Introduction

Those who have participated in this series for several years must have learned by now that economists do not view the world as others do. And those who have participated so far this year must have learned that this is especially true of those economists who specialize in health economics. Given that most of my career has been spent in Washington, it is my observation that economists play a very small role in the national health policy debate. While there is a grain of truth in the well-worn jokes about economists not being able to reach a conclusion, the reality is that health policy debates have been dominated throughout the 20th century by strong groups of providers who have a direct stake in the outcome of legislation. In the last few years these groups have been joined by other groups claiming to represent "consumers" or specific groups of patients. All of these participants in the health policy debate provide ample employment to a new army of pollsters and political analysts willing to tell the politicians and the public what kind of health policies they think we want. In the midst of all this noise, the hard-working academic or government health economist, trained to ask fundamental questions and seek answers based on factual information, and having earned a reputation for producing overly-technical and dull reports, has a difficult time being heard. The result is a political environment that has a higher probability of producing legislation based on emotion and wishful thinking than on the economist*s usual standard of economic efficiency.

Nowhere is this danger for bad policy more prevalent than in the current health policy debate. Politicians are striving to legislate improvements in medical quality without asking why our current market arrangements have put too little emphasis on quality and consumer satisfaction. They are beginning to seek ways to reduce the number of those without health insurance without considering the root causes of why most of the uninsured choose not to buy insurance. They are expressing more concern about the future of Medicare without considering how to get both consumers and providers more involved in the choice of productive and cost-effective medical plans and procedures. It is still my opinion that economists have a positive role to play in guiding health policies toward more efficient arrangements. Despite our reputation for dull reports, health economics is anything but a dismal science.

The purpose of this essay is to give my own interpretation of a rather large body of economic literature about how modern medical markets have developed. The emphasis will be on the role that tax policy played in shaping the distinctive form of health care institutions and incentives that characterize our market today. This background will then be used to comment on various tax policy changes being considered to make health insurance more available to the presently uninsured.

The Post-War History of the Development of Health Insurance[1]

World War II, a major event in the history of the 20th Century, provides a convenient demarcation in the history of medicine and health insurance. Numerous advances in scientific knowledge had been made prior to the war, but these advances were not generally available to the vast majority of Americans. Prior to the war there was little that the average physician could do for the average patient to change the course of a disease. This was to change dramatically in the two decades after the war, especially as a result of the development of penicillin during the war and more powerful antibiotics following the war. These new drugs gave physicians the power to fight infection and made possible many of the surgical operations that we now take for granted. New research on medical products and procedures, and the wide dispersion of this new knowledge to a growing number of health professionals made the average person more willing to seek medical care than they had been in the earlier part of the century. The increased availability of more effective medical care changed expectations about what physicians could do and increased the demand for medical services. As one indication of this change, Somers and Somers report that annual hospital admissions per 1,000 population increased from 56.7 during the period 1923-1943 to 99.4 for the period 1957-1958.[2]

Not only was the average citizen more willing to seek medical care, but they were increasingly able to afford it. The post-war period is known as one of rapid growth in population, employment, productivity, and personal income. In 1982 constant dollars, disposable personal income increased from $5,285 in 1945 to $8,944 in 1975, an average annual growth rate of 1.77 percent over this 30-year period.[3]

The increase in the demand for medical care was accompanied by a supply response from every factor market in the medical sector. In his history of the American hospital sector, 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."[4] During this same period, the number of physicians increased 74 percent while the number of nurses increased 116 percent.[5]

These structural changes in the health care sector also had a strong effect on the way health care was paid for. The post-war increase in medical productivity and the infusion of large numbers of personnel and capital resources meant that the cost of medical care increased, both absolutely and relative to average incomes. The average consumer had a stronger incentive to worry about the cost of medical care because they faced a higher probability of going into a hospital and utilizing the services of highly trained medical specialists. This change in the medical market created a desire on the part of consumers to protect themselves against the small probability that they would face a large medical expense. This increase in the demand for risk avoidance created the favorable conditions for the growth of the private health insurance industry following the war.

But, medical science and higher income were not the only forces attributing to the growth in the health insurance industry. In fact, tax policy contributed significantly to both the rate of growth of private health insurance and many attributes of its structure and performance. Again, the role of tax policy in health markets had its origins in World War II.

