Chapter 4: Competition Law: Hospitals

Analyses of the likely competitive effects of hospital mergers have been an important part of antitrust enforcement since the FTC issued its first hospital merger complaint in 1981.¹Most hospital mergers and acquisitions do not present competitive concerns.² The Department of Justice and Federal Trade Commission Statements of Antitrust Enforcement Policy in Health Care (Health Care Statements) specifically set forth a safety zone for hospital mergers that will be rarely (if ever) challenged by the Agencies.³ Indeed, since 1981, the Commission and DOJ have challenged relatively few hospital mergers, in some instances seeking relief only for part of the transaction.⁴ The Agencies have used consent orders to resolve competitive concerns about several of these mergers.⁵

Nonetheless, the Agencies have found some hospital mergers likely to have anticompetitive effects and had considerable early success in litigating hospital merger cases.⁶ From 1994 through 2000, however, when there were approximately 900 hospital mergers, the Agencies and state antitrust enforcers lost all seven cases they litigated.⁷ Some scholars have strongly criticized the courts' reasoning in these cases.⁸

The Agencies analyze hospital mergers using the same analytical framework they use for other mergers. The 1992 Horizontal Merger Guidelines (Merger Guidelines) specify that "mergers should not be permitted to create or enhance market power or to facilitate its exercise."⁹ Market power "is the ability profitably to maintain prices above competitive levels for a significant period of time."¹⁰ A merger also may "lessen competition on dimensions other than price, such as product quality, service, or innovation."¹¹

To identify mergers that are likely to cause competitive problems, the Merger Guidelines provide for the examination of several issues, including:

whether the merger, in light of market concentration and other factors that characterize the market, would be likely to have adverse competitive effects;
whether entry would be timely, likely, and sufficient either to deter or to counteract the competitive effects of concern;
whether there are efficiency gains from the merger that meet the Agencies' criteria for examination; and
whether, but for the merger, either party to the transaction would be likely to fail, causing its assets to exit the market.¹²

Merger analysis can begin with an assessment of direct evidence of likely anticompetitive effects.¹³ The Supreme Court has stated that "the finding of actual, sustained adverse effects on competition ... is legally sufficient to support a finding that the challenged restraint was unreasonable even in the absence of elaborate market analysis."¹⁴ A number of lower court decisions have followed this principle.¹⁵

Merger analysis also can begin with the identification of relevant product and geographic markets. A market is defined as a product(s) and a geographic area in which it is produced or sold, such that a hypothetical profit-maximizing firm that was the only present and future producer or seller of those products in that area likely would impose at least a "small but significant and non-transitory" increase in price.¹⁶ This market definition test is sometimes referred to as the "hypothetical monopolist" paradigm. A relevant market is a group of products and a geographic area that is no bigger than necessary to satisfy this test.¹⁷ Analysis typically starts with a narrow area that is broadened until a price increase by the hypothetical firm would be profitable because consumers have insufficient substitution alternatives available to defeat it.¹⁸

Hospital merger analysis raises a number of significant issues, including how best to define the geographic and product markets, assess the prospects for entry and the likelihood and magnitude of efficiencies, and determine the relevance of a hospital's institutional status (for-profit or nonprofit). This chapter considers each of these issues, and discusses relevant case law, academic commentary and research, and testimony and written presentations from the Hearings.

Chapter 4 also addresses the role of group purchasing organizations (GPOs) for health care systems, including the extent to which GPOs act as agents of their buyer-members or as agents of the sellers that pay the GPOs' administrative fees. This section also discusses the antitrust issues GPOs may raise and the applicability of the Health Care Statements to those issues. Chapter 4 concludes with a brief discussion of the antitrust implications of tiering and pay-for-performance.¹⁹

II. GEOGRAPHIC MARKET DEFINITION

The Agencies define hospital geographic markets using the process set forth in the Merger Guidelines. Panelists agreed that the Merger Guidelines provide an appropriate framework for defining and analyzing hospital geographic markets.²⁰ Although there is widespread agreement on the basic framework, two well-known health law scholars have written:

[T]he law concerning hospital [geographic] market definition is in a shambles. Common sense suggests that health care, like politics, is local. In the words of Judge Richard Posner, "People want to be hospitalized near their families and homes, in hospitals in which their own - local - doctors have privileges." However, courts have stretched the geographic boundaries of markets to strip merging hospitals of market power and thereby shield them from antitrust liability.²¹

In this section, we discuss the controversies about how to define relevant geographic markets for hospitals. We discuss the use of the Elzinga-Hogarty test and critical loss analysis to define hospital geographic markets, including the views of proponents and critics. We then describe alternative analytical techniques and evidentiary sources that Hearing panelists and researchers proposed for defining hospital geographic markets. The final subsection summarizes the Agencies' conclusions and recommendations concerning geographic market definition issues. This subsection includes the conclusion that, to date, the Agencies' experience and research indicate that the Elzinga-Hogarty test is not valid or reliable in defining geographic markets in hospital merger cases.

At the outset, we note that direct evidence of anticompetitive effects may make it unnecessary to define a relevant market. For example, consummated merger cases may present opportunities to assess competitive effects without using detailed market definitions.²²

A. Elzinga-Hogarty, Critical Loss, and the Alternatives

Since 1995, the Agencies have lost several hospital merger cases because the courts accepted the merging parties' use of patient flow data to perform either the Elzinga-Hogarty test²³ or critical loss analysis²⁴ to define the geographic market much more broadly than the plaintiff Agency.²⁵ Commentators and panelists observed that these cases reflect judicial acceptance of implausibly large geographic markets, judicial approval of mergers that would not be permitted in any other industry, and the lessening of competition in the hospital services market.²⁶

All panelists agreed that neither the parties nor the courts should use the Elzinga-Hogarty test as the sole basis for defining the geographic market.²⁷ As one panelist stated: "if [Elzinga-Hogarty] is the only tool that is being used ... it blurs everyone's vision as to who really are the competitors and the alternatives that matter."²⁸ Panelists and commentators identified numerous problems with the application of critical loss analysis, although panelists and commentators agreed that it can be a useful tool.²⁹

Several panelists offered alternative analytical tools and other types of evidence to use in defining the geographic market for a hospital. Most panelists agreed that no one piece of information is sufficient to define a hospital's geographic market.³⁰ In essence, panelists agreed the courts should apply the Merger Guidelines' hypothetical monopolist test in hospital merger cases, just as they do in merger cases involving other industries and products. The question is how to implement the hypothetical monopolist test, and what analytical frameworks and evidence should be used to do so.

