Forthcoming, J. Newhouse and A. Culyer, eds., Handbook of Health Economics

Several persons provided useful comments on previous drafts of this chapter: Joseph Newhouse, Edward Norton, Charles Phelps, Carol Propper, and William Vogt. This review benefited from seminars at the National Bureau of Economic Research and a work group organized by Burton Weisbrod. Remaining errors and omissions are my responsibility.

Frank A. Sloan

1. INTRODUCTION

A firm may be viewed as a nexus of contracts (Jensen and Meckling, 1976; Alchian and Demsetz, 1972). Some of these contracts are with suppliers of inputs, others with suppliers of financial capital, and still others with purchasers of the firm’s products. Conceptually, the appropriate decision as to ownership rights should be based on that combination that minimizes transactions costs between the firm and its various contractors (see, e.g., Hansmann, 1996).

In this general sense, health care is no different with one notable exception that drives the way health industries are organized, the nature of product purchasers. The product purchaser is distinctive in two respects. First, consumers are often not as well informed about their purchasers as are the suppliers. Second, consumers typically pay much less than marginal cost for health care services with parties other than direct consumers footing the rest of the bill. In general, the identity of this larger consumer is often not well-defined; nor is this larger consumer’s willingness to pay well revealed with nearly the precision as in "normal" markets.

For both reasons, it has been widely believed among experts in the health field that the for-profit organizational form may not minimize transactions cost in the health care sector. Because of asymmetric information, suppliers of care solely motivated by profit may be inclined to sell more care than consumers would demand if they possessed the information of the seller. Because of externalities in consumption, including but not limited to innoculations against contagious diseases (Pauly, 1971) and the value of maintaining excess capacity to accommodate stochastic variation in demand (Joskow, 1980), willingness to pay of others than direct consumers of care, including the care of those not able to pay, are part of total benefit. To the extent that there is direct payment for these community benefits, a profit-seeking organization

would provide socially optimal levels of care distributed appropriately. However, absent such subsidies, care for which marginal revenue falls short of marginal cost should be undersupplied. In the context of the hospital, such care includes services to the uninsured, excess capacity such as in the emergency room, various patient educational and social services, care for particular diagnostic groups, such as AIDS patients or drug addicts, teaching and research. As Gray and Schlesinger (1997) explained, the reasons for eschewing the for-profit ownership form for hospitals come down to trust and community benefits.

The for-profit hospital is clearly in the minority numerically in all developed countries. In the United States, such hospitals constituted only 12 percent of all nonfederal short-term general hospitals in 1994 (American Hospital Association, 1995). By contrast, 60 percent of hospitals were organized as not-for-profit hospitals with the remaining 28 percent of hospitals being operated by governments or special government authorities.

Although the for-profit share of hospitals has been reasonably stable historically, for-profit chains have grown both numerically and in influence since they first appeared in the late 1960s. Their growth has not been steady, but rather has been characterized by cycles of growth (Gray and Schlesinger, 1997). The growth of for-profit chains in particular has elicited substantial concern among health care experts that consolidation of hospitals under the aegis of publicly traded corporations will mean higher priced and lower quality care, lower rates of production of unprofitable outputs, and less accessible care to those with low ability to pay. The remaining for-profit hospitals are small individually-owned enterprises. They command less interest because of their small size, and they are a vanishing breed.

The percentage of for-profit hospitals in other countries in the 1980s and 1990s was generally low as in the United States, but the mix of public versus private not-for-profit varied considerably. Hoffmeyer and McCarthy (1994) presented detailed accounts of health care systems in various developed countries, including descriptions of the hospital sectors in each of the countries they described as of the late 1980s and early 1990s.

In Canada, for example, 98 percent of hospitals were public facilities. Private hospitals only had a two percent share. In France, 65 percent were public, 16 percent private not-for-profit, and 19 percent for profit. However, the public facilities tended to be larger. In Germany, each of the three ownership types had about one-third share each. However, in terms of beds, public facilities were 51 percent, private not-for-profit 35 percent, and for profit, the remainder—14 percent. In Germany, proprietary hospitals tended to be owned by physicians. In recent years, local public authorities have increasingly established private corporations as direct owners of hospitals to allow greater flexibility in wage setting for managers and some clinicians. In the Netherlands, private for-profit hospitals were prohibited by law. Most hospitals (about 60 percent) were private not-for-profit (religious) hospitals. In Switzerland, 46 percent were owned by public bodies (e.g., cantons and communes). Of the remainder, 32 percent were private not-for-profit, often supported in part with public funds, and 22 percent were for profit. The for-profit hospitals only cared for privately insured, and for those with insurance packages combining social insurance with a substantial private care element. In the United Kingdom, most National Health Service (NHS) hospitals were managed by public district health authorities. Since 1991, some of the NHS hospitals have become NHS trusts which means that boards of trustees are outside of district health authority control. A small for-profit sector emerged starting in the late 1970s, but by 1990-91, the number of beds under private control actually declined. Generalizing across countries on this topic is difficult in that payment systems vary markedly (Glaser, 1987), physicians less often practice as independent entrepreneurs in hospitals as they do in the U.S. (Pauly, 1987), and legal structures differ among countries.

In this chapter, the empirical evidence presently is largely limited to the United States. For one, the literature is far greater for the United States than it is for other countries. Also, the volume is written in the English language, and most readers (including this author) will find English language articles accessible, but will have difficulty with most other languages. However, as discussed more fully below, some of the results should generalize to other countries.

As a "Handbook of Health Economics," this review focuses on economic literature. In contrast to the health care experts, many economists have been suspicious of the view that private not-for-profit and public organizations perform better, measured in terms of transactions cost. For one, with the profit incentive attenuated, profits may be distributed to managers, physicians, and hospital personnel rather than to the poor and to provision of unprofitable outputs.

The literature on this subject addresses three fundamental questions. First, why do private not-for-profit arrangements dominate the hospital industry? Second, how do private not-for-profits differ from for-profits in their behavior? Third, is the private not-for-profit form more efficient in this industry? The first two are positive questions, and the third is a normative question.

The first two sections address the first question. Section II discusses legal differences among the various ownership forms. Section III reviews various explanations for the dominance of the not-for-profit form advanced by others. The next four sections address the second question. Section IV presents a capsule description of models of hospital behavior which provide a framework for discussion of empirical evidence that follows. This type of model underlies most of the empirical analysis on comparative performance discussed in the sections that follow. Section V discusses in order, cost, profitability, pricing and cost-shifting, uncompensated care, diffusion of technology, quality of care, and capital funds and investment. Section VI presents empirical evidence on hospital ownership conversions. Which types of hospitals convert and what happens when they do? In Section VII, I link ownership with competition among hospitals. Section VIII concludes the chapter and addresses the normative question about relative efficiency of hospital organizational forms and implications for public policy. More specifically, given all of the empirical evidence, does the private not-for-profit form fit the hospital industry? How certain can we be of the answer, and what types of research is needed to provide a more definitive answer?

.

II. LEGAL DISTINCTIONS AMONG OWNERSHIP FORMS

The main difference between for-profit and private not-for-profit organizations lies in the distribution of accounting profit. The latter do not distribute such profit to individual equity holders but rather in the form of a dividend to its sponsors or to whomever it designates. In principle, the community, however this is to be defined, is the equity holder. In practice, there are no rules requiring such organizations to define their "community." However, in recent years, conversions from public or private not-for-profit status have increasingly received regulatory scrutiny. Thus, at least for the first time, the definition of community ownership is beginning to be operationalized as the public policy debate about the merits of hospital conversions to for-profit status unfolds. Incorporation laws do not preclude private not-for-profit organizations from paying economic rents to employees, managers, or others who may exercise control over them, such as physicians. There are no statutory limitations on the amount of profit such organizations can earn. Some states in the United States do not restrict the scope of activities of such organizations. Other states limit their activities (Hansmann, 1980, especially p. 839).

