Briefing Paper No. 26 April 4, 1996

Briefing Paper

The Establishment Strikes Back
Medical Savings Accounts and Adverse Selection

by Peter Ferrara

Peter Ferrara is general counsel for Americans for Tax Reform and an associate policy analyst at the Cato Institute.

Executive Summary

None of the proposed structural reforms of Medicare has come in for as much criticism as have medical savings accounts (MSAs), which are designed to return control over health care spending to the consumer, reduce costs, and avoid the rationing of care inevitable under either government-run health care systems or managed care.

As MSAs have become increasingly popular in Congress, opponents have tried to discredit the concept. The most significant criticism is that MSAs would appeal only to the healthy, leaving traditional Medicare to serve the sick. Such adverse selection would, critics conclude, result in large cost increases to the system.

However, recent studies purporting to show adverse selection have been seriously flawed. In fact, both actuarial studies and experience in the private sector show that MSAs will appeal to a wide cross section of the elderly, regardless of their health status. Despite recent attacks, MSAs remain the best hope for reforming the Medicare system.


There are fundamentally only two ways to control health costs. One is for a government bureaucracy or some deputized third party, such as a health maintenance organization (HMO), to ration health care for consumers. That was the essence of the Clinton health plan considered in 1993 and 1994. It was also the reason the plan was so soundly defeated: third-party rationing means that consumers lose ultimate control over their own health care.

The alternative is to grant consumers control over their own health care funds and give them incentives to control their own health costs. Medical savings accounts (MSAs) do precisely that. Exactly the opposite of third-party rationing, MSAs increase consumers' controlover their own health care. As a result, MSAs have proved enormously popular everywhere they have been tried or explained.

For those reasons, the Republican budget plan included MSAs in the Medicare reform legislation. Retirees would have been free to choose to have their health coverage provided by MSAs rather than Medicare. The budget plan also provided for MSAs for workers as an alternative to traditional insurance coverage.

Not surprisingly, MSAs came under furious attack by established health policy interests and institutions. The basic argument of those who opposed allowing Medicare beneficiaries to use MSAs was that MSAs would appeal to and attract only healthy retirees and workers, leaving the sick with Medicare or traditional forms of insurance. Because healthy Medicare beneficiaries currently cost the system relatively little, it was argued that they would take more money with them to fund their MSA coverage than they would have spent in the old insurance systems. As a result, MSAs would increase Medicare costs.

That argument was reflected in the Congressional Budget Office's analysis of the Medicare MSA provision, resulting in an estimate that MSAs would increase Medicare costs by $3.5 billion over seven years.[1] A study by the consulting and lobbying firm of Lewin-VHI advanced the same argument and contended that MSAs would increase Medicare's costs even more.[2]

The ultimate point of those critiques of MSAs is that providing consumers direct incentives to control their own health care and its costs is not workable. The only viable approach to controlling health costs would then be the means overwhelmingly, and rightly, rejected by the public when it killed the Clinton health plan in 1994.

Fortunately, however, that critique of MSAs is not valid. This study will show, in fact, that the adverse selection criticism is grievously wrong.

MSAs for Medicare

The Republican Medicare reform plan, approved by Congress but vetoed by President Clinton, would have allowed retirees to use their share of Medicare funds to buy health coverage in the private sector in place of Medicare. Besides MSAs, retirees could have chosen any of a full range of private alternative plans, including HMOs, other types of managed-care plans, preferred provider organizations, provider service networks (local doctors and hospitals organized to provide health coverage directly), traditional fee-for-service plans, and plans sold by associations such as the American Association of Retired Persons or by unions.

The share of Medicare funds the elderly could have withdrawn to pay for those private plans would have varied depending at least on age, geographic location, and health factors. So the older and sicker people would have gotten more and the younger and healthier would have gotten less, reflecting what they would each have had to pay for private coverage.

For those who chose MSAs, the Medicare funds would have paid for catastrophic insurance covering all expenses above a high deductible chosen by the retiree. The remaining share of Medicare funds for the retiree would have gone into a savings account to pay expenses below the deductible.