Sherry Glied points out that the exclusion of health benefits provided by an employer "existed implicitly since the inception of the federal income tax in 1913."[6] But with little demand for health insurance prior to the war, this provision had little effect until labor market conditions changed during the war. In an effort to control the cost of war production, the federal government established wage and price controls during the war. With the expanding demand for labor, the wage controls created a classic case of excess demand and gave employers strong incentives to increase fringe benefits and other non-wage components of employment.

These inducements primarily took the form of the provision of pensions and health insurance, a practice that received official sanction by the War Labor Board in 1943. Marilyn Field and Harold Shapiro state that, "In a war economy with labor shortages, employer contributions for employee health benefits became a means of maneuvering around wage controls. By the end of the war, health coverage had tripled."[7] After some vacillation by the IRS following the war, the Congress made the exclusion of employer-based health insurance from taxable income a permanent feature of tax law in 1954.[8]

While this special tax treatment for employer-based health insurance was established for other reasons, it ended up having a profound effect on the development of health insurance, and in turn, on the development of the entire medical sector. While other forms of insurance were growing in the post-war period in response to the increase in consumer incomes and the desire to protect against financial losses, tax policy caused the health insurance industry to grow primarily by the growth in group policies rather than individual insurance. As is illustrated in Figure 1, group health insurance grew at a much faster rate than individual insurance in the post-war period covering 158 million persons relative to 43 million in individual policies by 1970.

Tax policy increases the demand for employer-based health insurance by making the insurance more attractive relative to wages. When wages are taxed and health insurance is not, employees have an incentive to favor additional health insurance over additional wages. This discount, or tax subsidy, for the purchase of health insurance is directly affected by one*s marginal tax rate (MTR) which means that the subsidy is greater for higher income people than for lower income. This also explains why the total amount of the tax subsidy has increased over time as increases in income have pushed more employees into higher tax brackets. To illustrate, Feldstein and Allison found that MTRs for federal taxes in 1969 ranged from 13 percent for incomes under $1,000 to 36 percent for incomes over $25,000.[9] They estimate that the total tax subsidy in 1969 was $2 billion which implies a 15 percent discount from $15.7 billion total health insurance premiums in that year.[10] In an update of these estimates, Feldstein found that the tax subsidy in 1978 exceeded $10 billion on insurance premiums totaling $42 billion, which implies a discount of 24 percent.[11]

Another indication of the growth in the value of the tax subsidy is the estimates of tax expenditures published by the Congressional Budget Office (CBO) in 1992 and presented in Table 1.[12] This study estimated the actual loss in federal and state tax receipts due to the exclusion of employer-based health insurance from 1969 through 1990 with projections for 1995 and 2000.[13] These estimates show that since 1969 the value of the medical expense deduction has declined in importance relative to employer-paid health insurance premiums. The table also illustrates the dominance of federal taxes over state taxes, although the MTR of states varies from zero to 10 percent.[14] This means that some taxpayers in the highest federal tax brackets and in some of the high-tax states may get tax subsidies exceeding 50 percent when they obtain health insurance through their employer. While Sheils and Hogan find that 68.7 percent of tax expenditures in 1998 went to those with incomes of $50,000 or more, their estimate of an average family tax expenditure of $1,031 implies that the average family receives a discount of approximately 2 percent of family income.[15]

Given the size of this tax subsidy and the fact that it could only be obtained by purchasing one's health insurance through one's employer, it is not surprising that it had a strong effect on the structural development of the health insurance industry. The subsidy gave incentives to employees and their unions to seek relatively more increases in health insurance benefits than wages since only the latter were taxed. Policies that covered primarily hospital stays in the 1940s gradually added surgical (major medical) benefits, out-patient physician services, and later coverage for such routine events as dental care and eyeglasses. In addition to the expansion of types of coverage, the degree of coverage also increased consisting in higher maximum benefits, the coverage of dependents, and lower levels of cost sharing (deductibles and co-payments). While such expansions obviously increased both the number of people covered and the extent of their coverage, they also had the usual moral hazard effects of insurance, the tendency of people with insurance to use more of the covered services since they were at least partially shielded from the effects of the cost of the coverage.[16] The result of this tax-induced increase in demand for both health insurance and medical care was an increase in both the quantity and cost of medical care beyond the levels that could have been expected in a less distorted market. Feldstein and Allison concluded their 1972 pioneering study of the effects of the tax treatment of health insurance by concluding 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."[17]

If the tax treatment of health insurance helps explain how we developed our present system of health insurance with its built in incentives for inefficiency, what role, if any, can tax policy play in solving the policy problems we now face? The next section turns to this perplexing issue.