1. Elzinga-Hogarty Test

The Elzinga-Hogarty test was designed to analyze commodity movements, not hospital mergers. It was proposed by two economists in an article critiquing the Agencies' geographic market definitions in two non-hospital merger cases.³¹ In one case, the government relied on LIFO ("little in from outside") data to argue that an entire state was the relevant geographic market for beer products. In the second case, the government relied on LOFI ("little out from inside") data to argue that the relevant geographic market for commercial banking was limited to a four-county area.³² Kenneth Elzinga and Thomas Hogarty argued that a proper geographic market analysis required the use of both LIFO and LOFI statistics, but observed that their analysis was not readily applicable to heterogenous goods or differentiated products.³³ Hospitals generally provide heterogenous or differentiated goods and services.³⁴

Nonetheless, the "Elzinga-Hogarty test" has been used extensively in hospital merger cases. The movement of a patient who resides within the provisional geographic market to a facility outside of that area for hospital services is considered an importation of hospital services into that provisional geographic market, measured as LIFO. The movement of a patient who resides outside of the provisional geographic market to a facility inside the provisional geographic market for hospital services is considered an exporting of hospital services outside of the provisional geographical market, measured as LOFI.³⁵ Thus, under the hospital application of the Elzinga-Hogarty test, evidence that few patients leave and few patients enter an area surrounding the merging hospitals is interpreted to support the conclusion that the area constitutes a relevant geographic market.³⁶

Conversely, if the patient flow data show large numbers of patients coming into or going out of the area for inpatient hospital care, then the geographic market is hypothesized to be broader than originally thought, and must include hospitals further away from the merging hospitals. A geographic market definition is usually described as "strong" if less than 10 percent of discharged patients from the merging hospitals' area come into or out of the area. If more than 10 percent (but less than 25 percent) of patients migrate in or out of the hospitals' core geographic area for in-patient services, the market definition is considered "weak."³⁷

Panelists identified a number of weaknesses with the use of the Elzinga-Hogarty test to define a geographic market for hospital services.³⁸ One panelist pointed out that the Elzinga-Hogarty test takes a leap in logic from a current level of patient migration to the conclusion that patients would respond to a small price increase by using hospitals outside of the merging hospitals' core geographic area - a leap not justified by either economic analysis or past experience.³⁹ Patients decide whether or not to travel for health care services for a variety of reasons, including perceived and actual variations in quality, insurance coverage, out-of-pocket cost, sophistication of services, and family connections.⁴⁰

Although patient flow data may show that patients go to hospitals beyond the core zip code area, this does not mean that their behavior reflects price sensitivity, or that other consumers would travel if prices increased.⁴¹ Stated differently, patient flow data can show existing hospitalization patterns, but offer no insight into what patients will do in response to a price increase by the merged hospital.

Another panelist described this phenomenon as the "silent majority fallacy."

The E-H [Elzinga-Hogarty] approach draws a conclusion about the entire market from the behavior of those consumers who express displeasure with their local sellers by traveling elsewhere. This is a valid logical leap when travelers and non-travelers have similar demands and related market experiences. However, if the two groups differ on dimensions other than location, then E-H gives rise to what we call the "silent majority fallacy." That is, if travelers and non-travelers display fundamentally different demand behavior, either because they differ in their taste for travel or their need for local/nonlocal services, then there is no necessary relationship between the market experiences of these two groups post-merger. If travelers differ significantly from non-travelers, then the presence of a minority of travelers does not imply that local firms lack market power vis-a-vis the majority of consumers who are non-travelers.⁴²

The silent majority fallacy is a particular problem with hospital merger analysis, because the goods and services are not fungible commodities, but are "highly differentiated by location and other dimensions."⁴³ Empirical evidence confirms that "the majority of patients are truly reluctant to travel and do not view distant hospitals as close substitutes for most services, even though a sizable percentage of their neighbors may travel for care. Those who do travel have distinct reasons for doing so and the fact that they travel would not inhibit merging local hospitals from increasing prices substantially."⁴⁴

One panelist also noted that in some circumstances, the Elzinga-Hogarty test cannot be satisfied. If the initial specification of the geographic market does not meet the required threshold for LIFO and LOFI, expanding the geographic market may not satisfy the required threshold either. The result is that the geographic market expands without limit.⁴⁵ This problem alone casts serious doubt on the utility of the Elzinga-Hogarty methodology for hospitals.

This same panelist suggested that the Elzinga-Hogarty test systematically leads to expansive geographic markets when zip codes are selected based on the absolute number of patients that come from a zip code.⁴⁶ There is tremendous variability in the number of individuals that live in a particular zip code. A hospital may have a small share of total admissions from a particular zip code, even though it gets a significant number of patient admissions from that zip code - and the Elzinga-Hogarty test, as used in hospital mergers, will include such distant zip codes in the market. According to this panelist, "a zip code that has 20,000 people, that's 40 miles away, might get included if the hospital gets 50 patients from there, whereas ten zip codes that are closer that only have a thousand people each, might send 40 people each, they would get excluded."⁴⁷ He suggested that such large and distant zip codes are particularly likely to be cities that have hospitals in them, which skews the results of the analysis from the outset.⁴⁸

2. Critical Loss Analysis

Critical loss analysis has the potential to provide a useful way to implement the hypothetical monopolist test, but it must be applied with great care.⁴⁹ Problems with its application have led some commentators to question the value of critical loss analysis as an antitrust tool.⁵⁰

Conventional critical loss analysis posits a particular price increase and asks what proportion of the hypothetical monopolist's sales would have to be lost to yield a net decrease in the hypothetical monopolist's profits.⁵¹ When critical loss analysis is used to delineate a relevant market, the first step is to calculate the percentage loss in sales that would make a given price increase unprofitable for a hypothetical monopolist over a candidate market. This calculation depends on the price increase posited and on the contribution margin (i.e., price minus marginal cost, all divided by price) on the sales that would be lost.⁵²

The second step is to estimate the likely actual loss in sales that would result from the hypothesized price increase, e.g., what percentage of patients likely would stop patronizing the hospitals in the candidate market in response to the price increase.⁵³ The estimated actual loss is then compared to the calculated critical loss. If the estimated actual loss exceeds the critical loss, it is inferred that the price increase would be unprofitable and the candidate market is too small to be a market.⁵⁴

One panelist described misuses of the critical loss technique that practitioners should avoid.⁵⁵ Notably, typical applications posit small (e.g., five percent) price increases. Yet, the Merger Guidelines' methodology for delineation of relevant markets asks whether the profit-maximizing price increase would be at least a small but significant amount (e.g., five percent). Even though a monopolist may find a five percent price increase unprofitable, it may find a larger price increase profitable.⁵⁶ This panelist presented an example based on the stylized facts of several hospital merger cases in which a five percent price increase would be unprofitable, but any price increase between 31 percent and 319 percent would be profitable, and the hypothetical monopolist would maximize its profits by increasing price 175 percent.⁵⁷ Thus, the candidate market was a market under the Merger Guidelines' hypothetical monopolist test, even though a five percent price increase was unprofitable.