The following example MSA, which closely follows the congressional reform plan, will be used in this paper. By the year 2002 Medicare is scheduled to spend approximately $7,200 per beneficiary. Actuarial studies show that a catastrophic insurance policy with a deductible of $3,000 could be purchased by seniors for approximately $5,700.[3] Accordingly, $1,500 would be left in Medicare funds each year for deposit in a beneficiary's MSA. The individual's health care expenses would be paid first from the MSA. After $1,500 of expenses, the person would face another $1,500 in potential out-of-pocket expenses, until the $3,000 deductible was reached. Once the deductible was reached, the individual would pay nothing further. The next year Medicare would deposit an additional $1,500 into the account regardless of whether the $1,500 from the prior year had been used up.

Such an MSA plan would provide better benefits than Medicare:

MSAs would be highly effective in controlling costs. Since retirees could keep what they did not spend out of their MSAs, they would seek to avoid unnecessary services or treatments and look for care at lower cost. Perhaps more important, doctors and hospitals would respond to revitalized consumer concern about cost by competing to reduce costs and help patients preserve their MSA funds. That would result in far lower costs than under Medicare-under which neither patient nor doctor shows much concern for costs, since the government is paying the bills.

Experience proves the effectiveness of that kind of cost control strategy. Over 1,000 employers across the country have already adopted MSAs for their employees. The cost reductions they are experiencing are much greater than the reductions targeted for Medicare under the congressional budget. While the budget sought to reduce Medicare's annual growth from about 10 percent to 7 percent, employers with MSAs have reduced annual health cost growth of 15 to 20 percent to 0 percent and even less.[5] Actuaries estimate that MSAs reduce health care use by close to 30 percent.[6] That means that the cost of the MSA plan described above, with better benefits than offered by Medicare, would over time be kept within the budget-limited amounts retirees could withdraw from Medicare to purchase private plans.

CBO's Analytical Fallacies

The Congressional Budget Office, however, failed to appreciate the cost-reducing power of MSAs. The CBO estimated that the MSA option in the Medicare reform plan would actually have increased Medicare costs by $3.5 billion over seven years because of adverse selection.

The CBO's analysis was based on the idea that MSAs would be chosen primarily by lower cost healthy retirees, not higher cost sick retirees. In theory, that would not cause any problem because Medicare would just pay lower, risk-adjusted amounts to the private plans for the healthier retirees, reflecting their lower costs. If that were done properly, the amounts those retirees withdrew from Medicare to pay for MSA plans would be no greater, on average, than what Medicare would have spent on them, leaving Medicare no worse off.

But the CBO posited that Medicare would not be able to fully adjust its payments to the private plans for the lower cost of the healthiest retirees who would be choosing MSAs, because risk adjustment methodologies are not sufficiently sophisticated to fully account for all cost variances. As a result, the CBO concluded, the share of Medicare funds that the healthiest retirees would withdraw from Medicare to pay for private MSAs would probably be more than what Medicare would have spent on them if they had stayed in the program, resulting in higher overall costs for Medicare.

That analysis is wrong because MSAs provide better benefits for the sick than does Medicare. Consequently, the sick as well as the healthy are more likely to choose MSAs than Medicare. With both sick and healthy people opting out of Medicare for MSAs, the Medicare payments for MSA plans would average out, and Medicare would not face higher costs due to adverse selection. Instead, Medicare would save costs to the extent beneficiaries chose MSAs because, again, the Medicare payment amounts for MSAs would be limited to grow no faster than the budget targets. The sick would prefer the better benefits of MSAs for a number of reasons:

The sick would prefer the complete catastrophic coverage of the MSA plan to Medicare's lack of catastrophic coverage.

In contrast to all those reasons for the sick to choose MSAs over Medicare, there are no reasons why they would choose Medicare over MSAs. Consequently, sick as well as healthy retirees would be likely to choose MSAs over Medicare, so there would be no adverse selection problem.

