Originally published in The Health Affairs
November/December 2006 - Volume 25, Number 6
Balancing The Goals Of Health Care Provision And Financing
An eminent economist suggests ways to improve on health savings accounts so that they help meet broader U.S. health system goals.
By MARTIN FELDSTEIN
ABSTRACT: A desirable system for providing and financing health care must balance three goals: (1) preventing the deprivation of care because of a patient's inability to pay; (2) avoiding wasteful spending; and (3) allowing care to reflect the different tastes of individual patients. This essay discusses the application of these goals and uses them to consider a reform of the system of health savings accounts (HSAs) that was enacted as part of the 2003 Medicare legislation and, separately, the challenge posed by the very expensive treatments for rare diseases that are becoming more frequently available. [Health Affairs 25, no. 6 (2006): 1603–1611; 10.1377/hlthaff.25.6.1603]
A desirable system for providing and financing health care would achieve three goals: (1) preventing the deprivation of care because of a patient's inability to pay; (2) avoiding wasteful spending; and (3) allowing care to reflect the different tastes of individual patients. In practice, however, it is not possible to realize fully all three of these goals. There are trade-offs among them, as there generally are in every aspect of life.We can achieve one goal more fully only by a less complete achievement of one or both of the others. But these goals should condition and inform the design of a good system for financing health care. In this paper I discuss the application of these goals in more detail and use them to consider a reform of the system of health savings accounts (HSAs) that was enacted as part of the 2003 Medicare legislation and, separately, the challenge posed by the very expensive treatments for rare diseases that are becoming more frequently available.
The Role Of Health Insurance
The role of health insurance is central to the delivery of health care and to the conflict among these three goals.Health insurance reduces patients' financial burden at the time of care and therefore reduces the deprivation of care that would otherwise be caused by inability to pay. But health insurance also leads to excessive spending, because patients do not face the full cost of care at the time that decisions on health spending are made. The controls that are often imposed to limit these costs interfere with the health care system's ability to reflect the varying tastes of individual patients.
The central role of insurance reflects the very skewed distribution of health care spending among households. Although most families use a relatively small amount of health services in a typical year, a small fraction of families experience one or more spells in the hospital and therefore use very costly health care. For most such people, the high cost of health care cannot be predicted in advance. This leads to the use of insurance (or some other form of prepayment) to finance such bills.
Insulating patients from true costs. An important feature of health insurance is that it drives a wedge between the cost of producing care and the cost to the patient at the time care is rendered. Since health insurance means that patients generally pay only a small fraction of the cost of the care at the time they receive the care, patients naturally want a much higher level of spending than they would if they were not insured. This is true even for moderate health events and for affluent people. Health insurance raises health care spending not because itmakes it possible for people to pay large health care bills (although it does that) but because it reduces the cost of health services relative to other goods and services that people consume.
So the widespread use of insurance is desirable because of the financial risks that result from the inherently skewed distribution of total costs per patient. But insurance is also damaging because it leads patients and their doctors to choose excessive levels of care that is, to increase spending to a level at which the last dollar's worth of care costs much more to produce than the value of that care to the patient's health. It is natural, therefore, to ask: What design of health insurance would best balance these two effects?
The uninsured. The financing of health care is also complicated by the fact that many Americans now have no health insurance. Because we as a society do not want people to go without care because they cannot afford it, the government steps in to provide free care for large groups of the population and to subsidize insurance for many others. What form should that government intervention take, in light of the three goals?
Emerging technologies. A new problem for health care finance is emerging because of the breakthroughs in pharmacology and biogenetics. The resulting new drugs make it possible, but sometimes very expensive, to treat some diseases that were previously not treatable or not treatable with as high a probability of success. Should private insurance cover such expensive treatments? Should such coverage be mandatory? If it is not, should the government pay the bill?
First Goal: Preventing The Deprivation Of Care
The goal of preventing the deprivation of care because of an inability to pay does not mean that care must be free for all. Although having free care would achieve that goal, it would conflict strongly with the goal of avoiding wasteful spending.
