Fortunately, there is a better way. We can have a system that provides a reasonable form of universal coverage for everyone and does so without spending more money and without intrusive and unenforceable government mandates. Here's how.

Characteristic No. 1: We should subsidize those who insure and penalize those who do not.

To the advocates of mandates, we can always ask the question: What are you going to do with people who disobey the mandate? As a practical matter, no one is suggesting that we put them in jail. So we are left with imposing a financial penalty (e.g., a fine). But a system that fines people who are uninsured, ipso facto is a system that subsidizes those who insure - the subsidy being the absence of the fine.

"We should subsidize those who insure and penalize those who do not."

For example, under the current system families who obtain insurance through an employer obtain a tax subsidy worth about $1,155, on the average.10 Since an uninsured family with an average income doesn't get this subsidy, the uninsured family will pay about $1,155 more in taxes than families that have employer-provided insurance. So instead of describing our current system as one that subsidizes employer-provided insurance, we could, with equal validity, describe it as one that penalizes the lack of employer-provided insurance.

We can describe any incentive system in one of two ways: (1) as a system that grants subsidies to those who insure and withholds them from those who do not; or (2) as a system that penalizes the uninsured and refrains from penalizing the insured. Either description is valid, since a subsidy is simply the mirror image of a penalty.

Characteristic No. 2: The subsidy/penalty should equal the value society places on insuring individuals, at the margin.

"An ideal system would have 10 characteristics."

We should decide how much we care (in money terms) whether a person is insured and that should determine the size of the subsidy/penalty.11 Any other policy would be indefensible and absurd. It would entail spending too much on subsidies and collecting too much in fines, or vice versa. Under an ideal system: [See Sidebar]

• We should never pay more for (subsidize) good behavior than the good behavior's benefit to us, and we should never collect more from (penalize) bad behavior than its costs to us.

• Conversely, we should never pay less for good behavior than its benefit or penalize bad behavior less than its cost.

"Those with the highest incomes would get the largest subsidies under current law."

Current policy violates this principle in several ways. Although the average tax subsidy is worth about $1,155 per family, households earning more than $100,000 per year receive, on the average, $2,638 per year in subsidies. By contrast, those earning between $20,000 and $30,000 receive only $599. [See Figure V] One reason is that those earning higher incomes are in higher tax brackets. For example, a family in the 40 percent tax bracket gets a subsidy of 40 cents for every dollar spent on their health insurance. By contrast, a family in the 15 percent bracket gets a subsidy of only 15 cents on the dollar.

A uniform subsidy would offer the same tax reduction to everyone who obtains private insurance, and that subsidy should reflect the value our society places on having one more person insured. But what is that value?

Characteristic No. 3: The revealed social value of insurance is the amount we spend on free care for the uninsured.

How do we know how much it's worth collectively for a given individual to insure? An empirically verifiable number is at hand, so long as we're willing to accept the political system as dispositive. It's the amount we expect to spend (from public and private sources) on free care for that person when he or she is uninsured.

To continue with the Texas example, if society is spending $1,000 per year on free care for the uninsured, on the average, we should be willing to offer $1,000 (say, in the form of a tax credit) to everyone who obtains private insurance. Failure to subsidize private insurance as generously as we subsidize free care encourages people to choose the latter over the former.

"The uninsured are not really uninsured -- they are participating in a different system."

One reason this principle is not generally understood is that many people think the uninsured are uncared for. In fact, they are not. They are simply participating in a different kind of health care system. For example, uninsured adults in Dallas County typically seek health care through the emergency room at Parkland Hospital. Uninsured children are typically treated next door, in the emergency room of Children's Medical Center.

Think of the system that provides these services as "safety net insurance," and note that reliance on the safety net is not as valuable to patients as ordinary private insurance, other things equal. The privately insured patient has more choices of doctors and hospital facilities. Further, safety net care is probably much less efficient (e.g., using emergency rooms to provide care that is more economical in a free-standing clinic). As a result, per dollar spent the privately insured patient probably gets more care and better care.

Characteristic No. 4: The penalties paid by the uninsured should be used to compensate those who provide free care to the uninsured.

