Chapter 1

The Tax Treatment of Health Insurance: Early History and Evidence, 1940-1970
Robert B. Helms, Ph.D.

Can there be any doubt that a tax policy that directly affects the financial well-being of 157 million Americans¹ and that has been in existence for over 50 years has made a substantial difference in the performance and efficiency of the U.S. health market? The answer is yes. But that doubt is very unevenly distributed across the American population. The complex way in which tax policy has affected the growth of health insurance and the behavior of consumers and providers is understood by a small number of health economists and other health policy specialists who have studied health markets and examined the empirical evidence. The rest of the population, including most politicians and policymakers, sees little connection between the arcane rules of the Internal Revenue Service and the policy problems we are experiencing in the health care marketplace.

Health care economists and other analysts have been writing about the effects of tax policy on health care markets for over 40 years.² Even though the effects of tax policy on the growth of private health insurance and the effects of health insurance on the behavior of consumers and providers are well understood by a few health policy specialists, it has not been easy to disentangle those effects from other forces affecting health markets in order to present unambiguous measurements. This has left room for academic skeptics, special interests, and some politicians to argue against tax changes designed to improve the performance of health markets. The purpose of this chapter is to review the early history and principal studies undertaken to measure the effects of tax policy on health insurance and economic behavior in health markets.³ To help understand those studies, the next section briefly reviews the history of health markets with emphasis on the changing nature of health insurance in the post-World War II period.

 

The Changing Postwar Health Market

To understand the effects of tax policy on health markets, it will be helpful to consider the rapid rate of change in almost all aspects of medicine and insurance after World War II. Historical accounts of the history of medicine and health care delivery all point to the nineteenth century and early twentieth century as a period of major change in the knowledge of medicine, in the role of hospitals and physicians, and in the financing of health care.⁴ All the major changes we associate with the period after World War II-the change in medical technology, the growth of employment-based insurance, and the changing nature of the hospital-had their antecedents in the period before World War I. What distinguishes the post-World War II period is the rapid rate of change of these factors. A look at some of the available data can help illustrate the magnitude of the changes.

The period following World War II is regarded generally as one of unprecedented growth and prosperity. Figure 1 shows the growth of the population after World War II. That bulge became known as the baby-boom generation. It was accompanied by a growth in national and personal wealth, as illustrated in Figure 2 by the growth of per capita gross national product and disposable personal income in 1982 constant dollars. It is not surprising that some of that growth in personal income went into the purchase of medical care and health insurance. Although we do not have comparable data for prewar and postwar per capita expenditures, Somers and Somers present data on per capita expenditures for medical care and health insurance for 1948 through 1958.⁵ Figure 3 illustrates the rapid growth in per capita total expenditures, payments made out of pocket by consumers, and those made through insurance. While total per capita expenditures went up 82 percent, consumers' out-of-pocket payments went up 49 percent and payments by the relatively small but growing insurance sector went up 442 percent.⁶

There was substantial progress in medical knowledge and the ability of health professionals to affect medical outcomes in the early half of the twentieth century, but the development of modern drugs and surgical and other medical procedures made possible a much more rapid increase in the application of scientific knowledge to medicine after World War II.⁷ The increased ability of medicine to provide valuable medical services and the public willingness to purchase those new services are reflected in the data on both expenditures and utilization. Table 1 is a partial reproduction of a Somers and Somers table showing a comparison of admission rates to general hospitals by diagnostic category for 1928-1943 versus 1957-1958. Considering that new technologies allowed some conditions to be treated in an outpatient setting, the increase in total admissions reflects the increase in hospital care for serious medical conditions requiring more intensive and expensive technologies and personnel. Looking at expenditures, Somers and Somers report that "In 1929 the proportion of disposable personal income spent for medical care was 3.5 per cent. In 1946 it was 3.8 per cent; by 1956, 4.9; and in 1958, 5.2. The increase between 1946 and 1958 was thus 37 per cent."⁸

This increase in the demand for medical care elicited a response from every sector of the health industry, including hospitals, physicians and other personnel, pharmaceutical and other medical product suppliers, and the health insurers. William White reports that "Between 1940 and 1965 the total number of general hospitals in the country increased by nearly 40 percent, while the number of beds increased by over 85 percent."⁹ During the same period, the number of physicians increased 74.2 percent (from 133 to 153 physicians per 100,000 population) while the number of nurses increased 115.8 percent (from 216 to 319 nurses per 100,000 population).¹⁰