Fortunately, most people can pay small and medium-size health care bills without any financial stress. It is the very large bills that can be a barrier to care for many people and a serious financial burden. That is why health insurance especially some form of major-risk insurance makes sense. (The special problems of reforming Medicare and Medicaid for the aged and indigent lie beyond the scope of this paper.)
Even the extremely expensive care for some forms of cancer and certain other diseases that have been made possible by new forms of drug therapy is sufficiently rare that the insurance premiums required to pay for the actuarially expected cost of such care is not likely to be large relative to individuals' incomes. A treatment that costs $100,000 but that is medically useful for only one insured person in 100,000 per year only increases the actuarial cost and therefore the needed insurance premium by one dollar per person per year. Such high costs of treatment appear in news stories when the individual's insurance policy does not cover these expenses, despite the low actuarial cost of doing so.
Pharmaceutical research is, of course, likely to increase the number of diseases that can be treated effectively but at high expense, adding to these actuarial costs. But at the same time, that research is working on developing ways to identify which patients can benefit from each type of treatment. These developments in targeted pharmacology will reduce the number of ineffective treatments and will thus reduce the cost of insurance. Targeted pharmacology should also reduce the cost of drug development (by reducing the sizes of the samples needed to prove efficacy), thus lowering the cost of the drugs themselves.
But there will inevitably be treatments that are very expensive and yet have a low probability of success. Avoiding deprivation of care because of inability to pay does not mean providing every possible treatment, regardless of how low the probability of success. When the cost of treatment is very high relative to the likely benefit, there would be general agreement that the cost of such treatment would violate the second goal (avoiding wasteful spending).
Moreover, individuals differ in their willingness to pay through insurance for treatments that have a very low probability of success, even if the incremental premium is relatively low. That's why it is important to allow care to reflect differences among individuals in preferences, a subject to which I return below.
Second Goal: Avoiding Wasteful Spending
There is widespread concern about the fact that health care spending has increased much more rapidly than gross domestic product (GDP) or personal incomes. It is important, however, to recognize that this rise in health care spending is not the same as an increase in the price of health care. The important difference is that the increase is for spending on new types of care, not higher prices for old types of care. Treatments have changed and become more effective. For many conditions, the cost of effective treatment has actually gone down because treatments are now more likely to be successful, because hospitalization has become unnecessary, or because hospital stays have shortened. And even for those conditions for which the cost of success has gone up, I would much rather be a patient now than ten or twenty years ago.
Value in health care. Estimates by my Harvard colleague David Cutler and others indicate that the value of the improved health over the past several decades has exceeded the increased cost of health care.1 But although that is true for the overall cost of care, it is not likely to be true for incremental care at the margin. There have been inframarginal gains in health and in the efficacy of health care, but the structure of current insurance means that at the margin we are undoubtedly spending more for care than the value to patients of the resulting health improvements.
If patients and their doctors increase their spending until the value to the patient of the "last dollar" of that care is equal to the additional net cost to the patient at the time of that care—that is, to the cost net of insurance—then the system is providing care that, at the margin, costs a dollar but is valued by the patient at only, say, twenty cents, or whatever the coinsurance ratemay be. When physicians are required by insurance rules or health maintenance organization (HMO) regulations to choose a standard of care that more closely reflects the total cost of the incremental unit of care, they create frustrated patients who feel (correctly) that they are being denied care that could help them.
Part of the solution to this problem is some form of major-risk insurance in which patients have a large deductible or high coinsurance rate. I proposed such a form of insurance in 1971 and examined it statistically with Jonathan Gruber in 1995.2 A deductible of $5,000—which is less than 10 percent of median family cash income—would leave most families paying for all of their health care out of pocket with no insurance reimbursement. The decisions of these patients, guided and advised by their doctors, would therefore not be distorted by insurance. And while no one would welcome a medical bill for the year as large as $5,000, most families would also not be deprived of care by an inability to pay, because their maximum annual payment under the deductible would be less than 10 percent of their income. The same principle would lead to lower deductibles for families with lower incomes. Special rules could apply to preventive measures that are shown to be cost-effective or to procedures that reduce contagious diseases.