Private not-for-profits enjoy some government-conferred advantages, including exemption from corporate income and property taxes, somewhat better access to tax-exempt bond financing, and eligibility for private donations. Until the end of the 19th century, U.S. hospitals were almost exclusively donative institutions. The more affluent received care in physicians’ offices and at home (Starr, 1982; Rosenberg, 1987). As discussed below, donations to hospitals have eroded as all private hospitals and even some public hospitals have changed from largely charitable institutions to large commercial enterprises. Further, the corporate tax advantage has decreased, and for-profit hospitals have access to tax exempt bonds in the form of industrial revenue bonds. These advantages now amount to no more than 3 percent of revenue and probably less (Becker and Sloan, 1985). For-profit hospitals have a substantial advantage in being able to raise equity capital through sale of stock.

Another important legal distinction concerns election of boards of directors. Since private not-for-profit organizations do not have shareholders, there must be an alternative mechanism for selecting the board. Typically, such boards are self perpetuating, even though charters may limit their latitude in appointing members. Whether this is an important difference in practice is debatable. Hansmann (1996, p. 239) speculated that managers of not-for-profit firms may not be that different from those in large publicly-traded investor-owned firms in which individual shareholders exercise no meaningful control. The real difference may be that the for-profit’s organizational goal is clearer. Whom the not-for-profit boards serve is an important policy issue, particularly in situations in which board members stand to gain from a transaction, such as the sale of a private not-for-profit to a for-profit hospital organization (Gray, 1997). Lack of precise definition of organizational goals and of identity of the owners open the door to personal gain.

Charters of private not-for-profit organizations generally contain clauses expressly forbidding private inurement on the part of management and boards of directors. Such provisions severely limit or even rule out explicit incentive compensation plans that are managers have in the for-profit sector. Since there is no stock, there is no possibility of offering stock options to managers and key employees in private not-for-profit organizations. For private inurement provisions to be effective, they have to be enforced, and it is unclear who the enforcers are and how much real oversight there really is. Although it might seem logical that offering managers assets in the enterprise motivates managers, this is not always true (De Meza and Lockwood, 1998; Hadlock, 1998).

Public hospital organizations are similar to private not-for-profits in several respects. They too are not taxpaying, and they have about the same access to tax-exempt bonds. However, any dividends are paid through the sponsoring governmental entity. The entity monitors output and appoints the board of directors of the enterprise. Often, public hospitals have less autonomy than their private counterparts, in particular regarding compensation and employment practices. Also, in contrast to even private not-for-profits, public hospitals are generally not in a position to refuse to care for patients, unless they do not have the facilities to care for them.

III. WHY THE NOT-FOR-PROFIT FORM IS DOMINANT

Transactions Costs and Assignment of Ownership Rights. As Hansmann (1998) explained, ownership of a firm involves two essential rights: (1) the right to control the firm and (2) the right to appropriate the firm’s net earnings. The defining characteristic of private not-for-profit firms is that individuals who control the firm do not have the right to that firm’s earnings. In this sense, such firms have no owners, although, increasingly, the legal interpretation is that the community, however this is to be defined, is the owner.

The efficient assignment of ownership minimizes the sum of (1) the cost of contracting—transactions cost between the firm and parties that deal with it either as customers or as input suppliers and (2) the cost of ownership—the cost of monitoring the managers and of risk bearing.

Hansmann identified several potential inefficiencies in contracting that may influence choice of appropriate organizational form. To cope with market power that arises in the case of natural monopolies, it may be desirable for the customers to be the owners, for example, consumer cooperatives for electric utilities. Contracting may be costly when the firm possesses better information that the parties with which it deals. To deal with asymmetric information, it may be desirable for consumers to be the owners, as in mutual life insurance companies. In some sectors, there are substantial transactions costs of switching after the initial investment has been made. Such transactions costs are often high in franchising. A response to franchisor opportunism is to make the franchisees, as a group, the owners of the franchisor. Worker-owned firms may protect personal investments made by employees in firm- or location-specific capital. Of course, shirking by employees is also a risk; this consideration may affect ownership rights in situations in which it is more likely. Suppliers of capital face the risk that the firm will be reckless with their funds and therefore face the risk of not being repaid; thus, there is a case for assigning monitoring rights to suppliers of capital.

On costs of ownership, some parties are better risk bearers than others depending on the ability to diversity away risk. Efficient assignment of ownership rights also depends on the ability of owners to be informed about the operations of the firm, and ability to see that owners’ decisions are actually carried out.

Hansmann explained why societies choose to assign property rights to ownership and why such patterns differ among industries. Several arguments for the dominance of the private not-for-profit form in the U.S. hospital industry deal with transactions cost of ownership: fiduciary relationships and complex output; implicit subsidies; and explicit subsidies.

Fiduciary Relationships and Complex Output. Kenneth Arrow (1963) explained the dominance of not-for-profits as a response to uncertainty and incomplete markets for risk in markets for medical care. Because such organizations are not pure profit-seekers, they would not fully exploit their market power vis-à-vis a patient who experienced a major health shock. In terms of the above framework, the Arrow argument suggests that private not-for-profit hospitals dominate because of lower costs of contracting with consumers.

Weisbrod (1988, Chapter 3) extended this concept to not-for-profits more generally. There are many classes of services for which the seller who may not be fully informed knows more than the patient, client, or donor. For such "customers," evaluating and rewarding performance is difficult. Often, for such services, the customer may never know for certain what would have happened if the service had not been performed or if it had been purchased from another seller. "If, in the case of health care, the patient improved, was it because of, or in spite of, the care received? Would an elderly, infirm resident of a nursing home have been better off at another home? The buyer of legal services will never learn whether he or she would have been better off dealing with another attorney; was a lawsuit won (or lost) because of or in spite of the attorney? When college education is purchased, would the student have received a ‘better’ education elsewhere? When a charitable contribution is made to an organization, what would have happened to the needy if the donation had not been made?" (p. 46).

Although this type of argument seems compelling, it does not explain why the private not-for-profit form would be dominant for some but not all types of health care providers. Many types of health services, such as nursing homes and physicians’ offices, are predominantly organized as for-profit enterprises. Nursing home care does not generally involve complex technology, but given the physical and mental condition of most patients, the potential for exploitation of patients exists. Physicians refer patients to hospitals. If the need for trust applies to hospitals, it applies with greater force to physicians. As is now much more commonly recognized than before, physicians are often uncertain ex ante about what the effects of their care is likely to have on their patients.

Complex output gives rise to asymmetric information (see, e.g., Easley and O’Hara, 1983). Not-for-profit organizations may be less prone to dupe consumers. Also, to the extent that consumers cannot monitor output, the profit-maximizing quality level may be lower than the level that consumers would demand if they were fully informed. However, again, physicians typically monitor hospital care, and they mostly work for for-profit firms.

Although the authors’ primary application was to prisons, they also commented (among several areas) on implications for hospitals. They argued that the gains from innovation in hospitals are potentially enormous, but so is the potential harm to quality from cost cutting. On balance, the ability of patients to choose their hospital (not a choice possible in all countries) strengthens the case for provision by a private contractor. They asserted that "a crucial difference is the limited ability of consumers to assess the quality of health care they receive. Consumers often cannot tell whether hospitals have failed to provide care to save costs, and hence would not so readily change suppliers in response to poor quality." (p. 1157). They concluded that this may be the reason that the public sector is an important provider of care in most countries.