This panelist discussed other problems that occur in some implementations of critical loss analysis. The standard formula presumes constant marginal cost and no avoidable fixed costs, but actual cost functions may differ significantly from this assumption. Also, the standard formula implicitly assumes proportionate increases in all prices, but the profit maximizing strategy for hospitals may involve highly disproportionate price increases.⁵⁸ This panelist also explained that critical loss calculations must focus on the margins for the patients that likely would be lost in the event of a price increase.⁵⁹

Much of the potential for abuse in critical loss analysis involves the second step -estimation of the actual loss. Some practitioners have relied in inappropriate ways on consumer surveys or patient flow data to estimate the actual losses in sales that would result from a price increase. For example, some practitioners use patient flow data to identify zip codes that are "contestable."

These practitioners then argue that the share of patients in these zip codes that would stop patronizing certain hospitals in a candidate geographic market in response to a given price increase would be greater than the critical loss, and that the geographic area must therefore be expanded in order to constitute a relevant geographic market.⁶⁰ Data on existing travel patterns for residents in a zip code, however, say nothing about why patients select specific hospitals or how a change in relative prices would affect patient migration.⁶¹ One cannot infer that just because some patients in a zip code currently choose more distant hospitals, others also would choose such hospitals if the prices of the merging hospitals increased.⁶²

Recent commentary, some of it published after the Hearings, has stressed a link between the first and second steps of critical loss analysis.⁶³ As a simple matter of arithmetic, the higher the contribution margin, the smaller the critical loss will be for a given price increase. The higher the margin, the more it costs the hypothetical monopolist to lose a sale, and so the smaller the sales loss required to offset the profit gain from making the remaining sales at a higher price.⁶⁴

Yet if firms are maximizing profits before the merger, high margins indicate that those firms face low price elasticities of demand.⁶⁵ Otherwise, these firms could earn greater total profits by reducing prices and expanding sales. Moreover, a hypothetical monopolist over any candidate market must face a lower elasticity of demand than the individual firms in that candidate market, so high margins must imply a very low demand elasticity for the candidate market.

Hospitals' experts commonly argue that merging hospitals' margins are high, which implies that the critical losses are low. They argue that post-merger price increases would be unprofitable because of the high per-unit foregone profits on lost sales. In essence, they argue that where the critical loss is low, the actual loss will exceed the critical loss. On this basis, they argue that relevant geographic markets for hospital mergers are broad.⁶⁶

Yet, as discussed above, high margins also imply low demand elasticities. Low demand elasticities indicate that the merged firm's actual losses of sales would be low. Because the actual losses may be less than the critical losses when margins are high, the relevant geographic market may in fact be narrow.⁶⁷

Other commentators also have described ways in which critical loss analysis has been carried out incorrectly, both for delineation of markets and for competitive effects analysis.⁶⁸ One article cites four key problems in how courts have applied critical loss analysis in recent hospital mergers.⁶⁹

First, courts have failed to consider whether a price increase greater than five percent would be profitable. Second, courts have failed to consider the fact that high margins often mean a firm faces inelastic demand and, therefore, actual losses would be low. Third, courts did not consider that, if prices increased, some consumers might be diverted to one of the merged firms. Finally, courts have assumed, contrary to economic theory, that firms in the area surrounding the merged firm would keep the same prices even though the merged firm raised its prices.⁷⁰ Thus, critical loss analysis may be useful in defining geographic markets and for competitive effects analysis only if it is applied appropriately.

3. Alternative Analytical Techniques

One panelist proposed an alternative analytical framework built on the observation that hospitals compete in two stages.⁷¹ According to this panelist, the Agencies typically focus on first-stage competition, in which hospitals compete to be included in the networks of health plans. At this point, health plans are the buyers, and prices may be constrained if a health plan can credibly threaten to, or actually, exclude the merging hospitals from its provider network and divert patients to alternative hospitals. The focus for defining the geographic market for this first stage of competition is on hospital locations, not patient locations.⁷² Once a hospital is in the plan's network, or in some cases even if it is not, the hospitals then compete at the second stage - for the individual patient.

This panelist suggested that defendants typically focus on second-stage competition for patients and argue for broader geographic markets based on patient flow data.⁷³ This level of competition differs significantly from the first stage.⁷⁴ A two-stage analysis may result in different geographic markets and different competitive effects for each stage, because the two stages involve different customers, different means of competition, and different evidence.⁷⁵ If anticompetitive effects are demonstrated at either stage, the merger should be enjoined, according to this panelist.⁷⁶

Another panelist disagreed with the two-stage analysis, noting that it might be worth looking at "as a stylized construct," but that "the appropriate model in which to analyze the factors that drive the pricing decisions and the profitability decisions of the hospitals are such that one cannot separate out the two stages."⁷⁷ She suggested that the distinction is even less relevant now, because most plans have inclusive provider networks. In these circumstances, network inclusion provides no assurance that patients will seek care at a particular hospital.⁷⁸

Another panelist submitted a joint statement proposing a different analytical framework for analyzing geographic markets in hospitals. The statement asserts that because potential patients select managed care organizations (e.g., health insurers) prior to knowing what their medical needs will be, the subsequent ex-ante pricing makes the connection between patient flows and pricing power tenuous.⁷⁹ For example, the statement suggests that "100% of patients place a high value on having access to a local hospital," but if they are part of the 20 percent of the group that develop a serious medical condition, these same patients may be willing to travel any distance to go to the best hospital for their condition.⁸⁰

As an alternative, the statement proposes a formal demand analysis model that would require data on patient and hospital characteristics in addition to the patient origin and destination data traditionally used. Although this model is more complex than patient flow analysis, the statement contends it provides "a measure of market power that, unlike patient flows, is theoretically valid for differentiated goods markets and is directly related to the prices that hospitals are able to charge."⁸¹

B. Other Evidentiary Sources

Panelists suggested numerous additional sources of evidence that should be used to establish the geographic market for hospital services. The recommended sources include types of evidence typically assessed in non-hospital merger cases: strategic planning documents and testimony from the merging parties and their competitors, and documents and testimony from major purchasers of services from the merging parties - here, third-party payors.

Panelists also suggested the use of evidence that casts direct light on the distances patients are willing to travel and the reasons they are willing to do so, and evidence that demonstrates the role, if any, physicians can play in defeating a hospital's post-merger, anticompetitive price increases. Each of these categories of evidence are considered below.

1. Hospital Strategic Planning Documents

The Agencies typically examine strategic planning documents from the merging parties and their competitors to assess relevant market and other key issues in merger analysis. Panelists suggested using strategic planning documents from the hospitals to help establish the proper geographic market.⁸² Such documents may specify the geographic regions in which a hospital is marketing its services and the hospitals it sees as its primary competition.⁸³ In addition, a hospital's strategic planning documents frequently disclose the hospital management's assessment of the extent to which the proposed merger will increase the hospital's negotiating power and its ability to raise prices. Hospital strategic planning documents can illuminate hospital competition for both inclusion in payor networks and for individual patients.