But a sizable fraction of families—and a much larger fraction of health care spending—would exceed a $5,000 deductible. With no out-of-pocket payment above that level, or only a modest coinsurance rate, there would be both wasteful spending and, to the extent that physicians restrict what patients would otherwise want, patient frustration as well. Raising the deductible to, say, $10,000 would reduce the problem of wasteful spending but would clearly create a large financial burden for many households and a barrier to appropriate care for some.
The first two goals (preventing deprivation and limiting wasteful spending) might be better accomplished by replacing deductibles with a high coinsurance rate. For example, instead of a $5,000 deductible, the insurance policy might take the form of a 50 percent coinsurance rate on the first $10,000 of spending. Patients would still be protected against paying more than $5,000 out of pocket. But with a 50 percent coinsurance rate on $10,000 of spending, there would be fewer patients and fewer dollars that face no out-of-pocket cost.
A constant 50 percent coinsurance rate on the first $10,000 of spending might not be the best structure for the insurance policy. Determining that requires more analysis than I have done. It would depend on the distribution of potential spending levels, the sensitivity of spending to different coinsurance rates, and the value that people place on limiting their out-of-pocket cost of health care. In principle, the optimal policy might involve a combination of deductibles and different coinsurance rates for different ranges of spending. The deductibles and coinsurance rates might also be related to family income.
Third Goal: Allowing Health Care To Reflect Patients' Preferences
It was not too many years ago that a physician could make decisions about medical care by asking him- or herself what would produce the very best health outcome for his or her patient. Economists, myself included, argued that that was not good enough and that doctors should take the cost of care into account, performing an implicit cost-benefit analysis to decide what care was appropriate. The rise in the cost of care, especially the cost of hospital inpatient care, brought about that change in physicians' thinking. Under pressure from hospital administrators and insurance companies, physicians developed protocols of appropriate care that reflected costs as well as outcomes.
Unfortunately, this approach has generally led to "one size fits all" medicine. A physician generally prescribes the same treatment for demographically similar patients who have a particular disease or who present with particular symptoms. Of course, doctors differ in their perceptions of the efficacy of different treatments or diagnostic procedures, and that leads to different behavior among doctors. But, with certain important exceptions, patients' preferences do not play an important role in this process. Those exceptions—for example, the treatment of prostate cancer or breast cancer—are generally about balancing risks and other outcome measures but not about balancing costs and outcomes.
And yet for every other kind of good or service, we assume that an important function of the market is to reflect differences in consumers' preferences. Of course, everyone wants good health. But some are more willing than others to make greater sacrifices to achieve that good health. This is not just a question of money or ability to pay.We all know that health is hurt by smoking, by being overweight, and by not exercising. And yet millions of Americans smoke, are overweight, and do not exercise. These habits might be hard to change, but millions have changed them. So addiction is not an excuse. It seems reasonable to conclude that some people enjoy the harmful behavior enough to accept the potentially adverse long-term health effects.
We should not be surprised, therefore, if taste differences about health also imply that some people are willing to pay more to get better health outcomes. This might involve paying more for more complete routine checkups, or for more complete diagnostic examinations when there are symptoms, or for more expensive care of adverse medical conditions (either at the time of care or in the choice of a particular insurance policy or HMO).
How does this fit with the other goals of health care provision? For patients whose spending is within a deductible limit, there is no conflict. If their physician and hospital are willing, they can buy whatever they want and are willing to pay for. It would certainly be a mistake to prevent them from doing so.
An advantage of the high coinsurance rate is that people can indicate their preferences by their willingness to pay. They may, of course, be paying only fifty cents to buy care that costs a dollar to produce, which suggests that it would be appropriate for providers to exercise some restraint on what consumers buy. But reflecting individual preferences implies that this should not be done to the point at which all people are forced to accept the same care.
The problem is more difficult when the coinsurance rate is at a low level or when the patient is not paying at all. Under those circumstances, there is no way to know patients' preferences, and the physician and institutional provider must determine what the patient gets. But that should be seen as an undesirable outcome, denying patients and their physicians the opportunity to adjust care to different preferences. That suggests that it would be desirable to design the financing system to avoid such situations. That might involve, for example, arrangements in which patients express their preferences by the type of coverage they select or the style of the HMO they join. Even if they are not paying out of pocket at the time of care, they can express a desire for more complete preventive care, diagnostic exams, or forms of treatment by the selection among different policies or different HMOs.
