to spending on health services. The subsidy granted to employer contributions to health insurance premiums under the individual income tax and the social security tax likewise increases the money pool. The very existence of health insurance apparently yields a measure of assurance to consumers that might lead them to spend more on all health services, insured or otherwise. An increase in the amount of income devoted to a particular service is tantamount to an increase in demand. At a given price, more units are taken or a given volume of services commands a higher price. Whether and when such changes in price or quantity may be deemed excessive is a separate question, one that does not yield unambiguous, meaningful answers.

IMPORTANCE OF UNIVERSAL ENROLLMENT

To equalize the financial burden of illness among a population while allowing ready access to care, enrollment must be universal. Medicare comes close to 100 percent enrollment for the aged-persons 65 and over. Voluntarily health insurance, though achieving volumes of enrollment beyond all forecasts made by friends and foes alike in the 1930s and 1940s, does not. For the population under age 65 the health insurance enrollment figures are not precisely known. Official estimates from the Social Security Administration vary between 80 and 91 percent for hospital care, which has the largest enrollment among the several objects of health expenditure. Enrollment for physician's services in the hospital is reported to range between 77 and 84 percent of the population under 65. For physician's services outside the hospital, out-of-hospital prescribed drugs, privateduty nursing, and visiting nurse service, the proportion of the population enrolled ranges between 50 and 60 percent, mostly under major medical insurance. It is noteworthy that for in-hospital services the percentage of the population enrolled has been virtually at a plateau since 1967.

It is not evident how much of a deviation from 100 percent enrollment is acceptable in the real world. Certainly, voluntary health insurance is unlikely to reach 100 percent enrollment, even if only because some people are gamblers, willing to take a chance that serious illness will not strike them, while others brought up in the Western tradition count on getting a free ride in case of a medical emergency. But-if it was not clear in the 1940s and 1950s as a matter of logic, it is clear now with our experience under Medicaid-medically indigent persons, once they are ill, are not insurable. And persons who are not yet ill and are not covered by insurance are liable to lose their eligibility for free care when government budgets for health programs are cut back. In places like New York City with a long-time liberal tradition of free care, some persons were worse off after Medicaid was instituted than before. Their condition was further aggravated by the higher prices they were now supposed to pay out-of-pocket.

In sum, it turns out, that under voluntary health insurance the deviation from 100 percent enrollment is appreciable. Thirty or forty years ago, the gap was masked sometimes by the cushioning effect of free or partly free care provided in government hospitals, usually local or federal; in voluntary, nonprofit hospitals subsidized by philanthropic funds; and in private physicians' offices, where a sliding scale of fees applied. In the course of time, the sliding scale has become eroded, owing to its essential incompatability with health insurance (which serves to enhance the patient's ability to pay) and the fact that the philanthropic dollar buys less than formerly. Local public hospitals, sponsored and operated by municipalities, have become dependent on federal and state funds and are compelled to reduce service when outside funds are cut back. Only the Veterans Administration system, owned and operated by the federal government, has remained intact in size, if not unchanged in program composition. It follows that a given deviation from enrollment of 100 percent of the population in a health insurance plan is a more serious concern today than it would have been a generation ago.

Unless the voluntary nature of health insurance enrollment possesses intrinsic merits that are not discernible to this observer, there are no reasons for departing from the principle of universal enrollment. There may be administrative difficulties that would have to be attended to during the period of transition in implementing such a goal. But the target must not be allowed to recede. Too many workers change jobs, too many people work part time to make health insurance enrollment dependent on permanent, full-time attachment to the labor force. Enrollment, not merely the opportunity to enroll, must be extended to all. Universality of enrollment, then, is the first of the proposed criteria for assessing the merits of a national health insurance plan.

BENEFITS PACKAGE

Next to consider is the benefits package. In this country, health insurance benefits have developed unevenly. Coverage of hospital care has always led coverage for other, related services. In consequence, some distortion in use has followed, with the insured service, even when more costly, sometimes substituted for cheaper, uninsured services. It is widely agreed now that a broad package of benefits for related health services is superior to a narrow package.