2. Payor Testimony

For non-hospital mergers, the Agencies regularly obtain the views of the merging firms' major customers to assess issues such as relevant market definition and competitive effects. These market participants typically have the most price negotiation experience with the merging firms, as well as the most to lose from price increases (or quality or other degradations) if the proposed merger were to create market power. On the other hand, major customers also have much to gain from reduced prices if the proposed merger would likely create efficiencies that would be passed on to customers.

Courts, however, have been skeptical about testimony from third-party payors in hospital merger cases, even though these payors routinely negotiate with hospitals about price and other aspects of hospital care. In Tenet, for example, the Eighth Circuit questioned the district court's reliance on payor testimony that they "would unhesitatingly accept a price increase rather than steer their subscribers to hospitals" outside of the core geographic area.⁸⁴ The Eighth Circuit believed that, although the testimony might have been truthful, the payors "spoke to current competitor perceptions and consumer habits and failed to show where consumers could practicably go for inpatient hospital services."⁸⁵

By contrast, panelists stated that payors can offer useful testimony on at least two distinct issues.⁸⁶ Payors have considerable insight into hospital geographic markets, because they must factor such matters into their decision whether to contract with a hospital in the first instance. Payors must strive to include a sufficient number of hospitals in each geographic market, because if they fail to do so, the plan is less appealing to purchasers, including benefit managers that must make recommendations and decisions for employers and other group purchasers.⁸⁷Accordingly, payors can offer useful testimony on the extent to which particular hospitals engage in price and non-price competition with one another.

Second, panelists suggested that payor testimony also would be helpful in determining whether payors can steer patients to a lower-cost hospital if prices increase post-merger.⁸⁸ Several panelists noted that payors used to create marketable plans with limited provider networks and thus could exclude a hospital if its prices were not acceptable to the plan.⁸⁹ Today, many consumers demand choice and open provider networks.⁹⁰ Therefore, payors frequently rely on mechanisms other than excluding hospitals to divert marginal consumers to lower-cost alternatives.⁹¹ For example, payors are currently experimenting with tiered networks that provide differing levels of coverage and co-payments based on the facility at which care is received.⁹² Testimony regarding the feasibility and performance of such strategies would be helpful in determining the alternatives available to payors in the event of post-merger price increases.⁹³

Panelists expressed different views on whether and to what extent payors can "steer" patients and the types of evidence that can help answer this question. One panelist noted that if payors actually can steer patients to (or away) from particular institutions, the distances traveled to hospitals should have grown in parallel with the rise of managed care.⁹⁴ In fact, the panelist noted, the distances patients travel to hospitals have not changed very much since the mid-1980s, and there is little distinction between the distances traveled for HMO versus non-HMO patients.⁹⁵ Based on this evidence, the panelist maintained that courts should not assume that payors can effectively steer patients in response to price increases.⁹⁶

Another panelist suggested that patient flow data may help show whether and, if so, how payors can steer patients.⁹⁷ This panelist asserted that payors have had enough success in moving marginal consumers to lower-cost hospitals that, in most cases, they can discipline hospital price increases.⁹⁸ She also concluded that in many cases, even if payors testify accurately that they must have the merging parties in their networks, that is not necessarily sufficient to give the hospitals unilateral power over price.⁹⁹

Other panelists were more skeptical about these claims. One panelist stated that, although in theory payors have mechanisms they could use to divert patients to other hospitals, in practice these tactics are often costly and counter-productive to a health plan's marketability and profitability.¹⁰⁰ This panelist argued that it is difficult (if not impossible) to target incentives to the insured consumers who are most likely to be affected. A pay or must consider the cost of providing a lower copayment to all patients, not just the marginal patients the pay or is trying to steer.¹⁰¹ Moreover, other hospitals may have higher prices than the merged hospitals, even assuming price increases as a result of the merger.¹⁰²

3. Patients' Willingness to Travel - How Far and Why?

Several panelists suggested that courts should give more weight to empirical studies of patients' willingness to travel to receive health care. Studies indicate that most patients prefer to be hospitalized close to their homes.¹⁰³ Some patients appear willing to travel long distances for very serious or complicated procedures, but many patients prefer to receive such care in their local hospital, even if their local hospital has higher mortality rates and less experience with such procedures.¹⁰⁴ Some patients are willing to receive care in a distant city because they work or have family in that city, or because of the hospital's religious affiliation.¹⁰⁵

Several panelists noted that such migration patterns are unlikely to be price sensitive, yet the application of the Elzinga-Hogarty test and critical loss analysis would result in a large geographic market in such circumstances, if enough patients traveled for these non-price reasons.¹⁰⁶

4. Physicians' Willingness and Ability to Steer Patients to Less Expensive Alternatives

Several hearing participants suggested that payors may be able to provide financial incentives to physicians to steer patients to less expensive hospitals.¹⁰⁷ Some of the proposals included requiring physicians to agree to a financial risk-sharing contract, threatening physicians with exclusion from a plan's network, imposing financial penalties on physicians who admit patients to the higher-priced hospitals, and providing bonuses to physicians who admit to lower-priced hospitals.¹⁰⁸

Even though such incentives are theoretically possible, it does not follow that payors would find them useful or desirable.¹⁰⁹ Indeed, such incentives could make a plan less marketable to employers and consumers who value open networks and unrestricted access to health care. Such incentives also could interfere with continuity of care, particularly if patients must use a different physician when they are diverted to a different hospital.¹¹⁰ These incentives also are unlikely to be effective if they require patients to travel long distances and physicians to travel those same distances to provide care.

C. Summary

The definition of a relevant geographic market has proven to be one of the most daunting components of a hospital merger case. Nonetheless, some guiding principles are clear. The hypothetical monopolist test of the Merger Guidelines should be used to define geographic markets in hospital merger cases. The types of evidence used in all merger cases - such as strategic planning documents of the merging parties and customer testimony and documents -should also be used to delineate relevant geographic markets in hospital merger cases. The Agencies believe that courts have given insufficient weight to payor testimony and documents in particular.

Empirical evidence is desirable on certain issues, such as the extent of patients' willingness to travel to distant hospitals in response to a small, but significant and non-transitory increase in price. Patient willingness to travel for non-price related reasons does not provide a sufficient basis to infer patient willingness to travel to distant hospitals in response to price increases.

The Agencies encourage further research to determine the circumstances in which patients will travel to distant hospitals in response to price increases. Empirical evidence also is desirable on the extent to which physicians can and will steer patients to lower-cost hospitals in response to price increases. To be persuasive, direct evidence should show that such steering by physicians is feasible, cost-effective, and likely.¹¹¹ The Agencies also encourage additional research to validate or refute the alternative analytical techniques discussed supra.

