Health Affairs, 23, no. 6 (2004): 129-135 doi: 10.1377/hlthaff.23.6.129 © 2004 by Project HOPE	New Online   Pay Cuts For Medicare Docs   Access To Care Woes   Public Coverage More Efficient   Empowering Consumers

This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Rosenthal, M. B. Search for Related Content PubMed PubMed Citation Articles by Rosenthal, M. B. Related Collections Insurance Market Business Of Health Chronic Care Health Reform Managed Care Medicare Health Spending

Health Tracking

MARKETWATCH

Doughnut-Hole Economics

Abstract

 
Both the recently enacted Medicare prescription drug benefit and a new cohort of consumer-directed health benefit models offer doughnut-shaped insurance coverage with large deductibles that begin around the mean annual spending for enrollees. These policies leave enrollees to bear more risk than policies with equal expected payouts that rely on first-dollar deductibles. This risk to enrollees is substantial, given the skewed distribution of health care spending and the placement of the typical deductible. I consider alternative explanations for this new benefit design trend and conclude that the desire to distribute tangible benefits to the largest number of constituents is most plausible.

There are a number of important differences between the Medicare drug benefit and consumer-directed health plans besides the scope of services covered. The Medicare plan is the product of the political process, while consumer-directed health plans are influenced by labor-market forces and regulatory regimes. In addition, the general shape of the benefit design for the Medicare drug plan will be the same for all enrollees, whereas employers and some employees may choose from among a variety of other health benefit models.

In both the employer-sponsored insurance and Medicare contexts, critics have expressed similar concerns about the insertion of sizable deductibles above a threshold of $1,000–$2,000 in spending. They worry that enrollees will face excessive financial risk. At the same time, it is widely believed that the Medicare plan will be very expensive. Similarly, there is considerable doubt that account-based, consumer-directed health benefit plans are equipped to reverse the current high rate of health care spending growth.² How can these benefits cost so much yet leave enrollees exposed to so much risk?

A number of thoughtful analyses examine the economics of the Medicare drug benefit and consumer-directed health benefit approaches. Most of the expert analysis of the Medicare plan has, understandably, focused on reimbursement and how competition for enrollees will work.³ These features, more than benefit design, appear to be associated with the high expected cost of the new plan. This is underscored by a recent paper examining the likely minimal effects of the new benefit in terms of spurring additional use of prescription drugs and increased spending.⁴ With regard to consumer-directed health plans, several recent papers have examined the anatomy of post–managed care, employer-sponsored insurance in the United States and the implications of the new health benefit models.⁵

In this paper I focus narrowly on the doughnut-hole provisions of the new plans. The goal is not principally to make the obvious point that the U.S. employer-sponsored insurance system and the politics of Medicare have wrought benefit designs that depart from optimal insurance principles, but to recognize explicitly the costs of this departure as well as the likely reasons for it. Using a few simple results from the economic literature on insurance, I show that for a given premium, first-dollar deductibles always offer more protection from financial risk than doughnut holes do. In this way, doughnut holes function similarly to benefit ceilings, to which economists have traditionally objected. These general results, which have been hinted at in earlier critiques of the new plans, apply equally to the Medicare prescription drug benefit and to account-based, consumer-directed health plans, despite other differences between the two contexts.⁶ I evaluate several possible economic and noneconomic reasons for the popularity of doughnut holes in health insurance, despite their failure to conform to the principles of optimal insurance. In conclusion, I look beyond the objective of efficient risk reduction to consider competing goals of plan sponsors that might be served by doughnut-shaped benefit designs.

