Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
The Nature of Technological Change: Incentives Matter! BURTON A. WElSBRODi Technological change involves two dimensions-invention and adoption. Much of the recent attention to cost containment in health care has focused on whether newly available technologies should be adopted, that is, whether they pass some form of cost-effectiveness test. It is clearly important to examine the question of whether a new health care procedure, diagnostic test, or drug is worth its cost relative to alternatives. Yet, by concentrating on the assessment of tech- nologies that already exist, one implicitly assumes that inventions of new tech- nologies are generated by exogenous factors over which society has no control. Are social choices limited to deciding when and whether to utilize newly developed medical advances? Or can society influence what kinds of advances are produced by the research and development process? In the economy generally, each year brings many new inventions that are simply ignored, temporarily or even permanently. Knowledge about how to produce new or improved products is not tantamount to actual provision; the cost may exceed what buyers are willing to pay. Does this separation of invention from adoption hold in the health care sector? Probably not! When life itself is at stake, society finds it excruciatingly difficult to withhold access to effective new technologies even when they are extremely costly, as is the case with organ transplants and mechanisms for sus- taining life for very low birthweight babies. Having to choose between making ~ Most of this paper was previously published as "The health care quadrilemma: An essay on technological change, insurance, quality of care, and cost containment" in the Journal of Economic Literature, 29:523-552, 1991. 8
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! i 9 such technologies available to all who can benefit from them, with enormous cost consequences, and restricting availability to persons with the ability to pray, or otherwise rationing access, poses choices with powerfully divisive social and political consequences. Society needs to recognize that to control these divisive forces society needs to intervene earlier in the process of technological change at the stage of re- search and development (R&D). This chapter highlights the means by which the health care insurance system sends signals to the R&D sector and how those signals affect incentives that generate tomorrow's new technologies. Those in- centives will determine whether the new technologies drive health care costs upward or downward. Incentives matter! Public policy can and does provide the R&D sector, where new technologies are developed, with incentives to pay more attention either to enhancing quality or to reducing costs. Depending on which of those is emphasized, tomorrow's health care technology arsenal will provide society with expanded opportunities either to extend life expectancy and enhance quality of life or to reduce costs. Public policy can choose the degree to which society concentrates on each of these social goals; unfortunately, they conflict. Throughout most of the post-World War II era the incentive signal sent from the health care finance system to the R&D sector has been to enhance quality of care regardless of costs. Over the past decade the incentive signal has been changing, shifting toward an emphasis on costs. It is not my purpose to deter- mine the socially appropriate balance between quality and cost. Rather, it is to emphasize the inevitability of such a choice and the instruments through which the choice can be made. One of the mechanisms currently being utilized to contain costs in hospitals .s the Medicare diagnosis-related group (DRG) system. The effectiveness of such a price control mechanism in a dynamic context of technological change has received little attention; it deserves more, however, since the mechanism, de- pending on how it is administered, can have a profound effect on the R&D sector. To see this, consider two alternative scenarios that differ dramatically in their response to technological change; they are polar opposites, and intermediate positions are possible, but they help to clarify social policy alternatives. In one, the DRG classification system is permanent and impervious to technological change; any new technology must fit into one of the existing DRG classes at prices that are not affected by the new technologies. In this case, an invention that would enhance quality of care but at an extremely high cost relative to the DRG price that a hospital would receive would have a weak market; a hospital considering adoption of such a quality-improving innovation would not be able to cover its costs unless it could supplement its resources with revenue from other sources such as gifts. Thus, it would be likely to reject utilization of the quality-enhancing invention. The rigid DRG system would discourage the R&D
10 BURTONA. WEISBROD sector from developing anything that raised costs, regardless of its effects on quality. An alternative scenario would posit a quite resilient DRG system. In this case every new invention that enhanced quality would result in either (1) estab- lishment of a new DRG class with an associated price that is high enough to make hospital utilization of the new invention profitable, or (2) a reinterpretation of an existing DRG class to encompass the new invention, together with an increase in the associated price. In this situation the R&D sector would see that anything that raised quality would be greeted with an effective price high enough to make it profitable for hospitals to adopt it. The cost-containment goal of the DRG system would be entirely undermined with respect to new technologies. Which of these scenanos, or which position between them, is likely to emerge from the political-regulatory process is unclear. The outcome, however, matters a great deal whether one's concerns are largely with cost containment or with the advance of medicine. At the Institute of Medicine conference on coverage and adoption decision- making about medical technologies, the principal focus was on the rapidly grow- ing area of technology assessment. This chapter, however, concentrates attention on the process that produces the new technologies requiring assessment. It high- lights some consequences of attempts by society to gain control over the health care system to achieve multiple goals encouraging technological change, ex- panding access to new technologies through insurance, sustaining or improving the quality of health care services, and controlling costs. INTRODUCTION Dunng the roughly four decades since the end of World War II, the health care system in the United States has experienced historically unprecedented changes in three dimensions. First, new technologies have revolutionized the ways in which health care is capable of being practiced. Almost all of today's armarnentarium of disease diagnosis and treatment devices and techniques were unknown 40 years ago. In the case of prescription drugs, for example, about 10 percent of the 200 top-selling drugs are new each year, and only 25 percent of the 200 top-selling drugs in 1972 remained in the group 15 years later (Cleeton et al., 1988~. Second, the role of health care insurance-private and public has expand- ed dramatically. By 1980, 82.5 percent of the U.S. population had some health care insurance, compared with fewer than 10 percent in 1940.2 2 Throughout the postwar period the expansion of private health care insurance has been spurred by federal tax policy. By making employer-financed health insurance nontaxable income to employees, federal policy distorted worker choice between health insurance and cash wages, encouraging excess health insurance (Feldstein and Allison, 1974; Pauly, 1974; Mitchell and Vogel, 1975; Mitchell and Phelps, 1976; Taylor and Wilensky, 1983; Chernick et al., 1987).
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 7 il Third, personal health expenditures have soared. From $300 per capita in 1950, they leaped to $1,493 in 1987 (all in 1982 dollars). The percentage-of the gross national product (GNP) devoted to medical care almost tripled over that period-from 4 to 11 percent (U.S. Bureau of the Census, 1979, p. 97; U.S. Bureau of the Census, 1989, p. 90; Letsch et al., 1988~. This chapter explains how the expansion of health care insurance has paid for the development of cost-increasing technologies and how the new technolo- gies have expanded demand for insurance. My goal is less to review the vast literature on the health care system and the rising level of real expenditures on it than to reflect on the dynamic interplay of incentives for the R&D sector to develop particular kinds of new technologies, the role of the insurance system in that process and, reciprocally, the long-run effects of new technologies (any new knowledge about health care) on the character of the health care insurance sys- tem. The broad model outlined here highlights the ways in which the quality of health care that can be supplied in a technically feasible manner at any point in time and the breadth of access to that care influence each other and the aggregate level of health care expenditures; but the model is not fully specified, nor is it tested rigorously. Thus, this chapter should be seen as a personal interpreta- tion largely positive, rather than normative, in character-of a period of enor- mous growth and massive change in both the practice and finance of health care. The central focus on technological change as an independent variable caus- ing changes in the form and extent of insurance coverage, and as a dependent variable, being influenced by incentives operating through the health insurance system highlights the impact of incentives; both the pace and types of research and development are functions of rewards that are endogenously variable, as are the comprehensiveness of insurance coverage and the breadth of access to it.3 The following propositions are set forward: (1) The amount of resources going into the R&D process and its direction during some time interval depend in part on the mechanisms expected to be used to finance the provision of health care in future periods, when the fruits of the research process become marketable. This is simply to say that R&D is influenced by expected utilization, which depends on the insurance system. Reciprocally, (2) the demand for health care insurance depends, in part, on the state of technology, which reflects R&D in prior periods. These relationships help to explain why (3) long-run growth of health care ex- penditures is a by-product of the interaction of the R&D process with the health care insurance system.4 I also examine briefly some effects of alternative forms of health care insurance on the quality of care, as distinguished from its quantity, 3 Other effects of health insurance, particularly on incentives for utilization of health services, have received considerable attention. For a recent and valuable review, see Pauly (1986). 4 Other forces also affect health care expenditures. Rising real income appears to have a positive effect on demand for health care; an income elasticity of +0.2 (or less) has been estimated from the RAND health insurance experiment (Manning et al., 1987).
2 BURTON A. WEISBROD and long-run changes in the definition of "health care" under insurance, as en- dogenous R&D alters the menu of technically feasible measures. To understand the markets in which health care is provided and financed, it is useful to consider ways in which health care differs from most other commodi- ties. First, it sometimes involves the preservation of life or, at least, major effects on the quality of life. Second, it is a technically complex commodity that abounds with informational asymmetries that are adverse to consumers (Arrow, 1963; Akerlof, 1970; Titmuss, 1971~. Third, and as a result of these two charac- teristics, "nonmarket" (government and private nonprofit) suppliers in the health care sector, especially among hospitals, nursing homes, and blood banks, play a large role in influencing the interaction between insurance and R&D.s Because health care affects the length and quality of life, many societies have come to accept the normative proposition that "high-quality" care ought to be made available widely, regardless of an individual's ability to pay. This assignment of property right the breadth of which is under continuing debate- results in pressure on government to finance access to some health care redistrib- utively. In the United States, private market financing of health care, by individ- uals and employers, has been supplemented by governmental resources- particularly through the Medicare and Medicaid programs and, to a smaller extent, through private charitable activities. Another reason in addition to providing widespread access-for society's willingness to intervene in private health care markets is the substantial informa- tional asymmetries, which give rise to economic and political demands for con- sumer protection (Arrow, 1963; Weisbrod, 1978, 1989; Hansmann, 19801. The claims that physicians "induce" demand (Arrow, 1963; Evans, 1974; Wilensky and Rossiter, 1983; Rossiter and Wilensky, 1984; Reinhardt, 1985; Cromwell and Mitchell, 1986; Stano, 1987), that they engage in "defensive" medicine- diagnostic testing and other practices that have no expected benefits for patient health but are defenses in "malpractice" suits (Gary et al., 1978; Zuckerman, 1984; Danzon, 1985)-and that they perform "unnecessary" surgery6 may or may not be valid; they are plausible, however, only if physicians are better in- formed than their patients (Pauly, 1979) and do not act as perfect agents.7 The importance of health care to life and well-being, combined with the limited abili- ty of consumers to make well-informed judgments about quality of care and with 5 Some readers may prefer the term nonprofit to nonmarket. Whatever term is used, the point is to distinguish private, profit-oriented organizations from the institutions of either government or the private nonprofit sector. To be sure, government and private nonprofit organizations operate in "markets," in the sense that exchange occurs. 6 A congressional subcommittee estimated that in 1977 there were 2 million unnecessary opera- tions, at a cost of $4 billion and with a loss of 10,000 lives ("Elective surgery: Cut it out," 1979). 7 Operationalizing the concepts of "induced" demand, "defensive" medicine, and "unnecessary" surgery-each of which reflects a market failure to the extent that it occurs-poses serious problems. These issues, however, are beyond the scope of this chapter.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 13 imperfect agency relationships with physicians, may help to explain why con- sumers of health care rely upon public and private nonprofit institutions~to an unusual degree. The remainder of the chapter proceeds as follows: The next section contains a brief outline of the recent history of the health care sector in the United States- its evolving technology, changing insurance/finance system, increasing level of real health care expenditures, and the advent of cost-control measures. Then I show how the constellation of services included in "health care" is endogenous, being affected by the interaction of the insurance system and the R&D process. The subsequent sections focus on the effects of R&D (technological change) on the health care insurance system, the reciprocal effects of the insurance system on the R&D sector, and the effects of alternative insurance systems on quality of care, with the state of technology fixed. I then summarize and point out some possible generalizations beyond health care. Finally, examining these interdependent relationships may help to explain some of the differences across countries in financing of health care and their roles in health care R&D, for the forces at work are not uniquely North American and the policy implications can be generalized. The United States is unusual, however, in the extent to which its actions as a producing and a consuming country influence the rate and direction of health care R&D. No other country is so major an actor in both the R&D (producing) sector and the health care (con- suming) sector. For most other countries, outputs of the R&D sector are essen- tially exogenous to their methods of financing health care, and their systems of health care finance are also essentially exogenous to their own R&D activities. Switzerland, for instance, is a substantial producer of health care R&D (especial- ly pharmaceuticals), but it is a small consumer; the United Kingdom and Japan, although they are not trivial elements in the R&D sector, are larger consumers of the outputs of that sector.8 It is the enormous size and, therefore, impact of both the producing and consuming elements in the United States that make it such a fine subject for study. A BRIEF RECENT HISTORY OF HEALTH CARE IN THE UNITED STATES: TECHNOLOGICAL CHANGE AND THE GROWTH OF INSURANCE COVERAGE One striking aspect of change in the U.S. health care system since World War II has been the dramatic increase in knowledge of means for diagnosing and treating illness. Fifty years ago, physicians were little more than diagnosticians, their activities being essentially "limited to identification of . . . illness, the pre- diction of the likely outcome, and then the guidance of the patient and his family 8 For a broader, European perspective on health care systems, see Organization for Economic Cooperation and Development (1990).