To date, and for the reasons discussed supra, the Agencies' experience and research indicate that the Elzinga-Hogarty test is not valid or reliable in defining geographic markets in hospital merger cases. In addition, if critical loss analysis is used, it must be used with great care to avoid the problems of application discussed in this section. The use of the Elzinga-Hogarty test and the misapplication of critical loss analysis has led some courts to find hospital geographic markets that are improbably large.

III. PRODUCT MARKET DEFINITION

The Merger Guidelines provide the framework for defining the relevant product market for hospital services. The product market has typically been defined as a broad group of medical and surgical diagnostic and treatment services for acute medical conditions where the patient must remain in a health care facility for at least 24 hours for recovery or observation.¹¹²

Over the past twenty years, many hospital merger cases have considered and rejected outpatient services as part of the relevant product market for hospitals. For example, in In re Hospital Corp. of America, 106 F.T.C. 361 (1985), the Commission noted that, although outpatient care for certain services might be a separate relevant market or markets, the evidence demonstrated "that the core and vast majority of an acute care hospital's business is acute inpatient care" and non-hospital outpatient providers could not defeat post-merger anticompetitive behavior affecting hospital inpatients.¹¹³

The Seventh Circuit agreed, observing that "although hospitals increasingly are providing services on an out-patient basis ... most hospital services cannot be provided by nonhospital providers; as to these, hospitals have no competition from other providers of medical services."¹¹⁴ Similarly, in American Medical International, 104F.T.C. 1 (1984), the Commission excluded outpatient services from the product market.¹¹⁵ The Eleventh Circuit also accepted inpatient acute-care services as the relevant product market in University Health.¹¹⁶ Only one court has included outpatient providers within the product market for inpatient services.¹¹⁷

Panelists agreed that providers of outpatient services, such as physicians' offices, urgent care centers, and ambulatory surgery centers, should generally not be included in the product market definition for hospital services.¹¹⁸ Panelists indicated that from the perspective of payors and patients, inpatient services are complementary and bundled.¹¹⁹ Even if hospital prices are increased, patients and payors cannot separate nursing care, diagnostic tests, and room and board from the other treatments provided as part of a hospital stay and out-source them.¹²⁰ Similarly, demand-side substitution is improbable; a cancer or heart attack patient is not going to substitute obstetrical care if prices for cancer care or heart attacks increase. Because outpatient treatment is generally not a substitute for inpatient care, there was agreement among the panelists that outpatient providers are (and were) correctly excluded from the product market.¹²¹

In the future it is likely that the Agencies will have to determine whether certain specialty hospitals should be included in an inpatient product market for particular proposed hospital mergers. Historically, the type of specialty hospital (children's, psychiatric, VA, military, and rehabilitation) justified its exclusion from the product market.¹²² In recent years, specialty hospitals focusing on cardiac or orthopedic care have emerged in numerous locations.¹²³ General acute-care hospitals view these specialty hospitals as competition in the provision of such services and have responded in a variety of ways.¹²⁴

Several panelists discussed an approach for defining an inpatient hospital product market more narrowly. Instead of treating acute inpatient treatment as an aggregated group, panelists suggested the possibility of grouping diagnosis related groups (DRGs) together, based upon the types of diseases and medical conditions treated by particular types of physicians.¹²⁵ In one study, this approach resulted in 48 service categories. Patient flow data can be separately analyzed for each category.¹²⁶ Panelists recognized, however, that payors generally do not disaggregate services this finely.¹²⁷

Conclusion. The Agencies continue to believe that inpatient acute-care services constitute a relevant product market. At the same time, the percentage of total health care spending devoted to outpatient care is growing, and the percentage devoted to inpatient care is declining. Over time, the level of payment and changes in technology may shift the provision of many inpatient services into the outpatient setting.¹²⁸ The Agencies will continue to examine whether services provided in outpatient settings may constitute additional relevant product markets, and if so, whether those services might be adversely affected by a hospital merger. The Agencies will also continue to examine the competitive significance of specialty hospitals, including whether and under what circumstances payors might discipline prices for cardiac or other services at general acute care hospitals by shifting a larger percentage of patients to specialty hospitals that provide such services.

Although the Agencies currently doubt the advisability and practicability of conducting separate product market analyses for many discrete markets - particularly when payors do not define the product they are purchasing in this fashion - the Agencies will continue to examine whether smaller product markets exist in addition to the traditional product market definition. For example, if more specialized medical procedures raise more competitive concerns than primary care services, there may be some circumstances in which the product market should be defined narrowly to include only a specific service or limited number of services. Similarly, it is possible that expertise in one or more specific specialities may make a hospital a "must have" hospital for a payor's network, which could justify a separate product market analysis.¹²⁹

IV. ENTRY

The Merger Guidelines provide that entry should be considered if it is likely to occur within two years and to be sufficient to deter or counteract anticompetitive effects of a proposed hospital merger.¹³⁰ Entry into the inpatient general acute care hospital services market by constructing a new hospital or adding additional beds to an existing facility is likely to exceed this time-frame. If the state requires that a Certificate of Need (CON) be granted before building a new hospital or increasing bed capacity, the approval of the CON can take anywhere from 18 months to several years.¹³¹ Compliance with other regulations will require additional time. Thus, the likelihood of timely and sufficient entry into the inpatient general acute care hospital services market is remote.

V. EFFICIENCIES

The Merger Guidelines make clear that efficiencies should be evaluated before determining whether a proposed merger is likely to be pro- or anti-competitive.¹³² Under the Merger Guidelines, the Agencies will not challenge a merger if cognizable efficiencies are of a character and magnitude such that the merger is not likely to be anticompetitive in any relevant market.¹³³ Efficiencies are cognizable when they are (1) merger-specific, (2) have been verified, and (3) do not arise from anticompetitive reductions in output or service.¹³⁴

Hospitals often claim that their merger will produce significant efficiencies, and some courts have given significant weight to these arguments. Claimed efficiencies have included avoidance of capital expenditures, reductions in management and operational support jobs, consolidation of specific services to one location (e.g., all cardiac care at Hospital A and all cancer treatments at Hospital B), and reducing operational costs, such as purchasing and accounting.

