Respondent.

Introduction

The conclusions of “All-payer ratesetting: Down but not out” are that “ratesetting was a successful cost-containment strategy [and] the experience of Maryland suggests that reconsideration of all-payer ratesetting for hospitals is warranted.” I argue that, on empirical grounds, hospital ratesetting has not proven to be better than nothing in the short and intermediate terms, and, on conceptual grounds, in the long term, it is considerably worse.

The literature on the effects of ratesetting in my view is more equivocal as to its effects on hospital costs than is implied in Anderson's statement that … “ratesetting was a successful cost-containment strategy.” Most ratesetting studies focus on costs per day or per admission, and typically they do indeed show statistically significant savings over the periods studied. But these studies look at the wrong measure. Ideally, ratesetting studies should examine total health costs per capita because payers are concerned with their total health expenditures, not just hospital spending. If ratesetting saved hospital costs but caused even higher out-of-hospital expenditures, it would correctly be judged a failure.

Estimates of total health care spending per capita by State (Levit, 1985), covering a period when ratesetting programs were in effect (1972-82), show that total per capita spending growth across States was surprisingly uniform. From 1976 through 1982, the latest year for which State per capita health spending are available, the average annual growth rate for the United States was 12.4 percent. Maryland's growth rate was 12.6 percent; the other three States with stringent hospital ratesetting programs—Massachusetts, New Jersey, and New York— had growth rates of 12.1 percent, 11.6 percent, and 11.3 percent, respectively. California, which has never had rate controls but historically has been a hotbed of private sector efforts to contain costs, had a growth rate of 11.8 percent.

Absent data on total spending per capita by State, even advocates of State ratesetting agree that the next-best measure of the effects of all-payer ratesetting is hospital spending per capita. Unlike the per admission measure, the per capita measure takes into account the volume of admissions, which, under ratesetting regimes, may increase as hospitals and physicians try to offset lower rates per case with a higher volume of cases. Two studies have measured the effects of ratesetting on a per capita basis and have used regression analysis to hold the effects of other variables constant. One, covering the period 1969-78 (Coelen and Sullivan, 1981), found a statistically significant reduction in the rate of increase in hospital costs. The other (Morrisey, Sloan, and Mitchell, 1983), covering a longer period, 1968-81, found two out of five mandatory State ratesetting programs, New York and New Jersey, to have lower hospital expenditures per capita after 6 years of operation. The regression coefficient for Maryland was a negative 1.5 percent, i.e., the effect of ratesetting was an annual reduction of 1.5 percent 6 years after implementation, but the Maryland savings did not come close to meeting standard tests of statistical significance: “The Maryland per capita effect by itself was significant at the 56 percent confidence level” (Morrisey, Sloan, and Mitchell, 1984).

More recent data comparing hospital expenditures show that the performance of the mandatory ratesetting States has been relatively poor. For the 10 year period 1979-89, and the latest 3 years for which data are available, 1986-89, three of the four States with the most stringent ratesetting programs, Massachusetts, New York, and New Jersey, had per capita growth rates and per capita costs higher than the national average (Table 1). Maryland's average annual growth rate during the 1979-89 period was lower than the national average by 0.8 of a percentage point, but that difference shrank to 0.3 of a percentage point during the 1986-89 period.

Table 1

Average annual percent change in hospital expenditures per capita in regulated and competitive States: United States, 1979-89

Area	1979-89	1986-89
	Percent
United States	9.7	8.5
Regulated States
Maryland	8.9	8.2
Massachusetts	9.7	9.7
New Jersey	10.9	11.5
New York	9.6	9.3
California	8.4	6.8

SOURCE: Lewin/ICF: Analysis of Hospital Expenditures and Revenues, 1979-89. Washington, D.C. 1991.

California's performance was considerably better than Maryland's. The former's average annual percent change in per capita hospital costs was 8.4 percent over the 1979-89 period versus 8.9 percent for Maryland. During the 1986-89 period, the difference grew. California's growth rate dropped to 6.8 percent, while Maryland's dropped only to 8.2 percent. If relative performance in controlling hospital costs per capita were the criterion for choosing a model State, the clear favorite would be not Maryland, but California, the home of leading-edge efforts to implement the concepts of managed care in an environment where competition is regulated, so that it takes place along price and quality dimensions, rather than patient selection.

