No. 11, September 2007
In January, the George W. Bush administration presented a two-part proposal to reform the tax treatment of health insurance. The current open-ended tax deduction of employment-based health insurance premiums would be capped and then replaced by a standard tax deduction for anyone who buys private health insurance, either through his employer or through the individual market. The reaction of the Democrat-controlled Congress was predictable. Representative Pete Stark (D-Calif.), chairman of the House Ways and Means Subcommittee on Health, said he would not hold hearings on the proposal, noting that it would help higher-income Americans more than lower-income ones. Ironically, that is exactly what the current system does. The effect of the administration's proposal would depend entirely on its details, which might have been explored in Congressional hearings. The proposal also provided an opportunity to explore the fairness and efficiency of the current tax treatment of health insurance. It appears that this opportunity will be lost for the moment, but perhaps other avenues can develop to advance these ideas.
The administration should be applauded for proposing any review of the favorable tax treatment of health insurance. This is one of the fundamental reforms of the U.S. health-care system that health economists have been recommending for decades. The economist's critique has been primarily that the favorable tax treatment of employer-sponsored health insurance (ESI) represents a distortion in consumer prices, leading to an inefficiently high level of health insurance coverage--that is, consumption beyond the point at which the marginal benefits of additional insurance are equal to the marginal cost. Overly generous health insurance, in turn, leads to high demand for health care and premiums that cost more than many Americans can afford, particularly those who have low incomes or are self-employed--for whom tax-deductible premiums are worth little or nothing. Economists also have questioned the fairness of the tax deduction for health insurance premiums, noting that the deduction is worth more to individuals in higher tax brackets than to individuals in lower tax brackets, and to individuals with ESI versus those who purchase individual insurance on their own.
The main deficiency with the administration's proposal is that it does not go far enough. There are two ways that the proposal might have been extended. The first would have been simply to withdraw the favorable tax treatment of health insurance premiums paid by either the employer or the employee, treating employer contributions to premiums and medical care as taxable income and the employee-paid portion like the purchase of any other taxed commodity. That alternative probably would have been attacked by the left and probably by some on the right, too. Incremental changes are more prudent politically.
The second possible extension is the one proposed in this Health Policy Outlook: a redistribution of the favorable tax treatment of health insurance premiums. Mark V. Pauly, Patricia M. Danzon, Paul Feldstein, and John S. Hoff proposed it sixteen years ago. This Outlook offers a different argument in support of it. The argument for the second extension proceeds as follows:
- Tax deductions are pervasive in the federal budget.
- Assuming that the purpose of the tax deduction for health insurance premiums is to encourage the purchase of health insurance, then that encouragement is equally desirable for all individuals, regardless of their income or how they purchase insurance (for example, ESI versus individually purchased insurance).
- Tax deductions, coupled with a progressive income tax, distribute the government's encouragement to purchase health insurance disproportionately to individuals in higher tax brackets.
- The presumably desirable features of a progressive income tax and tax deduction to encourage the purchase of health insurance both can be maintained by replacing the current tax deduction with a direct price subsidy (that is, a refundable and advanceable tax credit that is made equally available to all "tax-eligible" individuals to be used exclusively for the purchase of health insurance).
- This change would represent an improvement in fairness, and by reducing the current tax deduction for individuals in higher tax brackets, the change would improve efficiency, too.
The purpose of this proposal is not to decrease the number of uninsured Americans per se, though it might well have that effect. There are other and more direct ways to reduce the number of uninsured, such as mandating the purchase of health insurance (which was part of Pauly's original proposal). The primary purpose is to offer the same level of government assistance for the purchase of health insurance to all tax-eligible citizens who are not enrolled in public health insurance programs.
The tax deduction of health insurance premiums results in lost revenue to the federal government. This lost revenue is referred to as a "tax expenditure." The Congressional Budget and Impoundment Control Act of 1974 defines tax expenditures as "revenue losses attributable to provisions of the Federal tax laws which allow a special exclusion, exemption, or deduction from gross income or which provide a special credit, a preferential rate of tax, or a deferral of tax liability." Tax expenditures, like direct government price subsidies, often attempt to influence a taxpayer's consumption or production decisions by reducing the after-tax price of one type of consumption or investment relative to others. The Government Accountability Office (GAO) has expressed concern over the growing size of tax expenditures. On an outlay-equivalent basis, tax expenditures exceeded discretionary spending for most of the previous ten years, and they have been roughly equivalent to discretionary spending since 1981. Despite tax expenditures' importance, the GAO was concerned that they are not a customary part of budget presentations and are not subjected to the same reviews as other spending programs.