Brief Theory Of Insurance

Top Brief Theory Of Insurance The First (And Only)... Doughnut-Hole Placement And... Why Doughnut Holes And... Editor's Notes NOTES

 
Insurance obtains its value from the dual facts that life is uncertain and people are risk-averse. In practical terms, risk aversion means that people want to smooth out their fortunes in alternative possible states of the world. A simple example conveys the notion of risk aversion and, at the same time, demonstrates the power of insurance. Suppose a person is offered the opportunity to flip a coin and receive $1,000 if the coin lands on "heads." If it lands on "tails," the person receives nothing. So the expected value or average outcome over a large number of coin flips is $500. A risk-averse person, by definition, would strictly prefer $500 with certainty to the gamble. Moreover, the more risk-averse the person, the smaller the amount he or she would be willing to accept in place of the gamble (this amount is called the certainty equivalent). Insurance can just be thought of as a large number of people with independent coin flips pooling their winnings so that everyone gets $500 with near-certainty, which benefits everyone in the risk pool.

Of course, health care spending is not like flipping a coin. People make deliberate choices about when to go to the doctor or hospital and often participate with their physicians in weighing treatment alternatives. Because insurance pays for some or all of the costs of health care at the point of service, these choices are altered. In particular, people will elect to receive more services than they would without insurance, and these incremental services will be valued by the individual less than they cost the insurance pool. This perfectly rational behavior, which leads to inefficiency, is termed "moral hazard" (which, despite the name, connotes no moral opprobrium). Economists have shown that if people do behave this way—and there is indisputable evidence that they do in the health care setting—coinsurance should be part of the design of an insurance policy.⁷

A deductible is a dollar amount below which the insurance plan does not share in the cost of health care. Deductibles, like coinsurance, reduce the demand for health care services but are justified as part of an optimal insurance policy mainly because the administrative costs of insurance offset some of the benefits from insurance and thus make full coverage undesirable. More intuitively, deductibles may be thought of as the best way to alter an insurance policy to reduce the premium cost if one has a limited budget for health insurance. Although economists do not tend to think about designing an optimal insurance plan in the context of a limited budget, there are practical reasons that such a limitation might exist. For example, Congress might be politically able to authorize only a certain amount of funding for a new public insurance program. Minimum-wage laws in the private sector may limit the amount available for health benefits (which most economists believe are paid for out of workers’ wages).⁸ Thus, high-deductible or "catastrophic" plans are most commonly found in the individual and small-group insurance markets, where administrative costs are high and wages are typically lower. In the discussion that follows, I illustrate how deductibles alter insurance protection.

The First (And Only) Fundamental Theorem Of Doughnut-Hole Economics

Top Brief Theory Of Insurance The First (And Only)... Doughnut-Hole Placement And... Why Doughnut Holes And... Editor's Notes NOTES

 
Using the basic economic principles of insurance, it is relatively straightforward to show that the optimal placement of a deductible is at the first dollar; doughnut holes can always be improved upon at equal cost. Suppose that a person faces some distribution of possible levels of annual health spending and has a fixed sum to spend on insurance. The amount the person has to buy a policy is one dollar short of the premium to cover all of the risk (ignore moral hazard for the moment, since it does not change the result). Now consider two ways to make that policy affordable: a first-dollar deductible, and a doughnut hole. For illustrative purposes, suppose we start the doughnut hole at $100 and the probability of a person’s exceeding $100 at 25 percent. Let us also establish a 50 percent probability that a person has at least $2 of spending. So a $2 first-dollar deductible will save $1 in expected cost (premium). Because the likelihood of exceeding a higher spending threshold is less, however, a doughnut hole of $2 will not save a dollar. To save a dollar, the doughnut hole would have to be twice as large, or $4.⁹ The expected payouts of the $2 deductible policy and the $4 doughnut hole policy are equal, so the policies are, by definition, actuarially equivalent. That is, even though in some possible scenarios the policies would have different pay-outs, on average, given the likelihood of each possible outcome, they will result in the same total spending for the insurance plan. The individual, however, faces greater risk (variance of spending) with the doughnut hole than with the deductible. Out-of-pocket spending will vary between $0 and $4 with the doughnut hole and between $0 and $2 with the deductible. For any distribution of spending and starting point of the doughnut hole, this will be true. And because individuals are risk-averse, this increased variance of spending with the same expected (average) spending will result in a utility loss relative to the first-dollar deductible.