14 BURTON A. WEISBROD while the illness ran its full, natural course" (Report of the President's Biomedi- cal Research Panel, 1976, Appendix A, p. 3~. Today, the scope~of effective interventions includes kidney dialysis, organ transplants, polio vaccines, arthro- scopic surgical techniques, computed tomography scanners, nuclear magnetic resonators, and in vitro fertilization. As recently as a decade ago, heart and liver transplants were virtually unknown, but their numbers have soared, from 62 and 26, respectively, in 1981 to 1,441 and 1,182, respectively, in 1987 (U.S. Bureau of the Census, 1989, table 166~. At the same time that the technology of health care has been changing so dramatically, the system for financing health care has also been revolutionized. In the quarter century between 1950 and 1973 alone, the share of health care expenditures that was met by insurance more than tripled, from 12 to 41 percent (U.S. Bureau of the Census, 1975, table 105~. The mix of insurance between private and government insurance also changed during that period; while total private expenditures on health and medical services were growing almost six- fold, from $8.7 billion to $59.8 billion (current dollars), government expendi- tures (Medicare and, to some extent, Medicaid) were leaping fourteen-fold, from $2.5 billion to over $37 billion (U.S. Bureau of the Census, 1975, table 100~. Insurance coverage for "major" or "catastrophic" health care costs has also risen sharply, from 22 percent of the population in 1960 to 73 percent by 1984 (U.S. Bureau of the Census, 1987, tables 1, 2, and 137~. Initially, most health insurance was of one particular type-covering a limit- ed menu of only hospital services, perhaps after a small deductible and paying ("reimbursing") the hospital for the particular services provided to a patient, the payment being equal to the "actual" average cost of treating that patient with whatever technology was used (Stevens, 1989~. Included was an approximation of the average variable cost of any diagnostic or therapeutic procedures per- formed on the patient's behalf, plus a per diem payment for room, board, and basic nursing services and, in the case of for-profit hospitals, a markup. Thus, the payment received by the hospital was determined retrospectively and was a function of endogenous decisions by the hospital and physician as to the length of stay and the resources deployed in treating each specific patient. With hospital revenue being a function of the cost of services provided, there was little incen- tive to weigh costs against patient benefits. Any diagnostic or therapeutic re- source that had a positive expected value of benefits could be provided in a financially feasible manner, and even when there was great uncertainty about the probability distribution of benefits from a new, more costly technology, the ab- sence of a budget constraint encouraged its adoption. By the 1970s, however, the growth of real expenditures on medical care- reflected in rising private insurance premiums, Medicare budgets, and the share of GNP devoted to health care had become matters of growing public concern. Some attributed this "health care cost inflation" to the insurance system and its effect on demand; retrospective payment arrangements, operating through the
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 1 J insurance system, were encouraging "overuse" of medical resources (Feldstein and Friedman, 1977; Pauly, 1986~. The result was a spate of reforms designed to force health care providers to consider the cost consequences of their decisions. This was done by making more of providers' revenue "prospective." Health maintenance organizations (HMOs) and, beginning in October 1983, the Medi- care DRG system for pricing hospital services are the preeminent examples of this type of reform. Both HMOs and the Medicare DRG prospective payment system confront suppliers with the incentive to be more cost-conscious, but they differ in the comprehensiveness of that incentive. Under the current DRG system for paying hospitals, the "fixed" payment for a particular patient is supplemented by addi- tional payments to cover capital costs; thus, there is some incentive for hospitals to substitute capital for labor.9 In addition, under the DRG system, as under the previous retrospective pricing system, a hospital's revenue is a function of its admissions of patients; this produces an incentive to hospitalize rather than to utilize approaches that involve nonhospital inputs such as drugs, broad medical management approaches, and instruction of patients in ways to prevent and alle- viate problems through lifestyle and dietary measures. HMOs, which have a contractual responsibility to provide medical services, not simply hospital treat- ment, and receive a flat annual fee per member, maintain a greater financial incentive to utilize alternatives to hospitalization. To the extent that cost-based insurance has been at the root of the rising expenditures on health care, however, the causal mechanism is less clear than it seems. The "moral hazard" effect of insurance could cause patients and their physician-agents to utilize more health care resources, and therefore aggregate health care expenditures to be greater than they would otherwise be; yet it does not follow that insurance would cause expenditures on health care to grow more rapidly. Something had to be changing. That "something" could have been the state of technology that, as we will see, was expanding in a systematic direction as a consequence, at least in part, of the particular form of insurance that had been adopted. An expanding health care insurance system more widespread coverage of people and broader coverage of health care resources such as phar- maceuticals and chiropractic services- might also account for the growth of health care expenditures, but this explanation would pose the question of why insurance coverage would be expanding.l° The major theme of this chapter is that the demand for health care insurance and the process of technological change are interdependent. A shift away from insurance that paid hospitals and physicians on the basis of endogenously deter 9 I owe this point to an anonymous reviewer. 10 Even with constant technology, real costs of health care could increase if input prices rose for example, because of increased unionization of hospital labor and this could increase the demand for insurance, ceteris paribus (that is, other things being equal).
16 BURTON A. WEISBROD mined "costs incurred" and office visits to insurance that paid amounts that were largely independent of costs incurred on behalf of any particular patient repre- sented a major change. It altered incentives to use existing health care resources (that is, their rate of diffusion and utilization), and it altered earlier incentives for the R&D sector to invest in developing medical care techniques that were of higher quality but more costly. As noted above, the shift in the nature of health insurance has occurred in two principal forms-expansion of HMOs and adoption of the DRG system of hospital pricing. In the decade of the 1980s alone, enrollments in HMOs more than tripled, from 9.1 million in 1980 to 28.6 million in 1987 (U.S. Bureau of the Census, 1989, table 148~. Under the DRG prospective payment system, a hospi- tal receives payment (prices) for treatment (e.g., of appendicitis) based on indus- try-wide costs for each of the 468 DRG categories. Thus, conditional on admis- sion of a patient with a particular diagnosis, what a hospital faces is a price for treatment that is essentially independent of the actual resource cost it incurs (Hogan, 1988. Both HMOs and the DRG system of pricing hospital services are potentially revolutionary in their incentive effects on R&D. The fact that the principal objective of each of these forms of prospective pricing was fiscal control is not in doubt (Pauly, 1986~. Several related matters, however, are far from clear and deserve more research: Why did the shift in insurance mechanisms, from retro- spective to prospective, occur when it did? Why did the United States ever start with insurance based on retrospective and fee-for-service pricing; after all, the incentives that cost-based pricing generated were, or at least should have been, apparent long ago, and the fiscal problem, as manifested in the rising share of GNP devoted to health care, has been growing for decades. In some current research, Paul Boben (1989) presents a model in which retrospective pricing of hospital services and physician services (through fee-for- service payments to physicians on the basis of "usual and customary" fees) is allocatively efficient when there is little insurance coverage and health care pric- es are determined in relatively competitive markets, but diminishes as that cover- age spreads. In this model the discipline of prices on patient and provider behav- ior that prevails when few people have insurance gives way to growing price insensitivity (inelasticity) with the expansion of insurance. Thus, a "tipping point" is reached, at which the usefulness of market-determined prices as signals of opportunity costs becomes less than its cost in terms of distorted resource il The pricing system is not entirely rigid. For example, a hospital may collect from Medicare more than the DRG price for a limited number of unusually high-cost "outliers." 12 The DRG system of hospital service pricing initially applied only to Medicare patients. It has subsequently been expanded, however, through private arrangements, to a growing number of other patients who are not covered by the Social Security Medicare law.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 17 allocations (the moral hazard problem). Such modeling of the social choice of the insurance system is in its infancy. Many of the issues raised above have received scant attention in the litera- ture. The effect of advancing technology on health care financing arrangements, the incentives for R&D inherent in those financial arrangements, and the implica- tions of those arrangements for the quality of the care provided are each the subject of later sections, where I will also consider the inevitability that health care expenditures would soar in the post-World War II era. But, first, how do I define "health care"? How is it affected by technological change and how does its definition affect insurance coverage? DEFINING "HEALTH CARE" Up to this point I have been discussing the market for "health care" without defining that market carefully. The endogeneity of the definition of health care under insurance contracts has received some attention (Goddeeris, 1984a,b). Consider two nonmutually exclusive hypotheses concerning the causes and con- sequences of the definition of health care under insurance: (1) the operational definition of health care, under insurance contracts, is a function of the state of medical technology; (2) the state of medical technology today is a function of economic and political responses to prior definitions of health care coverage under insurance contracts. The way health care is defined under insurance contracts is important for a number of reasons, positive and normative. It affects the level of insured expen- ditures, the incentives to utilize resources that are covered relative to those that are not (Feldstein, 1988), and the incentives for the R&D sector to explore vari- ous potential health-promoting technologies. At the operational level, the defini- tion of health care is at issue when coverage for chiropractic care or for "experi- mental" drugs or other "new" technologies is debated. The effect of health care insurance on incentives for R&D depends on the operational definition of health care that is, on the boundaries of the insurance contract. Health insurance contracts do not offer the option of coverage only for particular subsets of technologies, such as those already available at a given point in time (Goddeeris, 1984b; Goddeeris and Weisbrod, 1985; Baumgardner, 1989~. A reasonable conjecture, however, is that health care expenditures today would be substantially lower than they would be if health care were being defined, for insurance purposes, as limited to the use of medical technologies available at the time the policy took effect or at some other fixed date. The more broadly health care is interpreted under the contract, and the more responsive it is to changes in technology, the broader the range of activities over which insurance will encour- age R&D. What determines how health care is defined? I suggest that the R&D pro- cess causes the definition of what is covered by health insurance to change in
18 BURTON A. WEISBROD systematic ways. Technological advances are not only expanding the range of medical capabilities for extending life and enhancing health status, as the latter term is customarily understood, they are also presenting opportunities to deal with problems not conventionally considered to be "illnesses," in ways not con- ventionally considered "health care."~3 An illustration of this causal process is the current debate over whether health insurance should necessarily cover in vitro fertilization. This has become an issue only in the past few years, when advances in medical capabilities made such fertilization technically feasible. An advance in medical technology has led to pressure to expand the traditional definition of insurance coverage, pressure being felt now through the political system; by 1988 such insurance coverage had been mandated in five states (U.S. Congress, 1988),14 and by the end of 1989, laws requiring insurers to cover such "advanced" treatments for infertility had been enacted in nine states and bills had been introduced in eighteen others (Nazario, 1989~. The effect of technological change on the health insurance market can also be seen with "experimental" drugs. The decision to term a drug experimental is often seen as a statement of the degree of professional knowledge about its safety and efficacy. It is, however, also a statement of whether the drug will or will not be deemed "health care" for insurance purposes, since insurance typically does not cover "experimental" technologies. For example, as long as the AIDS drug, zidovudine (AZT), was termed experimental, its exclusion from coverage under health insurance involved each patient with costs that, until 1990, were in excess of $8,000 per year, even though conventional hospital-based treatment was cov- ered in the traditional fashion. The hypothesis that the definition of health care is endogenous to the eco- nomic-political system in which health care insurance is defined, provided, and financed has important implications, to the extent that it is valid. If insurance coverage is defined, as it has been, to encompass new technologies regardless of the costs involved and to encompass an ever-widening concept of health care that is, itself, responsive to the development of new technologies, the R&D sector will continue to face incentives that reward costly new measures relative to cost- reducing innovations. Such a reward system may not be incentive compatible; 13 Another example of the need to decide, as a matter of public policy, how to define operationally what is health care involves people with physical disabilities. Surgery and physical therapy illustrate "traditional" health care resources employed to reduce the disabilities. City buses that are wheelchair accessible are unquestionably valuable to the disabled; whether their cost should be regarded as health care expenditures and covered by health care insurance is another matter. 14 As of May 1988, Arkansas, Hawaii, Maryland, Massachusetts, and Texas had enacted legislation requiring private insurers to provide some coverage for in vitro fertilization procedures. Delaware Blue Cross/Blue Shield began offering coverage voluntarily in response to legislative activity (U.S. Congress, 1988).