The large and relatively rapid growth of the health care industry in the postwar period raised the cost of care for the average consumer and created the conditions that were conducive to the growth of the health insurance industry. In discussing the history of hospitals in the early years of this century, White makes the point that "new technologies sharply increased the cost of care, so that in the absence of health insurance, a major illness spelled financial crisis for all but the most wealthy members of society."¹¹

The financial risk faced by a few consumers in the early part of the century was becoming increasingly important to the vast majority of postwar consumers. For the average consumer, the probability of having a major medical expenditure was relatively small, but if a serious medical problem developed, the costs of hospital and physician care could impose a significant financial burden. This concern among several million Americans with rising incomes about the potential costs of medical care helped create the desire to spread the risk of sudden financial loss.¹²

Looked at in this way, it is not difficult to understand why conditions after World War II were ripe for a rapid growth in the health insurance industry. At the same time, however, other types of insurance, such as fire, life, and automobile insurance, also were facing expanding consumer demand. To anticipate the following discussion, we will point out that these other forms of insurance expanded primarily through sales directly to consumers, while health insurance expanded primarily through provision by employers. The difference in the development of health insurance has had major effects on the performance and efficiency of health markets compared to other insurance markets, a topic discussed in more detail below, but first we will look at some characteristics of the growth of private-sector health insurance.

As the absolute growth in health insurance increased, there was increased enrollment not only in the traditional hospital coverage that dominated the prewar period, but also in coverage for surgery, physician services, and major medical expenses.¹³ Figure 4 presents the enrollment in health insurance plans by type of coverage from 1940 through 1959.

While individual purchase of health insurance was increasing during the postwar period, the major growth in enrollment was occurring in group policies being purchased through employers. Figure 5 shows the growth of Blue Cross/Blue Shield and group insurance policies sold directly to employers relative to the growth of individual policies before and after World War II. Somers and Somers point out that only 2 percent of the labor force (1 to 2 million employees) and their dependents (another 1 to 2 million) had any protection for health services in 1930. In contrast, by 1958, 123 million Americans had hospital insurance and 75 percent of them were enrolled through their employer.¹⁴

The absolute growth in the size of the health care sector obviously was affected by the rapid growth of population and wealth after the war. At the same time, the development of penicillin during World War II and more modern medicines in the next two decades led to an increase in the effectiveness of medical care and in consumer demand for medical services. The higher cost of more advanced medicine stimulated the demand for medical insurance. As a result, the rapid growth of employer-provided insurance after World War II set the U.S. health care system on a different course from most other developed countries. While Canada and the European "welfare states" adopted various forms of public health care delivery and financing, the United States relied on public programs for the elderly and the poor while retaining private-sector financing for the majority of working Americans. Compared with most other systems, those developments enabled the American health care system to perform better in terms of personal choice and technological and scientific progress but more poorly in terms of cost and number of people covered.¹⁵

 

The Tax Treatment of Health Insurance

A poll of Americans about the causes of high medical costs probably would elicit many answers regarding "greedy physicians" and "waste, fraud, and abuse," but almost no mention of tax policy. Yet health policy experts, especially health economists, almost unanimously implicate the special treatment of employer-provided health insurance as a major cause of economic inefficiency in medical markets. Those experts also believe that the special treatment explains the historical development of these markets in the postwar period.

One reason tax policy is so misunderstood is that it works in an indirect way through health insurance. The pioneering work explaining the complex and indirect effects of tax policy on medical market behavior was done in the early 1970s by Martin Feldstein in conjunction with Elisabeth Allison and Bernard Friedman.¹⁶ The following summary starts with their studies and then adds information from some more recent studies. To facilitate understanding of both the economic effects and the empirical estimates, the summary will be presented in three sections: (1) the effects of tax policy on health insurance; (2) the effects of health insurance on medical use and prices; and (3) the effects of higher medical costs on health insurance.