A corollary to the conclusion that doughnut holes are trumped by deductibles is that doughnut holes are less desirable than deductibles (in terms of risk): (1) the higher is the dollar value of the starting point (the point at which there is a gap in insurance), and (2) the lower is the likelihood that the insured person will have relatively high spending (in statistical terms, the more "skewed" is the distribution of spending). Next I use a few facts about the distribution of health care spending to illustrate these two points and give a flavor of the magnitude of the excess risk that the new doughnut-shaped health benefit models introduce relative to a traditional deductible.

Doughnut-Hole Placement And Spending Distribution

Top Brief Theory Of Insurance The First (And Only)... Doughnut-Hole Placement And... Why Doughnut Holes And... Editor's Notes NOTES

 
Consider first the consumer-directed health benefit models being offered in the commercial market. A recent survey found that the typical account-based, consumer-directed health plan has a $500 account and a $1,000 deductible.¹⁰ The latest figures from the Medical Expenditure Panel Survey (MEPS) suggest that half of Americans spent $500 or more on health care in 2001 and that 37 percent spent more than $1,000 (Exhibit 1).¹¹ Assuming a relatively uniform distribution of spending between $500 and $1,000, the doughnut hole in a typical consumer-directed health plan saves the plan sponsor (relative to no doughnut hole) about $220.¹² A rough approximation suggests that an actuarially equivalent plan (one that results in the same expected insurance payments) with a first-dollar deductible of about $300 could be offered.¹³

View larger version (18K): [in this window] [in a new window]   EXHIBIT 1 Annual Health Care Spending By The U.S. Population, 2001

View larger version (21K): [in this window] [in a new window]   EXHIBIT 2 Annual Spending On Outpatient Prescription Drugs By Medicare Fee-For-Service Beneficiaries, 2003

Top Brief Theory Of Insurance The First (And Only)... Doughnut-Hole Placement And... Why Doughnut Holes And... Editor's Notes NOTES

To encourage financial discipline? At first glance, it might appear that the doughnut hole is a good way to introduce some financial discipline among enrollees (that is, to control moral hazard). Theoretically, a person facing a doughnut hole may try to conserve even before spending has reached the gap between the first covered segment and the second, hoping to steer clear of the full-risk portion of the policy. So perhaps a $500 doughnut hole placed at $500 encourages restraint on $1,000. But unless moral hazard (price-responsiveness) is extremely intense in the $500–$1,000 range and much less so elsewhere, the best way to address this problem is through more continuous coinsurance.

Pursuing further the idea of moral hazard that varies by category of spending, health economists have suggested that optimal insurance should pay for basic nondiscretionary services, such as the standard treatment for a particular cancer, without cost sharing.¹⁵ The notion is simply that if there is some minimal level of treatment that every person with a condition must receive, there is no demand response to insurance coverage, and cost sharing just imposes unnecessary risk. But these payments would be contingent upon the unexpected occurrence of a given diagnosis, such as cancer, unlike the doughnut-shaped policies described here. And while it might be tempting to suggest that the designers of such policies believe that every enrollee "needs" a minimum level of services (for example, for annual screening tests or, in the case of Medicare, for medications to treat a chronic illness), if that were true, then there is also no risk. If there is no risk, then insurance does not provide any risk protection, which is the economic-efficiency reason for its existence.¹⁶

To pay for a set of services? Insurance coverage for the first $500 or so of health care spending may be intended to pay for a set of services that the sponsor is concerned will be forgone by consumers facing a sizable deductible. For example, perhaps these are preventive services that are valued by the plan sponsor more than by the individual, either because the services yield positive health externalities (such as immunizations) or improve productivity, or simply because consumers are myopic while payers are not. While this explanation seems especially plausible for consumer-directed health benefits, one then wonders why health benefit sponsors don’t just pay for the highly valued services separately rather than providing a lump-sum account that may just as well be spent by the enrollee on therapies of low value. Indeed, several of the more sophisticated consumer-directed health benefit models do make separate arrangements for recommended preventive care.¹⁷