Implicit Subsidies. A pure profit-seeking organization determines outputs where marginal revenue equals marginal cost. For some outputs of benefit to communities, private marginal revenue may fall short of marginal cost at the private firm’s optimum. An organization not fully motivated by profit may provide such socially beneficial services anyway.

This rationale is also not without its deficiencies. First, implicit cross-subsidies may not be the most efficient way to achieve socially optimal outcomes. One reason for this may be that incentives of principles and agents are not aligned. The boards and managers may be imperfect agents for the principal, the community, however this is to be defined. Second, penalties for failing to provide implicit cross subsidies are quite limited. There is no market for hospital boards in which communities can replace boards that do not act in their interest. Governments do not link tax-exempt status to the constellation of outputs provided. In fact, sanctions would have no legal standing since there are no explicit numerical standards specifying the outputs such organizations should provide.

Explicit Subsidies. A category of private not-for-profit enterprise is the donative charity. Individuals donate to such organizations with the intent that these organizations make private transfers to needy or worthy individuals. In effect, the donors purchase services from charities that in turn are given to third parties, the ultimate recipients of the gift. Here the rationale involves both costs of contracting and ownership, and the solution is to eliminate property rights to the organization’s residual (Hansmann 1998).

Even if there were tax advantages for donating to for-profit enterprises, few donors would probably want to do this since they would not want to take the chance of enriching the firm’s shareholders (see e.g., Hansmann, 1980, 1998). Also, some charities receive donations from thousands of individuals. Organizing such a dispersed group would be costly.

This also cannot be a major determinant of ownership mix of hospitals, except as a legacy of a time in which hospitals were truly donative institutions. As explained below, donations to hospitals, particularly as a percentage of capital funds, have decreased appreciably.

Given the lack of importance of donors to hospitals, it is hard to see them as a major factor in disciplining hospitals.

Inertia. Because private not-for-profit organizations, including hospitals with this ownership form, have no well-defined owners, transactions converting such firms to other forms require consent of the firms’ managers, who may have a personal stake in opposing such conversions. For this reason, the forces of market selection operate slowly, and the form remains dominant long after the rationale for the form has disappeared (Hansmann 1998, p. 741).

Cartel Theory. The next reason for the dominance of the not-for-profit form, if true, would be a form of market failure. The theory was not justified on the basis of minimizing the total cost of ownership. Rather, the private not-for-profit hospital is dominant because this in the doctors’ collective financial interest.

In the United States, most physicians who treat patients in hospitals are not employees of these organizations. Yet through their power to admit, they potentially exercise an important influence over hospitals (Pauly, 1980). Pauly-Redisch (1973) and Shalit (1977) have argued that hospitals are operated in the interest of a physician cartel. The Pauly-Redisch model is explained in some detail below.

Cartelization as a rationale for the dominance of the private not-for-profit hospital has some appeal for not-for-profits under local control, but is less applicable to major teaching hospitals and chain for-profit hospitals. For teaching hospitals, decision-making is shared among the physicians, their departments, the medical school, and the parent university. Local members of for-profit chains report to the home office, which in turn bears responsibility conferred upon it by the firms’ owners. Opposition to the growth of the latter on the part of physicians may reflect potential loss in the ability to cartelize when the hospital is owned by a for-profit chain.

Some would argue that private not-for-profit hospitals are less efficient than for-profits. The conceptual possibility of inefficiency stems from a lack of a well-defined residual claimant, which is an issue for many kinds of not-for-profit organizations. Yet, except for the problem of monitoring individual physician behavior at large hospitals, an issue Pauly-Redisch discussed, physician cartel members, as residual claimants, would not want to tolerate inefficiency since this would reduce their private returns. Thus, to the extent that hospitals are parts of cartels, it would seem unlikely that physicians would squander their income on perquisites for managers or other employees.

Profitability, For-Profit Entry and Industry Performance. Lakdawalla and Philipson (1999) focused on two questions. First, how does competition among firms lead to such marked differences in the share of for-profit activity across states? Second, how does competition influence the relationship between the relative size of the nonprofit sector and industry performance?

III. MODELS OF NOT-FOR-PROFIT HOSPITAL BEHAVIOR

Objectives. Ideally, a private not-for-profit hospital would make input and output decisions to maximize the welfare of the community subject to the constraints it faces. In practice, none of the economic models of hospital behavior specifies the maximization problem in this way. Zweifel and Breyer (1997, Chap. 9), for example, identified four hospital stakeholders: hospital-based physicians; hospital employees; owners--perhaps the community in this context; and nonphysician managers. Each stakeholder has distinct and sometimes conflicting objectives. Although hospital policies realistically reflect the results of a game, no one has formalized conflict in this way. In the Zweifel-Breyer model, the hospital objective function contained quality, profit, and labor slack, and capital slack as arguments. Profit in this one-period model is valued because, for one, such funds can be used for hospital investment in quality-enhancing technology. Slack or inefficiency may yield utility because nonphysician staff value a slower pace of work or having extra supplies on hand. Slack, however, is difficult to distinguish from quality and adding arguments to the objective function greatly decreases a model’s analytic tractability.

In an early and oft-cited model of hospital behavior, Newhouse (1970) assumed that not-for-profit hospitals maximize quantity and quality subject to a constraint developed from the loci of points where demand curves intersect average cost curves (points of zero profit). Eliminating slack as an argument from the hospital utility function simplifies the model in useful ways.

Comparative Statics Analysis. The Newhouse-type model is readily formalized (Sloan and Steinwald, 1980b, pp. 20-33). Let U=U(X,Y,p ) where X=output, Y=quality, and p =profit. The hospital faces a downward-sloping demand curve in its product market, P=P(X,Y;M), where P=price and M=any exogenous demand determinant. The hospital’s total cost function is C=C(X,Y;N), where C=average cost and N=any exogenous cost determinant, such as an input price.

Then according to the first-order conditions, the hospital sets quantity and quality, respectively, where the marginal utilities of quantity and quality equal the marginal disutilities of lost profit resulting from extending output and quality beyond their profit maximizing levels. Under various restrictive assumptions, in particular, a nonnegative U_xy (small income effect), an increase in demand (dM>O) raises both quantity and quality. Conversely, for an increase in cost (dN>O), both quantity and quality fall. If Y is dropped from the utility function, unambiguous results are obtained without the necessity of signing U_xy. That is, hospital output rises in response to outward shifts in demand and falls when input prices rise.

Retaining the simplified utility function and specifying a production function for output X=f(K,L), where K=capital and L=labor, one can analyze effects of not-for-profitness on input use. The effect of extending output beyond the profit-maximizing point is to raise input use. Yet the familiar textbook condition for the optimal input use ratio (capital and labor employed at points where the input price ratio equals the ratio of the marginal products of the inputs) is the same as for the profit-maximizing hospital. Only under the restriction that U_KL >0 are dL/dw (for wage) and dK/dr (for capital price) unambiguously negative. The effects of changes in w and r on L and K are unambiguously negative if one assumes a competitive rather than a monopolistic market for hospital care (Sloan and Steinwald, 1980b, pp.24-26).

This type of model has also been applied to study hospital responses to rate, revenue, and input regulations (see Sloan and Steinwald, 1980b, and Zweifel and Breyer, 1997, Chap. 9). For example, if the government fixes price at P*, then the only way for a hospital to secure additional patients is to raise Y. Under the assumption of U_yp*>O, an increase in the regulated price raises quality. Under revenue regulation, the governmental authority sets the hospital’s budget (R*). If U_xy is nonnegative, raising R* raises quantity and quality.