 

The Effects of Tax Policy on Health Insurance¹⁷

Tax policy subsidizes the purchase of health insurance in two ways: first, by allowing the deduction of a limited amount of medical expenses and health insurance premiums from taxable income; and second, by excluding the value of employer-provided insurance from the employee's taxable income for the income tax and Social Security and Medicare payroll taxes.¹⁸ Like wages and other costs, the cost of health insurance to the employer is deductible as a business expense. The tax exclusion is the more important.¹⁹

Both aspects of federal tax policy had their origins during World War II. The medical deduction of health expenses started in 1942. The exclusion of employer-provided health insurance from taxable income also started during the war as business firms offered fringe benefits as a way to compete for scarce labor under wartime wage controls.²⁰ This practice was recognized by a special ruling of the Internal Revenue Service (IRS) in 1943.²¹ In 1953, the IRS ruled that employers' contributions to individual policies were to be included in taxable income. Congress overturned the ruling the next year.²²

Tax policy increases the demand for health insurance by providing a government-funded discount for health expenditures and employer-provided health insurance. The higher the marginal tax rates (MTR) faced by an individual, the greater the amount of the discount.²³ Table 2 presents an example developed by Glied to explain how tax policy makes additional health insurance benefits more valuable than additional wages.²⁴ Because the tax subsidy lowers the net cost of a given level of health insurance, an employee rationally would prefer additional dollars spent on health insurance rather than on wages. An employee would have to pay taxes on the additional wages (at his or her MTR) but not on the health benefits. Under the assumption that employees value health insurance at least as much as an equal amount of wages, the employee with a 20 percent MTR is better off when the employer provides health insurance because the employee has the health insurance and $400 in additional after-tax income.

How high are MTRs and the resulting discounts on health insurance purchased by the employer? According to Feldstein and Allison, the effective MTR for 1969 ranged from 13 percent for incomes under $1,000 to 36 percent for incomes over $25,000.²⁵ They estimate that the total tax subsidy in 1969 was $2 billion, including $1.63 billion in reduced revenue and $339 million from individual insurance payments. This is a 15 percent discount from the $15.7 billion total health insurance premiums in 1969.²⁶ Writing in 1981, Feldstein reported that updates of these estimates indicate that the tax subsidy for 1978 exceeded $10 billion on insurance premiums totaling $42 billion, which implies a 24 percent discount.²⁷

Tax policy has several other implications, some of which are discussed by Feldstein and Allison, and others that follow from the history given above. For example:

• The value to the individual taxpayer and the cost to the government of the tax subsidy are greater when a person's income (MTR) is higher.
• Rising incomes and MTRs have increased the amount of the subsidy over time.
• The tax exclusion creates incentives to expand the scope of medical benefits (e.g., more coverage for dental, eyeglasses, and weight reduction programs) and to reduce the degree of cost sharing (smaller co-payments and deductibles).
• Tax policy was not the only factor affecting the rapid growth of health insurance and medical costs in the postwar period, but the distortions it created in labor and medical markets magnified the effects of the other factors, such as income, population, and medical technology.²⁸

 

Feldstein and Allison conclude their study by saying that the subsidy "causes a substantial revenue loss, distributes these tax reductions very regressively, encourages an excessive purchase of insurance, distorts the demand for health services, and thus inflates the prices of these services."²⁹

 

The Effects of Health Insurance on Medical Use and Prices

One of Feldstein's major contributions was his explanation of the effects of health insurance on the performance of medical markets. His analysis of the tax treatment of health insurance was preceded by his two large studies of the effects on hospital and physician markets.³⁰ Because conditions in physician markets made it difficult to estimate the demand for services, his study of health insurance concentrated on the hospital market.³¹ Feldstein argues that the effects are complicated by the special nature of nonprofit hospitals and by insurance.

As noted, the discount for employer-provided health insurance could be expected to increase the demand for it. The expansion of health insurance took three forms: an expansion in the absolute amount of insurance (more employees covered); an increase in the types of coverage (more extensive hospital and nonhospital coverage); and a reduction in cost-sharing (deductibles and/or coinsurance rates). For typical insured employees, expanded insurance reduced the observed price of a hospital stay or a physician visit, inducing them to increase their demand for these services.