The term "tax expenditure" is controversial. In 1999, the Joint Economic Committee (JEC), under then-vice chairman Representative Jim Saxton (R-N.J.), published a study that expressed concern over the assumptions that he viewed as implicit in the use of the term "tax expenditure." The study concluded that the term tax expenditure "rests on the assumption that tax rates should be applied to an expansive definition of taxpayer income so as to maximize tax revenue at any given tax rate." It recommended that discussion focus not on foregone tax revenue, but on the proper definition of the economic base for taxation.
The government's decision to forego tax revenue is certainly not the same--either philosophically or practically--as the government's expenditure of tax revenue, even if the intended behavioral influence is the same in both cases. The argument in this Outlook, however, rests neither on the assumption that these activities are equivalent nor on controversial definitions of the tax base. I assume merely that most people agree there are some activities--for example, pure public goods like national defense--that require the collection of tax revenue. Given that tax revenue must be collected to finance these activities, the relevant questions are whether some economic activities should receive favorable tax treatment, and, if so, how that favorable tax treatment should be structured. Current tax policy answers yes to the first question.
As to the second question, favorable tax treatment of specific economic activities is pervasive in the federal budget, and the associated foregone revenue is significant. But comparing the amount of foregone revenue associated with different tax expenditures is complicated. Any change in economic incentives is likely to produce changes in behavior. Some changes are intended, while others may be unintended, but unavoidable. The Office of Management and Budget (OMB) provides a good discussion of these issues in its Analytical Perspectives on the 2008 federal budget. For example, one cannot accurately estimate the amount of foregone revenue associated with the tax deduction for home mortgage interest simply by adding up the current deductions taken by all individuals, because any change in the tax treatment of home mortgage interest could alter the demand for home ownership, interest rates, home selling prices, and a host of other factors. In addition, favorable tax treatment of one activity could interact with favorable tax treatment of other activities, so treating the effects as additive could misrepresent the total effect of both tax expenditures.
Despite these complexities, simply adding up the current deductions is the way in which revenue losses from tax expenditures are actually calculated by the Treasury. Recognizing this limitation, the top five tax expenditures in fiscal year 2008 based on lost revenue are listed in table 1. Table 1 shows the importance of the tax exclusion of employer contributions to medical insurance premiums and medical care relative to other tax expenditures.
One would expect that the purpose of favorable tax treatment of any economic activity is to encourage that activity relative to those that receive standard tax treatment. That point, however, also turns out to be controversial. A second objection in the 1999 JEC critique of including tax expenditures in budget analyses is that some tax expenditures do not represent subsidies per se, but rather that they represent attempts to prevent income streams from being taxed twice or attempts to equalize the incentives for savings over consumption. This point might be valid for some types of consumption, but not health insurance premiums. The tax deduction of health insurance premiums makes health insurance a cheaper form of compensation than taxable wages (from the employer's perspective) and lowers the consumer's perceived price of health insurance relative to goods and services that receive standard tax treatment. It is difficult to imagine any purpose for the favorable tax treatment of health insurance other than encouraging the purchase of health insurance.
Progressive Taxes, Regressive Deductions
Given that some legitimate activities of government will require tax financing, what type of tax should be collected? Although federal, state, and local governments collect a variety of taxes, the most important at the federal level is the individual income tax. In fiscal year 2006, individual income taxes constituted 43 percent of federal tax revenue. Social insurance and retirement receipts represented 35 percent, and corporate income taxes 15 percent. The federal income tax is progressive with respect to the marginal tax rate as taxpayers move from one tax bracket to the next. All else equal, taxes represent a higher proportion of taxable income for individuals in higher tax brackets.
The tax deduction of health insurance premiums is regressive in the sense that the tax liability of the higher-income individual is reduced by a larger amount than the tax liability of the lower-income individual.