This could also be the framework for solving the problem of the very expensive treatments that are now becoming possible. There are some treatments that physicians (and patients, if they knew enough) would agree should always be done, even though they are very expensive, because they produce favorable outcomes with high enough probability relative to the cost of the treatment. Unless such treatments are required to be part of every standard health insurance policy (or are covered by a government catastrophic-risk plan), there would be a temptation for some people to reduce their insurance premiums by selecting insurance with- out such coverage in the knowledge that society would pay for the care if they met the medical conditions. Those are the policies that generate the news stories now about "impossibly expensive" forms of care.
But for those treatments that are more questionable—offering lower probabilities of success or only small increases in life expectancy in exchange for very large costs people could have the discretion in advance when they buy insurance or choose an HMO. Just as some insurance plans now include a wider range of drugs than other plans do, the same could be applied to the very expensive treatments for various diseases. How this is to be done in practice and what its consequences are for the cost of care remain important challenges for the future.
Implications For Reforming HSAs
There is a lesson in this for the possible reform of HSAs. I am an enthusiastic supporter of the HSA principle. I think, however, that it could be greatly improved and that failure to improve it might lead to its eventual rejection by the political process.
How HSAs now work. The rules creating HSAs were enacted as part of Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003. An HSA is similar to an individual retirement account (IRA) or 401(k) plan in that funds are deposited out of pretax income (by the individual or a combination of individual and employer) and enjoy the advantage that the income of the account (interest, dividends, and capital gains) accumulates tax free. Even better than an IRA or 401(k) account, the funds that are withdrawn from an HSA to pay for health care, broadly defined, are never subject to tax. The balance in the fund is carried forward just as an IRA would be. (The HSA rule is therefore different from those of earlier plans in which funds not spent within the year are not carried forward but are kept by the employer.)
The amount of money that can be deposited in an HSA each year is equal to the size of the deductible in a major-risk insurance policy that the individual chooses. More specifically, in 2006 a personmay contribute to his or her HSA nomore than the lesser of the deductible of the person's high-deductible health plan or $2,700. For family coverage, themaximum amount is $5,450. The policy must also provide protection against financial hardship by setting a maximum out-of-pocket amount that is incurred by a combination of the deductible and coinsurance payments above the deductible. For 2006, this maximum out-of-pocket amount for a couple is $10,500. An HSA policy might have a $5,000 deductible and then a 20 percent copayment until the total out-of-pocket spending reaches, say, $10,000, although typical policies have lower maximum out-of-pocket limits.
The tax advantage of the HSA provides a strong incentive for individuals and employers to shift away from the current common type of health insurance policy with a low deductible and low coinsurance rate. Such traditional policies largely distort the choice of care but provide a major tax benefit to the individual. The HSA option provides an opportunity to enjoy a similar or larger tax advantage by buying an alternative form of health insurance that provides financial protection while reducing the incentives that lead to wasteful spending.
Individuals could use the HSA balances to pay their uninsured expenses (that is, expenses up to the deductible amount plus the coinsurance payments up to the maximum out-of-pocket limit). Alternatively, they could treat their HSA as a type of IRA, paying uninsured expenses out of pocket and enjoying the maximum taxfree accumulation of funds.
Problems with current HSA structure. There are, however, two problems with the HSA legislation in its current form. First, the $5,450 maximum deductible will leave far toomany dollars of health spendingwithout an effective restraint. As I noted above, it would be more effective in controlling costs to have a 50 percent coinsurance rate on twice the amount of the current deductible ($10,900). Other combinations might be even better, but a 50 percent coinsurance rate is easy to understand.
