1. Factors underlying recent growth in health care expenditures

The tabulation below shows trends in the shares of a range of categories of health expenditures. There are a number of notable features.

National Health Expenditures ^8/

National Health Expenditures ^8/

		1965	1970	1980	1986
(Billions of dollars)
National health expenditures		41.9	75.0	248.1	458.2
(Percent of total)
Personal health care		85.7	87.2	88.6	88.2
	Of which:
	Hospital care	33.4	37.3	41.0	39.2
	Physician services	20.3	19.1	18.9	20.1
	Drugs and medical sundries	12.4	10.7	7.6	6.7
	Nursing home care	5.0	6.3	8.2	8.3
Program administration		4.1	3.7	3.7	5.3
Government public health activities		1.9	1.9	2.9	2.9
Research and construction		8.4	7.2	4.8	3.6
(Percent of GNP)
National health expenditures		5.9	7.4	9.1	10.9

View Table

National Health Expenditures ^8/

		1965	1970	1980	1986
(Billions of dollars)
National health expenditures		41.9	75.0	248.1	458.2
(Percent of total)
Personal health care		85.7	87.2	88.6	88.2
	Of which:
	Hospital care	33.4	37.3	41.0	39.2
	Physician services	20.3	19.1	18.9	20.1
	Drugs and medical sundries	12.4	10.7	7.6	6.7
	Nursing home care	5.0	6.3	8.2	8.3
Program administration		4.1	3.7	3.7	5.3
Government public health activities		1.9	1.9	2.9	2.9
Research and construction		8.4	7.2	4.8	3.6
(Percent of GNP)
National health expenditures		5.9	7.4	9.1	10.9

First, the sources of growth in health expenditures are broadly based. Even though the relative shares of the components fluctuate, with the exception of drugs and medical sundries and of research and construction, all categories grew relative to GNP. ^9/

Second, the share of hospital care expenditures in total health expenditures, which increased steadily over a number of decades, declined in the 1980s. This reversal, which may be temporary, is generally attributed to the 1983 reform whereby Medicare changed the way in which it reimbursed hospitals from a cost-based to a prospective per-admission basis. ^10/

Third, the share of physician services has been rising in recent years. However, in late 1989 Congress agreed to implement a significant change in the way in which physician’s are reimbursed under Medicare, replacing the current customary-prevailing-reasonable (CPR) system with a resource (or cost) based relative value scale (RBRVS). ^11/

Fourth, reflecting the aging of the population, the share taken by nursing home care has increased in the 1980s.

Fifth, as already implied, the share of health care expenditures devoted to drugs and medical sundries has declined sharply. This development may reflect the fact that the drug and medical sundries industries conform more closely to the competitive norm than do other components of the health care sector, resulting in a greater discipline in pricing policy over time.

For health care expenditures as a whole, it is estimated that 32 percent of the increase in spending between 1985 and 1986 could be ascribed to economy-wide price inflation; 22 percent to medical care price inflation in excess of the general rate of inflation; 11 percent to population growth; 35 percent to other factors, most notably changes in per capita consumption and in “intensity” due to rising income levels. ^12/ Changes in demographic structure play a relatively minor role, a result which has been observed over the past few decades irrespective of the sample period selected. Though subject to some variation, this pattern is repeated across the major categories of health expenditures. For example, taking the case of the 1.8 percentage point growth in the GNP share of in-patient care over the period 1965 to 1985, the bulk of that increase was due to a combination of growth in the “intensity” of care (i.e., real goods and services provided per in-patient day) and hospital price inflation over and above general price inflation with only about 8 percent of the increase being due to changes in the age and sex mix of the population. ^13/

The somewhat surprising result that demographic change may play a relatively minor role in driving up medical expenditures may be a function of the particular demographic breakdown used in the HCFA projections. In particular, the proportion of the population that is very elderly (age 85 and over) rather than elderly (age 65 and over) may be more important in explaining medical cost trends. This portion of the population is expected to grow by almost 4 percent a year over the next 20 years. ^14/

In sum, the large increases in medical care expenditures are due to a number of factors. In particular, the intensity of use and the prices of health care products have been steadily rising where the latter may reflect inaccurate measurement of quality changes. While of themselves they do not necessarily imply a need for reform, when these trends are considered in the context of the medical market’s structure, as shall be seen below, a number of policy issues do arise.