An inevitable result of combining a progressive income tax with deductions for certain economic activities is that the deductions represent larger tax savings to individuals in higher tax brackets than individuals in lower tax brackets, holding the level of the activity and other factors constant. Thus, excluding a $5,000 single-coverage health insurance policy from income taxes would save an individual in the 15 percent federal tax bracket $750 (relative to the same purchase in the absence of the tax deduction), whereas the deduction would result in tax savings of $1,750 for an individual in the 35 percent tax bracket. The tax deduction of health insurance premiums is thus regressive in the sense that the tax liability of the higher-income individual is reduced by a larger amount than the tax liability of the lower-income individual. If the commodity that enjoys the tax deduction is one for which expenditure increases with income, then the deduction of the commodity from taxation will become even more regressive using this definition. The total regressivity of the tax deduction for health insurance premiums is mitigated to a degree by the fact that premiums also are exempt from Medicare and Social Security taxes, which are flat percentages of wage income.
Suppose there were general agreement that it is desirable for the federal government to forego income to encourage the purchase of health insurance. Is there any reason to think the desirability of that policy increases with an individual's tax bracket, as suggested by current tax policy? It seems more likely that the assistance should be directed toward low-income individuals, or at least distributed equally to all with respect to income.
Bob Lyke and Julie M. Whittaker of the Congressional Research Service discuss several perspectives on the equity of the tax exclusion of health insurance premiums. They note that from one perspective, the distribution of the tax expenditure for health insurance premiums might not be unfair:
[T]o the extent that health insurance is a way of spreading an individual's catastrophic economic risk over multiple years, (then) basing tax savings on marginal tax rates might be justified. Under a progressive income tax system, economic losses ought to be deducted at applicable marginal rates, just as economic gains are taxed at those rates.
I find this argument a stretch, if not altogether untenable. Whereas actual economic losses should be deducted at the individual's marginal tax rate, it does not follow that expected losses also should be deducted in a similar manner, or at all. Insurance is a product that one purchases to shift the burden of risk from the individual to the insurer. That product can be purchased for any period of time ranging from the length of one airline flight to an entire lifetime. If, at the end of the contract period (usually a one-year health insurance contract), the individual has suffered no loss at all, the product has still completely fulfilled its function. One expects some correspondence between average losses and premiums, but losses at the individual level are irrelevant to the performance of the product. Furthermore, it is not clear why, for example, arthroscopic knee surgery should be treated as an "economic loss" in a way different from the time invested in the regular exercise that helps prevent knee problems. Splitting hairs, one could note that the premium consists not only of expected losses, but also of the insurer's administrative expenses, and if those expenses represent economic losses, then surely the purchase of any good or service is an economic loss.
But perhaps this definition of tax expenditure regressivity is overly simplistic. Presumably, when the federal government foregoes income, it means that there will be less of whatever the government would have spent the tax revenue on otherwise, or that the government must increase revenue from other sources. Under the heroic assumption that the government is providing the efficient level of services, income tax expenditures thus can be viewed as a tradeoff among a reduction in income taxes (which is desirable from the consumer's perspective) and borrowing, an increase in other taxes, or a reduction in government activities (all of which are undesirable from the consumer's perspective).
The question then is one of distribution. How are the benefits of reduced taxation and the costs of reduced government activities distributed among the populace? In order to argue that current tax policy provides any sort of match between the benefits of reduced taxation and the cost of reductions in government activities, one would have to argue that the cost of reductions in government activities or increases in revenue from other sources fall more heavily on individuals in higher tax brackets who receive the majority of the benefits of income tax deductions. Whether that is the case depends on exactly which government activities are reduced or which sources of revenue are increased when the government foregoes income tax revenue.
Clark C. Havighurst and Barak D. Richman point out that foregone revenue from regressive tax expenditures might be replaced by increased revenue from more progressive sources, thereby reducing the regressivity of tax expenditures. Andrew Chamberlain and Gerald Prante, however, found that the benefits of government spending are distributed disproportionately to low-income individuals. If tax expenditures are financed through reductions in progressive government spending, then those tax expenditures are even more regressive than an analysis based purely on revenues would suggest.