But, as Feldstein argues, that is not the end of the story.³² Nonprofit hospitals (which were more prevalent in the 1960s than today) responded to the increase in demand not only by raising their prices, but also by investing in more technology and labor to improve their services. That was how they competed for physicians and patients. Feldstein also noted that "this response does not depend on the usual mechanism by which increased demand raises price through higher profits and higher input costs. It is primarily by inducing a change in the hospital's product that insurance raises its price."³³

After measuring the welfare losses from excess health insurance, Feldstein uses his model to estimate the potential gains that could be achieved by higher coinsurance rates.³⁴ After allowing for the effects of changing hospital quality, and assuming a reasonable range of elasticity and risk-bearing values, Feldstein compared the welfare gains from reducing price distortion (raising the coinsurance rate) with the welfare losses of increased risk-bearing. Arguing that his estimates are likely to understate the net welfare losses, he anticipated "net gains in excess of $4 billion per year."³⁵ In 1969, this net welfare gain was 32 percent of the $12.6 billion in private hospital care expenditures.³⁶ This led Feldstein to conclude that an increase in the average coinsurance rate would increase welfare and that the net gain would probably be quite substantial. Moreover, a more general restructuring of the form of health insurance, reducing its role as a method of prepaying small and moderate hospital bills, and increasing its role in protecting against the major financial risks of very large health expenses, could produce even greater gains.³⁷

 

The Effects of Higher Medical Costs on Health Insurance

Incorporated into Feldstein's estimates of the consequences of tax policy are the feedback effects of an increase in medical costs on the demand for insurance. As explained above, the discounts made possible by the tax treatment of employer-provided health insurance increased the amount and type of coverage. That, in turn, increased both the demand for and the cost of hospital and physician services. The tax-induced rise in medical costs increased the risk of financial loss for the typical consumer. The desire to protect against this risk is the principal reason a rational consumer buys insurance. The result of rising medical costs was a further increase in the demand for insurance that, like the original tax-induced increase in demand, resulted in even higher medical costs and even more demand for insurance. Feldstein concluded that insurance is used more than it would have been and medical costs are higher than otherwise because the two are interdependent. The effect of any outside factor is magnified by that interdependence.³⁸

In historical terms, Feldstein pointed out that this two-way relationship between hospital behavior and the growth of insurance led to an "explosively unstable" period from 1958 through 1965, but that the system became more stable after 1966 even though hospital prices continued to rise at a rapid rate.³⁹

 

More Recent Studies and Evidence

The purpose of this chapter has been to look at the early history and evidence of the tax treatment of health insurance. A brief look at some data on tax policy and health insurance premiums, as well as a review of more recent studies, will help put the historical discussion into perspective.

The revenue lost by the federal and state governments because of the tax policy under discussion is computed annually. Table 3 presents the Congressional Budget Office (CBO) version of these so-called tax expenditures, compiled from annual budget documents as well as projections by the Joint Tax Committee.⁴⁰ The table covers selected years including 1969, the year of the Feldstein and Friedman estimates.

Comparing the composition of "tax expenditures" in 1969 and 1995, medical expense deductions have declined in importance (from 44 percent to 4.8 percent) while employer-paid premiums have increased (from 38.5 percent to 67.3 percent). The loss in state tax revenue has increased slightly (from 7.7 percent to 11.4 percent), reflecting the general increase in state taxation. As a result of higher federal income and payroll taxes, tax expenditures from the exclusion of employer-provided health insurance from federal taxation have grown, both in absolute terms and in relative importance, since 1969.

Another measure of the growth of private health insurance is total premiums paid. The available data covering the late 1960s, when Feldstein made his estimates, to the present are for total private insurance premiums paid by employers and individuals to commercial insurance companies and Blue Cross and Blue Shield. Because the data include premiums paid for individual policies, the amount is larger than the total premiums employers paid for group policies. Figure 6 displays the growth in private health insurance premiums from $10 billion in 1965 to $296 billion in 1993, a growth rate averaging 12.9 percent annually. If we assume that Feldstein's estimate of welfare loss as a proportion of total private insurance premiums remained constant at 30 percent (an admittedly crude assumption), welfare losses in 1993 would have been $89 billion and could be expected to have increased with the growth of insurance costs.