2. Prospects for medical care expenses

Based on historical trends and relationships as well as on recent experience, the Health Care Financing Administration (HCFA), which supervises the Medicare program, has projected aggregate health care expenditures through the year 2000. ^15/ In addition to its own technical assumptions concerning the medical sector, the HCFA projections adopt the general economic and demographic assumptions contained under Social Security Trustee’s alternative II-B. ^16/

A summary of the HCFA projections is presented in the tabulation below, which points to a further sharp increase in the share of GNP devoted to health care. It is noteworthy that demographic change cannot be the primary factor driving this result since, as already observed, demographic developments stand to be relatively favorable in the 1990s. ^17/ This serves to emphasize the importance of other factors such as the increasing relative price of health care goods in explaining the runup in health care costs.

National Health Expenditures ^18/

National Health Expenditures ^18/

		1986	1990	1995	2000
(Billions of dollars)
National health expenditures		458.2	647.3	999.1	1,529.3
(Percent of total)
Personal health care		88.2	88.6	90.1	91.4
	Of which:
	Hospital care	39.2	38.7	39.4	40.6
	Physician services	20.1	20.5	20.9	20.9
	Drugs and medical sundries	6.7	6.5	6.6	6.7
	Nursing home care	8.3	8.4	8.5	8.4
Program administration		5.3	5.4	4.5	3.8
Government public health activities		2.9	2.9	2.7	2.5
Research and construction		3.6	3.2	2.7	2.3
(Percent of GNP)
National health expenditures		10.9	12.0	13.4	15.0

View Table

National Health Expenditures ^18/

		1986	1990	1995	2000
(Billions of dollars)
National health expenditures		458.2	647.3	999.1	1,529.3
(Percent of total)
Personal health care		88.2	88.6	90.1	91.4
	Of which:
	Hospital care	39.2	38.7	39.4	40.6
	Physician services	20.1	20.5	20.9	20.9
	Drugs and medical sundries	6.7	6.5	6.6	6.7
	Nursing home care	8.3	8.4	8.5	8.4
Program administration		5.3	5.4	4.5	3.8
Government public health activities		2.9	2.9	2.7	2.5
Research and construction		3.6	3.2	2.7	2.3
(Percent of GNP)
National health expenditures		10.9	12.0	13.4	15.0

3. U.S. federal medical programs

Recent and prospective developments in federal health care expenditures indicate that this sector of the health care industry is expanding even more rapidly than the industry as a whole. For example, HCFA estimates suggest that the federal share of national health expenditures will rise from about 29 percent of the total in 1986 to about 33 percent in 2000. ^19/ Given the fiscal pressures faced by the United States, this prospect has to be viewed with concern.

The two major programs through which the federal government is directly involved in the health care industry are Medicare and Medicaid. ^20/ Since it has been the focus of most recent reform efforts, attention in this paper will be devoted primarily to Medicare. ^21/ Medicare is composed of two parts, hospital insurance (HI also referred to as Part A) and supplementary medical insurance (SMI or Part B).

The HI program pays for in-patient hospital care and other related care both of those aged 65 and over and of the long-term disabled. Disbursements in 1987 amounted to $50.3 billion. Expenditures under this program are financed primarily through payroll (social security) taxation. Since its inception in 1965, revenues have exceeded expenditures, with the balance being placed in a growing trust fund. ^22/ However, particularly in light of its mandate to care for the aged, the prospect for the financial status of the HI program is poor. As shown in the tabulation below, the most recent official projections suggest that the trust fund will be exhausted by 2005. ^24/ Underlying this development, the program will begin to experience operating deficits beginning around 1995 with those deficits increasing steadily over the balance of the 75-year projection period. These large unfunded liabilities imply that major changes in the program will be needed to place it on a sound footing. In particular, they provide an incentive for cost-containment reform efforts (see below).

The SMI program pays for physician services, out-patient hospital services, and other medical expenses of those aged 65 and over and of those who are long-term disabled. Disbursements in 1987 amounted to $31.67 billion. ^24/ The growth rates in this program have also been rapid, with outlays doubling in the five years ended 1985. The program is voluntary and is viewed by the authorities as being analogous to private group insurance. Although no long-run projections are made for this program—the trustees project an annual growth rate of 13 percent in benefits over the three years 1988-90—it is clear that further rapid growth in reimbursements is expected.

