Do high-deductible health insurance policies, coupled with tax preferred Health Savings Accounts (HSAs) championed by the Bush Administration and a number of health policy analysts, actually reduce rather than increase cost sharing for many groups? Dahlia K. Remler and Sherry A. Glied made this case in a Health Affairs paper which was quickly picked up by the advocates of HSAs to support the argument that HSAs will be beneficial to the insured.

The trouble with their argument is that it rests largely on one illustrative case, namely, for a family in a high marginal tax bracket and for an HSA with a relatively low annual deductible. As is explained more fully below, if one assumes lower marginal tax rates or higher deductibles, or both, Remler and Glied’s conclusions no longer hold as a general proposition.

Remler and Glied’s illustration. To make their case, Remler and Glied use as a baseline case the “prototypical comprehensive plan today” with a deductible of $350, a coinsurance rate of 20 percent, and an out-of-pocket maximum of $1,800, for an individual. Out-of-pocket spending under the comprehensive baseline plan is not tax deductible. They then compare after-tax out of pocket spending under this baseline plan with after-tax out-of-pocket spending under an HSA plan that has an annual deductible of $2,500 and zero coinsurance. Under the HSA plan, all out-of-pocket spending becomes tax-deductible. The illustration, however, is made only for a high-income individual assumed to face a marginal income tax rate of 40 percent. Such an individual pays out of pocket, on an after-tax basis, only 60 cents for every dollar he or she is billed by health care providers within the deductible range. Exhibit 1 is a slightly modified version of the author’s Exhibit 3.

High Deductible Plan With HSA Versus Comprehensive Plan:
Relationship between Medical Spending for an Individual
And Out-of-Pocket Spending (Deductible for the HSA is
$2,500 with 0% coinsurance.)

Exhibit 1

In Exhibit 3 of their paper, Remler and Glied present only the bottom two lines of Exhibit 1 above. These lines show that, under the authors’ assumptions, for total annual medical bills below $700 or above $6,000, the HSA actually saddles patients with less cost sharing in dollar terms than does the conventional, comprehensive baseline plan. It is the basis for the authors’ conclusion that “current HSA/high-deductible plans represent an increase in cost sharing for those in the middle of the spending distribution, but a decrease for those at the very low end and in much of the higher end.”

Alternative Scenarios

This generalization by the authors is much too broad, as it is valid only for the specific parameters assumed by the authors.

Lower tax rates: As is shown in the top of the three curves in Exhibit 1, the authors’ broad conclusion does not hold for individuals with incomes too low to require payment of federal income taxes. Such individuals pay only payroll taxes of 6.2 percent for Social Security and 1.45 percent for Medicare, which means that they pay out of pocket 92.35 cents for every dollar they are billed by health care providers within the deductible range. Because of these high after-tax payments, such low income individuals never pay less out of pocket under the HSA than they would under the comprehensive policy, as is shown by the top line in Exhibit 1.

Higher deductibles: Cost-sharing under the HSA relative to conventional policies is much higher still if the annual deductible exceeds $2,500. As Remler and Glied point out in their Exhibit 1, currently up to $5,000 a year may be deposited by individuals into tax-preferred HSAs, coupled with insurance policies with deductibles up to $5,000. The well-known “farmers’ market” for individually sold health insurance products, for example, features numerous insurance products with annual deductibles ranging from $5,000 per individual. Typically, these policies exclude maternity care, place limits on a number of other services and are medically underwritten.

As is seen in Exhibit 2 below, for such policies, for bills exceeding the non-tax-preferred deductible of $350 under the assumed conventional policy, the annual dollar amount of out-of-pocket payments under HSAs are higher for both, low-income and high-income families, relative to the conventional policy.

High Deductible Plan With HSA Versus Comprehensive Plan:
Relationship between Medical Spending for an Individual
and Out-of-Pocket Spending (Deductible for the HSA is
$5,000 with 0% coinsurance.)

Exhibit 2

Family policies: Total annual out-of-pocket spending under HSAs for families, rather than individuals, can be as high as $10,500 under current law. Not surprisingly, the website therefore features numerous health insurance products for families with annual deductibles of $10,000 or combinations of deductibles and health insurance with a maximum out-of-pocket limit of $10,000. Exhibits 3 illustrates cost sharing under these policies in a different format, showing the fraction of the total annual medical bill paid out of pocket by high- and low-income families.

Out-of-Pocket Spending under a High-Deductible Insurance
Plan for a Family With an Annual Deductible of $10,000
and with zero coinsurance.)