Hospital as a Physicians’ Cooperative. In an alternative formulation, the private not-for-profit hospital sets output so as to maximize the income per doctor of the doctors on the hospital’s medical staff (Pauly-Redisch, 1973). Then X, K, L, and the number of members of the hospital’s medical staff are set to maximize income per medical staff member. In this model, which is in the tradition of cooperative-collective models developed in other contexts, supply responses to changes in product and factor market conditions can be in reverse directions. That is, an increase in demand (M above) could boost price, reduce output, and result in lower hospital medical staff sizes. The model is silent about quality, community benefit, prestige and so on. They are not motives of the physician cartel.

So long as physician staff members cooperate to maximize income per doctor, hospitals should be technically efficient. However, as medical staff size increases, individualistic behavior on the part of doctor cartel members becomes increasingly difficult for the collective organization to monitor, leading to the possibility of some slack. In the more traditional models with U depending on X and Y, there is no room for slack either. However, as Y increases beyond the profit-maximizing level, so does cost.

IV. EMPIRICAL EVIDENCE ON THE EFFECT OF OWNERSHIP ON HOSPITAL PERFORMANCE

There is a rich empirical literature on hospital behavior which focuses on private not-for-profit hospitals, often in comparison to other ownership forms. In this review, I describe methods and findings on hospital cost, profitability, pricing and cost-shifting, uncompensated care, diffusion of technology, quality of care, and hospital capital funds and investment.

Cost. Cost may vary by ownership for several reasons: competitive advantages conferred by governments; community benefits, teaching, and/or research; slack; quality; and casemix severity. But mainly, the cost test for hospital ownership differences has been conducted to determine whether for-profit hospitals are more efficient. To state conclusively that for-profit hospitals are more efficient, it is necessary to hold these other factors, as well as input prices and scale, constant. Even if one could successfully do this, it would be difficult to distinguish whether cost differentials were due to slack or quality.

To the extent that the residual claimant is not well-defined, economists would expect that private not-for-profit and public hospitals would be less efficient. As seen above, under some circumstances the not-for-profit hospital’s residual claimant may not be well-defined. An allegation against the for-profit hospitals, especially the chain hospitals, advanced by some noneconomists, is that they "push" ancillary services, such as diagnostic procedures (Relman, 1980). This phenomenon too is difficult to measure with acceptable accuracy without conducting detailed reviews of medical records which has not been done. Higher rates of use of ancillary services do not necessarily imply abuse. A related criticism of for-profit hospitals is that they charge higher prices. This is an efficiency issue to the extent that there are deadweight losses associated with high taxes or premiums for public or private health insurance. Some studies have used price-based measures for comparing performance of alternative ownership forms.

In empirical work, an important methodological issue relates to measurement of cost. Economic cost includes a return on equity which reflects competitively-determined cost of capital. In fact, most studies have either used accounting cost which excludes such returns or a price measure which may include an economic rent. Cost may vary systematically by ownership because of differences in the tax laws. Tax-paying hospitals have an incentive to use accelerated depreciation, which others do not. Finally, costs could be related to style and amenities in care not captured by any explanatory variables.

Overall, the empirical evidence demonstrates no systematic differences in efficiency between for-profit and not-for-profit hospitals. Perhaps, until recently, the hospital market did not provide sufficient discipline for any hospital type to be efficient. Using accounting cost per case and not controlling for other factors, for-profit hospitals have the same or slightly higher cost. However, using per cost them or revenue-based measures per unit of output, for-profit hospitals tend to be more costly (Institute of Medicine, 1986). This is attributable in part to the shorter patient length of stay in for-profit hospitals.

The studies have used three distinct in methodologies: paired comparisons; regression analysis; and frontier production functions. Results from the paired comparisons run the gamut from for-profits are more costly (see, e.g., Lewin et al., 1981; Pattison and Katz, 1983) to for-profits and not-for-profits are equally or for-profits are less costly (e.g., Herzlinger and Krasker, 1987; Sloan and Vraciu, 1983). The largest difference reported by Lewin and co-authors and others was in the much higher use of ancillary services by for-profit hospitals. Part of the difference in findings reflects adjustment for taxes which logically should be counted as a community benefit.

Among the early regression studies, after controlling for a number of other factors including casemix and teaching status, Becker and Sloan (1985) found that free-standing for-profit hospitals (not part of a chain) were more expensive than other free-standing ownership types in terms of cost per adjusted patient day (adjusted to convert outpatient visits into inpatient days), but they were least costly in terms of cost per adjusted admissions (adjusted to convert outpatient visits into inpatient admissions). The chain for-profits that had been part of a chain for four years or more were six percent higher on cost per adjusted admission than were free-standing not-for-profit hospitals, but the coefficient on the binary variable for this group was statistically insignificant at conventional levels. Granneman et al. (1986), who focused on assessing scale and scope economies, included binary variables for hospital ownership. They found that hospitals operated by local or state governments had costs about eight percent lower and for for-profit hospitals had costs that were 15 percent higher than private not-for-profit hospitals.

Breyer et al. (1988), using a single cross section of 614 German hospitals, regressed cost per case on ownership, teaching status, various casemix indicators, and measures of output and capacity. Public hospitals had higher cost than for-profit hospitals. The difference between private not-for-profit and for-profit hospitals was not statistically significant at conventional levels, although the sign on the coefficient for private not-for-profit was negative and larger than its standard error.

Among types of cost, for-profit hospitals tend to use less labor but more nonlabor inputs (Sloan and Steinwald, 1980b). There is no difference in wage rates paid by hospitals of different ownership types, holding other factors constant (Sloan and Steinwald, 1980a).

Linear programming techniques, applied to individual hospital departments and to hospitals as a whole can be used to estimate a frontier production function. Inefficiency is then evaluated in terms of a hospital’s divergence from hospitals at or near the frontier. Wilson and Jadlow (1982) applied this technique to nuclear medicine departments of hospitals, finding that government hospitals were less efficient on average and for-profit hospitals more efficient than the omitted reference group, private not-for-profit hospitals. Focusing on one department probably reduces output heterogeneity, but these authors probably included too few covariates, which may be a source of serious omitted-variables bias.

Using stochastic frontier regression analysis of 219 community hospitals in New York, Vitalino and Toren (1996) found no effect of hospital ownership on hospital efficiency. With a much larger, and national samples and using the same technique, Zuckerman et al. (1994) found that for-profit hospitals were more likely to be inefficient than private not-for-profit hospitals. There were no statistically significant differences in inefficiency between public and private not-for-profits. Finally, Koop and co-authors (1997) used a Bayesian approach for estimating a stochastic cost frontier production function which they applied to a panel of U.S. hospitals. They found government-run hospitals to be the most efficient, followed by not-for-profit, and for-profit hospitals in last place.

In the same issue in which the Zuckerman et al. article appeared, Newhouse (1994), commenting on articles in the issue that used the frontier technique pointed to a weakness of the methodology, namely what is measured as inefficiency may be an amenity or unmeasured aspect of quality that consumers value. Are wider seats in first-class waste? Probably not. Consumers pay more for comfort on airlines because they value this. But cost per passenger-mile would be higher in first class. Of course, the same critique applies to any of the cost techniques economists have used since the analysis cannot capture quality differences of importance to consumers.

Profitability. If the term is taken at face value, private not-for-profit hospitals should earn a zero profit. But this applies to economic profit, and measurement has been done on accounting profit. Private not-for profit hospitals in the United States have consistently earned accounting profits. Judged in terms of such profit as a percent of revenue, in three years, 1988-90, such hospitals were even more profitable than for-profit hospitals (U.S. Medicare Payment Advisory Commission, 1998, p. 72).