Living with an Inefficient Health Care Market: Prospects of Efficient Reform

Since economists have been responsible for identifying of the distorting effects of tax policy, one would expect that they have also been the ones proposing changes in the tax treatment of health insurance. That has indeed been the case, starting with Martin Feldstein in the early 1970s and proceeding to the present.[18] Except for a brief period when a tax cap proposal was included in President Reagan's 1984 budget submission to the congress, a serious proposal to substantially reduce the tax subsidy to health insurance has not been proposed by any member of congress.[19] The politics of such a proposal are not hard to understand given that a relatively large subsidy that has gradually grown in size and importance over the last 57 years has created strong political support for the status quo.[20] The situation was best captured by Clark Havighurst when he wrote,

[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. Since the affected interests simply adjust their behavior to the incentives created, they have no occasion to complain or to call for political attention.[21]

But the world is never static and the existing policy is creating other problems which may be changing the reward structure for politicians. Tax policy, by increasing the demand for health insurance, has increased both the cost of health insurance and medical care. In addition, it has created incentives that have expanded covered benefits and reduced deductibles and co-payments. While tax policy can be credited with increasing health insurance coverage among the majority of labors who are in unions and work for larger firms, it has make it more difficult for lower-income workers, the self employed, and those who are more transitory in their employment arrangements, as well as the dependents of these workers. In addition, Pauly and Berger have argued that tax policy, by placing the choice of plans and cost-containment strategy in the hands of employers, has increased employees' dissatisfaction with managed care, the so-called managed care backlash.[22]

These somewhat complicated effects of tax policy have exacerbated the main policy problems of cost, lack of coverage, and concerns about quality that are the central issues in today's policy debates. The policy concepts that are being discussed cover a wide range of ideological beliefs about what causes the problem and what policies should be adopted to correct them. The specific proposals from each camp reflect these ideological beliefs. Liberal proposals have traditionally featured some form of federal mandates to assure universal coverage, either mandates for individuals to buy or employers to provide insurance. But the strong opposition to the mandates in the Clinton health proposal has made even the Democrats leery of this approach. Instead, they now lead with expansions of specific government programs such as Medicare for the near-elderly, Medicaid for the low-income, and increased subsidies for safety net providers treating the indigent. Expansion of the Medicaid eligibility provisions for children established in the State Children's Health Insurance Program (S-CHIP) in the Balanced Budget Act of 1997 (BBA) is a common feature in President Clinton's and Al Gore's proposals.[23]

Even among Democrats, direct subsidies to individuals or business firms are somewhat out of vogue when they cannot be tied into an existing program such as S-CHIP or traditional Medicaid. While not generally opposed to government administration of a program, they do seem sensitive to the current popular criticism of the ability of federal and state welfare bureaucracies to effectively run such programs. For this reason, some Democrats have turned to tax credits as the most expedient means to provide subsidies to the uninsured.

Refundable tax credits are the principle subsidy mechanism in two separate proposals by long-time Democratic advocates for universal coverage, Congressmen Pete Stark (D-CA) and Jim McDermott (D-WA).[24] These two proposals differ in several respects, but mostly in how they would determine the tax credit. Representative Stark would provide a 100 percent subsidy for the amount paid for qualified health insurance up to a cap of $3,600 for a family of four. It would only be available to those without other forms of coverage. Representative McDermott's tax credit would be 30 percent of the amount paid for health insurance limited by the person's income and Social Security tax liability. Of course, the more generous the tax credit and the more people who are eligible for it, the greater the cost of the program. Since Democrats have been less interested than the Republicans in using federal funds for defense or tax cuts, they have been more willing to propose more expensive tax credit proposals.