Financial Status of the HI Program ^25/

(Percent of taxable payroll)

Financial Status of the HI Program ^25/

(Percent of taxable payroll)

	Expenditures	Scheduled Tax Rates	Actuarial Balance	Trust Fund at End of Year
1988	2.52	2.90	0.38	69.6
1990	2.71	2.90	0.19	97.8
1995	3.11	2.90	-0.21	147.9
2000	3.42	2.90	-0.52	133.6
2005	3.68	2.90	-0.78	--
Averages
1988-2012	3.37	2.90	-0.47	…
2013-2037	5.50	2.90	-2.60	…
2038-2062	6.89	2.90	-3.99	…

View Table

Financial Status of the HI Program ^25/

(Percent of taxable payroll)

	Expenditures	Scheduled Tax Rates	Actuarial Balance	Trust Fund at End of Year
1988	2.52	2.90	0.38	69.6
1990	2.71	2.90	0.19	97.8
1995	3.11	2.90	-0.21	147.9
2000	3.42	2.90	-0.52	133.6
2005	3.68	2.90	-0.78	--
Averages
1988-2012	3.37	2.90	-0.47	…
2013-2037	5.50	2.90	-2.60	…
2038-2062	6.89	2.90	-3.99	…

SMI revenues come from participant premiums and from general revenues where premium income has recently yielded 25 percent of program costs. ^26/ This implies that the SMI program can be viewed as a subsidized insurance scheme.

In 1988, the Medicare Catastrophic Coverage Act (PL 100-360) was passed. This Act, which represented the largest expansion of Medicare benefits in 20 years, was intended to provide unlimited acute-care hospital coverage through HI and a cap on participant copayment costs for covered services as well as introducing coverage of outpatient prescription drugs through SMI. ^27/ An innovative feature of the package was that the new benefits were to be financed by a combination of a flat premium and additional income taxes levied on the participants. However, as the magnitude of the additional (progressive) surtaxes became evident—the surtax would rise to $850 p.a. for most participants earning over $35,000—a storm of protest rose amongst those most affected with the result that the 1988 Act was essentially repealed in November 1989. ^28/

4. International comparisons

Since health care expenditures are rising rapidly in the United States, the question naturally arises as to whether U.S. experience is unusual by international standards. The conventional way of answering this question has been to estimate cross country regression equations using data drawn from a range of developed countries gauging some measure of (per capita) medical care expenditures and (per capita) income. A relatively robust result to emerge from the literature is that variations in national income per capita explain more than 90 percent of the variation in medical care expenditures across countries. ^29/ This result implies that the U.S. experience is what might be expected, given the standard of living in the United States and, further, has led some to conclude that the level of health care expenditures in any given country has little to do with the degree of state involvement in the provision and/or financing of health care. ^30/ This is consistent with the fact that policy initiatives in many developed countries have been similar, with a focus on cost containment mechanisms such as mandating fee schedules and reimbursement rates. ^31/ A final point, the data used in the international comparisons literature have also been employed to estimate the income elasticity of demand for health care services. This has generally, but not always, been found to exceed unity. ^32/ This empirical issue is clouded by the fact that inter country cross-section estimates imply higher income elasticity values than intra-country estimates. ^33/

5. Value of health care

Before considering which aspects of the health care market are potential candidates for reform, it is important to consider the impact of health care expenditures on actual health. The consensus of the research in this area is that additional health care expenditures in high income countries have little impact on the actual health of the population of those countries. For example, data from the RAND Health Insurance Experiment suggest that, for the average “non-aged” American, the clinical effect of additional medical care induced by no cost-sharing insurance schemes was restricted to health improvements among poor adults with hypertension or myopia. ^34/ In other words, health care gains, to the extent they exist, seem to lie in specific targeted programs. While this may not necessarily imply great waste in health care programs—health care expenditures could afford other benefits such as providing reassurance, etc.—it suggests that curbs on health expenditures with a view to enhancing economic efficiency need not compromise underlying health standards.