Some hospitals claim that after the merger they will be able to provide better and more complex services to their patients. For example, in Tenet the merging hospitals claimed they would realize significant efficiencies, including: eliminating unused beds, bringing open heart surgery to Poplar Bluff, decreasing operating costs, consolidating services, reducing staff levels, and avoiding capital expenditures.¹³⁵ The district court rejected the hospitals' efficiency claims. The Eighth Circuit found that, although the district court may have properly rejected the hospitals' efficiencies, it should have nonetheless considered the claim that the merged entity would provide better care to its patients. The appellate court stated that "[t]he reality of the situation in our changing healthcare environment may be that Poplar Bluff cannot support two high-quality hospitals;" and admonished the district court for placing "an inordinate emphasis on price competition."¹³⁶

Some panelists were skeptical about efficiency claims. Several panelists pointed out that promised efficiencies may not materialize.¹³⁷ One panelist noted that efficiency studies are often conducted to support the HSR filing that the merging parties must make with the Agencies; this provides incentives for the parties to estimate unrealistically high savings.¹³⁸ Another noted that mergers can be great failures if hospitals do not have specific plans or are not willing to make tough decisions at the outset, such as closing facilities and consolidating hospital-based physician groups.¹³⁹ Institutional constraints can make it difficult for merged hospitals to combine and coordinate clinical operations.¹⁴⁰

For example, in Butterworth, the district court accepted the merging hospitals' claims that the proposed merger would result in efficiencies in excess of $100 million in the form of capital expenditure avoidance and operating efficiencies.¹⁴¹ One panelist reported, however, that Blodgett/Butterworth never closed Blodgett and consolidated services, at least in part because physician groups did not want the facility closed.¹⁴² Another panelist stated that, six years after the merger, Blodgett/Butterworth had realized less than half of the $100 million of claimed efficiencies.¹⁴³

Scholars have conducted numerous studies on the effect of hospital mergers on hospital costs.¹⁴⁴ The results are mixed: some studies have found that merged hospitals enjoy lower costs (or lower rates of cost increase) than nonmerging hospitals; others have found no differences in cost experience between merging hospitals and otherwise similar nonmerging facilities. One recent study found that the degree of cost savings that merging hospitals realize varies significantly depending on the extent of consolidation. According to this study, hospitals operating under a single license post-merger generate "significant, robust, and persistent" savings.¹⁴⁵ In contrast, those hospitals that conduct business under separate licences post-merger do not generate cost reductions. The authors attribute this difference to the ability of more fully merged hospitals to undertake substantial changes in they way they operate (including consolidation of services) that are not available to hospitals operating under separate licenses.¹⁴⁶

Even if a hospital merger is likely to create cognizable efficiencies, those cognizable efficiencies likely will not be sufficient to reverse a hospital merger's potential to harm consumers in the relevant market by preventing price increases in that market.¹⁴⁷

As discussed in detail in Chapter 3, supra, most studies of the relationship between competition and hospital prices generally find that increased hospital concentration is associated with increased prices.¹⁴⁸ Some panelists and commentators believe an important motivation for the creation of multi-hospital systems has been to gain market power to secure higher reimbursement from payors.¹⁴⁹ Indeed, one academic health economist reported that "I have asked many providers why they wanted to merge. Although publicly they all invoked the synergies mantra, virtually everyone stated privately that the main reason for merging was to avoid competition and/or obtain market power."¹⁵⁰

In several merger cases, hospitals have signed "community commitments" or agreements with State Attorneys General, promising not to raise prices for a specified period or to pass onto consumers a specified amount of money from the claimed efficiencies.¹⁵¹ Some State Attorneys General have signed these agreements in an attempt to translate merger-induced cost savings into price reductions to consumers. For example, in Butterworth/Blodgett, the merging hospitals agreed: (1) to freeze list prices for three years, (2) to freeze prices for managed care plans at pre-merger levels, (3) to limit profit margins by targeting a five-year rolling average for the merged entity that would not exceed the average of Moody's and Standard & Poor's upper quartile profit margin for other national health care providers, (4) to serve the medically needy, and (5) to ensure that the board of the merged entity would continue to reflect the interests of western Michigan.¹⁵² Similarly, the merging hospitals in Long Island Jewish Medical Center entered into an agreement with the Attorney General of the State of New York to "pass on to the community cost savings that will be achieved . . . [to] equal 100 million dollars during the five-year period commencing January 1, 1998."¹⁵³ The agreement further provided that up to 50 million dollars of the cost savings could be used "to fulfill its mission to provide high quality health care to economically disadvantaged and elderly members of the community."¹⁵⁴

Community commitments are temporary and may not represent a binding constraint even during the period they are in effect. Furthermore, such commitments do not solve the underlying competitive problem when a hospital merger has changed market circumstances in ways that increase the likelihood that market power will be exercised. Community commitments represent a distinctly regulatory approach to what is, at bottom, a problem of competition - and that problem will remain after the commitment has expired.¹⁵⁵

The Agencies do not accept community commitments as a resolution to likely anticompetitive effects from a hospital (or any other) merger. The Agencies believe community commitments are an ineffective short-term regulatory approach to what is ultimately a problem of competition. Nevertheless, the Agencies realize that in some circumstances, State Attorneys General may agree to community commitments in light of the resource and other constraints they face.

VI. NONPROFIT STATUS OF HOSPITALS

The significance of institutional form (nonprofit v. for-profit) has been an issue in several hospital merger cases. In three early cases, the Seventh and Eleventh Circuit Courts of Appeals rejected the claim that institutional form should figure in a merger analysis. Thus, in HCA, the Seventh Circuit noted that although "different ownership structures might reduce the likelihood of collusion, ... this possibility is conjectural," and that "adoption of the nonprofit form does not change human nature."¹⁵⁶ Similarly, in University Health, the Eleventh Circuit observed that "the Supreme Court has rejected the notion that nonprofit corporations act under such a different set of incentives than for-profit corporations that they are entitled to an implicit exemption from the antitrust laws."¹⁵⁷ Finally, in Rockford, the Seventh Circuit repeated and elaborated its position that institutional form was irrelevant to a merger analysis:

We are aware of no evidence - and the [appellees] present none, only argument - that nonprofit suppliers of goods or services are more likely to compete vigorously than profit-making suppliers .... If the managers of nonprofit enterprises are less likely to strain after that last penny of profit, they may be less prone to engage in profit-maximizing collusion but by the same token less prone to engage in profit-maximizing competition.¹⁵⁸

The relevant question for antitrust analysis is not whether nonprofit hospitals behave in a manner indistinguishable from for-profit institutions, but rather whether they would exploit merger-created market power in ways harmful to consumers.¹⁵⁹ Recently, some courts have asserted that institutional form should matter - and suggested that nonprofit hospitals, even if they acquire market power, will not harm competition or consumers. For example, in Butterworth, the district court relied on the nonprofit status of the merging hospitals as a reason why the merger would not have anticompetitive effects, and the Sixth Circuit emphasized this fact in its opinion affirming the district court.¹⁶⁰

Similarly, in Long Island Jewish Medical Center, the court believed that the merging hospitals were nonprofit organizations that "have a genuine commitment to help their communities," and "community service, not profit maximization, is the hospitals' mission."¹⁶¹