Democratic presidential candidates have developed extensive health proposals that include tax credits. Vice President Gore has a 25 percent credit that goes to small business whose employees choose to get their insurance through a purchasing cooperative. He also proposes a 25 percent refundable tax credit for people who are not covered by their employer and purchase individual insurance.[25] Senator Bill Bradley, in addition to providing full subsidies for health insurance premiums for the low income, proposes to use income-based refundable tax credits as the means for subsidizing both children and adults.[26]

Republicans have traditionally not been as active as the Democrats in proposing specific health care plans. They apparently reason that health care is a Democratic issue and they are best off politically playing a defensive role. But they have been forced by the public's demand for solutions to the growing inequities and other problems in the health care sector to become more active in this field in recent years. Proposals based on Medical Savings Accounts (MSA) and tax credits have been popular with Republicans because they are seen as ways to promote individual choice without expanding bureaucratic control. Specific tax credit proposals have been introduced by three Republicans, Representatives Nancy Johnson (R-CN), Dick Armey (R-TX), and John Shadegg (R-AZ). The tax credits in these plans are less generous than Stark's or McDermott's, for example, covering 60 percent of the amount paid for health insurance up to a limit of $2,400 for a family in the Johnson plan and 100 percent of the amount paid up to a $1,000 family cap in the Shadegg plan.[27]

The Republican presidential candidates face a dilemma when it comes to health care. The professional pollsters are telling them that health care is not an important issue among Republicans voters during the primaries, but it will be one of the leading issues in the national elections.28 So far, George W. Bush and John McCain have made only general remarks about their approach to health policy and have not come forward with detailed plans similar to the Bradley and Gore plans. Many observers believe that since tax credits were not included in Governor Bush's tax proposals, they have already been rejected in favor of a proposal to provide new block grants to the states for the purpose of expanding coverage to the uninsured. All we know for sure is that his advisers are still working on a plan to be used in the general election.

Senator McCain has mostly talked about new efforts to expand the coverage of children through Medicaid and S-CHIP and the expansion of MSAs and care for veterans.[29] His web site even contains a proposal to, ". . . use the tax code to provide powerful incentives for employers and individuals to obtain affordable coverage."[30] As yet, there does not seem to be a detailed proposal based on tax credits.

Since winning the nomination in the primaries is the first order of business for both candidates, it is not surprising that health policy is not on their front burner. What is important for our purposes is the position occupied by tax credits in the health policy debate. While both parties have approached health policy from different ideological directions, the most common feature in the serious proposals to reform the health insurance market all involve some variation of a tax credit. Whether tax credits will turn out to be solid grounds for compromise or just another pool of political quicksand, has yet to be determined. But there seems to be strong opposition to any set of alternative approaches, be they the expansion of existing programs (Medicare, Medicaid, or MSAs) or new ideas to expand insurance in the individual and small-group markets.

Issues in Designing an Effective Tax Credit for Health Insurance: The Devil is in the Details

If tax credits are to become the basis for a political compromise, a number of key issues will have to be addressed. In the last year a rather large, and in my view, overly-pessimistic, set of studies about the details of using tax credits to expand health insurance has emerged.[31] Analysis that is more optimistic is in a distinct minority.[32] The following will not try to debunk all the criticism, but will attempt to identify the issues and raise the possibility that we now know enough to give tax credits a try.

The first basic question about tax credits is, will they work? If the objective is to induce low-income working employees to choose to buy health insurance for themselves and their dependents, what kind of subsidy will it take to persuade a significant number to do so? And, is it best to subsidize the small employer as Vice President Gore proposes to do, or is it best to subsidize individuals as most of the other proposals do? Does the form of the tax subsidy matter? Are tax credits more effective than tax deductions? Should tax credits be refundable? Should they be a fixed dollar amount or a percentage of the cost of insurance? Is the IRS capable of adding such a scheme to an already complicated tax code? While all of these questions are difficult, it turns out that there is some analysis, and even some empirical research to give us some answers.