Hospital profits tend to be more variable for for-profit hospitals. Hoerger (1991) studied profit variability. He tested two hypotheses. First, various models of private not-for-profit hospital behavior specify that utility is maximized subject to a zero profit constraint for the hospital. If so, profits of such hospitals should be less variable. Second, the profits of not-for-profit hospitals should be less responsive to changes in such exogenous factors as payment levels of government payers. This would occur under conditions of cost-shifting, a phenomenon to be discussed next. He found empirical support for both hypotheses.

Hospital Pricing Patterns: Cost-Shifting in the Face of Demand Shifts. The notion that hospitals increase price to private payers when the government reduces the price it pays is widely accepted among health care experts outside the economics profession. The term given this type of behavior is "cost shifting" (Ginsburg and Sloan, 1984; Dranove, 1988, Morrisey, 1994). This type of behavior is to be distinguished from price discrimination that occurs when markets can be segregated by the seller and demand elasticities differ across markets.

Conditions for cost-shifting to occur are (1) the hospital has market power in its product market (private sector) and (2) such power was not fully exploited before the government reduced the price it paid hospitals. That is, the hospital did not set price at the profit-maximizing price before the government decreased price. The intuition is that hospitals raise price toward the profit-maximizing price to make up for the shortfall caused by the fall in the price paid for hospital care by government. One would expect nonprofit maximizing behavior from private not-for-profit but not for-profit hospitals.

More formally, following Dranove (1988), let the private not-for-profit hospital’s utility function be U=U(p _i(P_i,C) + p _j(P_j,C), X_i(P_i)X_j(P_j)). Market i is the private market in which the hospitals sets price _; P_I Market j is the public market. Quantity x_j in the public market is exogenous as is P_j. C is cost. The first-order condition requires that optimal P be set where the marginal utility of profit from a change in price equals the marginal utility from loss of output to the private market as P_i changes. If the marginal utility of the private market output is zero, which holds in the profit-maximizing case, P_i is set to maximize profit. Totally differentiating the first-order condition and imposing restrictions that (1) p _i>O and (2) the marginal utility of profit does not decrease with additional private output, then dP_i/dP_j<O, which is cost-shifting.

In his empirical analysis, Dranove found that a one-dollar decrease in hospital profits from government sources per private admission led to fifty-one-cent increase in price per private admission. This implies that about half of the revenue loss was recovered.

Morrisey (1994)’s book on hospital cost shifting is a very comprehensive review of theoretical arguments and of empirical analysis of the issue up to the date of the book’s publication. As he noted, the empirical evidence in support of cost-shifting is largely based on data before 1983 when the Medicare Prospective Payment System (PPS) was enacted. Other evidence he cited is inconsistent with cost shifting. For example, in one study of Blue Cross, it was found that Blue Cross saved money after Medicare PPS was implemented, presumably because the Medicare incentive to boost efficiency was also applied to patients with different payment sources. More recent studies showed that reductions in payment levels caused some hospitals to reduce or eliminate care for Medicaid patients. Studies of California hospitals, a leading state in introducing competition into the hospital sector, found that hospitals lowered prices in response to growth of managed care. These price cuts were larger when the hospital faced more competition. In Morrisey’s words, "None of these responses supports the idea that cost shifting works" (p. 85). He reasoned that it far more likely that if the indigent were to disappear, hospitals would spend their surpluses on other things, such as research, new technology, nicer offices for physicians and/or hospital administrators. Only if the hospital "likes" paying patients would prices charged such individuals fall. As he noted, "this seems like a thin reed on which to hang much public policy." (p. 86).

Other empirical evidence on private not-for-profit hospitals not setting price to maximize profits is from Lynk (1995). In this study, Lynk focused on the difference horizontal mergers of hospitals make on price, depending on the ownership of the merging organizations. When a merger occurs, price may fall in response to efficiency gains or may rise to the extent that the merged entity is more able (has more market power) and willing (wants to exploit this added market power) to raise price. Lynk argued that a private not-for-profit hospital may be devoted to its product which is hospital services than a government organization which may seek to earn a profit from health care to subsidize a nonhealth good. For-profit hospitals would want to set price to maximize the return to their owners. Empirically, using data from California, Lynk showed that mergers by private not-for-profit hospitals lowered price on average. By contrast, such mergers had the opposite effects for mergers involving the other two merger types.

The data in both studies were for the 1980s. To the extent that hospital markets have become more competitive since then, cost-shifting by private not-for-profit hospitals should be decreasing because of diminished ability to raise price, even in markets in which sellers’ output is concentrated. This is because group purchasers of hospital care have become more aggressive. In fact, an unpublished study by Keeler et al. (1999), using more recent data for California (through 1994), found that private not-for-profit hospitals located in market areas with higher hospital concentration charged higher prices, and the effect of concentration on price has increased over time. This is the "normal" relationship one would expect.

The growth of managed care has reduced hospitals’ ability to set price. Brooks and co-authors (1997) studied variation in the difference in price obtained and the minimum price that the hospital would accept to provide a privately-insured episode of care (for appendectomy). A large difference was interpreted as high bargaining power by hospitals. As one of many findings, the authors reported "rather paradoxically, for-profit hospitals had significantly less bargaining power than either public or voluntary hospitals" (p. 431). I am puzzled as well, but mention results such as this as a challenge for future research. One possibility is that lower markups at for-profits may reflect more elastic demand for their product.

Uncompensated Care. One public policy concern is that an increase in competition will place the poor and the uninsured at an even greater disadvantage than previously (see e.g., Mann et al., 1995). However, differences between private not-for-profit and for-profit hospitals in provision of uncompensated care were very minor, even before the 1990s (Sloan et al., 1986b). Uncompensated care is typically defined as the sum of charity care, services for which the hospital did not expect to be compensated at the time of the patient stay, and bad debt, services for which the hospital anticipated compensation but did not receive it.

Uncompensated care data for 1994 showed that, for private not-for-profits, uncompensated care was 4.5 percent of revenue. The corresponding percentage for for-profits was 4.0 percent (U.S. Prospective Payment Assessment Commission, 1996, p. 84). Differences in provision of uncompensated care by ownership for psychiatric hospitals appear to be much larger than this (Schlesinger et al., 1997).

There is evidence that ownership mix in a community affects hospitals’ provision of uncompensated care. In recent years, hospitals in markets with a higher for-profit market share provided less uncompensated care and were more likely to adopt policies to discourage admission of uninsured patients (Frank et al., 1990).

Morrisey et al. (1996) compared uncompensated care provided by private not-for-profit hospitals in California in 1988 and 1991 to the tax subsidies such hospitals received. For all but 20 percent of the hospitals, uncompensated care exceeded the tax subsidies. Yet this comparison is imperfect since the potential community benefit involves more than provision of uncompensated care . Several recent unpublished studies have concluded that for-profit hospitals provide less uncompensated care, but at least some of these studies were conducted to support advocacy positions of private not-for-profits (Kuttner, 1996a). Young and co-authors (1997) examined changes in the provision of uncompensated care by California hospitals that converted from private not-for-profit to for-profit status and found that conversion did not uniformly reduce provision of such care. With a national sample of data from 1981, Norton and Staiger (1994) found that when a for-profit and a private not-for-profit hospital are located in the same area, they serve an equivalent number of uninsured persons. However, for-profit hospitals more frequently locate in areas where better-insured persons live.

