Statement of Mark V. Pauly, Ph.D., Professor, Health Care Systems,
Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania

Testimony Before the Subcommittee on Health
of the House Committee on Ways and Means

Hearing on the Nation's Uninsured

April 4, 2001

Thank you, Madame Chairwoman, and members of the committee for inviting me today. I am Mark Pauly, Professor of Health Care Systems and Economics in the Wharton School of the University of Pennsylvania. I am happy to be here today to discuss the results of my research and policy analysis. Much of the background for my remarks was published in a recent issue of Health Affairs and represents joint work with Bradley Herring.

We analyzed options for the design of refundable tax credits intended to assist people in buying health insurance. We focus on the most numerous population group among the uninsured: those who are not poor but have family incomes too low to allow everyone to afford health insurance. In our analysis we defined this target group as families with incomes between 125 percent and 300 percent of the poverty line; more than 40 percent of the uninsured fall into this category. We also offer some comments on the uninsured with lower family incomes.

There are three important characteristics of information about this group of uninsured people. First, there is general agreement that they are uninsured primarily because the premiums for insurance--either explicit premiums they pay directly or implicit premiums they (like most of us) pay by making a sacrifice in cash income to take a job that carries health insurance--are high relative to their incomes. The problem is "affordability," and there is no better solution to this problem than a subsidy that lowers the net premium for insurance. Hence critics of this approach who say it will be ineffective cannot at the same time maintain that the problem is lack of affordability. Second, people in this group can nevertheless afford to pay something for their insurance, just not the whole premium. Most people in this income bracket do obtain private insurance, and even the uninsured on average pay substantial amounts out of pocket for medical care, money they could better use for health insurance premiums. For them, even a partial credit--what some critics of the refundable credit approach have called "a ten foot rope for someone in a thirty foot hole"--is of considerable value because they do have some resources; they have some rope themselves, and the trick is to figure out how to tie the pieces together. Finally, as we emphasize at great length in the article, all estimates of the impact of various tax credit plans on insurance purchasing behavior (including our own) are fraught with uncertainty; there is a wide range of plausible values, and no valid way to narrow it. To us, this means that any tax credit plan should be designed to deal with uncertainty, not (as is sometimes the case) by trying to regulate every possible thing that could go wrong, but rather by designing policies that will work reasonably well (though not perfectly) no matter what happens, and by planning to make adjustments in the program as it is phased in and information begins to be generated.

The most obvious and most important design feature of any credit plan is how generous it will be to the target population. In this case, however, there is more than just the general observation that offering larger credits for a given insurance policy will get more people to take it. Our research suggested that there is a very pronounced "notch" or "threshold," below which credits have small effects and above which effects become much larger. For example, we estimate that a credit of half the premium for an average policy will reduce the number of uninsured by half, whereas a 25 percent credit will only affect a few people, primarily those who are not wageworkers. There are two reasons why the credit has to be moderately large. First, all wage workers would be eligible for a subsidy associated with exclusion of employer provided premiums if they chose to take a job at a firm that offered coverage. For people who choose a job that does not pay all or most the premium, the value they attach to insurance must fall short of the group premium by an amount greater than the exclusion subsidy. Moreover, if they are to use the credit for a nongroup policy, the insurance premium will be higher than that of the rejected group option, because the administrative cost associated with that individual insurance is higher. Take someone in the lowest marginal income tax bracket. The value of the exclusion is about 30 percent (combining the payroll tax and the income tax), and the loading for individual insurance is 15 to 20 percent higher than for group insurance. So someone in this situation would need a credit of at least 45 to 50 percent just to be as well off as they would have been after taking the option they rejected. However, a credit of 50 percent for a policy with a loading of 30 to 40 percent means in effect that one pays less for the premium than the coverage of out of pocket payments the person expected to get back: for the person of average risk, insurance is a no-lose proposition. (The availability of charity care may temper this motivation).

A second important design feature is whether the credit is a fixed dollar amount (e.g., $1500 for self-only coverage) or pays a proportional share of the premium. In the individual market, premiums vary to some extent with risk--primarily with age and location, though not with health levels if the person did not wait to become insured before becoming sick. A fixed dollar credit will equal the premium for lower risk young people, but become a smaller proportion of the premium for higher risk middle-aged people. So the tradeoff, for a given average per person subsidy, is between covering a large number of lower risks or a smaller number of higher risks. Our estimates suggest that, at credits at about the 50 percent level, the same amount will cover up to 20 percent more of the uninsured if made available as a fixed dollar credit, although this difference shrinks as the credit grows. A possible design, present in the plan suggested by President Bush during the campaign, is to define the credit as the lesser of a fixed dollar amount or a proportional credit. At the modest $1000 level he proposed, the program would effectively provide a fixed dollar credit to almost everyone who bought a comprehensive plan.

Another design feature is the specification of the policies qualified for the credit. At one extreme, the required policy might be very comprehensive--assuring adequate (if not excessive) access to care, but imposing on any buyer a substantial payment, large enough to discourage many from purchasing. At the other extreme, there might be minimal restrictions in terms of coverage and costs sharing, effectively requiring only that the premium be at least has large as the credit. Then there should be virtually universal take-up, but of a policy that many will regard as parsimonious. We think it also important that qualified policies could be purchased either from private insurers or from a publicly contracted or operated plan.