Exhibit 3

Exhibits 2 and 3 illustrate clearly just how regressive the tax-preference accorded the HSAs can be. In effect, the construct makes the after-tax cost of health care much cheaper in absolute dollars for high-income families than for low-income families. One must wonder whether the implied distributive ethic conforms to prevailing American notions of fairness.1

Deductibles vs. Coinsurance

Remler and Glied correctly point out that what they call the “marginal price” of health care – that is, the incremental out-of-pocket payments due at any level of total annual medical spending — depends on the mix of deductibles and coinsurance in the HSA products. In their illustration, they posit a mix of deductible and coinsurance for the conventional insurance policy, but a deductible only for the HSA product, which means that, once the deductible is met, patients enjoy first-dollar coverage with zero out-of-pocket spending.

If the goal of the HSA/high-deductible construct is to force patients to have “[fiscal] skin in the game” – to use that dubious but common phrase — so that patients think more carefully about the benefits and costs of proposed treatments, then an argument can be made for lowering the deductibles and adding coinsurance instead.

For example, instead of a deductible of $2,500 with zero coinsurance, which makes health care free to the patient for bills exceeding $2,500, one might use a $1,000 deductible with 30 percent coinsurance and a maximum out-of-pocket stop loss of $2,500. Under that regime, health care would become free to patients only for annual medical bills exceeding $6,000. At the extreme, one might have zero deductibles and only coinsurance of, say 25 percent up to a maximum risk exposure of $2,500 a year, which would make health care free to patients only for annual bills exceeding $10,000.2

What particular mixture of deductible and coinsurance is preferable depends on (a) the maximum out-of-pocket spending one believes families ought to bear per year and (b) the size of the total annual medical bill beyond which one no longer wishes to visit cost sharing on patients. It is a matter worth further study and thought.

Implication for Health Policy

Those who believe that our health system will never be efficient until patients feel serious fiscal pain as they suffer physical pain, so to speak, will find the preceding analysis reassuring. A Bloomberg wire service story reacting to the Remler-Glied article, for example, treated the low cost sharing reported by these authors as a newsworthy discovery of a “flaw” in HSA design, since if less is spent out-of-pocket, incentives to conserve costs under HSA plans must be weaker.

Rationing by income class: As my analysis illustrates, however, the very high degree of out-of-pocket spending visited on individuals and families by many of the HSA products now on the market guarantees that especially low-income Americans are very likely to tighten their belts und forego health care they would have wanted under the more comprehensive, conventional insurance products. This self-rationing by price is doubly guaranteed, because the tax-preference accorded to HSAs effectively makes health care more expensive for low-income persons than for high-income persons.

On the other hand, as Remler and Glied, and the present analysis, show clearly as well, the HSA/high-deductible construct is not likely to alter significantly the health-care behavior of high income people – for example, the readers of the Bloomberg news wire. For one, high-income individuals and families will be able to absorb high deductibles with relative ease into their higher incomes. Furthermore, as noted, the tax code actually makes health care cheaper for high-income families, relative to Americans with lower incomes. With the government picking up close to half of the health care cost of high income people under HSAs, it can be doubted that self-rationing of health care by price will play any significant role among such families if they were forced to switch from the comprehensive baseline policy to an HSA product. It is fair to wonder whether the advocates of HSAs, who usually live in high income brackets, are aware of this asymmetric impact of HSAs on the rationing of health care by price and ability to pay.

These are ethical policy issues that economists can identify but on which they cannot offer normative dicta. Policy makers, on the other hand, might wish to explore these issues further, before shifting the nation wholesale toward the HSA/high-deductible health insurance, as this product is currently designed.

Making the HSA/high-deductible construct less regressive: Should policymakers want to make the HSA/high-deductible construct considerably less regressive, they could do so by modifying three parameters of that construct.

(1) The maximum out-of-pocket stop loss visited on individuals and families could be made to rise with disposable income (which raises the delicate question whether corporate executives who are paid annually in the millions of dollars should have any health insurance at all).

(2) Instead of making deposits into HSAs tax deductible, any American individual or family, regardless of taxable-income level, could be granted a refundable tax credit of, say, 30 percent.

(3) Chronically ill individuals and families could be granted risk-adjusted subsidies, so that the chronically ill are not forced to bear the entire maximum risk exposure under their health insurance out of their own resources.

With these modifications, the HSA/high-deductible construct probably would meet much less resistance than it has hitherto.


1To be sure, health insurance procured at the place of work includes a tax preference as well, as any premiums paid by employers are not taxable income to the employee but a tax-deductible business expense to the employer. Many economists consider that tax preference regressive as well, although the regressivity in that context is not nearly as certain as it is under the HSA construct. It is so because employers typically procure health insurance for their employees on a group basis, paying the insurer one lump sum premium for all employees. How that lumps sum is then shifted backwards into the take-home pay of individual workers is not clear, even to economists.

2Harvard economist Martin Feldstein, for example, endorsed that approach at a luncheon address during the American Economic Association meeting of January, 2006.