The exclusion of a nonrelated service, such as long-term care, is less harmful in the context of the benefits package and may be dealt with as a separate issue. Nevertheless the financing of long-term care on a systematic basis has been neglected too long, Usually required as a result of deterioration in health status, long-term care is a service properly encompassed within the health services industry. It may, however, be subject to different conditions of purchase, because it does not fall within the usual orbit of the physician-hospital relationship. Moreover, the components of long-term care are more susceptible to evaluation by the patient and his family than are those of acute medical and hospital care. In retrospect it seems that Medicare may have laid the foundation for the consensus on a desirable set of health insurance benefits which has apparently emerged in this country. Originally Medicare was proposed by successive Democratic administrations as a mandatory package of limited hospital care benefits. When a Republican counterproposal for voluntary health insurance was added to the bill, Congress enlarged the program to incorporate certain professional and other services and supplies used by the aged. Medicare benefits are both broad and deep. Most related medical services are included, either without limit as to quantity or with a high limit. In fact, extending a particular benefit without limits does not cost much beyond a certain point. It is a noteworthy advance in the public consensus concerning national health insurance policy when an identical benefit package is proposed for all population categories, as in the 1974 Nixon Administration bill. This is one necessary step toward curtailing, if not eliminating, a dual system of medical care.

It warrants recognition that not all the benefits that were offered under Medicare were promptly available. It took time and effort to organize home health services. Facilities for extended care did not then exist, and it is even questionable whether they exist today in skilled nursing homes, other than in name. There is reason to believe, too, that the general public misapprehended the extended-care benefit and that the ready availability of Medicaid as an alternative source of payment for long-term care served as a useful safety valve against widespread disappointment. With some allowance made for problems of transition, a broad and deep package of related benefits is, then, the second of the proposed criteria for assessing a national health insurance plan.

SINGLE SYSTEM OF CARE

The shortness of the list of criteria so far-universality of enrollment and a broad and deep package of related benefits-is in part attributable to the exclusionary argument followed earlier to divert the pursuit of certain goals or objectives to other instrumentalities. However, I incline to a small number of criteria for appraising national health insurance plans for another reason. Although a long list of criteria can facilitate the building of a broad coalition in support of a program, only a short list can yield measures of trade-off terms between the several objectives that might lend themselves to comparison by ranking or weighing. If some sense of the quantitative values of the trade-off terms is important, a long list of criteria lacks credibility.

I have already hinted at a third criterion, namely, the compatibility of a particular health insurance plan with a single system of medical care for the well-to-do and the poor alike. This criterion is a matter of ultimate value judgment, in my opinion, and is not susceptible to the usual canons of proof through evidence. It has been suggested that even the scientific aspect of medical care may not always be the best possible in the medical school setting, where the poor receive care because attending staff and residents combine to search for the rare and subtle manifestations of disease. A case can be made perhaps that patients who feel inconvenience, discomfort, or possibly discriminated against are not so responsive to a prescribed regimen as they might be. I have no reluctance, however, to rest this criterion on the self-evident desirability of equity. In the quest for a single system of care, a uniform package of benefits is necessary. Differences in fees according to source of payment, such as those

adopted under Medicaid, are unacceptable. Other factors, including the organization of care and the behavioral characteristics of the population, are perhaps even more important for achieving a single system of care for all. Therefore, I am inclined to regard this goal as a longer-range target, rather than as an immediate requirement for a national health insurance plan. The target should remain visible, to be always more closely approached, rather than to be allowed to recede.

PROVISIONS FOR PROVIDER REIMBURSEMENT

I turn now to the financing of national health insurance. Does financing yield one or more additional criteria? If everybody is to be enrolled and the package of benefits is sizable, not everybody will be able to pay for it out of his own resources. This conclusion is incontrovertible. What is at issue is whether the amount of the health insurance premium will be linked to the particular individual with a determination of the public subsidy required to be made in each case, or whether the level of premium will merely be an actuarial average, with sources of payment determined for the entire health insurance fund. Although the latter would be far simpler, it might raise questions about the distribution of the tax burden, which are always controversial in this society and are seldom resolved satisfactorily. As a practical matter, in order to facilitate the passage of a national health insurance plan, it may be wise to dissociate provisions concerning the sources of financing from the other aspects of the plan. My point is obviously not that financing is to be omitted from a national health insurance plan but rather that I am disposed to be tolerant toward, and choose among, a fairly wide range of alternatives. The issues, arguments, and decisions on sources of financing are essentially political-both in the sense of ultimate value judgments and in the sense of the distribution of power. I propose that at this time sources of financing need not yield a criterion for assessing a national health insurance plan.

Closely related, but separable, is the issue of the patient's participation in financing through cost sharing at the time of illness. Briefly, the bases for advocating cost sharing are three-fold. One, cost sharing serves as a deterrent to the use of services and possibly as an incentive to shop around for services at a lower price. Two, it serves as a means to reduce the amount of insurance premium. Three, it serves to reserve general tax revenues for those public expenditure purposes which only such taxes can pay for, such as national defense or cash transfers.