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 19 new technologies may be developed even though they are welfare decreasing, in the sense that the insured population is not willing to pay the real cost of develop- ing and applying the technology (Goddeeris, 1984b; Baumgardner, 1989~. EFFECTS OF RESEARCH AND DEVELOPMENT (TECHNOLOGICAL CHANGE) ON THE HEALTH CARE INSURANCE SYSTEM Advances in medical technology involving both diagnostics and treat- ment have been, at least arguably, a driving force behind the rapid growth of health care expenditures (Altman and Blendon, 1979; LaCronique and Sandier, 1981; Showstack et al., 1982; Aaron and Schwartz, 1984; Wilensky, 1987~. For example, the announcement for a conference cosponsored by the American Med- ical Association in October 1988 acknowledged the benefits from new medical technology but also cited the position that the growth of medical technology was a primary cause of the quadrupling of per capita health care costs between 1970 and 1986. Even if this causation occurs, however and existing research is far from conclusive on the matter the mechanism through which it works is not well understood. Neither is it apparent that technological advances would neces- sarily increase health care expenditures rather than decrease them. One mechanism through which technological change could foster increased expenditures on health care would be through its effect on the health care insur- ance system. If a previously untreatable condition becomes treatable, a possible outcome is that an individual could encounter a larger, but unpredictable, medi- cal care expense for treatment than was previously the case; thus, both the mean and the variance of an individual's health care expenditures associated with that condition could increase. Pooling of such risks is a logical response. In addition to the increased expected demand for private insurance, collective demand is also likely to in- crease; the fact that health care, particularly when it has a major effect on life expectancy or quality of life, is widely viewed as a "merit" good (or "altruistic externality" [Pauly, 19863) results in public pressure on government to ensure that the care is available to whomever needs it medically, regardless of their ability to pay. An example of such a merit good is organ transplant technology. Reacting to the lifesaving aspects of the new transplant technology, the federal govern- ment's Task Force on Organ Transplantation proposed that government pay for all organ transplant operations that patients cannot afford (Pear, 1986~. Some- what similar legislation, enacted in 1972 in response to the development of kid- ney dialysis (not transplant) technology, had the clear effect of increasing health care expenditures; no patient was rationed from access to the technology and the technology, while life-extending, was more costly in resource terms (although not necessarily in net benefit terms) than simply allowing the victim to go with
20 BURTON A. WEISBROD out treatment and, hence, to die.ls The interplay of financial and political forces following the development of the dialysis technology (Rettig, 1980; Rettig and Marks, 1983) and the massive public expenditures that ensued may help to ex- plain why there has been no subsequent U.S. legislation covering such complete treatment for any other disease, and why the British National Health System continues to restrict access to dialysis for persons over age 55. Life-extending technologies highlight the ambiguity of the concept of a tech- nology being "expenditure increasing." Total health expenditures over a per- son's lifetime are likely to increase if the person lives longer, although that is not necessarily the case. However, expenditures per year of life can decrease even if lifetime expenditures increase. A new technology that increases the cost of treat- ing a particular disease but that is successful in increasing life expectancy suffi- ciently to decrease expected health care costs per year of life could diminish the demand for health care insurance; my conjecture is that it would not, but this deserves more attention.l6 The point is that technological change need not in- crease demand for insurance, even if the change increases the expected cost of treating a particular illness. It could take forms that decrease either the aggregate expected health care cost for all illnesses or the variance. Demand for insurance would also decline even if a new technology increased the aggregate expected cost of treatment, if the variance decreased sufficiently.l7 If the focus is on the treatment of specific diseases, one finds that some innovations decrease the demand for insurance by decreasing both the expected cost of Beating that illness and the cost vanance. Administration of the Salk and Sabin polio vaccines, for instance, is quite inexpensive and by providing immu- nity to the ravaging effects of polio, they have reduced indeed, virtually elimi- nated- the variance in health care expenditures associated with contacting that disease and using costly treatment technologies. The potentially enormous ex 15 The view that dialysis and organ transplants are cost (or expenditure) increasing, ceteris paribus, deserves further comment as to what is embedded in the ceteris paribus assumption. One element is the set of probabilities of contracting all other diseases. The assumption that these probabilities are constant with respect to the organ transplant or dialysis decision may not be valid; a person whose life is "saved" through the use of one of these technologies may well face a greater probability of dying from other causes than do people who have not been victims of kidney disease. 16 The effect of increasing life expectancy on total health care expenditures as a percentage of GNP is yet another matter. This depends on the productivity of people whose lives are extended, as well as on longer-run effects on birth rates. 17 Even if technological change increases demand for insurance, it need not follow that the amount of insurance purchased would increase. Insofar as the technological changes were cost increasing, the price of insurance coverage would increase, which would diminish insurance purchases. In fact, the negative price effect of an increasing price for health care insurance appears not to have offset the positive demand-shift effect of technological change, judging from the growth in the fraction of the population with insurance; to be sure, however, much of the growth of insurance coverage over the last two decades has been through government rather than direct consumer purchases in private markets.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 21 penditures that have been eliminated, which include those associated with de- cades of use of an iron lung and the lifelong costs associated with being crippled, exceed the cost of providing the vaccinations (Weisbrod, 1971~. Thus, the polio vaccines, like many other vaccines, have the effect of reducing an individual's expected level of expenditures for treating the disease, as well as the variance around that mean. In the process they reduce the demand for health care insur ance. Organ transplant technology, on the other hand, is a technological advance that has increased both the mean and the variance of desired individual expendi- tures conditional on medical need. Before the new technology a person with serious liver malfunction, for example, simply died, with comparatively little health care expenditure.18 With the new technology it has become possible to spend vast sums on effective treatment. A single liver transplant operation can cost $200,000 or more, and subsequent medical attention and medication to pre- vent organ rejection typically total $10,000~$20,000 annually for life (Hudis, 1986~. Thus, a healthy person with some probability of developing liver disease faced a larger expected financial cost of treatment once the new technology was developed and a greater variance in cost; conditional on remaining healthy, the person would spend zero on treatment of his or her liver under either technologi- cal state with or without the transplant capability. Conditional on contracting liver disease, however, the person would spend a great deal more on treatment once the new technology became available. As a result, the development of transplant technology increased private demand for health care insurance, ceteris paribus. This is distinct from the increase in demand associated with the merit good-related desire to provide access to lifesaving technology to everyone re- gardless of ability to pay.19 These two cases of technological change polio vaccines and organ trans- plants illustrate several points: (1) some new technologies increase the expect- ed health care expenditures for victims of a given disease, ceteris paribus, while others decrease them; (2) some new technologies increase the variance of health care expenditures for victims of a given disease, ceteris paribus, while others decrease it; (3) a technology that increases the mean and variance of health care expenditures for a particular disease would tend to increase the demand for health care insurance, while one that decreased them would tend to reduce the demand for insurance. This latter proposition suggests the following conjecture: since 18 In fact, however, little is known systematically about the amount of health care expenditures associated with attempts to cope with the debilitating effects of liver dysfunction (or other terminal illnesses), even when life is not prolonged. 19 Positive income effects associated with rising income could also account for an increase in the demand for health care insurance. One might expect, however, that the income elasticity would be negative, not positive; increased income, ceteris Paribas, would increase the person's ability to self- insure (Mossin, 1968).
22 BURTON A. WEISBROD we observe growth in insurance coverage, private and public, the preponderance of technological change in recent decades has apparently increased-the means and variances of health care expenditures associated with various diseases rather than reduced them. Society has tended to develop a growing number of new technologies that permit higher levels of health care expenditures.20 Vaccines and transplants also illustrate stages in technical progress. Biolo- gist Lewis Thomas (1975) distinguishes among three levels of technology in medicine. "Nontechnology" tides patients over diseases that are poorly under- stood. It largely involves reassuring patients, providing hospitalization and nurs- ing, but with little hope; "tilt is what physicians must do now for patients with intractable cancer, severe rheumatoid arthritis, multiple sclerosis, stroke, and ad- vanced cirrhosis" (p. 37~. At a higher level is "halfway technology." This includes dealing, after the fact, with the incapacitating effects of diseases "whose course one is unable to do very much about" (p. 37~. It is a technology that adjusts to disease or postpones death. Examples include organ transplantations, the use of artificial organs, and treatment of cancer through surgery, irradiation, and chemotherapy. The cancer measures are halfway technologies because they are directed at "already estab- lished cancer cells, but not at the mechanisms by which cells become neoplastic" (p.39~. Finally "high technology," exemplified by immunization, antibiotics for bac- terial infections, and prevention of nutritional disorders, "comes as a result of a genuine understanding of disease mechanisms, and when it becomes available, it is relatively inexpensive . . . to deliver" (p. 40~. Thomas described the state of technology at a point in time not the process of change. If, however, one thinks of a dynamic process, in which knowledge tends to grow from the first of the three levels to the second and then the third, the cost function associated with any particular disease might be an inverted U shape, it is plausible, although certainly not verified, that health care costs are highest for the halfway technologies. In the extreme case of a nontechnology, when the knowledge base is so weak that there is nothing useful to be done, costs are likely to be low, as they are when the high-technology state of knowledge is reached. The evolution of knowledge about polio is a useful example. Two genera- tions and more ago, the nontechnology stage prevailed. Many victims of the disease died quickly as a result of paralysis; for them, the effects were disastrous, but the attendant health care costs were small. Development of the halfway (iron lung) technology prolonged life but at substantial cost. The high-technology 20 Treatment of heart attacks is another illustration. One study showed that between 1972 and 1982, treatment of myocardial infarction involving more complex technologies such as cardiac imag- ing, angiography, and coronary bypass graft surgery was associated with a tripling of physician costs per case (Sawitz et al., 1988).
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 23 polio vaccines (Sabin and Salk vaccines) dramatically reduced the costs associat- ed with polio, virtually eliminating it in the United States there were five cases in 1985, compared with over 38,000 in 1954, before the vaccines were devel- oped.21 Insofar as the inverted U-shape relationship holds between the state of tech- nology and the resource cost per case, there is an interesting implication. The aggregate effect of technological change on health care costs will depend on the relative degree to which halfway technologies are replacing lower, less costly technologies or are being replaced by new, higher technologies. The develop- ment of halfway technologies was implicitly encouraged by the cost-reimburse- ment insurance system that has dominated hospital and medical care until recent- ly, because there was little or no incentive for medical care providers to avoid costly technologies that were even marginally effective.22 Empirical research on how, and how much, the medical R&D process is now being affected by the shift to a prospective-pricing incentive system for cost control is in its infancy; there would seem to be an incentive for R&D to shift toward mechanisms that would bypass the high-cost, halfway states of technology. Depending on whether technological change is predominantly from nontech- nology to halfway, rather than from halfway to high or from nontechnology to high, the demand for insurance is likely to differ. With the demand for insurance being a function of uncertainty of loss, demand should tend to increase most rapidly when changes in technology are of the expenditure-increasing, halfway type. Costly new surgical techniques such as organ transplants and the use of artificial body parts spur the demand for insurance; low-cost vaccines diminish it.23 Why has there been relatively more development of technologies like organ transplantation than like the polio vaccines? Why, that is, has technological change in health care been "expenditure increasing"? Is it more than chance? To begin examining this issue, I turn to the effects of various kinds of insurance arrangements on incentives for the R&D sector to develop alternative types of technologies. For just as the forms of technological change affect the insurance system, so, too, does the insurance system affect the direction and pace of tech- nological change. Depending on the type of insurance available to consumers, the R&D sector faces differing incentives to search for cost-reducing, "process" innovations relative to quality-increasing but cost-increasing, "product" innova- tions. 21 Vaccines appear to be more cost reducing than they really are. If vaccination cost is, say, $5 per person, and if the incidence of the disease is 1 in 40,000, then the vaccine cost per case prevented is $200,000. That may or may not be resource-cost saving, at least with respect to health care costs. 22 Halfway technologies are not the only type of R&D encouraged by cost-based, retrospective insurance. Any technology with positive expected benefits is encouraged. 23 Thomas' typology applies to technologies used for prevention and treatment. While Thomas does not deal explicitly with technologies used for diagnosis, we can think of those as complements to treatment; that is, costs of treatment include costs of determining which treatment mode to use.