Frank and Salkever (1991) modeled provision of uncompensated care by private not-for-profit hospitals. They considered two alternative models: (1) a pure altruism model; and (2) an impure altruism model. In the first, hospitals have profit and unmet need in the community as arguments in the hospital’s utility function. In this model, increases in provision of uncompensated care by other hospitals in the market "crowd out" provision of such care by the hospital in question (except if the income or endowment effect is very strong). In the alternative, the hospital competes for public goodwill by providing free care. Here crowding out is less likely since hospitals compete for patients who do not pay for their reputational value. Using data from Maryland hospitals for 1980-84, the authors obtained mixed results for both pure and impure altruism models. Earlier work by Thorpe and Phelps (1991) found some evidence of crowding out. More specifically, private not-for-profit hospitals provided less uncompensated care when public hospitals were present in their markets.

In Gruber’s (1994) model, the private not-for-profit hospital maximizes utility which depends on profit and uncompensated care subject to a zero profit constraint. Under some restrictive assumptions, his comparative statics analysis showed that uncompensated care should fall as price consumption increases. Greater competition will affect the hospital by increasing the elasticity of its demand curve. Price competition in California increased markedly after 1982. Gruber’s empirical analysis of California data for 1984-88 showed relative decreases in hospital profitability in those areas with the least output concentration, that is, in the most competitive hospital markets. Likewise, and perhaps as a result, hospitals in such markets showed the greatest decreases in amounts of uncompensated care provided. Gruber also found that the price of hospital care rose less in less concentrated markets, a result that at first glance seems inconsistent with Lynk’s (1995) study. However, Lynk distinguished the effect of output concentration by hospital ownership. Gruber considered the effects of concentration on price irrespective of ownership.

Diffusion of Technology. A few empirical studies have investigated diffusion of technology in hospitals. However, there are few well-controlled studies, and most are now old. A large variety of innovations have been examined, including process and more often product innovations (Romeo, et al., 1983; Russell, 1979, Sloan et al., 1986a; Sloan et al., 1997). With rare exceptions, there were no statistically significant differences in adoption between private not-for-profit and for-profit hospitals. Schlesinger et al. (1997) studied technology adoption patterns among psychiatric hospitals. He found that increased competition had a greater impact on the for-profit psychiatric specialty hospitals’ propensity to adopt innovative services (measured by an index of four services--treatment of eating disorders, treatment of substance abuse, geriatric services, and adolescent treatment) than it did for private not-for-profit psychiatric specialty hospitals.

Quality of Care. In some hospitals models, quality (Y) is included as an argument for the reason that rather than distribute profit to stockholders, not-for-profit hospitals may use profit to cross-subsidize a higher than profit-maximizing level of quality. Although economists include quality in their conceptual work, noneconomists have conducted most of the empirical work. This review considers some of the more important studies on the effect of ownership on hospital quality. Several different measures are used to measure quality: structure or inputs to care; process or measures of how care is delivered; and outcomes. Most private and public regulation of quality to date has focused on structure and process. The lower staffing levels in for-profit hospitals, especially lower nurse staffing, discussed above, are structural indicators of lower quality in such hospitals. However, the relationship between hospital staffing and other measures of quality including outcomes has not yet been established (Institute of Medicine, 1996). On other structural measures, such as the percent of hospitals with national accreditation and percent of hospitals with cardiac and intensive care, ownership does not make a difference (Herzlinger and Krasker, 1987).

The most rigorous and extensive study of large-scale empirical study of quality published to date that permits comparisons of quality by hospital ownership is by Keeler and co-authors (1992). They used two process measures of quality based on reviews of 14,000 medical record for five diseases in five states. One of this "explicit process" gauged the extent to which the charts showed that specific diagnostic and therapeutic procedures were performed competently. Rather than focus on particular elements of care as explicit process did, a second process measure, "implicit process," measured the care process overall. For example, one of their implicit process questions to physician reviewers was: "Based on what you now know about this case, would you send your mother to this hospital?" In addition, they gauged quality on an outcome measure -- the difference between actual mortality and the rate that would be expected based on the patient’s characteristics.

They found no difference in quality between private not-for-profit and for-profit hospitals on two indicators, excess mortality and explicit process; public hospitals fared worse on both criteria. However, on a third measure, implicit process, there was a statistically significant difference between quality of care of private not-for-profit hospitals and the other two ownership types, indicating higher quality levels for the for profits. The authors appear to have been more persuaded by the results on the first two indicators, stating that "nonprofit and for-profit hospitals provide similar quality overall" (p. 1712).

In their national study of 981 hospitals in 1983-84, Shortell and Hughes (1988) found no difference in quality measured in terms of mortality by ownership. However, using fewer covariates, Hartz et al. (1989) did find that mortality was higher in for-profit than in private not-for-profit hospitals.

Sloan and co-authors (1998a,1998b, 1998c) examined outcomes of care of elderly persons hospitalized for one of four conditions: hip fracture; stroke; coronary heart disease; and congestive heart failure. They analyzed the first admission for these conditions since patients with a first unanticipated major health shock are less likely to shop among hospitals. Their outcome measures were survival, functional status, cognitive status, and living arrangements (probability of living in a nursing home). Although, on some measures, patients admitted to major teaching hospitals did better, a result consistent with Keeler et al. (1992), there were no statistically significant differences in outcomes between non-teaching private not-for-profit and for-profit hospitals. On some measures, elderly patients admitted to non-teaching government hospitals had worse outcomes, holding a large number of other factors constant.

Mark (1996)’s study of psychiatric hospitals assessed process measures of quality--violations and complaints reported to the Medicare and Medicaid programs. Comparing means, she found some differences in quality that were unfavorable to for-profits. She also estimated separate quality equations for profit and private not-for-profit hospitals. The signs on some of the coefficients differed between the profit and not-for-profit models that she interpreted as evidence that the two ownership types have different objective functions. For for-profits, market concentration was negatively related to the number of violations. As competition increased, violations increased. But for not-for-profits, the relationship was positive. In more competitive markets, quality improved. From these results, Mark inferred that not-for-profit hospitals compete by offering a higher level of quality. By contrast, for-profit hospitals respond to competition by lowering quality.

Schlesinger et al.’s (1997) study of psychiatric hospitals found that private not-for-profit hospitals provided greater access than for-profits when competition is limited, measured by uncompensated care. With increased competition, ownership-related differences in uncompensated care were reduced, a consequence worth noting as competition among hospitals increases.

Hospital Capital Funds and Investment. Ownership has an important bearing on sources of capital funds. Public and private not-for-profit hospitals are precluded from obtaining funds from the sale of equity. All equity comes from either philanthropic contributions or internally in the form of retained earnings. All hospitals have access to debt capital, but public and private not-for-profit hospitals have better access to revenue from sale of tax exempt bonds and to direct subsidies from private donors and from governments.

Two studies have examined capital structure of hospitals with emphasis on hospital ownership. In Wedig et al. (1988)’s analysis, for-profit hospitals solve a one period problem of value maximization. Not-for-profit hospitals maximize profits to serve a nonpecuniary objective, such as providing a higher level of service than the profit maximizing level. The former pays taxes on debt and equity income and benefits from non-debt tax shields. The tax advantage of debt is a primary motivation for issuing debt. The private not-for-profit hospital pays no tax. Thus other reasons, such as the extension of cash flow for nonpecuniary objectives must be a reason for issuing debt. Bankruptcy risk is a negative consideration in issuing debt for both ownership types. Both types receive payments from cost- and charge-based payers. A cost-based payer reimburses the hospital for "its" share of interest and depreciation expense. A charge payer bases payment to the hospital in part of the price billed to a covered patient.