It is unfortunate that the empirical literature to date on cost sharing is still so modest. It has been reported in a careful study at one side that co-insurance leads to a reduction in physician use and that the lowest socio-economic classes may incur the largest reduction in services. It is suggested by experience elsewhere that such reductions may not last. The first of these findings, namely that there is a large reduction in use associated with co-insurance, is the one that can be asserted with the most confidence at this time. If not eroded by the passage of time, the effect of co-insurance on the use of physician services is sizable. If we mean to overcome, or at least mitigate, the tendency of insurance to promote the use of services due to lower out-of-pocket net price that the patient faces at the time of illness, then some cost sharing must take place.

In practice that has not occurred under Medicare, and it is not likely to occur under national health insurance. Rather, the policy is to permit the purchase of supplementary insurance to defray the amount of cost sharing. When the consumer is poor, cost sharing is often assumed by Medicaid or is occasionally waived by the provider of services. I am impelled to conclude that deterrence of use is not meant seriously as an argument in favor of cost sharing. If it were so intended, however, income relatedness would become important. It is reasonable to suppose that a given price or amount of cost sharing imposes greater weight upon a person with low income than on a person with high income. Several economists have concerned themselves with the implications of health insurance for utilization and with the effects of cost sharing. None, however, has considered the practical implications of the policy to permit the purchase of supplementary insurance to defray the cost of cost sharing. Nor is there agreement on what is a proper measure of income in the context of illness in view of the possibility of sizable fluctuations and permanent changes in income subsequent to the onset of major illness, as well as uncertainty about the future level of medical care expenditures. How will people learn of a change in their health insurance and copayment status with respect to income? How will they cope with such a change administratively? If the change, or notification about 57-677-75-23

it, is retrospective, to what extent will it influence responses to price variation? How are we to avoid the so-called notch effect, whereby a consumer loses benefits when his income increases? Income related features also require a determination of the types of income to be assisted. In such a determination, is income in kind to be counted? In a given place, what is the point at which assistance from cost sharing ceases?

In light of the scarcity of empirical work on this topic, it seems incumbent upon one to be receptive to new evidence as it develops and to continue to be open to persuasion. Much about the administrative complexities inherent to an income-related cost-sharing scheme remains to be learned from experience. I believe that evaluation of experience in a large geographic area, like a state, can yield more useful knowledge than can analysis of the findings of a designed experiment in which several thousand families selected at random at three or four sites participate. The advantages of studying a large population in a contiguous area are the inclusion of possible responses on the supply side, acknowledgement of some of the emulative aspects of consumer behavior, and the greater stability of large numbers.

The second basis for advocating cost sharing is to keep down the level of premium. This consideration is obviously important in selling voluntary health insurance. Cost sharing under major medical insurance is readily understandable on the basis of this feature. It has always seemed to me that, if health insurance became mandatory, the level of premium could be determined by the size of the benefits package alone. Thus I am unable to understand the reason for cost sharing by the aged under Medicare. My understanding is not enhanced by recognition of the fact that many aged persons (approximately one-half) possess supplementary private insurance; another 10 percent are protected by Medicaid; and still others are not charged by their physicians for the coinsurance factor of 20 percent. Upon reflection, however, it is evident that even under mandatory enrollment, relying for financing on employer-employee contributions to premiums makes a lower premium desirable. Once the uniform benefits package is made broad and deep, it becomes necessary to supplement the insurance premium with private cost-sharing. In the context of financing through employer contributions, income relatedness is not pertinent.

The third basis of the rationale for cost sharing seems persuasive at a time of perpetual tightness in the federal budget. If we keep down the burden on the public fisc, not only can other things get done, which only the public fisc can pay for, but also the health services sector may enjoy greater latitude in spending. Certainly, the Social Security Trust Funds, which are usually treated as if they were quasi-private funds, have received authorized increases in benefits quite readily-much more so than they would have under regularly financed public programs. Ultimately, the decisive consideration may prove to be the balance between this factor-the desire to keep down the burden on the treasury-and administrative feasibility. If, in practice, cost-sharing payments turn out to be defaulted, they will have to be abandoned..

In summary, the rationale of cost sharing as a deterrent to use is not logically compatible with permitting the purchase of supplementary insurance to pay for cost sharing. Income relatedness would enter here, but not when cost sharing is intended to supplement employer contributions to premiums. Cost sharing for the purpose of saving tax funds for other public purposes is a new argument which must withstand the test of application and experience. It should be added that a policy decision adverse to substantial cost sharing need not preclude small copayments for extra services, such as night calls.