To my knowledge, we have no sophisticated update of Feldstein's estimates that accounts for changes in health and insurance markets and more recent estimates of the elasticity of demand and risk preferences of consumers. But the studies by Feldstein, Allison, and Friedman did elicit a number of follow-on studies and critiques of their methodology and results. This literature, much of it dealing with the medical industry and health insurance response to price changes, has been reviewed separately by Pauly, Morrisey, and Glied.⁴¹ The two most recent attempts to provide empirical estimates of the effects of these subsidies have been by the CBO in 1994 and, in preliminary work, by Gruber and Poterba.⁴²

The CBO study uses the 1987 National Medical Expenditure Survey (NMES) and its own tax model to estimate the distribution of tax benefits for 1994. For families with employment-based health insurance (61 percent of the population), CBO estimates that the tax subsidy averages 26 percent of health insurance premiums (which average $4,310). The subsidy increases with income, going from 11 percent for those with family incomes under $10,000 to 33 percent for those with incomes over $200,000.⁴³ When the subsidy is compared with after-tax income, the CBO finds that it averages 2.4 percent for families with employment-based health insurance but only 1.9 percent when taxpayers without such health insurance are included.⁴⁴ The CBO percentages can be compared roughly with Feldstein's and Allison's 1969 estimates of the tax reduction per family. When those estimates are expressed as a percentage of the midpoints of their family income ranges, the results range from 2.4 percent for families with incomes under $1,000 to 0.14 percent for families with incomes over $25,000.⁴⁵ The lower percentages reflect the lower MTRs and less extensive provision of health insurance by employers in 1969 than in 1994.

Gruber and Poterba use new data sources to show that previous studies have overstated the tax subsidy of employer-provided health insurance. Their data and methodology correct for two changes in the health insurance market: the growth in the proportion of employer-provided health insurance paid for by employees and changes in the use of the medical deduction for large medical expenses. Yet even with these new findings, their results seem to verify the large effects of the tax subsidy and the order of magnitude of the previous studies.⁴⁶

 

Conclusion

Milton Friedman has said that economists may not know much, but they know that if the government sets the price of any product below the market rate, a shortage will result. On the basis of this review of the tax treatment of health insurance, the noneconomist may be inclined to agree only with the first part of Friedman's statement. The natural doubt that most noneconomists have about the influence of tax policy on health care markets has not been helped by the highly technical and confusing literature on the subject. Working through the markets for labor, health insurance, hospitals, and physicians, the effects have been indirect and difficult to understand or measure.

Understanding the effects of tax policy and measuring these effects have been complicated further by a number of other strong forces impacting health care markets. Increases in population, income, and medical technology all have worked to increase the demand for both medical care and health insurance. The special tax treatment for health insurance purchased through employers has not been the only influence on those markets, but it has helped magnify the expansionary effects of the other forces. It has increased the demand for health insurance, reduced consumers' perceived cost of medical care, and induced the medical sector to supply a level and style of health care far beyond what consumers would value in the absence of this tax policy.

In what Clark Havighurst has called "flat-of-the-curve medicine," medical care markets operate beyond the point of maximum efficiency.⁴⁷ This obviously has left some people better off, but at the cost of leaving almost everyone else worse off. Economic efficiency means using valuable inputs for the goods and services most valued by consumers. "Net welfare losses" is what economists call their measurement of inefficiency after allowing for the offsetting effects of consumer gains and losses. It is a measure of what society loses from a known distortion-or, to turn it around, what society could gain if it eliminated the cause of the inefficiency.

Economic studies conducted over the past 25 years have found that the inefficiencies caused by the tax treatment of health insurance are real and large. Even though the tax exclusion directly affects those with employment-based insurance, the resulting increase in the cost of care touches all consumers and patients, especially those without health insurance who end up paying higher prices. Taxpayers also are affected because of the higher taxes they have to pay to make up for the lost revenue and the diversion of tax dollars from other uses. But, as Havighurst has argued, the effects are even more pervasive because of the politics of health care. The tax-induced expansion of health insurance lowers the perceived cost of medical care and insurance and, in the process, helps to create a "pervasive entitlement mentality" among consumers. Those who receive the subsidies see no direct personal costs and thus would be happy to have additional subsidies. There is little reward for any politician who looks to tax policy as a solution to the long list of complaints about our system.⁴⁸ In one especially telling passage, Havighurst says that a "tax subsidy is insidious precisely because, in addition to being an off-budget public expenditure, it can misallocate huge amounts of society's resources, yet be entirely painless at the level of individual producers and consumers."⁴⁹