Several elasticity studies have been done about the likely response of employees to tax subsidies to determine the take-up rate. Four of these studies are illustrated by the four demand curves in Figure 2. While not drawn to scale, they represent the order of magnitude of the estimates, ranging from the most inelastic by Chernew, et al, to the most responsive by Pauly and Herring.[33] Based on these elasticity estimates, projections are then made on the likely response of various groups of people to various tax credit plans. The results are difficult to compare because there is not a common definition of a tax credit proposal and they use different assumptions and data to make their estimates. Still, some useful lessons can be learned by comparing the results of three of the studies, illustrated in Figures 3-5.

Shiels, Hogan, and Haught use a price elasticity of -0.203, which, when applied to national numbers, implies a "loss of coverage for about 300,000 persons" for a one percent change in the price of insurance.[34] First, Figure 3 illustrates their estimates of what would happen to the number of uninsured if a tax deduction were given to those without access to employer coverage. They then show the effects of refundable tax credits of 30 percent, 59 percent, and 80 percent made available to all those without access to employer coverage. Their deduction is assumed to be "above the line," that is, it is subtracted from adjusted gross income so is available to all taxpayers for non-group premiums, even those who do not itemize deductions. They find that for the year 2000, 12.2 million of the 51.7 million who would be eligible for the deduction would take the deduction, but that 68 percent of these would be people who were already purchasing non-group coverage. Their relatively low elasticity assumption (compared to the others below) restrains their estimates of the reduction in the uninsured as the amount of the credit is increased. Meanwhile, the higher levels of tax credits raise the costs to the federal government, from $11.3 billion per year for the 30 percent credit to $50.3 billion for the 80 percent credit. Only by eliminating the present tax exclusion and mandating individual coverage, as is done in the Heritage plan, can universal coverage be achieved at a more reasonable cost to the federal budget.[35]

The next set of estimates are by Jonathan Gruber who estimates the effects of a tax deduction and different types of tax credits on various classes of insured and uninsured people.[36] He uses a base elasticity of -0.625, somewhat more responsive to price changes than that assumed by Sheils. But, he also adjusts this elasticity down for lower income persons.[37] To reflect the low participation in other tax credit programs he also assumes that only 50 to 90 percent of those who are now buying non-group policies will take advantage of the new tax credits.

For display purposes, his estimates of the reduction in the number of the uninsured are subtracted from the 43,450,000 uninsured used by Sheils. His results for the tax deduction are consistent with Sheils showing very modest effects. As he explains, this is not surprising since the deduction is worth very little to those with low incomes. The three tax credit proposals illustrated in Figure 4 are all refundable but are assumed to apply for only non-group insurance. But, since all persons are assumed to be eligible, each of these has a strong effect on inducing people with group insurance to switch to the non-group market. In the most generous case of a tax credit of $2,000 for an individual and $4,000 for a family, he estimates that almost 14 million people who previously had group insurance would switch to the non-group market.[38] While not illustrated in Figure 4, he also finds that this switching out of group policies would be reduced to only 4 million if the credits were restricted only to those not presently offered health insurance.[39]

Pauly and Herring concentrate their analysis on the 32.3 million workers and their dependents who are uninsured, that part of the population that tax credits are primarily designed to reach. They argue that their estimates are substantially different from other estimates because the elasticities used by Gruber, Sheils, and others have been obtained from studies of other working populations that are predominantly higher income than the low income population tax credits are designed to help.[40] As illustrated in Figure 5, they provide separate estimates of three levels of tax credits showing the results separately for those below and above 300 percent of the federal poverty level. No restrictions or caps are placed on eligibility for the credits.

They draw several conclusions from their estimates, most of which can be seen in Figure 5. First, refundable tax credits are going to be more effective on the low income than on higher income since by design they are targeted more to low-income people. A given percentage credit will have a stronger effect on the low income compared to higher income since it will move more of them over their "reservation price" and give them a reason to make a decision different from the one they had made without the subsidy. But, low levels of subsidy will not be as effective at higher incomes because most people who have already made a decision not to buy insurance will not be induced to change their minds until they see a substantial difference in their cost of insurance. They reach the same conclusion found by Sheils and Gruber that small tax credits are not likely to have a large effect on the number of the uninsured. But, unlike the previous studies, they find that higher levels of subsidy could have substantial effects on the uninsured, especially the low-income uninsured. Assuming a 40 percent loading factor, their 50 percent tax credit reduces the uninsured by 45 percent; if a larger individual insurance market would reduce leading factors to 30 percent, they estimate that the number of uninsured would decrease by 80.5 percent.[41] They also estimate that a tax credit equal to two-thirds of the premium would reduce the uninsured to zero.[42]