1. Health insurance: general issues

To many commentators, the major problems of the medical industry derive from the nature of insurance. In the area of individual health insurance, the current market solution relies mainly on cost reimbursement insurance. Arrow has demonstrated that in an ideal world where, in particular, incentives are not affected by the existence of insurance, and where insurance administrative costs (“loading”) are proportional to the total premium, then the optimal pattern of insurance will be full coverage above a deductible—a small amount of risk becomes tolerable to a consumer so as to save on the loading factor. ^39/

In reality, incentives are affected by insurance contracts—since costs rather than (exogenous) illnesses are reimbursed, insurance encourages individuals to spend more on medical services than they otherwise would. The result is that the demand for medical services is distorted. As an example, consider an insurance policy with a coinsurance rate of 20 percent (the insurance company pays for 80 percent of the medical costs associated with an illness). This case is shown in partial equilibrium terms in Figure 1, where DD refers to the demand curve for medical services and P₀B to the industry’s (infinitely elastic) supply curve to the individual. The supply curve faced by the individual is given by P₁D, the net of insurance price he pays for medical care. In equilibrium Q₁ is demanded, an inefficient outcome. To elaborate, the area ADQ₀ Q₁ which is the area under the demand curve, gauges how much the individual would be willing to pay for the increment Q₀ Q₁; AB Q₀Q₁, the area under the supply curve, gauges how much that same increment costs society to produce. The result is a net loss (excess burden) measured by ABD.^40/

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

This result is an example of the problem of moral hazard. ^41/ In the absence of first-best indemnity insurance against illnesses, what is involved is a tradeoff between economic losses from moral hazard and the gains from risk trading. ^42/ One branch of the literature has focused on this tradeoff and has considered the specification of optimal reimbursement schedules. For example, as can be seen from Figure 1, the more elastic the demand curve, the greater the potential for welfare losses, and therefore the higher the optimal coinsurance rates. ^43/

The nature of the optimum depends on the instruments assumed to be available to the insurance industry. Deriving an optimal coinsurance rate is analogous to deriving optimal excise taxes. Allowing simultaneously for deductibles, ceilings, and coinsurance rates, transforms the exercise into an optimal nonlinear pricing problem analogous to the optimal income tax problem. These more general problems have proved intractable. ^44/

When analyzing the impact and/or optimality of differing insurance contracts, the assumptions concerning the underlying information structure are crucial. To this point, it has been implicitly assumed that the insurance industry has no information about the consumer’s actions or efforts; the only information it receives is on outcomes. In reality, insurance companies do have additional information concerning the risks to which they are exposed. Concerning moral hazard, Shavell ^45/ demonstrates that if the insurer has information which reflects in some fashion the truth about the insured’s actions, then, no matter how imprecise the information, that information has value and should be incorporated in the insurance policy. ^46/ This useful result suggests that efforts on the part of insurers to reduce the information asymmetries between themselves and consumers may serve to mitigate the impact of moral hazard.

A particular problem that arises from the asymmetry of information between the insured and the insurer is that of adverse selection—an individual will generally have more accurate information than insurance companies about his expected health expenditures and will shop among alternative policies to his advantage. Such behavior has adverse implications and encourages insurance companies to use screening devices such as age, etc., to identify risk categories. ^47/ The potential for adverse selection raises the possibility of the nonexistence of a competitive equilibrium. ^48/

On balance, this section indicates that, in the absence of a first-best solution, there is a tendency for insurance markets to encourage overconsumption of medical services with obvious implications for trends in health expenditures. The specification of the first-best solution, by emphasizing the role of identifying illnesses rather than expenses, suggests the appropriate direction for reform. ^49/ Further, stressing the role of information asymmetries in generating distortions emphasizes the value of improving the flow of accurate information. ^50/

2. Medical malpractice

Just as the patient-insurance company relationship is associated with information asymmetries, so is the patient-physician relationship. Because of the high cost of accurately monitoring the heterogeneous services rendered by physicians, patients will tend to be relatively uninformed purchasers of those services and face possible disappointment in the quality of care they receive. The result is a demand for an ex post settlements mechanism. Given the desirability of ensuring confidence in the medical industry and therefore of insuring that the providers of health care face the proper incentives, an appropriate settlements mechanism would work through the establishment of negligence rules. ^51/

This raises the topic of malpractice insurance, and the question arises as to how this form of insurance influences health care costs. The theoretical literature has focused on the choice between, and the optimal properties of, the strict liability form and the negligence form of malpractice insurance. ^52/ The former requires the party who has caused the loss (the physician) to pay damages whether or not he was negligent; the latter requires that that party pays damages only if negligent. The focus of the analysis has been to determine whether the availability of insurance in these forms might inappropriately alter the incentives for accident avoidance.