The effects of the tax treatment of health insurance on health care markets are real and substantial, but they are also complicated, indirect, and difficult to separate from other influences. This creates the ironic situation in which we turn to direct control of prices and utilization to correct the problems of health care costs and access. It is indeed time for a fresh approach to health care reform. Empowering health care consumers through tax reform would be a good start.

---

Notes

¹ Employment Research Benefit Institute, Sources of Health Insurance and Characteristics of the Uninsured (Washington, D.C.: EBRI, February 1996), Table 1, p. 5.

² The principal studies and references to this literature will be presented subsequently.

³ For some previous reviews of this literature, see Ronald J. Vogel, "The Tax Treatment of Health Insurance as a Cause of Overinsurance," in Mark V. Pauly, ed., National Health Insurance: What Now, What Later, What Never? (Washington, D.C.: American Enterprise Institute, 1980), pp. 220-249; Mark V. Pauly, "Taxation, Health Insurance, and Market Failure in the Medical Economy," Journal of Economic Literature 24 (June 1985), pp. 629-675; Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance (Washington, D.C.: U.S. Government Printing Office, March 1994); Sherry Glied, Revising the Tax Treatment of Employer-Provided Health Insurance (Washington, D.C.: AEI Press, 1994).

⁴ For reviews of the history of American medicine and health care, see Paul Starr, The Social Transformation of American Medicine (New York, N.Y.: Basic Books, 1982); Herman M. Somers and Anne R. Somers, Doctors, Patients, and Health Insurance (Washington, D.C.: Brookings Institution, 1961); Randall R. Bovbjerg, Charles C. Griffin, and Caitlin E. Carroll, "U.S. Health Care Coverage and Costs: Historical Development and Choices for the 1990s," Journal of Law, Medicine & Ethics, Vol. 21, No. 2 (Summer 1993), pp. 141-162; William D. White, "The American Hospital Industry Since 1900: A Short History," in Richard M. Scheffler and Louis F. Rossiter, eds., Advances in Health Economics and Health Services Research, Vol. 3 (Greenwich, Conn.: JAI Press Inc., 1982); and Marilyn J. Field and Harold T. Shapiro, eds., Employment and Health Benefits (Washington, D.C.: National Academy Press, 1993), pp. 49-86.

⁵ Somers and Somers, Doctors, Patients, and Health Insurance, Table A-10, p. 546.

⁶ Ibid.

⁷ For one discussion of the change in medical technology, see ibid., pp. 17-24. Writing in 1961, the authors say (p. 24), "It is estimated that 90 per cent of the drugs prescribed in 1960 were introduced in the previous two decades; 40 per cent could not have been prescribed in 1954." For a discussion of the effects of technology on hospitals, see White, "The American Hospital Industry Since 1900," p. 162.

⁸ Somers and Somers, Doctors, Patients, and Health Insurance, p. 213.

⁹ White, "The American Hospital Industry Since 1900," p. 162. Hospital growth was also affected by the passage of the Hill-Burton Act, introduced in 1948. For an assessment of Hill-Burton, see ibid., pp. 162-163, and Judy Lave and Lester Lave, The Hospital Construction Act (Washington, D.C.: American Enterprise Institute, 1974).

¹⁰ Historical Statistics of the United States, Colonial Times to 1970, Part 1 (Washington, D.C.: U.S. Bureau of the Census, 1975), Series B275-290, pp. 75-76.

¹¹ White, "The American Hospital Industry Since 1900," p. 159.

¹² For readable explanations of the basic economics of insurance, see Mark A. Hall, Reforming Private Health Insurance (Washington, D.C.: AEI Press, 1994), esp. chapter 2, and Mark V. Pauly, "Overinsurance: The Conceptual Issues," in National Health Insurance, pp. 201-219.