24 BURTON A. WEISBROD EFFECTS OF THE INSURANCE/FINANCE SYSTEM ON RESEARCH AND DEVELOPMENT Theory suggests the probable direction of the health care finance system's effects on R&D. Depending on hospitals' and physicians' incentives to adopt new technologies (which are contingent on the insurance system through which providers are paid), the R&D sector can face quite different financial incentives for both the level and direction of research. Fiscal pressure on health care pro- viders to contain costs will affect the market for adoption of innovations, and by so doing will alter R&D in predictable ways. The effects of insurance on R&D are not simply based on the existing insur- ance system, but on the system expected to exist in the future. The process of developing new medical technologies involves years of planning and research and, when drugs and medical devices are involved, more years of clinical trials to obtain approval by the Food and Drug Administration; in the case of pharmaceu- ticals, a period lasting 12-15 years is typical between the initiation of a research process and the marketability of a drug. As a result of this lengthy process, the R&D process depends on forecasts of the health insurance system, for the form of expected insurance coverage will determine the strength of the market for new products. If, for example, decisionmakers in the R&D sector believed that devel- opment of a particular technology that was costly yet effective would cause gov- ernment to expand insurance to cover it as was done with kidney dialysis- there could be an incentive to develop the product even though it was not covered under existing insurance. By directing attention to the effect of health care insurance on R&D, I do not imply that insurance is the only force affecting R&D. Among other forces are the state of scientific knowledge, which affects the probability of scientific suc- cess from additional research; demographic variables, which affect the size of potential markets for new products; and political influences on the budget of the National Institutes of Health (NIH), which finances basic research. With respect to NIH, it would be useful to learn more about the way the size and allocation of its scientific research budget are influenced, perhaps quite indirectly, by the health insurance system through its impact on the eventual market for new tech- nologies of various types. Hospitals, physicians, and other health care providers select the resources used to treat any particular patient from those that are technologically feasible and subject to revenue constraints. These constraints depend partly on the insur- ance system, which influences both the diffusion of existing technologies and the expected profitability of potential new technologies (Newhouse, 1981, 1988; Goddeeris, 1987~. Thus, the following proposition requires testing: the insur- ance/finance system affects the incentives facing the R&D sector to develop new health care technologies of various "types." Since the demand confronting the health care R&D sector is derived from the demand facing health care providers,
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 25 alternative insurance/finance systems will have different long-run effects on the demand for innovations. In particular, insurance mechanisms can differ in the incentives they imply for reducing costs relative to enhancing quality. The two types of insurance payment mechanisms-"retrospective," which pays a provider on the basis of "costs" incurred, and "prospective," which pays sums that are independent of those costs incurred24 imply profoundly different incentives for both the development and diffusion of new technologies. The claim that hospitals operate according to some "technological impera- tive" that determines medical choices (Fuchs, 1986) and that drives hospitals to adopt the latest technology regardless of cost may well have been correct, but the reason may have been less mystical than the terns suggests. The economic incen- tives explaining the "rapid and indiscriminate adoption of "medical] innovations" (Fuchs, 1986, p. 29) and "the proclivity of doctors and hospitals to adopt almost any plausible new thing-drugs, surgical methods, equipment that increases capability in any dimension . . . without regard to cost" (Nelson, 1972, p. 56) have been documented for such technologies as intensive care units, cobalt thera- py, and the electroencephalograph (Russell, 1979~. One "explanation" offered for the insensitivity to cost is an alleged lack of training of physicians and hospi- tal administrators in weighing marginal benefits against marginal costs (Battistel- la, 1984~. Even if this is valid, the impact of insurance-based incentives may well be powerful; "methods of third party payment. . . [do] not give tdecision- makers] any inducement to acquire that ability" (Fuchs, 1986, p. 30~. Analyses of the effect of insurance on the adoption or diffusion of technolo- gies have tended to concentrate on technologies that have already been devel- oped. Less attention has been given to the implicit incentives for the R&D sector to develop various types of innovations. Retrospective pricing sends a clear signal to the R&D sector: develop new technologies that enhance the quality of care, regardless of the effects on cost. Careful analysis remains to be done to distinguish causation from spurious correlation, but it appears that in the post- World War II era this signal produced the two results that could be expected: historically unequaled improvements in medical care technology~lrugs, devic- es, diagnostics, clinical procedures, and so forth and unprecedented growth in health care expenditures.25 Transplantation of natural organs has already been mentioned as an example of a high-cost medical innovation made more likely by retrospective insurance. 24 Arrow (1963) identifies three types of insurance, the third being "indemnity." This type, howev- er, is a special case of prospective coverage in the sense that the insurer pays a fixed amount, conditional on a loss, but independent of the magnitude of the health care costs actually incurred. The indemnity might take the form of a fixed dollar payment for the loss of a limb or for a given illness. If it took the form of a fixed dollar payment per day of hospitalization, it would have the character of retrospective-type insurance. 25 Such increased costs might or might not pass a full benefit-cost test. The point, however, is that they contributed substantially to the accelerated growth of health care expenditures.
26 BURTON A. WEISBROD Another example is development of a wide range of implantable artificial joints and artificial organs. The human body has become increasingly like an automo- bile, with replacements available for an ever-growing number of parts an arm or a leg, at about $2,000, an elbow at $1,200, an ear at $10,000, and a heart at $50,00~$80,000. They are even available in small, medium, large, and extra- large sizes (Kleinfield, 1983~. "Installation," of course, is extra and, as with auto parts, is typically many times greater than the price of the part. Technological advances in recent decades have given us spectacular innova- tions, but with scant attention to the resource costs of utilizing them. Open heart surgery can replace clogged arteries (coronary artery bypass graft surgery) but at a cost averaging $46,000 (National Center for Health Services Research and Health Care Technology, 1988~. A baby born 2.5 months prematurely and weighing well under 2 pounds (907 grams) can be kept alive, but at a cost of $90,000 and with a 10 percent survival rate (French, 1989~. Ultrasound technol- ogy, computed tomography scanners, positron emission tomography (PET) scan- ners, and other diagnostic tools aid in disease detection but often at costs of tens of thousands of dollars per case detected-not counting the subsequent costs of surgery or other treatment. The PET scan, which aids in detecting heart disease at a cost of about $1,800 per test many times this for each case of heart disease detected has been argued to be only "slightly" better than single emission com- puted tomography, which costs less than half of a PET scan (Schiffman, 1989~. Under retrospective, cost-based financing, even small improvements have been adopted by physicians, hospitals, and other institutions that have had little or no incentive to balance social benefits against costs. Consider, now, the reward structure implicit in an alternative insurance/fi- nance system-prospective payment, in which payment to a service provider is exogenous to provider decisions, conditional on admission of a patient. The particular version that is being applied to hospitals' Medicare patients and, in- creasingly, to other patients as well, confronts a hospital (but not the patient's physician) with an exogenously determined set of prices, one for each of 468 diagnoses made at the time of admission.26 No longer is gross revenue for treating a particular patient a function of the hospital's decisions on use of re- sources. Financial incentives for hospitals under such a prospective payment arrange- ment differ diametrically from the incentives under a retrospective payment ar- rangement. With a hospital's revenue being exogenous for a given patient once admitted, and an HMO's revenue being exogenous for a member for the given year, the organization's financial health depends on its ability to control costs of 26 In some instances diagnostic categories can be altered after admission, on the basis of informa- tion not available at admission. This produces some degree of revenue endogeneity, because the hospital and physician can decide on the amount of exploratory effort.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 27 treatment.27 Thus, under a prospective payment finance mechanism, the health care delivery system sends a vastly different signal to the R&D sector, with priorities the reverse of those under retrospective payment. The new signal is, develop new technologies that reduce costs, provided that quality does not suffer "too much." (The meaning of "too much" will be examined below.) When a ceiling was placed on government payment for kidney dialysis, the direction of technical change was affected; large surface dialyzers were devel- oped that cut the time required per session nearly in half, from ~8 hours down to 3.5~.5 hours. This led to substantial savings in professional labor costs, which are a major cost component (Rettig, 1980~. The shift to a prospective payment system (PPS) under Medicare appears to have brought about some of the expected changes in utilization of health servic- es. PPS has not diminished use of intensive care units, but it has apparently decreased use of such diagnostic procedures as chest X-rays; in the three years prior to PPS, 198(}1983, the mean annual change in the number of chest X-rays per Medicare patient discharge was zero, whereas for the 1983-1985 period it decreased by 8 percent (Sloan et al., 1988~. HMOs also present providers with an incentive to increase attention to costs relative to medical benefits. HMOs which are, in effect, mergers of health care providers and insurers-can be expected to adopt more slowly any new technolo- gy that is cost increasing, even if it is more effective, than would a provider facing a retrospective pricing system. HMO members have been found to have lower costs per patient-year than do nonmembers whose insurance was based on retrospective costs, largely attributable to a 30 percent lower rate of hospitaliza- tion (Luft, 19811; but the rate of introduction of new technologies does not ap- pear to differ, at least as that is reflected in rates of change of per capita costs. The growth rate in total costs per person (including out-of-pocket costs) in the 1960s and 1970s appears to have been about the same (Newhouse et al., 1985) or "only slightly lower" (Luft, 1980) for people in, and those not in, HMOs, after making some adjustments for selection bias. 27 While hospital revenue is largely exogenous once the patient is admitted, a hospital can influ- ence both its gross and net (of cost) revenues through a variety of mechanisms for controlling admis- sions. A nongovernmental hospital may, for example, choose not to provide particular services such as an emergency room; it can decide which physicians may serve on its medical staff and, hence, which may admit patients; and it can provide its affiliated physicians with subtle but clear signals to "encourage" patients with complex problems to utilize governmental hospitals. Recent research is disclosing that, with the advent of prospective pricing for Medicare patients at most nongovernmental hospitals in 1983, there has been an increase in admissions to Veterans Affairs hospitals, which are not included in the prospective DRG pricing system; therefore, we might expect them to receive more of the patients with illnesses likely to constitute financial "losers" to for-prof~t and voluntary nonprof- it hospitals (Wolfe, 1989). In the long run, when the location of a hospital is variable, there is additional discretion for nongovernmental hospitals to locate in areas that are less likely to generate unprofitable cases.
28 BURTON A. WEISBROD The longer-term effects of PPS and ElMOs on the R&D sector are more difficult to discern. There has been no formal modeling of the long-run effects on technical change of alternative payment systems for hospitals and physicians. Early literature attempting to explain the rising level of health care expenditures did not identify an important role for technological change. Subsequent literature sometimes directed attention to the effect of technological change on health care costs (Altman and Blendon, 1979), but that change in "quality and style of hospi- tal care" was assumed implicitly to be exogenous-captured econometrically, perhaps, by a time trend (Feldstein, 1971~. The rate of diffusion of a number of existing technologies has been found to be responsive to insurance-related incentives (Russell, 1979; Romeo et al., 1984; Lee and Waldman, 1985; Sloan et al., 1986~. There has been little study, however, of the effects of insurance on the R&D sector-private, governmental, and non- profit where new technologies are developed, although the linkage between the insurance system and incentives for the R&D sector has been noted (Joskow, 1981; Goddeeris, 1984a,b; Goddeeris and Weisbrod, 1985; U.S. Congress, 1985~. The effect of prospective payment insurance on R&D is illustrated by experience in the late 1980s with the cochlear implant for hearing-impaired individuals; scien- tif~cally promising research was discontinued as a consequence of its expected unprofitability, which resulted from application of the DRG pricing system. The 3M Company, the manufacturer of the first Food and Drug Administration- approved single-channel cochlear implant model, halted research on a multichan- nel device because of hospitals' financial disincentives (Kane and Manoukian, 1989~. Similarly, R&D on assistive communication devices for speech-impaired individuals appears to have been retarded by the lack of insurance coverage; Medicare's payment policy favors inpatient over outpatient care, and there was "an administrative decision that the [communication] devices are not prosthetic devices needed for the functioning of a malformed body member" (U.S. Con- gress, 1984b, p. 30~. The current climate and incentives facing the R&D sector are not conducive to the development of costly new technologies. Another example is the newly emerging diagnostic procedure PET, which produces three-dimensional images that portray the metabolic and chemical action of tissue. PET is in clinical trial, but the General Electric Company, its developer, "isn't making the kind of in- vestment it did to rush CT (computerized tomography scanners) and MRI (mag- netic resonance imaging devices) to market" (Schiffman, 1989~. According to a General Electric official, "The government is very cautious about approving re- imbursement for PET. In the past, if a technology improved patient care, it would be approved. Now it must also be cost-effective" (Naj, 1990, p. Bib. There are some further implications of the new incentives for hospitals to reduce costs rather than to increase quality. In the new era of prospective pricing of hospital services, there will likely be a diversion of R&D resources away from new surgical techniques and toward lower-cost substitutes, frequently pharma
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 29 ceuticals. Surgical advances can be cost-reducing, especially when they substi- tute for other halfway technologies; angioplasty, for example, substitutes for more costly coronary bypass graft surgery, and kidney transplantation substitutes for years of dialysis. When surgical advances substitute, however, for nontreat- ment, they are likely to increase the cost of treating the specific illness; since life expectancy may increase, though, the effect on mean annual health care costs per capita is less clear. Surgery is costly, relative to nonsurgical interventions, be- cause it is labor-intensive, "custom" production performed on a single patient; as such it has limited capacity for taking advantage of scale economies.28 Increased use of surgery over the 1972-1982 penod, during which retrospective pricing of hospital services dominated, was the primary source of rising treatment costs for patients admitted to a teaching hospital for acute myocardial infarction, respira- tory distress syndrome of the newborn, and other intensive treatments for the critically ill (Showstack, et al., 19851.29 New surgical interventions are likely to be less attractive in a cost-containment environment. By contrast with surgery, research on those pharmaceuticals that decrease expenditures, relative to those that increase quality but increase expenditures, is more attractive under prospective pncing. This is because demand patterns by hospitals (and HMOs) reflect the search for cost-reducing modes of treatment, including substitutes for costly surgical interventions; in particular, the advent of prospective pricing has increased the expected profitability of R&D on drugs than can prevent the onset of costly treatments vaccines, for example (Huston and Weisbrod, 1988) and of R&D on drugs that substitute for surgery cxam- ples are beta-blockers, which can substitute partially for coronary bypass surgery, and cimetidine, which substitutes for ulcer surgery (Geweke and Weisbrod, 1982~. The effects of PPS insurance on the pharmaceutical industry will not, how- ever, be entirely favorable. Pharmaceuticals are not always substitutes for sur- gery; they are sometimes complements. Development of a new drug that com- plements surgery can increase the efficacy of surgery and thereby increase the demand for surgery with major cost implications. In a cost-containment insur- ance/finance environment, pharmaceutical industry R&D faces an incentive to develop drugs that substitute for surgery rather than enhance its efficacy. 28 Cost reductions are likely to result, however, from experience learning by doing which is a function of total accumulated volume, even if not a function of the rate of surgery per unit of time. In a study of six surgical procedures, including coronary artery bypass and hip replacement, between 1984 and 1986, it was found that mortality declined with volume for five of the six procedures, but current cost per case declined with volume for only two of the procedures. Data covered between 646 and 4,738 hospitals, depending on the procedure (Project Hope, 1988). 29 Much of the medical literature reports findings for a single hospital. Whether the findings are generalizable to the entire hospital system is not clear.