Characteristic No. 3 furnishes the basis for answering an important question: what should be done with the penalties collected from people who choose to remain uninsured? The answer: the funds should be used to compensate providers who give free care to the uninsured - no more and no less.12

As noted above, under the current system the uninsured pay higher taxes because they do not enjoy the tax relief given to those who have employer-provided insurance. These higher taxes are a "fine" for being uninsured. The problem is the extra taxes paid are simply lumped in with other revenues collected by the U.S. Treasury in Washington, D.C., while the expense of delivering free care falls to local doctors and hospitals.

Under an ideal system, the government would offer every individual a subsidy. If the individual obtained private insurance, the subsidy would be realized in the form of lower taxes (say, in the form of a tax credit). Alternatively, if the individual chose to be uninsured, the subsidy would be sent to a safety net agency in the community where the individual lives. [See Figure VIa]

One way to think of such an arrangement is to see it as a system under which the uninsured as a group pay for their own free care. That is, in the very act of turning down a tax credit (by choosing not to insure) uninsured individuals would pay extra taxes exactly equal to the average amount of free care given annually to the uninsured.13 [See Figure VIb]

Characteristic No. 5: The subsidies for the insured should, at the margin, be funded by reducing spending on free care for the uninsured.

"Subsidies for insurance should be funded by reducing spending on free care."

Point three also furnishes the basis for answering a related question: how should we fund the subsidies for those who choose to move from being uninsured to insured? The answer: at the margin, the subsidy should be funded by the reduction in expected free care that person would have consumed if uninsured - no more and no less.

"The federal government should make $1,000 available to every individual -- either in the form of a tax credit (insured) or a grant to a local safety net (uninsured)."

Suppose everyone in Dallas County chose to obtain private insurance, relying, say, on a refundable $1,000 federal income tax credit to pay the premiums. As a result, Dallas County no longer would need to spend $1,000 per person on the previously uninsured. So all of the money that previously funded safety net medical care could be used to fund the private insurance premiums [See Figure VIc].

How could this scheme be implemented? Since much of the safety net expenditure already consists of federal funds, the federal government could use its share to fund private insurance tax credits instead. For the remainder, the federal government could reduce block grants to Texas for Medicaid and other programs.

"Individuals who are uninsured would pay $1,000 more in taxes to find a local safety net."

One way to think about this arrangement is to see it as a system in which people who leave the social safety net and obtain private insurance actually furnish the funding needed to pay their private insurance premiums - at least at the margin. They do this by allowing public authorities to reduce safety net spending by an amount exactly equal to the private insurance tax subsidy.14

Characteristic No. 6: Subsidies for being insured should be independent of how the insurance is purchased.

"Individuals who decide to cease being uninsured would get a $1,000 tax credit funded by reducing the budget of a local safety net."

The American health care system is largely an employer-based system, in which more than 90 percent of people with health insurance obtain it from an employer. In recent years many have questioned the wisdom of having employers choose health plans for their employees. Increasingly, there is interest in a system of personal and portable insurance in which individuals take their insurance with them as they travel from job to job and employers make defined-contribution premium payments to the plans their employees choose.

These issues should be resolved in the marketplace, rather than by the tax-writing committees of the U.S. Congress.

"The role of the employer should be determined in the marketplace."

Figure VII shows that a typical middle-income family with employer-provided coverage gets a tax subsidy equal to about half the cost of insurance. By contrast, families who purchase their own insurance get virtually no relief under the tax law. An ideal system would give the same tax relief, regardless of how the insurance is purchased [see Figure VIII]. If the playing field were level under tax law, the role of the employer would be determined through competition and choice in the marketplace.

Characteristic No. 7: The optimal number of uninsured is not zero.

"An ideal system would give the same tax relief, regardless of how the insurance is purchased."

The goal of health insurance reform is not to get everyone insured. (Indeed, everyone is already in a loose sense insured.) Instead, the goal is to reach a point at which we are socially indifferent about whether one more person obtains private insurance as an alternative to relying on the social safety net. That is the point at which the marginal cost (in terms of subsidy) to the remaining members of society of the last person we induce to insure is equal to the marginal benefit to the remaining members of society (in terms of the reduction in cost of free care). Once we satisfy this condition, it follows that the number of people who remain uninsured is optimal, and that number is not zero.

