You are here

Administration Defense of Health Savings Accounts Rests on Misleading Use of Statistics

February 16, 2006

To encourage wider use of Health Savings Accounts (HSAs), tax-free individual accounts for health-related expenses that must be coupled with a high-deductible insurance policy, the Administration is proposing significant new HSA-related tax breaks that it estimates would cost $156 billion over ten years. Several important concerns have been raised about HSAs: most notably, that they are most attractive to people who are in better health and have higher incomes and that they would undermine employer-sponsored health coverage. In recent days, Administration officials, including the President, have argued that these concerns are belied by actual experience with HSAs over the past two years. This analysis examines the claims made by these officials and HSA proponents outside the Administration and explains that they rest on the misleading use of statistics.[1]

Claim #1: HSAs are not disproportionately attractive to high-income households.

Allan Hubbard, director of the White House's National Economic Council, stated at a press briefing on February 1 that 40 percent of HSA enrollees have incomes below $50,000. Treasury Secretary Snow repeated this statement at a Senate Finance Committee hearing on February 7, and President Bush repeated it in a speech in Ohio on February 15. Hubbard, Snow, and the President implied that this refutes claims that HSAs are disproportionately attractive to the affluent. But they did not fully or accurately represent the data.

First, data are available from three recent surveys of HSA enrollees. These surveys found, respectively, that 29 percent, 33 percent, and a little over 40 percent of HSA enrollees had incomes below $50,000. Mr. Hubbard and Secretary Snow ignored the two surveys with the lower figures and presented the 40 percent figure (from a survey by an online insurance broker) as though it were the only one available.

Second, and more important, the survey cited by Hubbard and Snow was restricted to HSA enrollees in the individual health insurance market, who tend to have lower incomes than HSA enrollees who have employer-based coverage. Excluding the latter group makes HSA enrollees appear to have lower incomes than is actually the case.

People who purchase insurance in the individual market, where the administrative costs are higher and premiums can spike from one year to the next if you become sick, are primarily lower- and moderate-income people who are unable to obtain coverage through an employer. Leonard Burman, co-director of the Urban Institute-Brookings Institution Tax Policy Center, has written that, "Even for healthy people, nongroup insurance tends to be much more expensive than employment-based health insurance because of high costs for administration, marketing, and underwriting."[2]

Census data show that in 2003, beforeHSAs were available, insured households with incomes below $25,000 were nearly twice as likelyas households with incomes above $75,000 to have obtained coverage through the individual market. Since it makes sense for people who already are purchasing high-deductible coverage in the individual market to establish HSAs, a survey of HSA users that is limited to the individual insurance market will necessarily show them to have lower incomes than if HSA users with employer-based coverage were included as well.

In addition, the survey cited by Hubbard, Snow, and the President may suggest the opposite of what they claim. The survey shows that 40 percent of HSA enrollees have incomes below $50,000. But Census data show that in 2004, the percentage of non-elderly individuals with private insurance obtained in the individual market who had incomes below $50,000 was even larger: 52 percent. This suggests that a disproportionate proportion of HSA enrollees do, in fact, have higher incomes. As discussed below, other HSA data indicate this is the case.

Most important of all, the current survey data do not answer the critical question in this area -- whether healthy, higher-income workers with employer-based coverage will shift in disproportionate numbers to HSAs and high-deductible policies if their employers begin to offer such packages as an option. If that process (known as "adverse selection") were to occur, the workers left in comprehensive employer-sponsored health plans would, on average, be less healthy. That, in turn, could cause premiums for comprehensive coverage to spiral upward.

There can be no question that the tax benefits of HSAs are tilted toward high-income households and would be tilted still more strongly under the Administration's HSA proposals. High-income households benefit much more from HSA tax deductions than lower-income households because they are in higher tax brackets. Also, high-income households can afford to contribute much larger amounts to HSAs than people of more limited means. (For example, only higher-income households generally would be able to take full advantage of the President's proposal to raise the annual HSA contribution limit to $5,250 for individuals and $10,500 for couples.) Moreover, for affluent individuals who do not expect to incur significant health-care costs, HSAs provide unprecedented tax-sheltering opportunities: they are the only savings accounts that feature both tax-deductible deposits and tax-free withdrawals.

Not surprisingly, other surveys not cited by Administration officials find early evidence of an HSA tilt toward more-affluent households.

  • A survey recently conducted by the Employee Benefit Research Institute and the Commonwealth Fund found that nine percent of individuals with HSAs (or HRAs, a similar type of account) and high-deductible policies had incomes of $150,000 or more, while only four percent of people with comprehensive health insurance did.

  • The Blue Cross and Blue Shield Association has presented a similar finding with respect to people with incomes over $100,000.

  • A new survey by the Government Accountability Office (GAO) finds that 43 percent of federal employees who receive insurance through the Federal Employee Health Benefits Program (FEHBP) and are enrolled in an HSA had incomes over $75,000. That is nearly twice the percentage of enrollees in all other FEHBP plans who had incomes over $75,000. The GAO reported that the HSA enrollees "had consistently higher incomes across all age groups."

Claim #2: HSAs help reduce the number of uninsured Americans.

Mr. Hubbard also said in his February 1 press briefing that 37 percent of people with HSAs were previously uninsured. Several days later, in a congressional hearing, Secretary of Health and Human Services Michael Levitt rounded this figure up to 40 percent.

Here, too, Administration officials ignored the results from other surveys that found a smaller percentage of HSA users were previously uninsured. Mr. Hubbard cited only the results from a survey, conducted by America's Health Insurance Plans, that produced the higher number.

In any event, the statistic that 37 percent of HSA enrollees were previously uninsured does not show that the Administration's HSA proposals would reduce the ranks of the uninsured, because this figure is from data that are limited to HSA enrollees in the individual insurance market.