30 BURTONA. WEISBROD Organ transplants illustrate the issue. Liver transplantation, a surgical tech- nique, is effective today largely because of a recent technological advance in pharmaceuticals. The drug cyclosporine is crucial because it suppresses the body's immune system reaction to the transplanted organ; yet, unlike earlier immunosuppressant drugs, it does not stop the body from fighting off infections. The good news about this technological breakthrough is that cyclosporine permits people with liver, kidney, and heart failure to be kept alive, living essen- tially normal lives. The bad news is that the resulting increase in the efficacy of organ transplant surgery has brought sharp increases in the usage of these very costly procedures. Only 26 liver transplants were performed in 1981 and, while increasing to nearly 1,200 by 1987, some 4,000 - ,700 people per year could benefit from the procedure. At a cost of about $200,000 each, plus annual main- tenance costs, meeting all the medical needs implies an annual cost of $1 billion for this one procedure (National Organ Transplant Act, 1983; Pear, 1986; "Cy- closporine turns five," 1988~. Heart, kidney, and other organ transplants suggest many times this level of potential expenditures as a consequence of the pharma- ceutical breakthrough. It has also produced political pressure to ensure access to this lifesaving technology, regardless of a patient's ability to pay pressure that is still being suppressed in part by the expedient of terming the procedures exper- imental. The enormous expenditure potential of technological advances in drugs is currently highlighted by the drug AZT for patients with AIDS. In 1990 it was estimated that some 600,000 people might benefit from this drug, which, even with reductions in price and dosage at that time, costs about $3,000 per year of treatment (Lublin, 1990), for a total potential cost of nearly $2 billion. Given the current financing environment, why are cost-increasing drugs such as AZT, cyclosporine, and tissue plasminogen activator (t-PA) being developed? Cyclosporine came onto the market prior to the advent of PPS, and given the lengthy research and regulatory process in pharmaceutical research, it is reason- ably clear that work related to AZT and :-PA was well under way by the time prospective-type insurance incentives became powerful in the 1980s. Today, the fiscal pressures operating through Medicare, HMOs, and state Medicaid "formu- laries"-lists of drugs that will be paid for are reducing drug company incen- tives to develop drugs for which "high" prices would be required to make the R&D effort profitable. The form of insurance affects the direction of R&D not only in terms of quality relative to cost. It also affects the incentive to search for methods to treat the ill rather than to prevent their illnesses.30 In general, health insurance has primarily covered treatment in hospitals, with preventive measures having quite 30 In the long run, the price of private health insurance depends on the state of the technology. Even so, risk-spreading over all the insured may make it privately profitable for the R&D sector to develop technologies for which the value (willingness to pay) is less than the social cost (Goddeeris, 1984a,b; Baumgardner, 1989).
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 31 limited coverage. As a result, the R&D sector has had less incentive to focus effort on prevention than on treatment, with the exception, perhaps, of vaccina- tions, for which government subsidization is common. Insofar as preventive measures are covered by insurance, they tend to involve technologies that utilize the "health care sector"-especially physicians and hospitals even though other measures, such as better diet and exercise, might improve health at lower cost.3~ I do not intend to imply that a reallocation of resources toward prevention would necessarily be efficient, given the existing state of knowledge (Russell, 1986, 1987~. Indeed, the concept of efficiency is itself controversial; it certainly can be defined in terms of either patient willingness to pay or some measure of health status and in either private or social terms.32 The point is that today's state of knowledge about measures for preventing illness and for treating it reflects the historical incentives for R&D of both types, and those incentives have been shaped by the insurance system. A number of the relationships discussed above may be summarized in the following five diagrams. Figure 2-1 shows that the state of technology, which determines what is possible for the health care system to do, interacts with the health care insurance system, which determines the prices and other incentives that providers and consumers confront, to determine the level of health care expenditures. Figure 2-2 portrays the interaction between the scientific know- ledge base and the incentive structure that operates through the health insurance system to affect "tomorrow's" technological base. Figure 2-3 shows that the technological base that exists at one point in time affects the insurance system, since the more that base presents opportunities to spend large sums of money on health care, the greater will be the demand for insurance. Relatedly, Figure 2-4 points up the likelihood that actual increases in health care expenditures will increase the demand for insurance. Finally, Figure 2-5 summarizes the relationships shown in Figures 2-1 through 2-4. Even at this simplified level it portrays a complex system of inter- dependent relationships. Society cannot expect to manipulate any one of the variables- for example, by altering the health care insurance system, incentives for developing new technologies, or the level of aggregate health care expendi 31 Thus, prevention has its halfway technologies, too (as pointed out by an anonymous referee). For many forms of prevention, insurance is inappropriate because there is little uncertainty about the financial expenditure involved (thanks to Mark Satterthwaite for noting this). 32 Aaron and Schwartz (1984) define efficiency in medical terms but using a Pareto-like approach: "Medical resources are efficiently used when a given total expenditure cannot be reallocated to alternative kinds of care to achieve an improved medical outcome.... [Thus] it would not be possible to increase total medical benefits by taking some money away from one service, for example cancer chemotherapy, and spending it on another, say x-ray" (pp. 79, 89). Ellis and McGuire (1986) define efficient supply of care as existing when the physician acts as a perfect agent, weighing a dollar of hospital profit equally with a dollar of benefit to the patient.
32 BURTON A. WEISBROD Techn3 1 1 Health Care Insurance System 1 Health Care Expenditures FIGURE 2-1 The state of technology and the incentives to use the technology deter- mine total expenditures: The short run. Science Base Health Care Insurance System ~Technology FIGURE 2-2 The health care insurance system establishes incentives for the R&D sector. Health Care Insurance System - I Technology FIGURE 2-3 The technical capability for delivering health care affects the form and coverage of the health care insurance system.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! Health Care Insurance System Health Care Expenditures FIGURE 2-4 The level of health care expenditures affects the demand for insurance. | Science Base l O' ~ ,~ ~ ~ , Health Care Insurance System - - _ 03 ~ Technology Health Care Expenditures 33 FIGURE 2-5 The dynamic system of interaction of the health care insurance system, technological change, and health care expenditures. Circled numbers 1 to 4 refer to Figures 2-1 to 2-4, respectively. lures without altering other variables and, ultimately, without affecting out comes of the system in perhaps quite unexpected ways. EFFECTS OF INSURANCE ON CHOICE OF TECHNOLOGY AND QUALITY OF HEALTH CARE IN THE SHORT RUN, WITH TECHNOLOGY GIVEN In addition to its potential to influence R&D, the health insurance incentive structure also influences the deployment of existing medical technology, with implications for quality and access to care. A retrospective, cost-based reward
34 BURTONA. WEISBROD structure and a prospective reward structure such as a DRG system and an HMO33 offer very different incentives for provider choice between-increasing quality and decreasing costs (Momsey et al., 1984~. For a given state of technological knowledge, a prospective payment insur- ance system provides encouragement, at the margin, to use production processes that reduce cost rather than improve quality. This is particularly so when quality is affected in dimensions that are costly for consumers (or regulators or insurers) to observe. The central point is that in a world of asymmetrically high informa- tion costs to consumers relative to service providers (e.g., hospitals and HMos),34 it is harder to detect reductions in quality in some forms than in others, and the finance system can influence provider incentives to choose among input combi- nations that differ in the relative importance of effects that are more and less costly for nonproviders to monitor. Every commodity health care or anything else~an be thought of as a bundle of attributes that vary in the cost of monitoring them as well as in their importance to buyers. To simplify, consider two classes of attributes- type I, which are costless to monitor, and type II, which are costly to monitor. If con- sumers respond largely to the observable, type I attributes, then sellers will find price to be essentially independent of quality in the type II dimensions, and quality in the latter forms will be low (Weisbrod, 19883. Price will be a poor gauge of overall quality. A prospective payment reward structure such as a DRG system is a price control mechanism. It poses the problem of how to ensure that real prices are not raised through the expedient of reducing service quality, especially in the type II dimensions.35 The potential effects of price-setting by a governmental agency or pnv ate insurer, when quality is asymmetrically costly to monitor, can be seen by com- panng the setting of prices for electricity and for care in a hospital or nursing home. A kilowatt-hour of electricity is far more homogeneous and easier to monitor than is a day of care (or any number of other potential measures of output) in a hospital or nursing home.36 Thus, regulating price does not pose a 33 There are important differences between an HMO- and a DRG-type payment system at least as these operate now. For example, the DRG system applies currently only to hospital services, while HMOs cover a wider range of medical services. HMOs may operate their own hospitals, but they typically subcontract with independent hospitals for treatment of HMO members; such subcontracts can take many forms, and with either retrospective or prospective pricing. 34 An HMO, which vertically integrates a provider group with an insurer, reduces the informational asymmetry between the two, though not between either of them and consumer-patients or regulators. 3s Throughout this discussion the role of physicians as agents for patients has great importance. I assume that physicians act as imperfect agents, which leaves patients asymmetrically underinformed. 36 There are other elements of the electric power regulatory process for example, the "appropriate" levels of inputs that involve asymmetric costs. The literature on the Averch-Johnson effect focuses, in effect, on the difficulty regulators have in determining the degree of overcapitalization of public utilities under rate-of-return regulation (Averch and Johnson, 1962; Baumol and Klevorick, 1970).
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 35 serious risk that quality of output will be compromised in unobservable ways by the regulatory process. Because of the more complex attributes of the health care system, opportunities are greater for providers to reduce output quality in dimen- sions that, being costly to monitor, are difficult to embody in a performance contract. The Joint Commission on Accreditation of Health Care Organizations (JCAHCO) recognizes implicitly the distinction between type I and type II char- acteristics for assessing the quality of a hospital. In his testimony at the 1973 Senate hearings, the executive director of JCAHCO said it was concerned with whether a hospital had the physical environment to permit high-quality medicine to be provided for example, an operative sprinkler system (a type I attribute)- not with the actual clinical practices that is, how carefully surgery is performed (a type II attribute) (Lohr et al., 1988~.37 I remarked earlier that under a prospective payment system, financial incen- tives are to cut costs, provided quality does not suffer "too much."38 There are consequences, of course, of cutting quality, and they constrain health care pro- viders: tort law liability for medical malpractice, loss of patients to competitors (Hirschman, 1970), loss of donations and volunteer labor, penalties for violating regulatory rules (Weisbrod and Schlesinger, 1986) and professional ethics codes, and, in the case of HMOs, possibly greater costs of treating member-patients in the future.39 Thus, the financial incentive to reduce costs by cutting quality is presumably equated at the margin with the effects of reduced quality on these revenue and cost variables (Woodward and Warren-Boulton, 1984; Ellis and McGuire, 1986~.4° Little is known about the quantitative importance of each of these constraints, but because of them a prospective payment price control sys- tem implicitly encourages health care providers to cut resource use in the type II dimensions-which would minimize revenue losses and other penalties-not in ways that would be socially efficient.41 Consumer-patients and donors cannot respond to changes in quality they cannot observe. Thus, given the imperfections in agency relationships (Ellis and 37 John Porterfield, the JCAHCO executive director, reportedly said that a hospital reviewer would observe whether the hospital's sprinkler system worked and whether certain medical committees functioned and kept adequate records, but if a surgeon on the staff decided that good-quality care required taking out the appendix of all blue-eyed males over age 16, that was none of the JCAHCO reviewer's business. 38 Morrisey and colleagues (1984) model the effects on quality of care in a hospital confronted by downward price pressure. 39 For HMOs, the latter effect is attenuated by the uncertainty that the person will remain a member. 40 Because HMOs involve a prospective payment to cover all "needed" care for the stipulated period, the incentives facing HMOs are analytically very similar to those facing hospitals under DRG pricing; thus, in general, propositions in this section referring to hospitals will also apply to HMOs, mutatis mutandis (that is, the necessary changes having been made). 41 This is analogous to "skimming" and "creaming" of program participants.