A common conclusion from these studies is that tax credits will be more effective in reducing the number of uninsured than will a tax deduction. And, a refundable tax credit will be more effective in reaching relatively more of the low-income than will a flat dollar tax credit. In addition, there seems to be agreement that low levels of tax credits will have relatively small effects. There is less agreement about both the effects of high levels of tax credits and the policy prescription that tax credits should be our next major initiative in health policy. Jonathan Gruber has proposed that we should conduct a large and relatively sophisticated demonstration in order to learn more about what it will take to change our tax-subsidized health insurance market.[43] Mark Pauly and Brad Herring, the authors of the most optimistic estimates of the effects of tax credits, have reminded us that we presently give large tax subsidies to most American workers so we should not look at a policy that shifts these subsidies as necessarily a net loss in economic efficiency. A policy that both increases government expenditures and decreases government revenue may have distributional effects without imposing a real economic loss.[44]

Prospects for the Future: Positive Economics and Dismal Politics

To paraphrase one of the oldest jokes about economists, we have now lined up all the economists working on tax credits end-to-end and they can't reach a conclusion. So, one might ask, if the economists can not agree, how would we ever expect politicians to agree? The answer is that tax credits might become the basis for a political compromise to expand health insurance coverage if enough politicians believe that there is no chance to get the alternative policies they prefer. The next election of the president and the congress will have a big influence on that possibility.

Without having a comparative advantage in making political predictions, let us look at three possible outcomes. The first is that the Republicans win both the White House and the House of Representatives (the Senate is assumed to remain in the hands of the Republicans in all three cases). This result will likely bring about less overall interest in health policy, and certainly not much interest in achieving universal coverage. There will be some interest in expanding the use of MSAs, health marts, and block grants to states, but there will not likely be enough agreement on premium support to do much about Medicare. The Republicans are likely to try to pass some low-cost, income-based prescription drug benefit for Medicare. Their interest in cutting taxes and controlling the growth in federal expenditures will not leave much room for expensive new health programs like tax credits.

The second outcome to consider is that the Democrats win both the White House and the House of Representatives. Even without control of the Senate, they will be in a position to carry out a substantial part of their agenda, including expansions of S-CHIP to families of uninsured children, higher income eligibility for Medicaid, and more direct subsidies to safety-net providers. More likely too would be an expansive Medicare drug benefit and a patient protection act with a strong right to sue. Expanding insurance coverage in the private sector through complicated and expensive tax credits are not likely to be very popular.

The third possibility is, of course, that party control is split between the White House and the House of Representatives. Regardless of how this happens, and assuming that one party does not gain the power to overcome a veto, it leaves us in somewhat of a political stalemate, not unlike the situation we have mostly been in for the last 20 years. In such a situation, neither party has the ability to carry out it's preferred agenda. Tax credits, which already enjoy some bipartisan support, might emerge as the basis for action, but only if there is a strong desire to do something about private coverage. Two current trends, if continued, could help bring that about. If the cost of health care and health insurance premiums continue to increase, this could increase the number of the uninsured as some employers drop coverage and others cut back on the proportion of the premium that they cover causing more employees to decline coverage for themselves and/or their dependents. In addition, if more firms decide to convert to a defined contribution approach, this could improve the efficiency of the individual market making the cost-effectiveness of a tax credit approach more acceptable to the congress.[45]

That is anything but a definitive prediction, but I have lived in Washington too long to have much faith in anyone's ability to predict the future of political events. If politicians decide that tax credits are the only way to achieve a desired political objective, then there is enough positive economic analysis around to tell them how to make them work. If they were accompanied with some limit on the tax exclusion, we could make a faster start on the road back to a more efficient health care market. Politics at this time is too dismal for that much change.

Robert B. Helms is a resident scholar at AEI.