"Individual purchase and employer purchase should be on a level playing field under the tax law."

This is achieved by taking the average amount spent on free care and making it available for the purchase of private insurance. In our example, the government guarantees that $1,000 is available, depending on the choice of insurance system. From a policy perspective, we are indifferent about the choice people make.

Characteristic No. 8: The principles of reform apply with equal force to all citizens, regardless of income.

"The subsidy for being insured should be independant of income."

None of the first six points is in any way dependent for its validity on the income of the person who elects to be insured or uninsured. As a practical matter, one could argue that the high-income uninsured are likely to pay more out-of-pocket (get less free care) than the low-income uninsured, and therefore their subsidy/penalty should be smaller. Against that is the observation that high-income people (because of greater sophistication) are more adept overall at spending other people's money once they enter the health care system.

Waiving these considerations, a $100,000-a-year family can generate hospital bills it cannot pay almost as easily as a $30,000-a-year family. For that reason, our social interest in whether someone is insured is largely independent of income. For this reason as well as practical considerations, the tax credit should be independent of income.15

Characteristic No. 9: Health insurance subsidies need not add to budgetary outlays.

A common misconception is that health insurance reform costs money. For example, if health insurance for 40 million people costs $1,000 a person, some conclude that the government would need to spend an additional $40 billion a year to get the job done. What this conclusion overlooks is that we are already spending $40 billion or more on free care for the uninsured, and if all 40 million uninsured suddenly became insured they would - in that act - free up the $40 billion from the social safety net.

At more than one trillion dollars a year, there is no reason to believe our health care system is spending too little money. To the contrary, attempting to insure the uninsured by spending more money would have the perverse effect of contributing to health care inflation. As Gene Steuerle has shown, we can simply make some portion of people's tax liability contingent on proof of insurance.16 Getting all the incentives right may involve shifting around a lot of money - i.e., reducing subsidies that are currently too large and increasing subsidies that are too small. But it need not add to budgetary outlays. What follows are some possible ways of implementing this idea.

"There is no reason to believe our health care system is spending too little money."

Tax Subsidies for Middle-Income Taxpayers. For families who already pay substantial federal income taxes, the trick is to make some portion of tax liability contingent on proof of insurance. For example, the child credit (currently $500 per child, proposed to be $1,000 under President Bush's tax plan) could be tied to proof of insurance. Families that fail to provide proof would lose the credit and pay an additional $1,000 per child in taxes. Similarly, $1,000 of the personal exemption could also be tied to proof of insurance.

Tax Subsidies for Low-Income Taxpayers. Families with children and earning between, say, $10,000 and $30,000 a year generally qualify for the Earned Income Tax Credit (EITC). Even though they owe no income tax, these families can complete a tax return and get a "refund" from the Treasury, amounting to as much as $3,000 or $4,000 per year. The payments could be contingent on proof of insurance - whether through Medicaid, S-CHIP, an employer plan, or direct purchase.

Characteristic No. 10: The federal government's role should remain strictly financial.

Currently, the federal government "spends" more than $141 billion a year on tax subsidies for employer-provided insurance. However, there is virtually nothing in the tax code about what features a health insurance plan must have in order to qualify for a tax subsidy.17 Insurance purchased commercially, around two-thirds of the total, is regulated by the state governments. But the federal tax subsidy applies to whatever plans state governments allow to be sold.18

In this sense, the federal role is strictly financial. That is, the current tax break is based solely on the number of dollars taxpayers spend on health insurance, not on the features of the health plans themselves.

"There is no reason why government should dictate the content of health insurance plans."

This practice is sensible and should be continued. Aside from an interest in encouraging catastrophic insurance, there is no social reason why government at any level should dictate the content of health insurance plans. To continue our example, the role of the federal government should be to insure that $1,000 is available. It should leave the particulars of the insurance contract to the market, and it should leave decisions about how to operate the safety net health care to local citizens and their elected representatives.