36 BURTON A. WEISBROD McGuire, 1986), the shift to a DRG-type prospective payment insurance system can be expected to cause reductions in quality in precisely those forms that are difficult for insurers to monitor (Weisbrod, 1989~. This prediction requires test- ing, which needs to recognize that in a competitive market there can be simulta- neous decreases in type II dimensions of quality and increases in type I dimen- sions. For example, increased "quality" in easily observed forms such as hospital candlelight dinners for maternity patients and their spouses can attract patients to a hospital, and free dental or eye checkups can attract members to an HMO, even while quality of medical care is being reduced in more subtle, hard-to-detect forms (Weisbrod, 1988, chapters 2, 3, 8~. The reuse of "disposable" items by hospitals illustrates the potential for cutting quality in ways that are difficult for consumers to monitor and an effect of prospective pricing on the choice of production technology. Until the late 1940s, hospitals reused most medical devices; tubing, syringes, needles, etc., were made to be used, sterilized, and used again. When the new technology of disposables was introduced after World War II, it was quickly adopted by a health care finance system that encouraged the greater convenience and safety of disposables and that Reemphasized the cost consequences. The expanding system of retro- spective-pay health insurance that covered all "reasonable" hospital costs spurred both the development and the adoption of disposable items along with any other technology that was arguably quality enhancing. Today, with the shift to prospective pricing, sterilization and reuse is return- ing. This change may or may not be efficient allocatively or medically. What is striking is that hospitals are reusing items that are labeled by the manufacturers for "one-time-use only" (Otten, 1984~. Even "disposable" filters for kidney dial- ysis machines are being reprocessed and reused (U.S. Congress, 1984a). These practices reduce hospital costs. They may have no effect on revenues, for they are difficult for consumers (but presumably not their physician-agents) to observe. Thus, the financial consequences are relatively unambiguous. At the same time, the effect on health risks of reusing disposables is not currently known (Chu et al., 1986; National Center for Health Services Research and Health Care Technology, 1986~. While the safety debate proceeds, the dispute is being re- solved in favor of the cost-reducing technology. This is in sharp contrast to the situation in the 1950s, when the incentive structure was reversed; at that time, single-use disposables replaced the prior use-sterilize-reuse technology, despite the absence of strong evidence of favorable health effects. In general, the switch to prospective payment can be expected to bring changes in the technology of health care of just that type: they have clearly favorable effects on costs, but subtle or uncertain, yet presumptively nonpositive effects on the quality of care. I say "presumptively" nonpositive because, given the state of technical knowledge, any change in resource use that is made after a change in incentives could have been made before; disposables could have been reused earlier.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 37 Another quality-related dimension of hospital behavior likely to be affected by a shift to prospective pricing for hospitals is the length of a patient's stay. Confronted, under a DRG pricing system, by a fixed price for treating each patient, hospitals have a financial incentive to discharge patients earlier (Lave et al., 1988~. Even if they do so, however, it is difficult for a patient to determine whether he or she has been discharged "quicker but sicker" (Heinz, 1986~. Here, once again, a crucial question is how well asymmetrically underinformed pa- tients are represented by physician-agents. A reduction in use of hospital inputs is not necessarily inefficient, in economic or medical terms; the cost savings may exceed the loss in benefits (although placing a value on the benefits is difficult), and in some situations there might be no medical benefits at all from, say, a longer hospital stay. Neither, though, is a reduction in inputs necessarily efficient. Input substitutions and cost reductions that may result from the shift from cost-based to prospective insurance cannot be assumed to be efficient or inefficient in a world of asymmetrically underinformed patient-consumers who confront prices that often bear little relationship to real marginal costs. Public policy, if it is to increase allocative efficiency, clearly demands understanding of the effects of pricing and other interventions on both quality and cost, not simply on costs. In particular, there should be attention to the tendency of a prospective payment insurance pricing system to cause input substitutions that overvalue reductions in easily observed expenditures and un- dervalue attention to reductions in quality that are more costly to observe. The response of the health care sector to financial incentives may not be the same for its various institutional elements private enterprise, government, and private nonprofit. In the hospital industry, 65 percent of all short-term beds are in private nonprofit hospitals; 26 percent are in government hospitals. Thirty percent of nursing home beds are in nonprofit (22 percent) or government (8 percent) facilities. Of kidney dialysis centers, 48 percent are nonprofit and an additional 12 percent are governmental (Weisbrod, 1988~. The key question is this: In response to a public policy shift from cost-based to prospective payment to providers, is there a different response, quantitatively or qualitatively, depend- ing on the institutional ownership mix of the industry?42 Confronted by the incentives that prospective payment provides to discharge patients earlier and to engage in other forms of quality-shaving actions in the type II dimensions, do for-profit, nonprofit, and government organizations respond differently?43 Does institutional form matter? 42 A related issue is how competition among organizations of various ownership types affects long- run equilibrium, and whether one form of institution can be expected to drive out the others (Schiff and Weisbrod, 1987). 43 Whether earlier discharge of a hospital patient is a type I or a type II attribute is debatable. I regard it as type II. While the length of stay for any patient is easily observable, what is difficult for the patient to observe is whether the length of stay was lower than it would have been if the physician and hospital were not responding to the altered financial incentive of PPS.
38 BURTON A. WEISBROD Finding the answers to these questions requires modeling the behavior of each form of organization and the process of competition among them. There has been some attention to the conditions of equilibrium in institutionally mixed industries (Schiff, 1986; Marmor et al., 1986; Phelps and Sened, 1989), but strong conclusions have not been reached. Economic behavior may differ across ownership forms because of differenc- es in objective functions, constraints, or both. Profit maximization is typically assumed for the pry ate enterprise components of the health care sector, but a variety of objective functions have been suggested for the nonprofit sector (New- house, 1970; Davis, 1973; Pauly and Redisch, 1973; James, 1983; Young, 1983), as have venous constraints on the distribution of profit44 and access to public subsidies and private donations of money and time (lIansmann, 1980; Rose- Ackerman, 1982; Easley and O'Hara, 1983; Holtmann, 1983; Clotfelter, 1985; Steinberg, 1986; Weisbrod and Dominguez, 1986~.45 DRG pricing provides the same financial incentive for all hospitals to dis- charge patients earlier than would a retrospective pricing system, but because of differences in objective functions and constraints, the behavioral responses may differ among institutional forms. There have been studies, for example, of the effect of prospective payment on the condition, at discharge, of elderly patients with hip fractures (Palmer et al., 1989; Fitzgerald et al., 1988) in two nonprofit hospitals, but they have not examined differences across ownership forms.46 More generally, neither theory nor empirical tests have resolved the question of whether there are systematic differences among institutional forms. Econo- metnc evidence, while mixed, is growing that when for-profit, nonprofit, and government organizations coexist in a given industry as they do in hospitals and nursing homes, for example-they do behave differently. Differences have 44 Nonprofit organizations are not legally restricted from engaging in profitable activities; they are, however, restricted in what they may do with the profits. Little explicit attention has been devoted, however, to the enforceability of this constraint (Weisbrod, 1988). This is relevant to the "managerial discretion" models of Williamson (1967), Alchian and Demsetz (1972), and Migue and Belanger (1974). 4s All organizations, regardless of ownership, confront the same technological constraints, but they face different financial constraints in such forms as nonprofits' exemptions from property and sales taxes and eligibility for postal subsidies. Charitable contributions of time and money to a nonprofit hospital (but not to a proprietary hospital) might respond positively to the amount of unprofitable services it provides to low-income, uninsured, or other "deserving" people. The relationship between donations to nonprofit organizations and the tax and expenditure behaviors of government" the "crowding out" effect has also received attention in the public finance literature. At the theoretical level, see Warr (1982); Roberts (1984), Bergstrom et al. (1986), and Andreoni (1988); for empirical studies, see Abrams and Schmitz (1978, 1984) and Schiff (1985). 46 Palmer and colleagues (1989) found no change in ambulation status, comparing patients dis- charged from one nonprofit hospital in the several years before and after the change in price incen- tives. Fitzgerald and colleagues (1988), studying a single "community" hospital (presumably also a nonprofit), found significantly reduced mobility.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 39 been examined in four principal dimensions: (1) access to care, as reflected by admission of uninsured patients (e.g., provision of "uncompensated" care) and the use of waiting lists rather than prices, (2) quality of care, (3) cost-efficiency, and (4) extent of opportunistic behavior toward asymmetrically underinformed consumers. Systematic behavioral differences between private firms and nonprofit orga- nizations have been found in some studies (Gray, 1986 twhich summarizes a number of studies]; Herzlinger and Krasker, 1987; Lewin et al., 1988; Weisbrod, 1988; Selden, 1989), but not in others (Clark, 1980; Sloan and Vraciu, 1983; Gaumer, 1986~. Nonprofit providers of health care, especially the church-affili- ated nonprofits, appear to utilize a somewhat greater proportion of their resources to care for the indigent; they provide a wider range of services (and in this sense, higher quality); and they take less advantage of their informational advantages over patients. Neither the underlying theory nor the available, nonexperimental data, how- ever, are yet strong enough to justify confident generalizations about differences in institutional behavior. Measuring quality of service in a hospital (Shortell and Hughes, 1988), controlling for differences in patient conditions, and distinguish- ing care of the indigent from "bad debts" associated with poor management all remain subjects for future research, as does any differential responsiveness to the development of new technologies.47 There is also a question of the appropriate estimation modeling; many econometric efforts to detect differential behavior across institutional forms may have misspecified their models, controlling erro- neously for variables such as organization size, which are endogenous to the choice of institutional form (Weisbrod and Mauser, 1990~. CONCLUDING REMARKS Economists' concerns about skyrocketing health care expenditures have fo- cused heavily on insurance and its encouragement of inefficiently great utiliza- tion. Yet it is clear that much of the growth in health care expenditures during the post-World War II period has resulted not from increased prices for existing technologies but from the price for new technologies. Newly developed technol- ogies have driven up both costs of care and the demand for insurance, while also expanding the range of services for which consumers demanded insurance. At the same time, expanding insurance coverage, which includes more people as well as a growing array of health care inputs, has provided an increased incentive to the R&D sector to develop new technologies and a growing incentive for subsets of consumers, who could benefit from particular new technologies, to 47 In a related study of rapidity of introduction of new technologies in HMOs relative to fee-for- service providers, the RAND Corporation health insurance experiment found an apparently slower rate of introduction in HMOs (Newhouse et al., 1985).