Two other economic theories. Two other economic theories can be offered for doughnut-shaped, consumer-directed health benefits, but these are not applicable to Medicare. The employer-sponsored account that covers the first $500 or so of the consumer-directed health benefit can typically be rolled over, and if the plan qualifies as a health savings account (HSA) as defined by the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003, it is portable. Thus, the account offers a tax-preferred way for people who use few health services to save for future needs. The gain from this opportunity depends on a number of factors, including a person’s marginal tax rate. It is difficult to judge the exact value of the ability to use current account dollars for future health care needs, but for many people it seems unlikely to exceed the value of reducing financial risk in the near term. This is because people tend to discount, often substantially, the importance of future needs compared with current ones.

Another possible explanation for the appeal of a doughnut hole in the context of private insurance markets is risk selection. If people who expect to see lower health care spending are differentially attracted to doughnut-shaped policies, health insurers may find these models especially profitable to offer (at least in the short run). In fact, it should be true that consumers with reasonably high expected health spending will care little about the placement of the deductible, whereas those who expect only a few office visits will care most about whether the deductible starts at $0 or $500. Whether selection will be favorable or adverse depends, however, on the precise benefit design and features of employers’ alternative offerings. The selection story also raises the question of why employers would want to introduce a new plan that skims off their healthiest beneficiaries, possibly causing a premium spiral among their preferred provider organization (PPO) or health maintenance organization (HMO) offerings.

Doughnut holes as political constructs. None of the alternative theories considered here suggests an explanation for doughnut holes based on the economics of insurance. It is often the case, however, that insurance serves purposes other than risk protection. These potentially competing goals include attracting and retaining productive employees, currying favor with key interest groups, and redistributing income. In the case of account-based, consumer-directed health benefits, the intent may be to increase the perceived value of the compensation package, particularly to healthier and presumably more productive workers. For Medicare, policy-makers may have intended the new benefit either as a means of achieving otherwise needed intergenerational transfers or simply to court the vote of those who are eligible for Medicare.

In this light, doughnut-shaped health benefits, under Medicare and employer-sponsored insurance alike, do have one important property: They ensure that the maximum number of people see some point-of-service benefit from the policy. Most beneficiaries will thus perceive that a doughnut-hole policy is more generous than one with a first-dollar deductible, despite the gap in risk protection. This may be the case because although employees and Medicare enrollees should theoretically prefer first-dollar deductibles, they may be more concerned about getting their "fair" share of annual benefit spending than about risk protection. They may also greatly underestimate their risk of having health care spending in the doughnut hole.

Of course, it is hardly news that the design of U.S. health benefits is determined more by perception and politics than by principles of health economics. It is nonetheless important to recognize that doughnut holes are primarily political constructs, not economic ones. There is no such thing as doughnut-hole economics.

Editor's Notes

Top Brief Theory Of Insurance The First (And Only)... Doughnut-Hole Placement And... Why Doughnut Holes And... Editor's Notes NOTES

 
Meredith Rosenthal (mrosenth{at}hsph.harvard.edu) is an assistant professor of health economics and policy in the Department of Health Policy and Management, Harvard School of Public Health, in Boston, Massachusetts.

The author is grateful for financial support from the Robert Wood Johnson Foundation’s Changes in Health Care Financing and Organization (HCFO) Initiative, the Agency for Healthcare Research and Quality (AHRQ Grant no. P01-HS10803-01), and the Alfred P. Sloan Foundation. Helpful suggestions on earlier drafts were provided by David Cutler, Eric Rosenthal, and two anonymous reviewers.