A Reform That Encourages Health Insurance and Preserves Progressive Tax Treatment
Is there any way to maintain the progressivity of the income tax and favorable tax treatment of health insurance without having regressive tax expenditures? The answer is yes, and the way to do it is to replace the current tax deduction of health insurance premiums with a refundable and advanceable tax credit that consumers can apply exclusively to the purchase of health insurance. "Refundable" in this case means that individuals are eligible to receive the tax credit assistance from the government even if they pay no taxes. "Advanceable" means that the credit is available to insurers in a way that does not disrupt the supply of insurance to the individual.
The logic of the proposal begins with the current tax expenditure on health insurance. I assume that the current tax expenditure on health insurance represents the amount of revenue the government is willing to forego in order to encourage the purchase of private health insurance by--or for--individuals who are not enrolled in public insurance programs. The proposal is to take the current tax deduction for health insurance premiums and, rather than distributing it disproportionately to individuals in higher tax brackets, offer an equal level of assistance to all tax-eligible individuals. The tax credit would be equal to the current tax expenditure on health insurance divided by the total number of citizens who are tax-eligible.
This proposal embodies a specific concept of equality--namely, equality of opportunity for purchasers of private health insurance to receive government assistance.
This proposal embodies a specific concept of equality--namely, equality of opportunity for purchasers of private health insurance to receive government assistance. Because some people may be more likely to take advantage of government assistance than others, the proposal does not guarantee that the final distribution of actual government assistance will be equal across all individuals.
Offering the same level of government assistance to the entire tax-eligible population is not the only alternative. The tax expenditure could be distributed across the population according to the price elasticity of demand for health insurance in an attempt to maximize the induced purchase of health insurance per dollar of tax expenditure. In addition to the practical problem of calculating those elasticities at the individual level, however, that approach has the potential disadvantage of penalizing people who would be more inclined to purchase health insurace even in the absence of the tax deduction.
John Sheils and Randall Haught estimated that total federal tax expenditures for health benefits in 2004 amounted to $188.5 billion, including the foregone FICA payroll tax revenue not included in OMB's tax expenditure calculations. The Agency for Healthcare Research and Quality estimates that in 2004 there were approximately 224 million noninstitutionalized Americans under the age of sixty-five who were not covered by Medicare or Medicaid. Dividing the Sheils and Haught health-care tax expenditure estimate by 224 million people produces an estimated refundable tax credit of roughly $840 per person in 2004. Sheils and Haught estimated that the current tax deduction of premiums allocated over 70 percent of the 2004 tax expenditure to families with incomes over $50,000 per year (a little over 40 percent of all families in 2004). The $840-per-person estimate above is based on the assumption that all eligible individuals actually claim the refundable tax credit, and thus the estimate might be accurate only in the context of a legislative mandate that requires all individuals to purchase health insurance. If a substantial number of individuals failed to claim the credit, then the government could either reduce the total revenue it foregoes to encourage the purchase of health insurance, or it could hold that amount constant and increase the per-person allocation in a subsequent year.
Senator Tom Coburn (R-Okla.) and four other Republicans recently introduced the Every American Insured Health Act, which also proposes replacing the current tax deduction for health insurance premiums with a refundable, advanceable tax credit of $2,160 per individual and $5,400 per family. The sponsors state that their proposal is budget-neutral, but it involves a different mix of offsets and assumptions. (For example, it uses 2009 dollars, and it does not simply eliminate the entire current tax expenditure for health benefits and reallocate it on a fixed per-capita amount across the tax-eligible population.)
A refundable tax credit is not devoid of administrative costs, either to individuals or to the government. To claim the credit, individuals would have to submit proof of insurance that could be verified in a tax audit. Employer contributions to health insurance premiums would be converted to taxable wages, however, and employers would no longer have to keep track of contributions to premiums for tax purposes.
There are a number of possible objections to this proposal. Tax credits will not help low-income people very much. Timing mismatches between the period when the tax credit funds would be made available to such individuals and the points at which their premiums will be due are likely. The uncovered portion of the premium in any case still will be unaffordable for many lower-income people.
But the primary purpose of the proposal is not to decrease the number of uninsured Americans. Its main objective is to move toward greater equalization of access to government assistance for the purchase of health insurance. The proposal will be of greater help to low-income people than the current system, and it will be more equitable. The government is gaining experience with refundable tax credits for health insurance through the Trade Adjustment Assistance Reform Act of 2002, and this approach is likely to improve over time.