40 BURTON A. WEISBROD seek a wider definition of what would be covered by insurance. Both the re- source costs of health care and the technical ability to prolong life and enhance its quality have risen sharply. The interactive process involving insurance and R&D is still evolving. It is increasingly being influenced by the recent change in incentives associated with the shift from retrospective, cost-based insurance cov- erage to prospective, exogenously determined pricing. Although this chapter has focused on the health care sector, the kinds of incentive effects it has examined are quite general. As an example of the poten- tial effect of insurance on incentives facing the R&D sector, consider another major area of public policy and expenditure education. Unlike health care, which has been financed for decades by a retrospective, cost-based finance sys- tem, elementary and secondary education has been financed traditionally through what amounts to a prospective payment system; roughly speaking, state and local governments have given the schools a fixed grant per child. This is roughly analogous to a DRG system with a single DRG, so that every patient (child) entering a hospital (school) brings a fixed sum of revenue to the provider. A school district can also be thought of as, like an HMO, providing "comprehen- sive" services to all "members" (students) in return for a fixed annual fee. By examining how the interaction of finance mechanisms and R&D incentives have operated in the health and education areas, one can gain insight into what the health care system would be like today, had the country taken an alternate route for financing it, as well as how a change in school finance would be likely to have an impact on the education system. Assume that public schools had been financed differently in the way hos- pitals have been financed until recently: (1) school revenue was determined through a retrospective (cost-based) pricing system, in which (2) teachers were empowered to decide what resources should be used (a) to diagnose a particular child's educational "needs" and (b) to meet those needs, and (3) a bill for the cost of the resources used for each child was sent to government or a private insurer and subsequently paid to the school district. Two questions arise: If such a system had been adopted after World War II for schools, what would have happened over the subsequent 40 years to the level of education expenditures? What would have happened to the pace of technolog- ical change in education? The lessons from health care suggest conjectures: if schooling had been "insured" on the basis of retrospective costs, expenditures would have increased far more rapidly than they did and the pace of technologi- cal innovation in schools would have been far greater than it was. Since education actually utilized a prospective pricing system, while health care utilized a retrospective pricing system, it is interesting to compare the two programs in terms of expenditure growth and technological change. First, with respect to expenditures, the share of GNP devoted to public elementary and secondary education has changed little over several decades (in which enroll- ments have remained relatively constant); between 1960 and 1985, for example,
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 41 years of virtually identical school enrollments (36.7 million and 36.6 million, respectively) public school expenditures increased from 3.03 percent of GNP to 3.42 percent (U.S. Bureau of the Census, 1987, tables 186, 190 and 698~; mear~- while, health care expenditures were rising from 4.6 to 10.7 percent of GNP (U.S. Bureau of the Census, 1975 and 1987~. Second, with respect to the pace and nature of technological change that might have occurred in education had retrospective pricing prevailed, one can do some informed speculating. To begin, compare, impressionistically, the techno- logical change that has occulted in health care and in education. The typical hospital, for example, is barely comparable to its counterpart several decades ago, with entirely new techniques and facilities for diagnosis and treatment. The typical school, however, differs far less from its post-World War II counterpart, utilizing similar classrooms, teachers trained in similar ways, and instructional techniques that, despite some computerization in recent years, employ capital- labor ratios that have changed relatively little. One can predict that if retrospective reimbursement had prevailed for schools, the private sector would have devoted more resources to development of "improved" educational diagnostic and reaming technologies; had that been the case, society would probably find now that education, like health care, had im- proved dramatically, but that society was paying a great deal more for it. Today, the public policy "problems" in health care and education are per- ceived to be sharply different, and in ways that correspond to the differences in finance mechanisms (although other forces are doubtless also at work). In health care, the central policy focus is on control of expenditures, with quality of care not generally being seen as a problem.48 In education, it is the reverse the policy focus is on "low" quality of education, with control of school expenditures receiving relatively less attention. The ideas presented above are a mixture of solid knowledge, soft knowl- edge, and hypotheses requiring testing. In order to expand knowledge about health care and provide financial access to it, society needs to understand more fully the dynamic process through which the health insurance sector, private and public, interacts with the R&D sector. This area offers a rich research agenda with enormous potential, for the policy implications extend far beyond health care and across geographical boundaries. REFERENCES Aaron, H. J., and Schwartz, W. B. 1984. The Painful Prescription: Rationing Health Care. Washington, D.C.: Brookings Institution. 48 Problems of the uninsured are serious, but are receiving less attention than is the general prob- lem of cost containment.
42 BURTON A. WEISBROD Abrams, B. A., and Schmitz, M. D. 1978. The crowding-out effect of governmental transfers on private charitable contributions. Public Choice 33~1~:29-37. Abrams, B. A., and Schmitz, M. D. 1984. The crowding-out effect of governmental transfers on private charitable contributions: Cross-section evidence. National Tax Journal 37~41:563-568. Akerlof, G. 1970. The market for lemons: Qualitative uncertainty and the market mecha- nism. Quarterly Journal of Economics 84~3~:488-500. Alchian, A., and Demsetz, H. 1972. Production, information costs and economic organi- zation. American Economic Review 62~5~:777-795. Altman, S. H., and Blendon, R., eds. 1979. Medical Technology: The Culprit Behind Health Care Costs? DH;EW Publication PHS 79-3216. Washington, D.C.: U.S. Government Printing Office. Andreoni, J. 1988. Privately provided public goods in a large economy: The limits o altruism. Journal of Public Economics 35~11:57-73. Arrow, K. 1963. Uncertainty and the welfare economics of medical care. American Economic Review 53~5~:941-973. Averch, H., and Johnson, L. 1962. Behavior of the firm under regulatory constraint. American Economic Review 52~5~:1053-1069. Battistella, R. 1984. Review of The PubliclPrivate Mixfor Health, by Gordon McLachlan and Alan Maynard, eds. Inquiry 21~1~:98-99. Baumgardner, J. R. 1989. Forms of Lnsurance and the Pattern of Aggregate Technical Change in Medical Care. Working Paper, Duke University, Department of Economics. Baumol, W. J., and Klevorick, A. K. 1970. Input choices and rate of return regulation: An overview of the discussion. Bell Journal of Economics and Management Science 1(2): 162-190. Bergstrom, T., Blume, L., and Varian, H. 1986. On the private provision of public goods. Journal of Public Economics 29( 1 ):25-49. Boben, P. 1989. Modeling the Shift from Retrospective to Prospective Pricing of Health Care. Discussion Paper, University of Wisconsin-Madison, Center for Health Eco- nomics and Law. Chernick, H. A., Holmer, M. R., and Weinberg, D. H. 1987. Tax policy toward health insurance and the demand for medical services. Journal of Health Economics 6~1~:1- 25. Chu, F., Novak, N., Radany, M. H., and Perry, S. 1986. Reuse and reprocessing of disposable medical devices: Legal liability issues. Journal of Health Care Technol- ogy 3~1~:5-12. Clark, R. C. 1980. Does the nonprofit form fit the hospital industry? Harvard Law Review 93~7~:1416-1489. Cleeton, D., Goepfrich, V. T., and Weisbrod, B. A. 1988. The Consumer Price Index for Prescription Drugs: Dealing with Technological Change. Working Paper, Univer- sity of Wisconsin-Madison, Center for Health Economics and Law. Clotfelter, C. T. 1985. Federal Tax Policy and Charitable Giving. Chicago, Ill.: Univer- sity of Chicago Press. Cromwell, J., and Mitchell, J. B. 1986. Physician-induced demand for surgery. Journal of Health Economics 5~44:293-313. Cyclosporine turns five. 1988. Science 242~4876~:198.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 43 Danzon, P. M. 1985. Medical Malpractice. Cambridge, Mass.: Harvard University Press. Davis, K. 1973. Theories of hospital inflation: Some empirical evidence. Journal of Human Resources 8~21: 181-201. Easley, D., and O'Hara, M. 1983. The economic role of the nonprofit firm. Bell Journal of Economics 14~2~:531-538. Elective surgery: Cut it out. 1979. Science News 115~1~:9. Ellis, R., and McGuire, T. 1986. Provider behavior under prospective reimbursement, Journal of Health Economics 5~2~: 129-151. Evans, R. G. 1974. Supplier induced demand: Some empirical evidence and implica- tions. In: M. Perlman, ed. The Economics of Health and Medical Care. London: Macmillan. Feldstein, M. S. 1971. Hospital cost inflation: A study of non-profit dynamics. Ameri can Economic Review 61~54:853-872. Feldstein, M. S. and Allison, E. 1974. "Tax Subsidies of Private Health Insurance" in The Economics of Federal Subsidy Programs. Part 8. Papers submitted to the Subcom- mittee on Priorities and Economy in Government, Joint Economic Committee, 93rd Congress, 2nd session, July. Feldstein, M. S., and Friedman, B. 1977. Tax subsidies, the rational demand for insur- ance, and the health care crisis. Journal of Public Economics 7~2~:155-178. Feldstein, P. J. 1988. The Politics of Health Legislation: An Economic Perspective. Ann Arbor, Mich.: Health Administration Press. Fitzgerald, J. F., Moore, P. S., and Dittus, R. S. 1988. The care of elderly patients with hip fracture: Changes since implementation of the prospective payment system. New England Journal of Medicine 319~211:1392-1397. French, H. W. 1989. Tiny miracles become huge public health problem. New York Times, Feb. 19, pp. 1, 22. Fuchs, V. R. 1986. The Health Economy. Cambridge, Mass.: Harvard University Press. Garg, M. L., Gliebe, W. A., and Elkhatib, M. B. 1978. The extent of defensive medicine: Some empirical evidence. Legal Aspects of Medical Practice 6~1~:25-29. Gaumer, G. L. 1986. Medicare patient outcomes and hospital organization mission. En: B. Gray, ed. For-Profit Enterprise in Health Care. Washington, D.C.: National Academy Press. Geweke, J., and Weisbrod, B. A. 1982. Clinical evaluation of a new drug. Medical Care 20~8~:821-830. Goddeeris, J. H. 1984a. Insurance and incentives for innovation in medical care. South- ern Economic Journal 51 (2~:53~539. Goddeeris, J. H. 1984b. Medical insurance, technological change, and welfare. Eco- nomic Inquiry 22~1~:56~7. Goddeeris, J. H. 1987. Economic forces on hospital technology: Lessons from the U.S. experience. International Journal of Technology Assessment in Health Care 3~21: 223-240. Goddeeris, J. H., and Weisbrod, B. A. 1985. What we don't know about why health care expenditures have soared: Interaction of insurance and technology. Mount Sinai Journal of Medicine 52~9~:685~91. Gray, B., ed. 1986. For-Profit Enterprise in Health Care. Washington, D.C.: National Academy Press.