¹³ Somers and Somers, Doctors, Patients, and Health Insurance, Table A-12, p. 547. "Regular medical expense" was the term used to refer to coverage of outpatient physician expenses and physician services not covered under hospital insurance. "Major medical" refers to supplemental and catastrophic coverage not covered by other policies.

¹⁴ Somers and Somers, Doctors, Patients, and Health Insurance, pp. 228 and 230. Historical Statistics of the United States, Series B 401-412, the source for Figure 5, gives the total of persons covered in 1958 by Blue Cross/Blue Shield and group policies by insurance companies as 103.1 million.

¹⁵ For a sample of the rather extensive literature on international health systems, and for additional references, see Health Care Financing Administration, Health Care Financing Review, 1989 Annual Supplement, Vol. 11, No. 1 (Fall 1989), and Robert B. Helms, ed., Health Care Policy and Politics: Lessons from Four Countries (Washington, D.C.: AEI Press, 1993).

¹⁶ Their articles are conveniently reprinted in Martin Feldstein, Hospital Costs and Health Insurance (Cambridge, Mass.: Harvard University Press, 1981), chapters 6-8. For additional explanations by economists of the role of tax policy, see Glied, Revising the Tax Treatment of Employer-Provided Health Insurance, pp. 2-4; Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance, March 1994, pp. 5-27; and Vogel, "The Tax Treatment of Health Insurance Premiums as a Cause of Overinsurance," pp. 228-233. For explanations by a legal scholar, see Clark C. Havighurst, Deregulating the Health Care Industry (Cambridge, Mass.: Ballinger Publishing Company, 1982), pp. 388-391, and Clark C. Havighurst, Health Care Choices (Washington, D.C.: AEI Press, 1995), pp. 100-103.

¹⁷ Because the above material has relied extensively on the comprehensive historical analysis and data presented by Somers and Somers, it is interesting to note that, to the best of my knowledge, they at no time explicitly acknowledge tax policy as a contributing factor in the growth of health insurance in the postwar period. They come close to acknowledging it in their explanation for the reason that 75 percent of those with hospital coverage received it through employee benefit plans when they say that, "For numerous reasons most Americans now appear to prefer the extension and improvement of 'fringe benefits' to direct wage increases." Somers and Somers, Doctors, Patients, and Health Insurance, p. 228; emphasis added. They place major emphasis on the growth of unions and collective bargaining; ibid., p. 226.

¹⁸ For a study of the role of the medical deduction, see Bridger M. Mitchell and Ronald J. Vogel, Health and Taxes: An Assessment of the Medical Deduction (Santa Monica, Cal.: RAND Corporation, R-1222-OEO, August 1973).

¹⁹ See Table 3.

²⁰ Sherry Glied points out that the exclusion of health benefits "existed implicitly since the inception of the federal income tax in 1913" but grew in importance due to the expansion of fringe benefits to avoid wage controls beginning in World War II. Glied, Revising the Tax Treatment of Employer-Provided Health Insurance, p. 5. See also Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance, p. 5.

²¹ Vogel, "The Tax Treatment of Health Insurance Premiums as a Cause of Overinsurance," p. 223. Vogel's cite is Special Ruling, October 26, 1943, 433CCH, Federal Tax Service, paragraph 6587. The IRS ruling apparently followed a ruling by the War Labor Board that health and pension benefits did not count as wages. See Field and Shapiro, Employment and Health Benefits, p. 70.

²² Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance, p. 5. The exclusions of the employer's contribution for health insurance from the federal income tax and payroll taxes are contained, respectively, in Sections 106 and 3121 of the Internal Revenue Code.

²³ The marginal tax rate refers to the ratio of the additional tax burden for an additional dollar of income.

²⁴ Glied, Revising the Tax Treatment of Employer-Provided Health Insurance, p. 4.

²⁵ Feldstein and Allison, in Feldstein, Hospital Costs and Health Insurance, Table 7.2. Their estimates do not include the effects of state taxes.

²⁶ Ibid., pp. 174 and 208-215. Feldstein and Allison argue that these estimates are likely underestimates of the actual subsidies due to limitations of their data and model.

²⁷ Ibid., p. 174. Some additional updates are discussed later.