A strong conclusion that emerges from the literature is that, given certain assumptions, the availability of liability insurance of either form, while it does change the incentives created by liability rules, does not do so in a manner which would make it socially beneficial for the Government to intervene in the operation of competitive liability insurance markets. ^53/ In the case of strict liability insurance, a first-best outcome can be achieved except when physicians are risk averse and the liability insurers cannot observe the physician’s level of prevention activity. This result parallels that of the previous subsection. That is, when there are information asymmetries between the insured and the insurers, moral hazard problems arise, and less than complete coverage is optimal. ^54/ In the case of negligence insurance, with damages and the due care standard set at optimal levels, a first-best solution is again attainable and the demand for insurance is zero. The essence of this provocative result is that if the correct incentives are in place, the physicians will choose to be non-negligent. ^55/

As evidenced by the widespread purchase of expensive and complete malpractice insurance, the actual market solution deviates substantially from this theoretical benchmark. The data indicate that both the number of malpractice suits and the size of settlements have increased rapidly in the United States in recent decades. ^56/ Moreover, malpractice insurance policies typically lack some of the features one would expect in insurance policies to encourage accident avoidance—there are no deductibles or coinsurance. ^57/

These observed trends appear to be the result of the interaction of a number of interdependent influences which violate the assumptions underlying the theoretical results. The assumption that there be no errors or uncertainty as to legal outcomes needs to be relaxed. Danzon, for example, argues that optimal liability insurance contracts under a negligence rule must recognize the existence of imperfect competition, court error, ^58/ and costly legal defense. Monitoring-cost economies of scale imply that damages and legal defense costs should typically be covered by the same policy. The optimal contract then involves a tradeoff between providing physicians with the incentive to reduce accidents and the insurer the incentive to defend claims—the latter militates against positive coinsurance rates.

Also in this connection, the market solution for malpractice insurance appears to encourage excessive testing. Stiglitz reports that the fear of malpractice suits led to excessive treatment prescription and testing of between $15 and $40 billion in 1983. ^59/

On balance, malpractice insurance as practiced clearly contributes to escalating health care costs. However, the complexity of the market for malpractice insurance rules out simple policy prescriptions. One exception to this conclusion is that further improving the accuracy of court decisions would appear to be desirable.

A final point: as with any market, the characteristics of the malpractice market are endogenous. One response to developments in malpractice costs has been the creation of physician-owned mutual insurance companies in some states, internalizing the costs of legal defense. ^60/ This reduces the need for monitoring costs. Another development—one which is a cause for concern--has been the recent finding that 12 percent of physicians who bill Medicare have ownership or investment interests in facilities to which they make patient references. These physicians order 45 percent more tests for their Medicare patients than do the patients of other physicians. This form of vertical integration would appear to exacerbate some of the negative impacts of the malpractice situation on health care costs.

3. Imperfect competition

The existence of information asymmetries and the heterogeneity of the services produced by the health care industry suggest that the market for these services will be noncompetitive. The literature on this subject has concentrated, though not exclusively so, ^61/ on the physician services sector, taking as its starting point the observation that patterns of medical treatment are generally determined by physicians rather than by patients.

Currently, there are two competing frameworks for characterizing how the physician’s profits from the physician-patient agency relationship might be constrained. ^62/ On the one hand, there are those who argue that the ethics of the medical profession will temper the tendency for physicians to exploit the agency relationship for personal gain. ^63/ The proponents of the supplier-induced demand (SID) hypothesis, whereby physicians are presumed to induce demand for their services as increases in doctor availability reduce the pool of patients per doctor, ^64/ tend to be of this view and maintain that the greater the economic pressures on physicians, the less will ethics restrain their behavior. The alternative view holds that the information asymmetries are not so great and that therefore patients can evaluate physician services with sufficient accuracy to allow market forces to perform the constraining role. ^65/ It has been argued that, since, irrespective of the model employed, physicians will balance their own welfare against that of their patients, the outcome need not result in a serious misallocation of resources. ^66/ However, that result depends on the incentive mechanisms, i.e., the insurance systems in place.