44 BURTON A. WEISBROD Hansmann, H. 1980. The role of non-profit enterprise. Yale Law Journal 89~5~:835-901. Heinz, J. 1986. The effects of DRG's on patients. Business and Health 3~-8~:17-20. Herzlinger, R., and Krasker, W. 1987. Who benefits from nonprofits? Harvard Business Review 65~11:93-106. Hirschman, A. O. 1970. Exit, Voice, and Loyalty. Cambridge, Mass.: Harvard Univer- sity Press. Hogan, C. 1988. Urban and Rural Hospital Costs: 1981-85. DHHS Publication (PHS) 88-3419. Hospital Studies Program Research Note 12, National Center for Health Services Research and Health Care Technology Assessment. Rockville, Md.: Public Health Service. Holtmann, A. G. 1983. A theory of non-profit firms. Economica 50~200~:439~49. Hudis, J. 1986. The organ game. Health 18~10~:61~6. Huston, J., and Weisbrod, B. A. 1988. Benefits and Costs of Human Vaccines. Working Paper, University of Wisconsin-Madison, Center for Health Economics and Law. James, E. 1983. How nonprofits grow: A model. Journal of Policy Analysis and Management 2~31:35~365. Joskow, P. 1981. Controlling Hospital Costs: The Role of Government Regulation. Cambridge, Mass.: MIT Press. Kane, N. M., and Manoukian, P. D. 1989. The effect of the Medicare prospective payment system on the adoption of new technology: The case of cochlear implants. New England Journal of Medicine 321~201:1378-1383. Kleinfield, N. R. 1983. Companies in search of bionic man. New York Times, Nov. 20, Section 3, pp. 1, 2~27. LaCronique, J. F., and Sandier, S. 1981. Technological innovation: Cause or effect of increasing health expenditures. In: C. Altenstetter, ed. Innovation in Health Policy and Service Delivery. Cambridge, Mass.: Oelgeschlaher, Gunn, and Hain. Lave, J. R., Frank, R. G., Taube, C., Goldman, H., and Rupp, A. 1988. The early effects of Medicare's prospective payment system on psychiatry. Inquiry 25~3~:35~363. Lee, R. H., and Waldman, D. M. 1985. The diffusion of innovations in hospitals. Jour- nal of lIealth Economics 4~41:373-380. Letsch, S., Levit, K. R., and Waldo, D. R. 1988. National health expenditures, 1987. Health Care Financing Review 10~2~:109-122. Lewin, L. S., Eckels, T. J., and Miller, L. B. 1988. Setting the record straight: The provision of uncompensated care by not-for-profit hospitals. New England Journal of Medicine 318~18~: 1212-1215. Lohr, K. N., Yordy, K. D., and Thier, S. O. 1988. Current issues in quality of care. Health Affairs 7~1~:5-18. Lublin, J. S. 1990. 1990. FDA lets Wellcome cut dosage of drug for AIDS, easing pressure for price cut. Wall Street Journal, Jan. 17, p. B2. Luft, H. S. 1980. Trends in medical care costs: Do HMOs lower the rate of growth? Medical Care 18~1~:1-16. Luft, H. S. 1981. Health Maintenance Organizations: Dimensions of Performance. New York, N.Y.: John Wiley & Sons. Manning, W. G., Newhouse, J. P., Duan, N., Keeler, E. B., Leibowitz, A., and Marquis, M. S. 1987. Health insurance and the demand for medical care: Evidence from a randomized experiment. American Economic Review 77~3~:251-277.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 45 Marmor, T., Schlesinger, M., and Smithey, R. W. 1986. A new look at nonprofits: Health care policy in a competitive age. Yale Journal on Regulation 3~1~:313-349. Migue, J.-L., and Belanger, G. 1974. Toward a general theory of managerial discretion. Public Choice 17~41:27-47. Mitchell, B. M., and Phelps, C. E. 1976. National health insurance: Some costs and effects of mandated employee coverage. Journal of Political Economy 84~31:553- 571. Mitchell, B. M., and Vogel, R. J. 1975. Health and taxes: An assessment of the medical deduction. Southern Economic Journal 41 (4~:660 672. Morrisey, M. A., Conrad, D. A., Shortell, S. M., and Cook, K. S. 1984. Hospital rate review: A theory and an empirical review. Journal of Health Economics 3~11:25- 47. Mossin, J. 1968. Aspects of rational insurance purchasing. Journal of Political Economy 76~4,:553-568. Naj, A. K. 1990. Diagnostic equipment field in squeeze. Wall Street Journal, September 6, pp. B 1, B4. National Center for Health Services Research and Health Care Technology. 1986. The Reuse of Hemodialysis Devices Labeled for Single Use Only. Washington, D.C.: National Center for Health Services Research and Health Care Technology. National Center for Health Services Research and Health Care Technology. 1988. New technologies result in increased physician services and costs. In: Research Activi- ties. Washington, D.C.: National Center for Health Services Research and Health Care Technology. National Organ Transplant Act. Hearings before the Subcommittee on Health and the Environment, July 29, Oct. 17, 31, 1983. Washington, D.C.: U.S. Government Printing Office. Nazario, S. L. 1989. Infertility insurance gains backing. Wall Street Journal, December 5, pp. B1, BS. Nelson, R. R. 1972. Issues and suggestions for the study of industrial organizations in a regime of rapid technical change. In: V. R. Fuchs, ed. Policy Issues and Research Opportunities in Industrial Organization. New York, N.Y.: National Bureau of Economic Research and Columbia University Press. Newhouse, J. 1970. Toward a theory of nonprofit institutions: An economic model of a hospital. American Economic Review60~1~:64-75. Newhouse, J. 1981. Erosion of the medical marketplace. Advances in Health Economics and Health Services Research 2:1-34. Newhouse, J. 1988. Has the erosion of the medical marketplace ended? Journal of Health, Politics, Policy, and Law 13~2~:263-279. Newhouse, J. P., Schwartz, W. P., Williams, A. P., and Witsberger, C. 1985. Are fee-for- service costs increasing faster than HMO costs? Medical Care 23~8~:960-966. Organization for Economic Cooperation and Development. 1990. Health Care Systems in Transition. Paris: OECD. Otten, A. L. 1984. Reuse of devices by hospitals and doctors stirs debate over cost savings and safety. Wall Street Journal, Sept. 14, p. 27. Palmer, R. M., Saywell, R. M., Zollinger, T. W., et al. 1989. The impact of the prospec- tive payment system on the treatment of hip fractures in the elderly. Archives of Internal Medicine 149~10~:2237-2241.
46 BURTON A. WEISBROD Pauly, M. V. 1974. Overinsurance and public provision of insurance: The roles of moral hazard and adverse selection. Quarterly Journal of Economics 88~1~:44-62. Pauly, M. V. 1979. What is unnecessary surgery? Millbank Memorial Fund Quarterly 57:95-117. Pauly, M. V. 1986. Taxation, health insurance, and market failure in the medical economy. Journal of Economic Literature 24~2~:629-675. Pauly, M. V., and Redisch, M. 1973. The not-for-profit hospital as a physicians' coopera- tive. American Economic Review 63~1~:87-100. Pear, R. 1986. Federal payment for transplants for poor studied. New York Times, May 18,Section 1,p.31. Phelps, C. E., and Sened, I. 1989. Market Equilibria with Not-For-Profit Firms. Working Paper 191, University of Rochester, Department of Political Science. Project Hope. 1988. Trends in the Concentration of Six Surgical Procedures Under PPS and Their Implications for Patient Mortality and Medicare Costs. Chevy Chase, Md.: Project Hope, Center for Health Affairs. Reinhardt, U. 1985. The theory of physician induced demand: Reflections after a de- cade. Journal of Health Economics 4~2~:187-193. Report of the President's Biomedical Research Panel. Washington, D.C.: U.S. Govern- ment Printing Office, 1976. Rettig, R. A. 1980. The politics of health cost containment: End-stage renal disease. Bulletin of the New York Academy of Medicine 56~1~: 115-138. Rettig, R. A., and Marks, E. 1983. The Federal Government and Social Planning for End-Stage Renal Disease: Past, Present, and Future. Santa Monica, Calif.: RAND Corporation. Roberts, R. D. 1984. A positive model of private charity and public transfers. Journal of Political Economy 92~1~:136-148. Romeo, A. A., Wagner, J. L., and Lee, R. H. 1984. Prospective reimbursement and the diffusion of new technologies. Journal of Health Economics 3~1~:1-24. Rose-Ackerman, S. 1982. Charitable giving and "excessive" fundraising. Quarterly Journal of Economics 97~2~: 193-212. Rossiter, L. F., and Wilensky, G. R. 1984. Identification of physician-induced demand. Journal of Human Resources 19~2~:231-244. Russell, L. B. 1979. Technology in Hospitals. Washington, D.C.: Brookings Institution. Russell, L. B. 1986. Is Prevention lRetter Than Cure? Washington, D.C.: Brookings Institution. Russell, L. B., ed. 1987. Evaluating Preventive Care: Report on a Workshop. Washing ton, D.C.: Brookings Institution. Sawitz, E., Showstack, J. A., Chow, J., and Shroeder, S. A. 1988. The use of in-hospital physician services for acute myocardial infarction: Changes in volume and complex- ity over time. Journal of the American Medical Association 259~161:2419-2422. Schiff, J. 1985. Does government spending crowd out charitable contributions? National Tax Journal 38~4~:53~546. Schiff, J. 1986. Nonprofit Behavior in the Long Run. Working Paper, Tulane University, Department of Economics. Schiff, J., and Weisbrod, B. A. 1987. Competition Between For-Profit and Nonprofit Organizations. Working Paper, University of Wisconsin-Madison, Department of Economics.
THE NATURE OF TECHNOLOGICAL CHANGE: INCENTIVES MATTER! 47 Schiffman, J. R. 1989. Debate grows over clinics pushing costly PET scans. Wall Street Journal, Apr. 3, p. B1. Selden, T. M. 1989. Does Institutional Form Matter? Economic Behavior in the Inpa- tient Mental Health Care Industry. Research Paper 6, University of Wisconsin-Madi- son, Mental Health Research Center. Shortell, S. M., and Hughes, E. F. 1988. The effects of regulation, competition, and ownership on mortality rates among hospital inpatients. New England Journal of Medicine 318~171:110~1106. Showstack, J. A., Shroeder, S. A., and Matsumoto, M. F. 1982. Changes in the use of medical technologies, 1972-1977: A study of 10 inpatient diagnoses. New England Journal of Medicine 306~12~:70~712. Showstack, J. A., Stone, M. H., and Shroeder, S. A. 1985. The role of changing clinical practices in the rising costs of hospital care. The New England Journal of Medicine 313~191:1201-1207. Sloan, F. A., and Vraciu, R. A. 1983. Investor-owned and not-for-profit hospitals: Ad- dressing some issues. Health Affairs 2~1~:25-37. Sloan, F. A., Valvona, J., Perrin, J. R., and Adamache, K. W. 1986. Diffusion of surgical technology: An exploratory study. Journal of Health Economics 5~1~:31-62. Sloan, F. A., Morrisey, M. A., and Valvona, J. 1988. Medicare prospective payment and the use of medical technologies in hospitals. Medical Care 26~9~:837-853. Stano, M. 1987. A clarification of theories and evidence on supplier-induced demand for physician services. Journal of Human Resources 22~4~:611~20. Steinberg, R. 1986. Should donors care about fundraising? In: S. Rose-Acke~man, ed. The Economics of Nonprofit Institutions. New York: Oxford University Press. Stevens, R. 1989. In Sickness and in Wealth: American Hospitals in the Twentieth Century. New York, N.Y.: Basic Books. Taylor, A. K., and Wilensky, G. R. 1983. The effect of tax policies on expenditures for private health insurance. In: J. A. Meyer, ed Market Reforms in Health Care. Washington, D.C.: American Enterprise Institute. Thomas, L. 1975. The Lives of a Cell. New York: Bantam Books. Titmuss, R. M. 1971. The Gift Relationship: From Human Blood to Social Policy. New York: Pantheon Books. U.S. Bureau of the Census. 1975. Statistical Abstract of the United States, 1975. Wash- ington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census. 1979. Statistical Abstract of the United States, 1979. Wash- ington, D.C.: U.S. Government Printing Office. U.S. Bureau of the Census. 1987. Statistical Abstract of the United States, 1987. Wash- ington, D.C.: U.S. Government Printing Office. . Bureau of the Census. 1989. Statistical Abstract of the United States, 1989. Wash- ington, D.C.: U.S. Government Printing Office. U.S. Congress, Office of Technology Assessment. 1984a. The Hemodialysis Equipment and Disposables Industry. Washington, D.C.: U.S. Government Printing Office. U.S. Congress, Office of Technology Assessment. 1984b. Medical Technology and Costs of the Medicare Program. Washington, D.C.: U.S. Government Printing Office. U.S. Congress, Office of Technology Assessment. 1985. Medicare's Prospective Pay- ment System: Strategies for Evaluating Cost, Quality, and Medical Technology. Washington, D.C.: U.S. Government Printing Office.
48 BURTON A. WEISBROD U.S. Congress, Office of Technology Assessment. 1988. Infertility: Medical and Social Choices. Washington, D.C.: U.S. Government Printing Office. Warr, P. G. 1982. Pareto optimal redistribution and private charity. Journal of Public Economics 18~1~: 131-138. Weisbrod, B. A. 1971. Costs and benefits of medical research: The case of poliomyelitis. Journal of Political Economy 79~3~:527-544. Weisbrod, B. A. 1978. The Private Nonprofit Sector: Facts in Search of Theory. Discus- sion Paper 501, University of Wisconsin-Madison, Institute for Research on Poverty. Weisbrod, B. A. 1988. The Nonprofit Economy. Cambridge, Mass.: Harvard University Press. Weisbrod, B. A. 1989. Rewarding performance that is hard to measure: The role of nonprofit organizations. Science 244~49041:541-546. Weisbrod, B. A. 1991. The health care quadrilemma: An essay on technological change, insurance, quality of care, and cost containment. Journal of Economic Literature 29:523-552. Weisbrod, B. A., and Dominguez, N. O. 1986. Demand for collective goods in private nonprofit markets: Can fundraising expenditures help overcome free-rider behavior? Journal of Public Economics 30~11:83-95. Weisbrod, B. A., and Mauser, E. 1990. Partial vs. Total Effects of Institutional Form: Disentangling Endogenous from Exogenous Variables. Working Paper, Northwest- ern University, Center for Urban Affairs and Policy Research. Weisbrod, B. A., and Schlesinger, M. 1986. Public, private, nonprofit ownership and the response to asymmetric information: The case of nursing homes. In: S. Rose- Ackerman, ed. The Economics of Nonprofit Institutions. New York: Oxford Uni- versity Press. Wilensky, G. R. 1987. Technology as Culprit and Benefactor. Paper presented at a conference on health care policy, November 12-14, The College of William and Mary. Wilensky, G. R., and Rossiter, L. F. 1983. A re-examination of the use of physician services: The role of physician initiated demand. Inquiry 20~2~:162-172. Williamson, O. 1967. The Economics of Discretionary Behavior. Chicago, Ill.: Mark- ham Press. Wolfe, B. 1989. Assessment of the Effect of Medicare's PPS on the Demand for V.A. Inpatient Services: An Examination of Externalities of a Public Policy. Paper pre- sented at the meetings of the Association for Public Policy and Management, Wash- ington, D.C. Woodward, R. S., and Warren-Boulton, F. 1984. Considering the effects of financial incentives and professional ethics on "appropriate" medical care. Journal of Health Economics 3~31:223-237. Young,D. 1983. IfNotforProfit,for What? Lexington, Mass.: Lexington Books, D.C. Heath. Zuckerman, S. 1984. Medical malpractice: Claims, legal costs, and the practice of defensive medicine. HealthAffCairs3~31:128-133.