The authors’ theoretical analysis showed that when cost-based revenue as a share of total hospital revenue increases, private not-for-profit hospitals will issue more debt. Because of more complex offsetting effects, including the effects of cost-based payment on the value of debt as a tax deduction, the direction of effect from an increase in the cost-based revenue share on capital structure could not be determined theoretically for for-profit hospitals.

Empirically, the study showed that the effect of the share of revenue from cost-based sources on the hospital’s debt to asset ratio was positive and higher for for-profit than for not-for-profit hospitals. Other than the effect operating through the cost-based payer share, there were no differences in hospital capital structure according to ownership. Perhaps, as my discussions with persons in the hospital industry suggested was true at the time, for-profit hospitals were more adept at maximizing reimbursement. The payment system rewarded maximizing reimbursement rather than efficiency.

Although private not-for-profit hospitals do not benefit from using debt to shield profits from the corporate income tax, they do benefit from the personal income tax reductions reflected in the yields they pay on municipal bonds. In the second article on capital structure, Wedig et al. (1996) found that the not-for-profit hospital selects taxable and tax exempt debt levels jointly with fixed investment to maximize end-of-period returns which in turn are used to fund its nonpecuniary objectives. The hospital can earn an indirect arbitrage by issuing low-yield tax exempt debt rather than spending its accumulated cash reserves. Offsetting this incentive to issue debt are several factors causing the marginal cost of borrowing to increase with higher leverage, such as agency costs and bankruptcy risk. The balance of these factors leads to an internal optimum consisting of an optimal mix of equity and debt. Also, use of debt may be constrained by a "project financing" rule that requires that fixed investment exceed the value of tax-exempt debt flows. Consequently, the hospital may be induced to undertake low-yield investments to increase access, to tax-exempt debt. Conversely, the number of high-yield investments may exceed debt capacity. In such cases the constraint against issuing equity may cause the hospital to issue tax-exempt debt beyond the optimal leverage point. Empirically, the authors found that private not-for-profit hospitals behave as if they have target levels of tax-exempt debt. Debt targeting is constrained by availability of capital projects. Excess debt capacity increases investment.

Payment policies of insurers have not only influenced hospital capital structure, but also have affected hospital ownership composition. Wedig et al. (1989) assessed effects of variations in cost recovery policies of insurers on hospital location. They found that for-profit hospital market shares were higher in states in which payment policies were more generous and conversely for private not-for-profit hospitals. That is, the latter satisfied patient demand by locating in those areas that for profit hospitals were not willing to serve.

In principle, having access to private donations could be an advantage for private not-for-profit hospitals. But in recent years, philanthropy has become an unimportant source of revenue for private not-for-profit hospitals. In 1983, such hospitals derived only 0.4 percent of revenue from this source. For state and local government hospitals, the corresponding percentage was 1.0 (Institute of Medicine, 1986, p. 100). In earlier years, this was a much more important revenue source of revenue (Sloan et al. 1990).

One reason for the decline in private giving to hospitals is the growth of health insurance coverage. However, as Sloan and colleagues (1990) demonstrated, the direction of effect of increased insurance coverage on donations to hospitals cannot be deduced. On the one hand, increased coverage crowds out donations by providing an alternative revenue source. However, insurance also raises patient willingness to pay for hospital output, and high output is plausibly valued by donors. Thus, increased coverage allows a dollar’s worth of donations to buy more hospital output. In effect, the insured patient provides matching funds to the donor’s gift. Lump-sum subsidies by governments should crowd out donations, but matching grants may increase them. Using both time series and cross section analysis, Sloan and co-authors found strong evidence that increased health insurance coverage crowded out private donations to hospitals.

Given that the only source of private external equity to private not-for-profit organizations such as hospitals is philanthropy, it is difficult to know what the cost of such capital is. Several authors have argued that such cost equals the return on equally risky tax-free-securities (Conrad, 1984, 1986: Silvers and Kauer, 1986; Herzlinger and Krasker, 1987). But determining "equally risky" is not a straightforward task. One approach used by Sloan and colleagues (1988) was to take the beta (from the capital asset pricing model) for publicly-traded for-profit hospital companies, compute the unlevered beta, and then compute a levered beta for private not-for-profit hospitals, using their debt-to-equity ratio, and finally adjust for the fact that not-for-profits do not pay taxes. They found that the cost of equity capital was 1.1 to 2.5 percentage points lower for not-for-profit than for for-profit hospitals. The weighted cost of capital (cost of debt and equity capital weighted in proportion to their shares of total capital) was from 0.5 to 1.0 percentage points lower. This is not a very large competitive advantage for not-for-profits. Morrisey et al. (1996), using a sample of California hospitals, calculated that interest rate subsidies per not-for-profit hospital in that state amounts to about one fourth of the total tax subsidy such hospitals received. Corporate income and property tax subsidies accounted for the rest.

There is very limited published empirical research on hospital investment (see Sloan et al., 1989; Calem and Rizzo, 1995), and this research did not explicitly compare investment behavior by hospital ownership type. An unpublished study by Hoerger (1995) investigated hospital investment decisions during the 1980s in California, Florida, and Tennessee. The most noteworthy finding on ownership was that cash flow, the sum of accumulated hospital accounting profits and accumulated depreciation, had a much greater positive effect on investment by not-for-profits than for for-profit hospitals. This result is plausible because the former cannot issue stock.

Bottom Line. Given the vast amount of evidence on hospital performance, one can find some contradictory results on about any of the dimensions. Overall, one is struck by the similarity between private not-for-profit and for-profit performance, except in areas, such as capital structure where there must be differences for institutional reasons. In reviewing the literature, I have emphasized comparison between the types of private hospitals rather than between public and private hospitals. Public hospitals clearly have a different orientation, for example, in treating much higher proportions of patients without insurance. Some of the results on quality differences between public and private hospitals—those showing the public facilities to be worse—are troubling to the extent that one expects single-tier care. Quality, even of the sort that only experts observe, does not appear to be a sound reason for public provision of hospital care in the United States. This result, as many of the others, such as on uncompensated care, are likely to be country-specific.

Sloan and co-authors (forthcoming) analyzed effects of hospital conversions on various outcomes. They found that private not-for-profit hospitals that converted to for-profit status experienced an increase in their profit rate, but so did hospitals that converted in the opposite direction. Hospitals that converted from not-for-profit to for-profit status increased the uncompensated care they provided, but this also occurred for some other types of ownership conversions. The pattern of adoption of specific services was mixed, even for services that have important public health benefits. The empirical findings were based on a sample of hospitals in three states. Further empirical analysis is needed to determine whether these results generalize and to better document the stylzed facts about changes in hospital behavior following conversions.

An important issue in conversions is what the transactions price should be. There is widespread concern that sellers of hospitals obtain too little for their assets. An evaluation of rates of return on hospital purchases by Sloan and co-authors (forthcoming) found that private rates of return to acquirers were mostly at competitive levels, especially if the acquirer was a for-profit firm. However, whatever the private value is, the social value may be different to the extent that some services are provided at a price below marginal cost. Goddeeris and Weisbrod (1998) noted that to approximate social value for the for-profit sector in a conversion from public or private not-for-profit, one should add private bids, the present value of future tax liabilities, and adjust for the social value of obligatory community services. As they indicated, estimating the marginal social value of unprofitable social services is far from a trivial task.

Competition in the hospital sector is complex for several reasons. First, at least for some types of patients, price is set by a government agency. Second, patient out-of-pocket payments for hospital care are minimal. Third, in recent years, public and private insurance plans have increasingly selected hospitals on behalf of their insured (see e.g.. Robinson and Phibbs, 1989; Vistnes, 1994). To the extent that price is set administratively by an outside governmental authority and/or the cost to the patient of selecting a more expensive hospital is minimal, competition among hospitals must take place on a nonprice basis. Bilateral negotiations between insurers and hospitals are likely to involve both price and nonprice.