The empirical literature has focused on evaluating the SID hypothesis. The impression that physicians can and do manipulate patients’ demand for health services is supported by the observation that there are very large interregional differences in the per-capita utilization of certain elective medical procedures. ^67/ Much of the original interest in the SID hypothesis, however, was generated by empirical research which found a positive relationship between per-capita utilization indices and, ceteris paribus, physician density ratios. ^68/ However, the bulk of the empirical literature in this area has been criticized on the grounds both that the results as already implied are consistent with other theoretical frameworks and that the empirical tests lack discriminating power. ^69/ Some recent empirical research, which uses tests with greater discriminating power, may suggest that the market for physician services, though not perfectly competitive, is more competitive than had been believed. ^70/ It seems fair to conclude that the evidence is inconclusive. ^71/

On balance, the most important observation would appear to be that the implications of the structure of the market for health care costs will depend on the incentives faced by the various agents. More generally, the issue of physician-induced demand raises the central question of whether the appropriate strategy for containing health care costs should focus on the demand side by creating the correct incentives for consumers to discipline providers or on the supply-side through regulation. ^72/ This question underlies the analysis of the next section on government policy.

A final point: this subsection has not discussed other sources of noncompetitive behavior in the health care sector. These sources include barriers to entry to the medical profession ^73/ and the existence of not-for-profit hospitals. The discussion in this subsection, therefore, is only an example of the possible implications of imperfect competition for health care.

1. Exclusion of employer contributions to medical insurance and health care

The ways in which the U.S. Government directly enters the health care market have already been described. An important additional indirect way is through the tax system. In this connection, the current practice in the United States is that an employer’s contributions to an employee’s health care costs are excluded from individual taxable income, although the contributions are a deductible expense for the employer. This constitutes a tax subsidy to medical insurance.

The issue this tax subsidy raises is that, even before the introduction of the tax subsidy, there is a pre-existing distortion in the health insurance market due to reimbursement practices. The tax subsidy is therefore in effect a subsidy of a subsidy. Returning to the simple example described in Figure 1 above, one effect of the tax subsidy would be to reduce the equilibrium coinsurance rate relative to that which would be generated in a no-tax-subsidy equilibrium. ^74/

This outcome is illustrated in Figure 2. In that figure P₀λ₀ corresponds to the price (coinsurance rate) that would result in the absence of the tax subsidy, whereas P₀λ₀ is the price (coinsurance rate) that would prevail after the introduction of the tax subsidy. The marginal impact of the tax subsidy on the size of the distortions—measured by the area BEDF—is large, technically a first-order effect.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

• Download Figure
• Download figure as PowerPoint slide

• Download Figure
• Download figure as PowerPoint slide

The size of the tax distortion depends on the elasticity of demand. Since Feldstein and Friedman’s original work on this subject, a large literature has emerged in which the empirical evaluation of the significance of this theoretical point has been steadily refined. One representative recent estimate concludes that the tax subsidy could increase employer-sponsored health insurance demand by about 27 percent and aggregate medical services demand by about 5 percent. ^75/ This is potentially a major effect and suggests that tax reform in this area merits serious consideration as a means of restraining health care costs.

2. The role of Medigap policies

Another important example of how the interaction between government policies and the private market can result in overinsurance is afforded by the case of the privately purchased insurance policies designed to supplement Medicare coverage (otherwise known as “Medigap” policies).

As background, Part A Medicare has a deductible (set at $560 in 1989), copayments (at one fourth of the deductible per day for days 61 through 90 during a hospital stay) and, subject to minor reservations, the beneficiary faces full costs for days above the covered limits. Part B has a $75 annual deductible and thereafter Medicare pays 80 percent of the “reasonable charge.”

Elderly beneficiaries supplement Parts A and B with private medigap policies. While the premiums on these policies are often high, the policies from a beneficiary’s point of view are highly leveraged--if the beneficiary is induced to consume more medical services because of a Medigap policy, the premiums on that policy are theoretically based on 20 percent of the incremental expenditure with 80 percent being picked up by Medicare. ^76/ It has been argued that, in the absence of catastrophic care, medigap policies do not necessarily result in overinsurance, given the presumed high degree of risk aversion of the elderly. ^77/ However, while it received little attention at the time, had catastrophic health care under Medicare remained in place, the interaction between Medicare and private supplementary insurance would likely have resulted in inefficient overinsurance, with the demand-limiting effect of the cost-sharing implied by the Medicare structure being weakened. ^78/ As it is, the possibility of overinsurance cannot be ruled out. ^79/

3. Health Maintenance Organizations (HMOs)

To this point, the focus has been on the impact of government intervention in the demand side of the health care market. It has been argued that there are a number of ways in which governmental actions have interacted with private health insurance markets with the effect of excessively reducing the optimal cost-sharing provisions of health insurance. The cases considered are examples of the general policy dilemma discussed earlier of the tradeoff between risk spreading and establishing appropriate incentives to restrain expenditures. ^80/

This suggests that cost containment efforts might more profitably be targeted at the supply-side of theindustry. An example of this type is afforded by Health Maintenance Organizations (HMO) which provide comprehensive medical care on the basis of a fixed periodic prepayment. Unlike insurance companies, HMOs directly provide health services and, as necessary, ensure payment for medical services rendered by outside organizations.