Experience with MSAs confirms that analysis. More than 1,000 employers are already using MSAs, and everywhere a choice is allowed, the overwhelming majority of workers, the sicker as well as the healthier, choose MSAs. And those who become sick later on show no tendency to leave the MSAs. For example, Golden Rule Insurance Co. offers MSAs as well as a traditional fee-for-service plan to its 1,300 employees. Over 90 percent of its workers, the sick as well as the healthy, have chosen the MSA option, and there is no experience of those who become sick leaving the MSAs.[7]

But discussions with CBO staff and a letter from CBO director June O'Neill to Sen. Jon Kyl (R-Ariz.) revealed that the CBO did not consider any of the above factors in its analysis.[8] It did not consider that MSAs provide complete catastrophic coverage while Medicare does not. It did not consider that MSAs provide a maximum cap on out-of-pocket expenses while Medicare does not. It did not consider that MSAs pay for medical expenses not covered by Medicare. It did not consider that MSAs offer greater freedom of choice of doctors, hospitals, services, and treatments than does Medicare-probably the most powerful reason the sick would choose MSAs over Medicare. Moreover, the CBO did not research and was not aware of any of the actual experience with MSAs, which shows no adverse selection problem.

Instead, the CBO analysis of MSAs was based on the assumption, as stated in O'Neill's letter to Kyl, that "the average Medicare beneficiary could anticipate facing higher out-of-pocket costs under this option [MSAs] compared to either traditional Medicare or the other capitated [private] options."[9] Since in the CBO's view, those who get sick would have to pay more out of their own pockets with MSAs than with Medicare or other private options, the sick would not choose MSAs. Only the very healthiest Medicare beneficiaries who would be sure not to incur any out-of-pocket costs with MSAs would choose them. In fact, the CBO concluded that, for that reason, only the healthiest 1 to 2 percent of Medicare beneficiaries would choose MSAs.

But the analysis presented in this paper shows that MSAs would reduce, not increase, the out-of-pocket costs faced by retirees under Medicare:

There is no way the out-of-pocket costs borne by the elderly with MSAs would be greater than those they pay under Medicare.

The CBO's analysis, therefore, could not be more wrong or upside down. In reality, MSAs reduce out-of-pocket costs compared to Medicare, but the CBO estimated the impact of MSAs assuming that they would increase out-of-pocket costs. That and the failure to consider the factors and experience discussed above, which show that the sick would be likely to choose MSAs over Medicare, are the reasons why the CBO's estimates of the likely impact of MSAs are so grossly inaccurate.

The CBO's notion that only 1 to 2 percent of the elderly would choose MSAs is also seriously mistaken. Whenever workers have been able to choose MSAs, the overwhelming majority has done so.[10] That is because MSAs give workers and consumers much greater control over health care funds and services, allowing them to benefit directly from cost savings and to exercise greater control over their own health care. Because MSAs are a better system for the sick as well as the healthy, most people pick them. Most retirees, too, will end up choosing the MSA option in place of Medicare.

Finally, the CBO suggests that sick retirees would be more likely to choose HMOs than MSAs because HMOs offer lower out-of-pocket costs. But that, again, flies in the face of all logic and experience. Both in the employment context and in the experience with an already established HMO option for the elderly under Medicare, the reality is that the healthy tend to choose HMOs while the sick tend to flee. That is because with HMOs patients have to give up ultimate control over their own health care. For the healthy that is often not a major concern because they do not see doctors very often. They may consequently choose a lower cost HMO. But for the sick, not having a permanent doctor often becomes unbearable, and they will either leave HMOs or avoid them in the first place. MSAs are, in fact, most appealing to the sick because they grant patients more control over their health care than any other option, as the MSA funds are theirs to spend on whatever health care they choose without third-party interference.

The sick will tend to choose MSAs over any other fee-for-service option as well, because the other options will be so much more expensive. Other fee-for-service plans will have, not only the added costs of lower deductibles, but also much higher costs due to the lack of the MSA cost control incentives, or any other cost control system. They will consequently offer the elderly the same runaway costs as Medicare does.