Conceptual problems

All-payer ratesetting essentially is resource allocation by bureaucratic fiat and/or by negotiations between government and providers. The history of economywide administered pricing, wherever it has been tried and however long it has been tried, suggests that it does not work over the longrun. I contend that in like manner, sectorwide economic planning and bureaucratic micromanagement—however well intentioned—will not work because it cannot work. By “work” I mean:

Allocate resources so that they provide maximum feasible satisfaction to consumers of the service being provided, given the resources available.

Stimulate continuous improvement in the value to patients of the services provided.

Ratesetting eventually fails because of three flaws that are inherently impossible to overcome. The first is that ratesetters can never acquire information that is timely enough and detailed enough to enable them to set prices that maximize satisfaction; reasonably well-working markets generate this information automatically. A good example of this problem is the prospective payment system for hospital operating costs. Each year the system becomes more baroque as yet another group asks for special exceptions or adjustments because the general payment rules fail to account for special circumstances; each year payments become more divorced from economic reality; each year the bureaucracies seek more information that is out of date by the time it is collected, much less analyzed. The result is growing cynicism, anger, and frustration on all sides, even though the vast majority of all the parties concerned are acting with the best of intentions.

The second inherent difficulty with ratesetting is that prices set by a regulatory process inevitably fail to reflect consumer preferences because regulatory authorities cannot possibly know what those preferences are in sufficient detail to ensure that value to patients is maximized. What prevails instead is some mix of bureaucratic preferences and the preferences of providers and user groups with disproportionate political power. Because consumer preferences are highly varied and influenced by specific circumstances, no set of general rules, no matter how complicated, can accommodate them satisfactorily.

Third, under ratesetting, decisionmakers' accountability for their decisions is tenuous at best. The effects of poor pricing and/or poor expenditure-control decisions do not show up for a long time, but their effects—because they affect entire systems—are large. Because accountability is not clear and direct, and response times are long, mistakes are corrected very slowly and only after they have become so obvious that even the most artful denials of responsibility are met with public disgust.

The universal lesson from economic history is that consumers receive the greatest value for their money, the most progress, and the most choice from markets that are highly competitive. The goal, therefore, should be to make health care markets workably competitive, rather than embrace an approach that has already been shown to be a failure. Health care markets today are not nearly as competitive as they can and must be, in large part because of government policies that encourage perverse behavior, such as the exclusion from taxation of employer-paid health insurance, excessive State-mandated health benefits, State laws banning restrictions on patient freedom of choice, insurance regulations that make insurance unavailable and unaffordable to many, inadequate subsidization of the poor uninsured, lack of pooling mechanisms for small employers and individuals to enable them to exercise market power, etc.

One promising approach, in my view, is “managed care,” which Alain Enthoven, in an unpublished 1991 paper, has summarized as follows:

“Limitation of tax-free employer contributions to employee health benefits to the price of Uniform Effective Health Benefits [a uniform basic benefits package based on the best available evidence of appropriateness and effectiveness] in the employee's area of residence as evidenced by the premium charged by the lowest-priced plan contracting with the Health Insurance Purchasing Corporation (HIPC) in the employee's area of residence….” [An HIPC is a voluntary, not-for-profit collective purchasing agent for small employers that would offer health care coverage by contracting with organized medical plans—Enthoven calls them Accountable Health Partnerships (AHPs)— to offer enrollment to all employees of participating employers.]

“Mandating all employers to … make a defined contribution equal to an amount in the range of 50 to 100 percent of the price of the lowest-priced AHP covering Uniform Health Benefits in the applicable HIPC area, and make payroll deductions and payments to the HIPC or the AHPs ….”

“A 'play or pay' payroll tax on all non-covered employment … combined with opportunity for part-timers to make subsidized purchases of coverage through a HIPC. A similar tax will be placed on the Adjusted Gross Income of nonemployed people with incomes …, with proceeds used to subsidize purchases of coverage through a HIPC ….”

“The federal government would collect a 'play or pay' payroll tax (equalized among states) …. States would contract with HIPCs to enrol] all people not covered through employment, Medicare, Medicaid, CHAMPUS or CHAMPUSVA [sic] ….”

“Public programs would have the opportunity to contract with HIPCs for coverage of their beneficiaries …. The Medicare program must evolve from its present fee-for-service orientation to purchasing coverage through HIPCs and AHPs. Initially, this should be done by creating per capita cost equivalents … in each HIPC area, and making capitation payments on behalf of those beneficiaries who wish to choose coverage through AHPs ….”