If employment-based insurance or other group insurance is more efficient than individual coverage, group insurance will continue to compete favorably with individual coverage.
Another critique of the proposal is that the tax credit would have to be adjusted for geographic differences in health-care costs. In response, the tax credit could be adjusted for differences in the price of inputs to the cost of care using a "cost of care" index such as the Medicare Geographic Practice Cost Indicator (GPCI). The GPCI is used to adjust Medicare physician fees for variation in the price of inputs such as rent and wages of office staff--not variation in the use of services.
A final objection is that employers or healthy employees will drop insurance. But there appears to be political support for equalizing the tax treatment of employment-based insurance and individual insurance even in the absence of this proposal. Regardless of tax issues, group insurance is probably more efficient than individual coverage due to lower marketing and underwriting costs. If employment-based insurance or other group insurance is more efficient than individual coverage, group insurance will continue to compete favorably with individual coverage. Employers could still "sponsor" health insurance for their employees, and employees could arrange for automatic payment of premiums from their paychecks as they do for other services.
Although the estimated $840 per person per year was, in 2004, far less than the level of assistance provided under current tax law to those who benefit the most from it, it was the amount of help that would result from even distribution of the existing level of government assistance across the entire eligible population. Whether that figure is too low or too high is a matter of political debate, but that debate needs to take place in the context of two tax policy questions: why should wealthy Americans receive any government assistance at all for their purchases of health insurance, and, given the government's decision to provide assistance in the form of foregone tax revenue, why should the bulk of that assistance be distributed to individuals in higher tax brackets?
Federal, state, and local governments all offer assistance to poor people. Beyond help to the poor, however, governments also use the tax code, as well as direct expenditures, to encourage or discourage various economic activities. The fundamental question raised in this Outlook is under what circumstances that encouragement or discouragement should be a function of income. Government's encouragement of the purchase of health insurance is a function of income, but in a perverse way: more encouragement is given to the wealthy than to the poor who are not eligible for public insurance. Why does this arrangement persist?
This topic deserves extended debate. The Bush administration has expressed willingness to discuss variations on its proposal. It should be commended for encouraging both greater equity in the tax code and debate over our current system.
Bryan E. Dowd (firstname.lastname@example.org) is a professor and director of graduate studies for health policy and management at the University of Minnesota School of Public Health. He thanks Thomas P. Miller, Robert B. Helms, and Roger Feldman for their helpful comments. The views expressed, and responsibility for any errors, are solely his.
1. The paper does not address individuals who are enrolled in public insurance programs such as Medicare and Medicaid or individuals in the military and their family members.
2. Julie Appleby, "Bush Unveils Health Plan Tied to Tax Deduction," USA Today, January 24, 2007. See also Pete Stark, "Unworthy of Consideration," USA Today, January 29, 2007.
3. Throughout this Outlook I use the terms "employer contributions" and "employer-paid" portion of the health insurance premium to distinguish the portion of the premium that is deducted from compensation by the employer. Employee contributions, or the employee-paid portion of the premium, may also be excluded from taxable income if an employer offers a "premium conversion" option under tax-advantaged cafeteria plan benefits provided to employees. The employee theoretically pays the entire insurance premium, as well as any other employer contributions to medical care itself, in the form of foregone wages.
4. For simplicity, throughout this Outlook I generally refer to the favorable tax treatment of employer- and employee-paid contributions to health insurance premiums and medical care as the "tax deduction of" or "favorable tax treatment of" health insurance premiums. However, it may also be referred to as the "tax exclusion" within the context of annual federal budget documents prepared by the Treasury Department and the Office of Management and Budget.
5. Mark V. Pauly, Patricia M. Danzon, Paul Feldstein, and John S. Hoff, Responsible National Health Insurance (Washington, DC: AEI Press, 1992), available through www.aei.org/book220/; and Mark V. Pauly, Patricia M. Danzon, Paul Feldstein, and John S. Hoff, "A Plan for 'Responsible National Health Insurance,'" Health Affairs 10, no. 1 (1991): 5-25.
6. By "tax-eligible individuals," I mean U.S. citizens who have to pay U.S. income taxes or who would have to pay U.S. income taxes if their taxable incomes were high enough, and who have not enrolled in a public insurance program.