The practical significance of a hospital's institutional form has been studied extensively. One panelist (who was an expert for the defendant in the Butterworth/Blodgett case) stated that economic incentives made it likely that a typical nonprofit hospital's pricing behavior would differ systematically from that of a typical for-profit hospital.¹⁶²

This panelist argued that a number of studies, including work he had performed, indicated that nonprofits that attain market power behave differently from for-profits when it comes to pricing.¹⁶³ This panelist qualified this observation, noting that the observed price effects in these studies are averages and do not predict whether or not a particular nonprofit hospital merger will have an effect on price and do not preclude the possibility of price discrimination against certain customers.¹⁶⁴ Moreover, this panelist acknowledged that the empirical evidence of a price effect is mixed.¹⁶⁵

By contrast, several panelists maintained that the best available empirical evidence indicated no significant differences between the pricing behavior of for-profit and nonprofit hospitals.¹⁶⁶ For example, one panelist stated that "the preponderance of the empirical evidence indicates that nonprofit hospitals use their market power in roughly the same fashion as for-profit hospitals."¹⁶⁷ Another panelist similarly reported that the "literature suggests that, on average, nonprofit hospitals do use market power to obtain higher prices."¹⁶⁸

Recent empirical studies of pricing behavior paint a fairly consistent picture. One study found that there was no significant difference in how for-profit and nonprofit hospitals exerted market power; for-profit hospitals generally had higher prices in 1986, but nonprofits increased their prices faster from 1986 to 1994.¹⁶⁹ A case study of a nonprofit hospital merger in Santa Cruz, California, found significant evidence of post-merger price increases.¹⁷⁰ Another study noted that "the most interesting result for antitrust policy is the finding that nonprofit hospital mergers lead to higher prices, not lower ones, and that the price increases resulting from a nonprofit merger are getting larger over time."¹⁷¹

Merger simulation studies have produced a similar picture. One study found nonprofit status did not lead to lower prices in urban markets, but did result in modestly lower prices in rural markets.¹⁷² Other studies found no differences in pricing behavior resulting from institutional status.¹⁷³

One panelist asserted that even if there are no pricing differences between for-profit and nonprofit hospitals, there can be other differences.¹⁷⁴ Nonprofit hospitals may have different long-term missions and have a different level of public accountability because of their long-term community obligations.¹⁷⁵ There is some empirical evidence that institutional status affects the mix of services provided by a hospital.¹⁷⁶

This panelist also suggested that board members of a for-profit hospital had fiduciary duties to a different group of individuals than would be the case if the hospital was nonprofit.¹⁷⁷Another panelist responded that "ownership variations are distinctions without a significant difference [and that all hospitals, irrespective of ownership] have the same mission: to provide the highest quality, appropriate medical care possible to the patients they serve, irrespective of the patient's ability to pay for such care."¹⁷⁸ Government statistics indicate that on average, uncompensated care accounts for a similar percentage of total costs at for-profit and nonprofit hospitals.¹⁷⁹

Although institutional status has loomed large in debates and legal disputes, the best available evidence indicates that nonprofits exploit market power when given the opportunity to do so. Accordingly, the profit/nonprofit status of the merging hospitals should not be considered a factor in predicting whether a hospital merger is likely to be anticompetitive.

VII. GROUP PURCHASING ORGANIZATIONS

A group purchasing organization (GPO) negotiates contracts with vendors of medical supplies on behalf of its members. GPO members include hospitals, nursing homes, home health agencies, and other health care systems. Some Hearing participants and industry commentators assert that GPOs, acting as their members' buying cooperatives, can be tremendous engines of efficiency, allowing medical buyers to pool their purchasing power to lower health care costs.

Nonetheless, others assert that certain GPO contracting practices may raise competitive concerns related to tying, bundling, and exclusive dealing. The Senate Judiciary Committee, through efforts by Chairman Mike DeWine and Ranking Member Herb Kohl of the Antitrust, Competition Policy and Consumer Rights Subcommittee, and the U.S. General Accounting Office have examined this issue in depth,¹⁸⁰ and the issue was an important topic in the Hearings and the Commission's Health Care Workshop.¹⁸¹

In the sections that follow, we explain what a GPO is, describe its role as a purchasing intermediary, and provide an overview of the GPO industry structure. This section then discusses the various organizational structures GPOs may adopt, the potential incentives created by each, and the various contracting practices used by either GPOs or their suppliers and their potential impact on competition.

Finally, this section addresses concerns expressed during the Hearings and elsewhere that Health Care Statement 7, which governs GPOs, impedes the Agencies' ability to challenge GPO practices when, and if, they are anticompetitive. For the reasons discussed below, the Agencies believe these concerns are misplaced, and it is not necessary to revise Health Care Statement 7.¹⁸² This statement does not provide a safety zone for the specific types of conduct that some commentators have criticized, including tying, bundling, or exclusive dealing. In such situations, the Agencies would analyze the conduct on a case-by-case basis to determine whether it may violate the antitrust laws.

A. What is a GPO?

GPOs are entities that aggregate health care providers' purchasing volume and contracting functions to negotiate discounts with manufacturers, distributors, and other vendors of medical products and services.¹⁸³ According to the Health Industry Group Purchasing Association (HIGPA), 96 percent of all acute care hospitals in the United States use the services of a GPO, and on average, hospitals use at least two GPOs.¹⁸⁴ More than 70 percent of hospital purchases are made through a contract negotiated by a GPO.¹⁸⁵

GPOs negotiate contracts with manufacturers of products that fall into two general categories - commodities and medical devices.¹⁸⁶ Cotton balls, bandages, and linens are examples of commodities for which hospital clinical staffs generally do not have strong preferences about the manufacturer. High technology medical devices such as pacemakers and stents are examples of medical devices for which hospital clinicians may have a preference as to the manufacturer.¹⁸⁷

GPOs are not wholesalers or distributors, and they do not take possession of, or title to, the products for which they negotiate contracts.¹⁸⁸ Vendors of medical supplies and services generally submit bids to a GPO in response to a "Request For Proposal."¹⁸⁹ One panelist stated that GPOs "simply negotiate a contract with a supplier that all members of the GPO can access. This guarantees the GPO member that it will receive a price no worse than the pre-negotiated price on the GPO contract."¹⁹⁰ Hospitals and other health care providers then purchase products and services directly from the vendor pursuant to the prices and contract terms specified in the GPO's contract with that vendor.¹⁹¹

Others note that in many cases, the GPO's contract does not bind the health care providers and they are free to negotiate separately with the vendor.¹⁹² According to one commentator, "GPO members have substantial freedom to purchase alternative products and do so in significant volumes, particularly where the products in question are differentiated."¹⁹³