Lewin's Fairy Tales

The same analytical fallacies, and more, were uncovered in the study of MSAs under Medicare by Lewin-VHI. That study, conducted for the National Committee to Preserve Social Security and Medicare, is interesting because it reveals even more clearly the analytical fallacies underlying the misestimation of the effects of MSAs.

Lewin's analysis was based on the proposition that only people who expect their health costs for the year to be less than the amount contributed to the MSA for the year could be expected to choose MSAs.[11] In other words, given a plan with a $3,000 deductible for catastrophic coverage and $1,500 contributed to the MSA each year, only people who expected to have health costs below $1,500 for the year would choose MSAs. Lewin assumed that anyone who expected to have health costs above the amount contributed to the MSA would not choose the MSA plan. So anyone who expected to have health costs above $1,500 for the year would not choose the MSA.

That analysis is transparently wrong to the point of silliness. At a minimum, the correct comparison is between the expected out-of-pocket costs under the MSA plan and the expected out-of-pocket costs under Medicare. Anyone who expected to spend less out-of-pocket under the MSA plan than under Medicare could be expected to choose the MSA plan, even if his health care costs would be greater than the amount contributed to the MSA for the year. Under the MSA plan described above, someone with $1,501 in health expenses for the year would spend $1 out-of-pocket. But that person would have spent several hundred dollars out-of-pocket under Medicare (due to the deductibles under Part A and Part B and the 20 percent coinsurance fee under Part B). Yet Lewin's analysis assumes that such a person would not financially benefit by choosing the MSA and would stay in Medicare.

As another example, take someone with $10,000 in yearly medical expenses, half of which would be covered by Medicare Part A and half by Part B.[12] Under the MSA plan described above, that person would have out-of-pocket costs of $1,500 for the year. Under Medicare, the person would have out-of-pocket costs of almost $2,000. Yet Lewin's analysis again assumes that all such sick people would not benefit financially from MSAs and would stay in Medicare.

Lewin's analysis quite clearly asks the wrong question, so it quite clearly gets the wrong answer. And it, again, completely overlooks every one of the reasons why MSAs would appeal to the sick.

The inconsistency of the Lewin analysis is shown best by the authors' concrete specification of how an MSA plan would work. They actually estimate that the amounts withdrawn from Medicare would finance an MSA plan with a $3,000 deductible on the catastrophic policy and $2,500 paid into the MSA each year. Yet their analysis concludes that only 3 percent of Medicare beneficiaries would choose such an MSA. The maximum out-of-pocket cost with such an MSA plan is $500 per year. The deductible for one hospital stay under Medicare is well over that. A retiree on Medicare with $2,100 in expenses under Part B would have to pay $500 out-of-pocket. With the MSA, the entire $2,100 would be covered by the MSA with no out-of-pocket expenses. Every retiree in America, and certainly all of the sickest retirees, would be better off with such an MSA than with Medicare.

The Real Effect of MSAs on Medicare

In contrast to the analyses by the CBO and Lewin, Milliman and Robertson, the nation's top actuarial firm, conducted a study for the National Center for Policy Analysis.[13] Milliman and Robertson examined a Medicare reform plan similar to the one adopted by Congress. They estimated that, were a full range of private options, including MSAs, available to the elderly, 50 percent of Medicare retirees would choose one of the private options in the first year, and 80 percent would do so by the seventh year.[14] That estimate was based on Milliman and Robertson's highly informed and experienced judgment in advising real-world insurance companies on their products. Such a result would reduce Medicare spending by almost $200 billion over seven years, Milliman and Robertson calculated. That would occur because the amounts that could be withdrawn from Medicare for the private options are limited to grow no faster than the budget targets.