7. Congressional Budget and Impoundment Control Act of 1974, U.S. Code 2 (1974), § 622.
8. Government Accountability Office (GAO), Government Performance and Accountability: Tax Expenditures Represent a Substantial Federal Commitment and Need to Be Reexamined, GAO-05-690 (Washington, DC: GAO, September 2005), available at www.gao.gov/new.items/d05690.pdf (accessed July 31, 2007).
9. Joint Economic Committee (JEC), Tax Expenditures: A Review and Analysis, 106th Cong., 1st sess., August 1999, available at www.house.gov/jec/fiscal/tax/expend.pdf (accessed August 8, 2007).
10. Ibid., 1.
11. Office of Management and Budget (OMB), Analytical Perspectives: Budget of the United States Government, Fiscal Year 2008 (Washington, DC: OMB, 2007), available at www.whitehouse.gov/omb/budget/fy2008/apers.html (accessed August 17, 2007).
12. The Treasury Department also calculates "outlay-equivalent" estimates based on the amount that the federal government would have to compensate individuals in order to hold their after-tax incomes constant.
13. JEC, Tax Expenditures: A Review and Analysis.
14. These receipts included revenue from the Social Security Old-Age, Survivors, and Disability Insurance (OASDI), Medicare Hospital Insurance (HI), railroad retirement, unemployment insurance, and federal employee retirement (employee share) programs.
15. OMB, Historical Tables: Budget of the United States Government, Fiscal Year 2008 (Washington, DC: OMB, 2007), table 2.2, available at www.whitehouse.gov/omb/budget/fy2008/pdf/hist.pdf (accessed August 13, 2007).
16. Annual income subject to the OASDI portion of the Social Security payroll tax is capped. The cap in 2007 is $97,500. Payroll taxes for Medicare HI apply to all annual wage income.
17. Bob Lyke and Julie M. Whittaker, Tax Benefits for Health Insurance and Expenses: Overview of Current Law and Legislation, Congressional Research Service Report for Congress no. RL33505, March 29, 2007.
18. Clark C. Havighurst and Barak D. Richman, "Distributive Injustice(s) in American Health Care," Journal of Law and Contemporary Problems 69, no. 4 (2006): 36-37.
19. Andrew Chamberlain and Gerald Prante, "Who Pays Taxes and Who Receives Government Spending? An Analysis of Federal, State and Local Tax and Spending Distributions, 1991-2004" (working paper 1, Tax Foundation, Washington, DC, March 2007), available at www.taxfoundation.org/files/wp1.pdf (accessed July 31, 2007). Nonetheless, John Holahan and Sheila Zedlewski estimated that when direct spending on health care, employer premium contributions, tax benefits, and tax spending are all taken into account, individuals in the lowest income decile in the United States spend about 20 percent of their income on health care, while individuals in the top decile spend about 8 percent. John Holahan and Sheila Zedlewski, "Who Pays for Health Care in the United States? Implications for Health System Reform," Inquiry 29, no. 2 (1992): 231-48.
20. Alternatively, one could define "health insurance units" along the lines of the Current Population Survey.
21. John Sheils and Randall Haught, "The Cost of Tax-Exempt Health Benefits in 2004," Health Affairs web exclusive (February 25, 2004), available at http://content.healthaffairs.org/cgi/reprint/hlthaff.w4.106v1.pdf (accessed August 7, 2007); and OMB, Analytical Perspectives: Budget of the United States Government, Fiscal Year 2008.
22. Agency for Healthcare Research and Quality, Medical Expenditure Panel Survey, "Table 5: Trends in Health Insurance Coverage, Poverty Status, Gender and Age for Community Based Population, 1996-2008," 2007, available at http://meps.ahrq.gov/mepsweb/data_stats/download_data/pufs/pmep04/
pmep04_table5.pdf (accessed August 8, 2007).
23. John Sheils and Randall Haught, "The Cost of Tax-Exempt Health Benefits in 2004": W4-106-W4-112.
24. Office of Senator Tom Coburn, "Dr. Coburn, Colleagues Introduce 'Every American Insured Health Act,'" news release, July 26, 2007.
25. Internal Revenue Service, "Health Coverage Tax Credit Overview," 2007, available at www.irs.gov/individuals/article/0,,id=109960,00.html (accessed August 1, 2007).