NOTES

Top Brief Theory Of Insurance The First (And Only)... Doughnut-Hole Placement And... Why Doughnut Holes And... Editor's Notes NOTES

 

1. A.J. Culyer and J.P. Newhouse, eds., Handbook of Health Economics (Amsterdam: North-Holland, 2000).
2. M.B. Rosenthal, C. Hsuan, and A. Milstein, "Design, Implementation, and Early Experience of Consumer-Directed Health Benefit Models: Thirteen Case Studies" (Unpublished manuscript, Harvard University, 2004).
3. J.P. Newhouse, "How Much Should Medicare Pay for Drugs?" Health Affairs 23, no. 1 (2004): 89–102.[Abstract/Free Full Text]
4. M.V. Pauly, "Medicare Drug Coverage and Moral Hazard," Health Affairs 23, no. 1 (2004): 113–122.[Abstract/Free Full Text]
5. J.C. Robinson, "Reinvention of Health Insurance in the Consumer Era," Journal of the American Medical Association 291, no. 15 (2004): 1880–1886; and [Abstract/Free Full Text]Rosenthal et al., "Design, Implementation, and Early Experience."
6. Pauly, "Medicare Drug Coverage and Moral Hazard"; and Robinson, "Reinvention of Health Insurance."
7. D.M. Cutler and R.J. Zeckhauser, "The Anatomy of Health Insurance," in Handbook of Health Economics, 564–629.
8. J. Gruber, "Health Insurance and the Labor Market," in Handbook of Health Economics, 645–680.
9. The expected values of spending in the deductible and doughnut hole are not quite right, since there is positive probability of spending between zero and $2 and between $100 and $104, but this difference is not important to the overall conclusion.
10. M.B. Rosenthal and A. Milstein, "Awakening Consumer Stewardship of Health Benefits," Health Services Research 39, no. 4 (2004): 1055–1070.[CrossRef][ISI][Medline]
11. Data from the Medical Expenditure Panel Survey (MEPS) for various years are available online at Agency for Healthcare Research and Quality, "MEPSNet," www.meps.ahrq.gov/MEPSNet/HC/MEPSnetHC.asp (16 August 2004).
12. This is calculated as follows from Exhibit 1: 37 percent of the population would incur the full $500 in the doughnut hole, so the plan would save 0.37 x $500 on these people, or $185. If the 14 percent of the population (51 percent – 37 percent) who spend between $500 and $1,000 were uniformly distributed in this range, then the plan would save 0.5 x 0.14 x $500 on these people, or $35. Total savings to the plan from the doughnut hole would be $185 + $35, or $220.
13. From Exhibit 1, about 60 percent of the population would spend the entire $300, and about 25 percent of the population spends $1–$300, so if the spending were approximately uniformly distributed between $1 and $275, the expected value of the deductible would be about 0.6 x $300 + 0.125 x $300, or $217.50. In fact, because many account-based plans allow enrollees to roll over any unspent funds, the probability that the plan would pay out $500 is equal to 1.
14. Congressional Budget Office, "Projected Spending for Prescription Drugs by and on Behalf of Medicare Enrollees," February 2003, ftp .cbo.gov/51xx/doc5197/02-03-PrescripDrugs Memo.pdf (16 August 2004).
15. M.E. Chernew, W.E. Encinosa, and R.A. Hirth, "Optimal Health Insurance: The Case of Observable, Severe Illness," Journal of Health Economics 19, no. 6 (2000): 585–609.[CrossRef][ISI][Medline]
16. There may, however, be an argument for redistribution, which I consider in the final section.
17. Rosenthal et al., "Design, Implementation, and Early Experience."

                      What's this?

This article has been cited by other articles:

				V. Gray, D. Lowery, and E. K. Godwin Public Preferences and Organized Interests in Health Policy: State Pharmacy Assistance Programs as Innovations Journal of Health Politics Policy and Law, February 1, 2007; 32(1): 89 - 129. [Abstract] [PDF]

				W. F. Gellad, H. A. Huskamp, K. A. Phillips, and J. S. Haas How The New Medicare Drug Benefit Could Affect Vulnerable Populations Health Aff., January 1, 2006; 25(1): 248 - 255. [Abstract] [Full Text] [PDF]

				M. Schlesinger Getting a Piece of the Pie Journal of Health Politics Policy and Law, August 1, 2005; 30(4): 555 - 562. [PDF]