The individual market consists heavily of people who are unable to secure, or have lost, employer-based coverage and consequently have been forced into the individual market after a period of being uninsured. A survey of people who have purchased any type of policy in the individual market would show that many of them had previously been uninsured. The fact that 37 percent of HSA enrollees in the individual market previously were uninsured tells us little about whether HSAs are reducing the ranks of the uninsured.

Moreover, if HSA use grows dramatically, as the Administration hopes, the percentage of HSA users who previously were uninsured will fall sharply, since many of the new HSA enrollees will be workers who have shifted from comprehensive employer-based coverage to high-deductible coverage -- and thus previously were insured. In fact, a Blue Cross/Blue Shield study that covered bothemployer-based coverage and the individual market found that only 12 percent of individuals who purchase high-deductible plans that qualify for a HSA previously were uninsured.

More importantly, while the availability of HSAs (including the Administration's proposed new tax breaks) should enable some uninsured people to afford coverage, the Administration's proposals also would induce some employers -- especially small-business owners -- to drop coverage (or not to offer it in the first place), since the proposals would eliminate all of the tax advantages that employer-based insurance now has relative to coverage purchased in the individual market. That would expand the ranks of the uninsured, because some of the individuals who lost employer-based coverage -- especially those who are sicker -- would be unable to obtain affordable coverage in the individual market. People with serious medical conditions generally are refused coverage in the individual market or are offered policies only at extremely high cost.

If more people lost coverage (because their employer ceased to offer it) than gained coverage (because of the new tax breaks), the net effect would be to increase the ranks of the uninsured. The 37-percent statistic cited by the White House sheds no light on the question of whether the President's proposals would make a dent in the size of the uninsured population or, conversely, cause it to rise still higher. But a new study by M.I.T. economist Jon Gruber, one of the nation's leading health economists, estimates that the President's proposals would cause a net increase in the uninsured population of 600,000 people. The proposed HSA tax breaks would cause 3.8 million uninsured people to gain coverage, according to Gruber's analysis, but also would cause 4.4 million people who previously had employer-sponsored coverage to wind up uninsured.[3]

Moreover, even if as many people gained coverage as lost coverage as a result of the Administration's proposals, the overall effect would be negative, because the people who would gain coverage would tend to need it less than the people who would lose coverage. The people most likely to gain coverage as a result of the Administration's proposals are healthy people who would be able, with the help of the new tax breaks, to obtain affordable coverage in the individual market. In contrast, the people most likely to become uninsured as a result of the proposals would be less-healthy employees in small businesses that dropped coverage, since less-healthy people have the most difficult time obtaining affordable coverage in the individual market.

Claim #3: HSAs are not disproportionately attractive to healthy individuals.

On this issue, the Administration officials cited above have been silent. But some HSA proponents outside the Administration have cited a national Blue Cross/Blue Shield survey that found that the average self-reported health status of people who are enrolled through the individual market in high-deductible health insurance plans (and are eligible for HSAs) was similar to the self-reported health status of people with traditional comprehensive coverage purchased in the individual market. These HSA supporters have cited this survey as showing that HSAs do not disproportionately attract healthier people.

In fact, this comparison says little about whether HSAs attract healthier people because it was limited to people who have coverage secured in the individual market. As noted, the individual market is accessible primarily to healthier individuals, since insurers can (and generally do) decline to offer coverage or charge much higher premiums to less-healthy people. Thus, both the people who have an HSA in the individual market and the people who purchase a more comprehensive plan in the individual market are likely to be relatively healthy and to have similar average health status.

The Blue Cross/Blue Shield survey did not use control groups or employ other standard statistical methodologies to ensure it could obtain an unbiased answer to the question: "Given a choice, will healthy people be more likely than less-healthy people to choose HSAs?" To answer that question would require comparing the choices of healthier and sicker individuals who are offered the sameset of health-care options. For example, if an employer offers both comprehensive insurance and a high-deductible plan tied to an HSA, will healthier employees migrate to the HSA while less-healthy employees remain in comprehensive coverage?

Past studies have consistently found that plans with higher deductibles tend to attract a disproportionate number of healthier people. (This is not surprising, given that people with significant medical conditions and high medical costs would end up paying large out-of-pocket medical costs under high-deductible plans.) Preliminary studies indicate that this seems to be occurring with high-deductible policies tied to HSAs:

  • The survey conducted by the Employee Benefit Research Institute and the Commonwealth Fund, discussed above, indicates that high-deductible plans attached to HSAs or HRAs attract a healthier population, on average, than comprehensive coverage does.
  • Similar findings were reported by a study by Blue Cross and Blue Shield of Minnesota, as well as two recent GAO studies. For example, a GAO study found that federal employees enrolled in the Federal Employee Health Benefits Program who choose high-deductible plans and HSAs are younger on average -- and thus are likely to be in better average health -- than those who choose other coverage.

Serious concerns have been raised about HSAs, and the experience of HSAs thus far has not dispelled those concerns. It would be unwise to enact additional costly, deficit-financed tax cuts aimed at promoting HSAs until these concerns are resolved.

End Notes:

[1] For more detailed discussion of these issues, see Edwin Park and Robert Greenstein, "Latest Enrollment Data Still Fail to Dispel Concerns About Health Savings Accounts," Center on Budget and Policy Priorities, Revised January 30, 2006.

[2] Leonard E. Burman, "President Bush's Costly and Counterproductive Health Care Tax Incentives," Tax Notes, Feb. 13, 2006.

[3] Jonathan Gruber, "The Cost and Coverage Impact of the President's Health Insurance Budget Proposals," Center on Budget and Policy Priorities, February 15, 2006,