A third design feature is eligibility for the credit by income level. For both fiscal and administrative reasons, it seems sensible to offer large credits to lower income families, and smaller or zero credits to others (e.g., people with incomes above the median). The latter group currently are rather generously subsidized by the exclusion, so modest credits for them would rarely be taken up. But reducing credits with income does offer some disincentive to earning more income, and may be complex to administer. These observations suggest that any credit should not phase down too rapidly, and that perfect targeting by income might be compromised in favor of administrative simplicity.

The fourth design feature is whether the credit is offered to everyone who obtains a qualified insurance policy at a given income level, or whether those currently insured in an employment-based group plan should be ineligible. If the credit is fairly generous but is restricted to those not in groups, there will be an incentive to employers and employees to drop group coverage in order to claim a larger credit (as long as the qualified insurance is reasonably priced relative to group insurance); there will be "crowd out." The economically efficient policy here is somewhat counterintuitive: the best credit is a neutral one made available to all regardless of how they obtain coverage, and the higher budgetary cost for such a plan (relative to one targeted only to those not offering or taking employment-based coverage) is not a real cost to the economy, but only a tax reduction, and one that improves both equity and efficiency. It is, after all, manifestly unfair to offer a credit to someone who has neglected to obtain insurance while denying it to someone else at the same income level who already made the sacrifices needed to take the job that provided coverage. The correct economic view of a credit as a tax cut is to be distinguished from the erroneous "target efficiency" view that is often taken of this problem.

The final design feature is the form of administration, encompassing mechanisms for identifying who is eligible for credits, making them aware of the program, getting the credit to them in a way that minimizes cash flow problems when they buy insurance, and assuring that they do take advantage of the program for which they are eligible. For people who pay federal taxes, both income taxes and payroll taxes, a credit can be applied to those liabilities. Since such taxes are withheld periodically from wages, the net effect of such credits would be increase cash wages, thus furnishing the disposable income needed to pay any remaining premiums. Since all workers are required to make an estimate of their tax liability when they fill out a W-4 form for employment, the Treasury Department does have this information. In the interest of simplicity, it is probably better to offer credits of only one or two different dollar amounts to people within a given income range. Once the W-4 form is used to identify those eligible for credits, they need to be informed of the program. It should be easy for them to participate. Existing Medicaid and CHIP programs involve complex processes of eligibility determination that cause more than a third of eligibles to fail to apply. One simple device would be to offer workers not already covered annual "$1500 off" coupons, which could be redeemed by insurers or insureds for periodic credit payments. Any deviations in end-of-year income from the original estimated income could be adjusted as part of the income tax process. Private insurers would be expected to seek out people who have a substantial subsidy for the purchase of their product.

The most problematic feature of proposals to make credits available for private insurance is the current rather unimpressive state of the private individual insurance market in the United States. As already noted, the main problem in this market is that administrative costs are high. There is also some risk rating, though the presence of guaranteed renewability features and other protections required by the HIPAA law mean that few in this market pay higher premiums because of chronic illness. One reason why the product is expensive is that it is bought only by a small fraction of the overall insurance market, only by about 6 percent of all private insurance purchasers. The small scale which makes the breakeven premium high and the need for substantial selling efforts for a costly product would both be much changed if a large-scale program of credits were available. If 18 million new buyers, armed with substantial credits, entered this market, it is very likely that product quality would improve, selling costs would fall, and such risk screening as there is would diminish greatly. Putting in place a high-risk pool should allow the few high risks who fall through the cracks to be supported.

What do these issues imply for policy design? The most obvious need is to have some serious considerations of the tradeoffs just outlined--between the size of the credit and the extent of effect, between covering many low risks or fewer higher ones, between being fair in treatment of employment based insurance and treating credits as government expenditures rather than taxes, and between requiring comprehensive but unaffordable coverage and incomplete but affordable insurance. Our article provides some empirical estimates to help with these tradeoffs, but the data do not uniquely anoint a single best plan, so that some policy debate and decision making is needed.

My own preferences in this matter are to suggest an adequately funded, minimally restrictive credit plan and to be prepared to learn from experience with such a program. In particular, I would suggest credits on the order of $1500 for self-only coverage and $3500 for family coverage, and permit those credits to be used for health insurance offered either by a public or a private insurer. These credits could be made available in the form of redeemable coupons, with reconciliation with total income at the end of the tax year. All people with incomes in the target range would be eligible for credits, even if they obtained employer-paid coverage, but the value of the tax exclusion would be netted out of the credit in the latter case. It would also be desirable to offer larger credits, equal to the premium for a comprehensive policy, for people with incomes below 125 percent of poverty; they also could use these credits either for a CHIP, Medicaid, or government contracted plan, or for a private plan of equivalent coverage.

Before fully opening the private market to the poor, it would probably be best to wait to see if the improvements in functioning we expect that follow from substantial credits to lower middle income people do materialize. It would also be desirable to encourage (or at least not obstruct) the emergence of alternative group purchasing arrangements for individuals and small employers, such as "Health Marts" and the like.

The most fundamental point, however, is that, after years of talking about helping the uninsured, tax credits provide a way to do something. They furnish a vehicle that can be kept free of regulatory restrictions, that does not require public decisions about exactly what insurance people should have, and that gives people the wherewithal to afford the health insurance that they feel is best. We ought not to make the unattainable perfect the enemy of the feasible good.