This type of arrangement offers a number of potential advantages. In particular, by treating many patients, an HMO can insure itself in a way an individual cannot by averaging high- and low-cost patients (ex post), thereby mitigating the tradeoff between risk spreading and incentive creation. Moreover, the incentive structure is more appropriate since in this case health care is produced in a set of firms controlled, so to speak, by the insurer. ^18/ Any benefit of efforts to control costs will accrue entirely to the relevant HMO plan. ^82/

HMO and the related preferred-provider insurance (PPI) plans have proved popular and now provide coverage for more than 60 million people. ^83/ The bulk of the empirical literature concerning their impact on health care costs has concluded that HMOs have worked to reduce those costs. However, the precise manner in which they restrain health care costs is not clear. The conventional wisdom has been that the mechanism primarily involves a reduction in hospital admissions, ^84/ but some recent work points to a reduction in the length of hospital stay as a major source of the cost containment. ^85/ Irrespective of how the cost reductions are being achieved, evidence indicates that the reductions have not been at the expense of health status. ^86/ Finally, despite the cost reductions, it is worth noting that HMOs appear to experience the same high rate of inflation as do other health care producers, indicating that they represent at best a partial solution to soaring health costs.

Thus far, the effects of government policies have been analyzed individually. The case of HMOs provides an example of how they interact. In particular, the medigap policies that emerged in response to Medicare may be frustrating the intent of Medicare’s expenditure containment policies as they bear on HMOs. Beginning in 1985, Medicare began paying HMOs and qualified prepaid health plans for Medicare beneficiaries. However, increasingly these institutions cannot compete with a combination of Medicare and supplementary private (medigap) insurance. Similarly, the tax subsidy discussed earlier also may tend to reduce the competitive advantage of HMOs. Indeed, this has led Senator Durenberger and Representative Gephardt to introduce legislation with a view to abolishing the tax subsidy for health insurance to encourage HMO growth. ^87/

4. The Medicare prospective payment system

Prior to the passage of the Tax Equity and Fiscal Responsibility Act of 1982, Medicare reimbursed hospitals for the “reasonable costs” of care. With the passage of that act, however, the Medicare program began a four-year transition to payment of hospitals on a prospective per-admission payment system (known as PPS) ^88/ Under this system, any Medicare admission must be classified into one of the 500 or so Diagnosis Related Groups (DRG). Once an admission has been assigned to a DRG, the Medicare reimbursement for that admission is set, since each DRG has an associated predetermined reimbursement amount. In the transition period, the amounts paid for each DRG will be allowed to vary by region. In the long run, however, there will be a “national price” for each DRG, though adjustments will be made to reflect urban/rural and area wage differences. Finally, an amount is set aside to make outlier payments to hospitals for extremely long length of stay and/or costly cases.

This change represents a major reform. By paying only the prospective amount, Medicare requires the hospital or other provider to bear the marginal costs of treatment. In other words, the PPS represents an attempt to introduce an extreme form of supply-side cost sharing. ^89/ Conceptually the PPS appears to be the supply-side analogue for demand-side indemnity insurance for specific incidents of ill health.