The HSA approach will only succeed if people find it attractive. For lower income families, the risk of a $5,450 deductible (or of out-of-pocket payments of 50 percent on the first $10,900 of health spending) might be larger than they are willing to accept. Under the HSA rules, such a family could select a policy with a lower deductible and put less money into their HSA. A family with an income of $30,000 could decide to have a deductible of only about, say, $3,000 and therefore put only $3,000 into their HSA (including their employer's contribution).With a deductible of only $3,000, such a family would be even more likely to spend above the deductible amount, removing any effective discipline on such spending. It would be better to allow such a family to deposit up to $3,270 in an HSA if they have a policywith a 30 percent coinsurance rate on $10,900 of spending. Although there must be a lower limit on the coinsurance rate to make sure that it has a favorable incentive effect, a 100 percent coinsurance rate (that is, a deductible) on a low amount is certainly wrong.
In principle, it would be possible to remove the favorable tax treatment of health insurance completely, substituting a specified level of the employer/employee contribution of pretax dollars to an HSA while leaving people free to select any combination of deductible and coinsurance they want. This would eliminate the current distorting effect of taxes on the form of insurance, provide a special savings account to meet health care bills, and let people optimize their deductibles and coinsurance to match their personal risk preferences. One danger with such a plan is that some people, particularly those with relatively low income and assets, might choose a plan with no maximum out-of-pocket limit in the knowledge that the government or hospitalwould pay for the very expensive care.
The second problem with the HSA legislation in its current form is suggested by the fact that hospitals all across the country are experiencing a large volume of bad debts caused by patients who do not pay their hospital bills. This is particu- larly true for uninsured patients, but it is also true for insured patientswho do not pay the coinsurance and deductibles called for by their policies.
Such nonpayments could become much more severe with HSAs. A person with a $5,000 deductible might not have cash on hand to pay the bill when he or she is discharged from the hospital. Although many people would accept the obligation and pay the bill promptly after that, perhaps by drawing on their HSAs, others might simply put off payment and eventually not pay.
The advantage of the out-of-pocket payment as a discipline on excessive spending would of course be lost if people simply did not pay the deductible or copayment. The impact on hospitals' financial soundness of not collecting the first $5,000 of each hospital bill would be very serious. And yet, unlike the sellers of other goods and services, hospitals cannot deny care to emergency patients who do not have the cash to pay at the time of care.
One simple remedy for this would be to allow hospitals (and other health care providers) to have easy access to the HSA balances of people who have not paid within, say, three months of the time of care. This might be arranged through the insurance provider or by using an HSA debit card as collateral at the time of care. It is important to the attractiveness of the HSA system to allow people the choice of paying out of pocket, if they prefer, rather than from their HSA accounts. But easy access to the HSA accounts by providers without the usual legal procedure of collecting bad debts would be desirable in itself and a strong incentive for people to pay their bills.
In summary, the design of health care financing involves balancing three goals: preventing the deprivation of care, avoiding wasteful spending, and allowing care to reflect the different tastes of individual patients. The trade-offs among these three goals are changing as the cost and potency of care increase. The system of HSAs provides a new framework within which to balance these goals, but it requires modification if it is to strengthen incentives and avoid increasing the bad-debt problems of hospitals and physicians.
This paper is based on remarks presented at the meeting of the American Society of Health Economists session of the Allied Social Sciences Associations, Boston, Massachusetts, 7 January 2006.
1. D.M. Cutler, Your Money or Your Life:Strong Medicine for America's Healthcare System(NewYork: Oxford University Press, 2004); and D.M. Cutler andM.McClellan, "The Costs and Benefits of Intensive Treatment for Cardiovascular Disease," NBER Working Paper 6514 (Cambridge, Mass.: National Bureau of Economic Research, 1998).
2. M. Feldstein, "A New Approach to National Health Insurance," Public Interest 23, no. 7065 (1971); and M. Feldstein and J.Gruber, "A Major Risk Approach to Health Insurance Reform," in Tax Policy and the Economy, vol. 19 , ed. J. Poterba (Cambridge, Mass.: MIT Press, 1995).
Martin Feldstein (firstname.lastname@example.org) is a professor of economics at Harvard University and is president and chief executive officer of the National Bureau of Economic Research in Cambridge,Massachusetts.He is on the board of directors of Eli Lilly and Company, the Hospital Corporation of America, and the American InternationalGroup and is a member of the corporation of Partners HealthCare System in Boston,Massachusetts.