in the cost of basic insurance for taxpayers in different age and sex categories.