However, the PPS reform has been superimposed on a market characterized by imperfect competition and information asymmetries, raising questions about the precise nature of the reform’s incentive effects. One approach to analyzing these incentive effects is to view the physician as the key decision maker who acts as the “agent” for the two “principals”; the patient and the hospital. ^90/ The PPS disturbs the commonality of interest for greater health care expenditures that used to be shared by patients and hospitals under the old cost-based reimbursement system. If physicians end up placing greater weight on hospital profits than on patient benefits, the tendency will be for an under-supply of medical services to occur. ^91/ Other issues raised by the nature of the market include the possibility that PPS may induce cost shifting between public and private patients (for which there appers to be little clear evidence) ^92/ and the possibility that the relatively broad nature of some of the DRGs might lead to adverse selection problems. ^93/

As far as pinpointing the significance of these various potential incentive effects is concerned, the conclusion must be that the empirical literature to this point has not been particularly helpful. ^94/ What is clear, however, is that the introduction of the PPS was associated with a remarkable reduction in hospital costs. ^95/ However, there is some evidence that this may have been a once-for-all effect with the growth in costs subsequently increasing again. ^96/ Nonetheless, the PPS can offer valuable lessons for other parts of the health care system. At a minimum, it suggests that incentives, particularly those associated with insurance schemes, have a powerful role to play in determining the course of health care costs.

5. The Resource Based Relative Value Scale (RBRVS)

The analysis earlier of the effects of imperfect competition on physician behavior concluded that while the empirical evidence is at best sketchy, the importance of setting the correct incentives for physicians is clear. The current Medicare practice for determining physician fees is known as the customary-prevailing-reasonable (CPR) system. ^97/ (Private insurers use the similar usual-customary-reasonable (UCR) system.) The Deficit Reductions Act (DEFRA) of 1983 required physicians to choose either to participate or not to participate in the Medicare program, where a ‘participating’ physician agrees to accept Medicare reimbursements as payment in full. ^98/ The intent of this measure was to reduce the practice of balance billing whereby physicians charge patients for more than what Medicare reimburses. Subsequent legislative measures, by favoring participating over nonparticipating physicians, have greatly enhanced the incentive to elect to participate. (The overall assignment rate increased from 51 percent in 1983 to 74 percent in 1987.) When this policy is viewed in conjunction with other measures to freeze or constrain Medicare reimbursement rates, the ad hoc effect is to institute for physicians a scheme analogous to the hospitals’ PPS.

The present system, however, seems generally to be presumed to have a couple of serious flaws. First, expenditures for physician’s services are growing more rapidly even than spending for other medical care. At least in the case of Medicare, where fees have been constrained, this presumably reflects volume rather than rapid price increases. Second, it has been widely perceived that CPR is in some sense distortionary. In particular, it is often argued that the prices charged by physicians bear a weak relationship to the underlying costs of providing the relevant services. Specifically, surgeons’ fees are substantially higher than internists’ fees. ^99/

It is against this background that the Physician Payment Review Commission, established in 1986 to advise the Government on the reform of Medicare’s physician payment methods, has recommended that the CPR system be replaced with a fee schedule that would reflect resource costs. ^100/ The fee schedule would be based on a revised version of the Resource-Based Relative Value scale (RBRVS) of Hsiao et. al. ^101/

In November 1989, Congress passed legislation that in essence adopted the RBRVS scheme in conjunction with a number of cost containment measures. Specifically, the salient features of the package include the following. First, Medicare is to create a uniform national fee scale for each of 7,000 different types of doctor services, allowing for some minor regional cost variations. This scale is to be phased in between 1992 and 1996. Second, the fees are to be based on a relative value scale reflecting the time, effort, physical labor, overhead, etc., embodied in each procedure. Third, by 1993 balance billing in excess of 15 percent of the Medicare allowed amount would be forbidden. Fourth, if Medicare expenditures in any given year exceed Congress’ expenditure target for that year, Medicare fees would be cut in the following year. The fee cuts that can automatically be imposed if “volume-performance standards” exceed guidelines will be phased in slowly, reaching 3 percent by 1996.

Concentrating on the potential impact of the RBRVS component of the package, it is useful to distinguish between its distributional implications within the medical profession and its potential influence on costs over time. On the former, in the short run an RBRVS is likely to eliminate at least some of the perceived inequities across medical specialities. ^102/ The longer-term implications, however, depend on whether existing income disparities reflect compensating differentials or entry barriers. ^103/ This is an empirical matter which has yet to be resolved.

Concerning the potential impact of RBRVS on trends in health care costs, the proposed system does not appear to share with the PPS obvious cost-saving characteristics. In fact, given the recent practice with the CPR system, it can be argued that to have replaced the CPR basis with an RBRVS alone would have been tantamount to replacing one fee schedule with another fee schedule, signifying nothing more than a redistribution of monopoly rents or quasi-rents among doctors. ^104/ This serves to emphasize the importance of the other elements of the package which seek to contain costs.