Pope (1989) analyzed nonprice competition among hospitals when the government sets a fixed price per case and patients pay nothing for hospital care out-of-pocket. In his model, hospitals value profits and slack. Profit is a proxy for objectives funded with excess of revenues over costs. Hospitals compete for patients by offering higher quality, which given the fixed price, reduces profit. There is a critical difference depending on whether hospital profits are positive or zero. When profits are positive, increased competition among hospitals (more competitors in the market area) raises expenditures on quality, thereby raising costs, but does not affect efficiency (i.e., slack). When profits are driven to zero, increased competition continues to raise quality expenditures , but reduces slack while not affecting costs. To raise quality, hospitals are forced to become more efficient. To encourage additional quality, the government can offer a mixed reimbursement system in which a combination of a fixed price and a payment based on hospital cost is paid (also see Ellis and McGuire, 1986). Although paying partly on a cost basis may improve quality, it also subsidizes (increases) slack. Substituting any other nonpecuniary objective for slack would yield the same conclusions. This implies that increased competition should force private not-for-profit hospitals to be increasingly similar to their for-profit counterparts.

In fact, the empirical evidence reviewed above suggests that this is so. Norton and Staiger (1994) found no difference between private not-for-profit and for-profit hospitals in provision of uncompensated care, once they controlled for hospital location. A study by Banks and co-authors (1997) obtained empirical support for the view that for-profit hospitals provide some uncompensated care as a cost of doing business. Most directly to the point, is Gruber’ s (1994) finding of higher rates of decrease in uncompensated care in the face of increased competition in California. The stability of the shares by ownership--private not-for-profit hospitals were 60 percent of nonfederal short-term general and other specialty hospitals in 1965 and in 1965 attests to the flexibility of the private not-for-profit sector. Of course, an alternative interpretation is that for-profit hospitals, which were 15 percent in 1994 and 14 percent in 1994, were no better in satisfying consumer wants (Gray and Schlesinger, 1997, p. 16).

The hospital industry provides an interesting laboratory for studying the impact of ownership form on performance. Overall, the evidence suggests that for-profit and private not-for-profit hospitals are far more alike than different. If private not-for-profit hospitals are to distinguish themselves in terms of some nonpecuniary objective, they will have to define specifically what that focus is. In an environment of increased competition, to finance this objective, they will have to secure sources of funding other than patient dollars. On the other hand, given the transformation of health care in the United States toward price and quality competition that has occurred during the 1980s and 1990s, one might have anticipated much more relative growth of the for-profit sector than has actually occurred. Thus the evidence also calls into question the "knee-jerk" reaction that for-profit automatically means greater efficiency.

One of the questions we posed at the outset was whether or not other ownership forms than the for-profit form are more efficient ones for the hospital sector. The answer depends on much more than technical efficiency and allocative efficiency in choice or inputs. Viewed in such narrow terms, the for-profit form performs about as well as private not-for-profits. The answer also depends on whether or not ownership has a bearing on whether or not hospitals with particular ownership forms produce the socially optimal combination of outputs. On this score, my review also suggests not much difference and, if anything, under competition, differences may be expected to narrow. Private not-for-profit hospitals will have less latitude than previously to produce outputs they deem to be socially worthy. In the United States, private donations are too small a part of the entire hospital revenue stream to make individual donors efficient monitors.

My review has intentionally not focused on hospitals with major missions of medical education and research. Such hospitals produce public goods that are not likely to be produced by a for-profit institution, unless, of course, dedicated subsidies exist for these activities. Also, I did not explicitly consider public hospitals, especially those serving disadvantaged populations in large cities. If such hospitals were to go, it would be necessary to invent new organizations to serve these roles, especially in the United States, where large numbers of persons lack health insurance.

This chapter’s overall conclusion about the effect of ownership is supported by empirical comparisons from other sectors that have questioned the supposition that the profit motive leads to greater efficiency. In fact, within an industry, firms tend to be more alike than they are different or at least the evidence is mixed (Borcherding et al. 1982; Boardman and Vining 1989; Peters 1993; and Vining and Boardman 1992). For example, based on their literature review, Vickers and Yarrow (1988) concluded that competition in the firm’s product market may be a more important determinant of performance than is ownership in itself.

Although much is known about effects of ownership on hospital behavior, there is much to be learned. These are important priorities.

First, one response to this chapter’s results is that it failed to consider some valuable outputs that, if considered, would reverse the findings. Such responses are common among noneconomists in health policy discussions, but among some very knowledgeable economists as well. One issue that comes up is local control as a benefit of not-for-profit ownership. If the hospital were owned by a stock company, local control would be lost. Perhaps local control leads to a better perception of community preferences. But more broadly, the question about local control is who the potential beneficiaries are. Possible candidates are the managers and employees; less likely beneficiaries are patients. More generally, the challenge is identify these heretofore unmeasured outputs and evaluate the effects of ownership on their provision. Just believing that the outputs are out there somewhere is not terribly compelling. Perhaps it would be useful to start with an informal survey of practitioners in the health field. I leave the particulars to you, the reader.

Second, empirical evidence on the performance of hospitals in countries other than in the United States is badly needed. There is considerable heterogeneity among countries in how care is organized and financed. In some, for example, patient choice of hospital is very limited. Complete lack of competition that exists in some countries where patients do not have a choice of hospital should plausibly affect performance. The results reviewed above for the United States probably do not generalize. Nevertheless, the studies do provide a useful point of departure for scholars of other health systems.

Third, research should not be confined to differences in performance. Although there has been some interesting new research on the topic, there still is no really definitive answer to the question as to why, in countries throughout the world, the for-profit ownership hospital is in the minority. To understand ownership composition of hospital sectors, it may be necessary to think about other sectors as well.

Fourth, if there is one dimension of performance that is likely to be persuasive, in comparing hospitals, it is likely to be quality. As often noted, quality of health care is difficult for consumers to monitor. But it is even difficult for the experts to measure, although much progress has been made in this field in recent years. To the extent that ownership indeed affects quality of care, this is critical to know. The better studies of quality use several alternative measures and are able to track outcomes for some time after admission. To isolate effects of hospital characteristics on quality, it is necessary to hold other factors, such as casemix, constant.

Fifth, empirical research to date has varied in sophistication of the econometrics employed. There is probably no pressing need for another case-control study of hospital ownership. Yet even some of the better studies may be plagued by endogeneity and omitted heterogeneity. Patient choice of hospital or hospital choice of organizational form are not exogenous to the behavior that has been studied. Data from randomized controlled trials will never be available for this topic. Therefore, it will be necessary for economists to deal with endogeneity in some other way. Precise ideas are beyond the scope of this chapter. Readers are encouraged to read the chapter in this volume on econometric applications in health economics.

Finally, there are important links between ownership and the constraints under which hospitals operate. As stressed above, increased competition may be expected to seriously limit the latitude hospitals have to engage in nonprofit maximizing behaviors. In this sense, hospital ownership as a topic in the United States may be a declining industry. On the other hand, hospitals are integrating vertically in the United States, largely because of market changes. In many cases, the partners may be of a different organizational form than the hospital. This opens up the possibility of all kinds of blended combinations, for example, a for-profit health maintenance organization linked to a private not-for-profit hospital which may produce a variety of outcomes. Studies of effects of health system ownership on various dimensions of performance should be conducted.