Milliman and Robertson did not estimate how many people would choose each of the private alternatives. But their other estimates suggest that over half of those who choose the private alternatives to Medicare will choose MSAs. That estimate is based on both analysis and experience. MSAs offer consumers the most direct control over health care funds and services. Unlike any other alternatives, MSAs enable them to benefit directly from controlling health costs, by keeping the unspent funds in their MSAs. MSAs also allow consumers the greatest freedom of choice in health care providers and services, because consumers may spend the MSA funds as they choose without third-party interference, and because the effective cost controls of MSAs make minimally restricted fee-for-service catastrophic coverage financially feasible. MSAs are consequently the best choice for both the sick and the healthy. Moreover, wherever an MSA option has been allowed, it has been overwhelmingly chosen over the other options.


Established interests that dominate health coverage today fear competition from the newly emergent MSAs. Consequently, they have produced a blizzard of criticism to short-circuit the potential competition. While the criticism has been echoed by those ideologically opposed to markets, it cannot withstand scrutiny.

MSAs will inevitably be the dominant form of health coverage in the 21st century. Twenty years from now virtually all Americans will have some form of MSA, whether in a fee-for-service or a managed-care context. That is because MSAs are by far the most effective means of controlling costs and the only means consistent with maintaining, and even increasing, consumers' control over their own health care. The great health care debate of 1993-94 showed that the American people demand such individual control and will not allow the alternative-third-party rationing.


1. Congressional Budget Office director June O'Neill, letter to Sen. William V. Roth Jr (R-Del.), October 20, 1995

2. John F. Sheils, Gary J. Claxton, and Randall A. Haught, Changes in Medicare Program Spending under Alternative Medical Savings Account Models," Prepared by Lewin-VHI, Inc., for the National Committee to Preserve Social Security and Medicare, September 22, 1995.

3. Mark Litow, "Saving the Medicare System with Medical Savings Accounts," National Center for Policy Analysis, Dallas, Texas, Policy Briefing no. 199, September 1995. The study was conducted by Milliman and Robertson, the nation's top actuarial firm. While this study did not analyze specifically the reform plan passed by Congress, the data in the study show how the MSAs in the reform plan would work.

4. "Medicare: Hospital Insurance & Supplemental Medical Insurance," National Academy on Aging Fact Sheet, August 1995.

5. For examples of companies successfully using MSAs, see Peter J. Ferrara, "More Than a Theory: Medical Savings Accounts at Work," Cato Institute Policy Analysis no. 220, March 14, 1995.

6. Litlow.

7. The Golden Rule experience with MSAs is discussed in Peter J. Ferrara, "The Health Policy Debate: Options for Reform," National Center for Policy Analysis, Dallas, Texas, Policy Backgrounder no. 132, July 7, 1994, pp. 29-30; and Peter J. Ferrara, "Medical Savings Accounts: The Private Sector Already Has Them," National Center for Policy Analysis, Dallas, Texas, Brief Analysis no. 105, April 20, 1994.

8. Congressional Budget Office director June O'Neill, letter to Sen. Jon Kyl (R-Ariz.), November 8, 1995.

9. Ibid., p. 2.

10. For examples, see Ferrara, "More Than a Theory."

11. "In general, persons who find that the MSA contribution amount would be greater than what they expect to spend on health care during the year are persons who could potentially benefit from the program. Conversely, persons who, because of their health status, expect to spend more on health services than the MSA contribution amount would not expect to benefit under the program. Thus, the MSA model would attract healthy individuals who expect to spend little on health care while discouraging enrollment by individuals in poorer health who have high health care costs." Sheils et al., p. 2.

Later the study says, "We assumed that individuals would enter the MSA program only if their expected level of health spending is less than the MSA contribution amount that they could receive" (p. 3).

12. Medicare Part A covers hospitalization. Part B covers physician care, diagnostic tests, and other outpatient services.

13. Litow.

14. Estimate by Milliman and Robertson, cited in Peter J. Ferrara, "Medical Savings Accounts for Medicare: The Adverse Selection Issue," National Center for Policy Analysis, Dallas, Texas, Brief Analysis no. 183, October 17, 1995.

1996 The Cato Institute
Please send comments to webmaster