At this time I am not disposed to take a stand on cost sharing as a criterion for appraising a national health insurance plan. Too much remains to be learned about how it works and what its effects are. A sound policy to adopt is one that is reversible at a low or moderate cost.

It is, nevertheless, useful to note the importance of nonfinancial factors in influencing the use of health services. There is the physician-hospital relationship previously alluded to. Not to be overlooked is the sought-for patient-physician relationship, one of mutual confidence and trust. Of course, the fiduciary role of the professional in which he acts in behalf of the patient as if the latter's interests were his own, is not confined to medicine. These factors are, however, highly important in medicine. Financial incentives are not necessarily best adapted to the fostering of such a relationship, one of mutual trust and voluntary delegation of decision-making to the other party.

Preventive services pose a problem here, if their efficacy is granted. With or without insurance supplementation and with or without income relatedness, cost sharing should not apply to specified preventive services, such as maternal and infant care or well child care. For young families such payments are likely to be burdensome, even if they could be met in installments. It might be argued that such services are schedulable and can be planned for, so that they are not truly insurable risks. The controversy between health insurance and prepayment is an old one, however, and has not proved to be particularly fruitful. Perhaps more important, the number of personal health services that are currently presumed to be effective as preventive measures is quite small.

In the absence of cost sharing there would be no need to set a maximum limit on an individual's or family's expenditures for medical care. In the presence of cost sharing, it is necessary to set such a limit. It is worth noting that any computation of maximum liability on the part of the patient-individual or family is bound to be an understatement to the extent that it omits the contribution to the insurance premium, excludes payments on covered benefits in excess of the stated fee, and, of course, excludes expenditure on items that are not among the covered benefits.

So far the discussion of financing has not yielded any criteria for assessing national health insurance plans which I am prepared to propose in light of existing knowledge. Appropriate provisions for reimbursement mechanisms and formulas are, however, a fourth criterion for assessing a national health insurance plan, in my judgment. I am aware that other students of the post-1965 rise in costs and expenditures do not attach as much importance to the influence of reimbursement as I do. Nevertheless, there is a widespread consensus that retrospective cost reimbursement of hospitals, which has been the major method for paying hospitals since 1965, must be ended. Since I can see no practical way to pay hospitals at uniform rates for a given service, it will be necessary to devise and adopt payment formulas that yield a fixed rate to individual hospitals for the period ahead. A fair summary of experience to date is that automatic reimbursement formulas do not work and that usually negotiations between provider and third-party payers are necessary. The several sources of payment acting separately are ineffective at controlling cost. It will be necessary to bring all major sources of payment together in negotiating both reimbursement formulas and amounts with individual institutions. In this country, experience with negotiated rates in behalf of all major sources of payment is lacking. The diverse and extensive experience of the several Canadian provinces in paying for hospital care since 1958 should be examined with care. Regulatory mechanisms. Beyond the development of formulas and appropriate data sets lies the organization of effective regulatory mechanisms. The administrative problems of operating a national health insurance plan to cover a good part of an industry that is now spending more than $100 billion a year cannot be overestimated. There is bound to be a struggle for control by levels of government, between the private and public sectors, and between institutions and individuals. Personally, I should be inclined to look hard at the evidence on past performance as an indicator of future potentialities for contributing to this awesome task. Congressional hearings are highly useful for eliciting this type of information. Yet, it is doubtful that Congressional hearings can convey the atmosphere in which regulation actually takes place.

Speaking from personal knowledge, I must confess to a sense of disappointment over the extent to which the regulatory process in the health field seems to have taken on some of the adversary characteristics of criminal court proceedings. It is not only the consumer who is suing for malpractice today, but the provider is suing the health insurance plan plus the state regulatory agency, while the regulatory agency sues a health insurance plan plus its individual board members. Power and its exercise have displaced the ordinary civilities based on a comity of interests that usually accompany relationships between legitimate governors and the governed. Economists have traditionally tended to distrust regulation on the ground that the regulated tend to become the regulators. What has not been foreseen by proponents of regulation is the possible runaway nature of regulatory power when carried out with full discretion. As much as possible, discretionary authority must be limited, even as ministerial functions are extended, with both subject to the constraints of a free flow of information that is equally accessible to all. In a complex society such as ours, it it obviously not practical to dispense with regulation. But the time has come