²⁸ A number of analysts identify medical technology as the leading cause of the rapid rise in medical costs. Technology obviously added to the cost of medical care (and, it is hoped, improved outcomes), but the tax-induced rise in the demand for insurance increased the demand for medical technology and resulted in a larger effect on medical costs than would have existed without the subsidy. Health insurance induced hospitals, physicians, and consumers to adopt a more intensive use of technology, and insurance companies to pay medical claims without questioning whether technology was used efficiently. For discussions of the role of technology in explaining medical-cost increases, see Joseph P. Newhouse, "An Iconoclastic View of Health Cost Containment," Health Affairs, Supplement 1993, pp. 152-171, and Burton A. Weisbrod, "The Health Care Quadrilemma: An Essay on Technological Change, Insurance, Quality of Care, and Cost Containment," Journal of Economic Literature, Vol. 29, No. 2 (June 1991), pp. 523-552.

²⁹ Feldstein and Allison, in Feldstein, Hospital Costs and Health Insurance, p. 216.

³⁰ Martin S. Feldstein, The Rising Cost of Hospital Care; Martin Feldstein, Hospital Costs and Health Insurance, chapters 1-5; and "The Rising Price of Physicians' Services," Review of Economics and Statistics, Vol. 52, No. 2 (May 1970), pp. 121-133.

³¹ Feldstein, Hospital Costs and Health Insurance, p. 179.

³² See Martin Feldstein, "Hospital Cost Inflation: A Study of Nonprofit Price Dynamics," American Economic Review, Vol. 61, No. 5 (December 1971), pp. 853-872, reprinted in Hospital Costs and Health Insurance, chapter 3.

³³ Feldstein, Hospital Costs and Health Insurance, p. 187.

³⁴ In any economic change there usually are winners and losers. In its simplest meaning, welfare losses (sometimes referred to as dead-weight losses) refer to net losses from a change. If the gainers cannot compensate the losers, then it can be said that there is a net loss in welfare for society. For Feldstein's discussion of five aspects of the welfare effects of health insurance, see Hospital Costs and Health Insurance, pp. 97-98.

³⁵ Feldstein, Hospital Costs and Health Insurance, p. 203.

³⁶ In a review article published in 1974, Feldstein summarizes his estimates of welfare losses by saying that the "estimated welfare cost for 1969 was about $2.4 billion, or 30 percent of total insurance benefits and 20 percent of total private hospital expenditure." Feldstein, Hospital Costs and Health Insurance, p. 99.

³⁷ Ibid., p. 203.

³⁸ Ibid., p. 173.

³⁹ Ibid., p. 18.

⁴⁰ Congressional Budget Office, Projections of National Health Expenditures, October 1992, p. 56.

⁴¹ Pauly, "Taxation, Health Insurance, and Market Failure in the Medical Economy"; Michael A. Morrisey, Price Sensitivity in Health Care: Implications for Health Care Policy (Washington, D.C.: NFIB Foundation, 1992); Glied, Revising the Tax Treatment of Employer-Provided Health Insurance. For a criticism of Feldstein's assumptions about hospital behavior, see Peter Temin, "An Economic History of American Hospitals," in H. E. Frech III, ed., Health Care in America (San Francisco, Cal.: Pacific Research Institute for Public Policy, 1988), pp. 92-97.

⁴² Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance, and Jonathan Gruber and James Poterba, "Tax Subsidies to Employer-Provided Health Insurance," National Bureau of Economic Research Working Paper No. 5147, June 1995.

⁴³ Congressional Budget Office, The Tax Treatment of Employment-Based Health Insurance, Table 4, p. 30.

⁴⁴ Ibid., Table 4, p. 31.

⁴⁵ Calculated from Feldstein and Allison, Hospital Costs and Health Insurance, Table 7.2, p. 208. They use $40,000 to represent all incomes over $25,000.

⁴⁶ Gruber and Poterba, "Tax Subsidies to Employer-Provided Health Insurance."

⁴⁷ Havighurst, Health Care Choices, pp. 92-96.

⁴⁸ Ibid., pp. 101-103. At one point (p. 103), Havighurst says that "capping the tax subsidy is a notion that only a policy wonk could love, a meritorious policy idea with no natural political constituency."

⁴⁹ Ibid., p. 102.