B. GPO Industry Overview

The Hospital Bureau of New York, established in 1910, is the first known hospital GPO.¹⁹⁴ According to HIGPA, "[f]rom 1974 to 1999, the number of GPOs grew from forty to 633 . . . [and] it is estimated that approximately 200 GPOs contract directly with suppliers, and that twenty-six of these operate on a national level."¹⁹⁵ One commentator asserted that "when markets in this industry are properly defined, no GPO has a market share as high as 20%. Further, there are many GPOs, and hospitals can and do join multiple GPOs or switch memberships."¹⁹⁶

In contrast, the GAO's pilot study focused on seven national GPOs, each with purchasing volume of more than $1 billion. The GAO stated that the seven GPOs collectively accounted for purchases totaling approximately $43 billion, or "more than 85% of all hospital purchases nationwide made through GPO contracts."¹⁹⁷ Moreover, according to the GAO, the two largest GPOs in its study accounted for approximately 66 percent of total GPO purchasing.¹⁹⁸

One panelist explained that the numbers may differ depending on the study, the years measured, and whether percentages are based on all hospital purchases or only on hospital purchases made through a GPO.¹⁹⁹ For example, this panelist noted that the largest GPO accounts for 15 percent of total purchases by hospitals, but 30 percent of purchases made by hospitals through a GPO. Similarly, the second largest GPO's market share goes from 12 percent of all purchases to 25 percent of purchases made through a GPO.²⁰⁰

C. Structure and Incentives

The GAO report explained that "GPOs differ in their corporate structures and their relationships with member hospitals."²⁰¹ Member hospitals own some GPOs; in other cases, shareholders that are independent of the member hospitals own the GPO.²⁰² In some instances, suppliers finance GPOs by paying administrative fees that often are calculated as a percentage of each member's purchases of each supplier's products.²⁰³ These fees are designed to "cover [a] GPO's operating expenses and serve[] as its main source of revenue."²⁰⁴ GPOs may distribute surplus fees to their member hospitals as well.²⁰⁵ GPOs may be for-profit or nonprofit organizations.

Because of these differing structures, some panelists and commentators question the extent to which GPOs act as the agents of their buyer-members, or as the agents of the sellers that pay the GPOs' administrative fees. Because suppliers pay GPO fees, some worry that GPOs may operate to increase suppliers' revenues - and, correspondingly, GPO fees - rather than to minimize members' purchasing costs.²⁰⁶

Some panelists stated that when GPO members play important decision-making roles in the GPO, the GPO may be more likely to act as the agent of its buyer members. As one commentator put it, "[m]any GPOs are owned by their members, who sit on their boards, and are operated as cooperatives. These boards have no interest in procuring overpriced or substandard products [on] behalf of their own institutions."²⁰⁷ Similarly, one CEO stated that in his GPO the buyers

make all of the [GPO] contracting decisions ... [award] all of the contract[s] ... decide which suppliers get the contracts, what their compliance requirements are going to be, ... [and] the type of contract that's going to be awarded, whether it's a sole source contract, a dual source contract, or a multi-source contract... [and that each health care system] has a seat on [the] Board of Directors ... see[s] financial statements every month, ... help[s] us set the budget... [and has] a seat on every single contracting body.²⁰⁸

Another panelist stated that hospitals in such GPOs have "multiple opportunities through surveys, through advisory boards, advisory groups ... to have input into the suppliers that are selected for contract in [their] group purchasing organization."²⁰⁹

Other panelists asserted, however, that some GPOs act as the agents of the suppliers. One panelist asserted that the majority of GPOs "are financed and thereby controlled by large medical product companies rather than by the hospitals they are supposedly the agents for .... Fees and other incentives running from large medical manufacturers to GPOs allow such manufacturers to inappropriately influence the buying policies of the GPOs, because the compensation of most GPO management is almost always based on this fee income rather than on the real savings to hospital members."²¹⁰ As a result, another contended, GPOs "are selling protected market share to dominant suppliers in exchange for fees."²¹¹ Such seller payments "may reflect side-payments being made in exchange for the GPOs conferring a de facto exclusivity that enhances the market power of the incumbent device maker."²¹²

D. Contracting Practices

At the Hearings, panelists focused a significant portion of the discussion on whether certain GPO contracting practices - principally alleged tying, bundling, or exclusive dealing practices - injure competition. Such contracting practices include allegations that GPOs negotiate sole-source contracts with certain privileged manufacturers; require hospitals to purchase given volumes of certain supplies; bundle contracts that offer price discounts to purchasers of particular product groups; and enter contracts with manufacturers that last five years or more.²¹³

GPOs' critics stated that some suppliers, in league with GPOs with sufficiently large market share, can insist upon a variety of anticompetitive GPO contracting practices to exclude rival suppliers from serving the buyers.²¹⁴ They argue that such practices can discourage competitors from entering to bring down prices, and can discourage the research and development efforts necessary to produce innovative health care products that may improve on the incumbent's product.²¹⁵ Some charge, for example, that "if a large GPO negotiates a sole-source contract with a manufacturer, the contract could cause an efficient, competing manufacturer to lose business and exit from the market and could discourage other manufacturers from entering the market."²¹⁶

Similarly, GPOs' critics challenge hospital "commitments" to purchase a given volume to obtain a better price.²¹⁷ According to one panelist, under such a commitment, a hospital that buys an unauthorized product not only loses its better price on the complying product, but also must repay savings earned from having enjoyed that better price for years.²¹⁸ Critics also challenge contracts that offer bundled price discounts to purchasers of particular product groups, and contracts of five years or more that "can direct business to manufacturers for an extended period."²¹⁹

The economic literature on tying, bundling, and exclusive dealing practices indicates that they can be efficient, although under certain circumstances they may be harmful to competition.²²⁰ Scholarly legal commentary in recent years also has called into question the anticompetitive explanations for these practices and has focused on efficiencies and the potential welfare-enhancing aspects of these business arrangements.²²¹ Thus, courts typically engage in a fact-intensive inquiry to evaluate the competitive effects of such tying, bundling, and exclusive dealing claims.

Courts reviewing tying claims generally require that "(1) two separate products or services are involved, (2) the sale or agreement to sell one is conditioned on the purchase of the other, (3) the seller has sufficient economic power in the market for the tying product to enable it to restrain trade in the market for the tied product, and (4) a not insubstantial amount of interstate commerce in the tied product is affected."²²²

Courts reviewing the competitive consequences of exclusive dealing contracts typically analyze factors such as:

the degree of exclusion flowing from the restraint, its duration and terminability, the percentage of the market foreclosed and other indicia of the likely effect on competitors' ability to operate, the availability of alternative access routes to supplies or customers, rivals' ability to employ countermeasures to defeat the attempted exclusion, and, ultimately, the likely impact of raising rivals' costs on competition in a relevant market, including consideration of any procompetitive justifications.²²³

As a threshold matter, some panelists and commentators questioned whether allegations of exclusive dealing, tying, and bundling are true. For example, one panelist stated that "very few GPO contracts today are, in fact, exclusive," and unlike true exclusive dealing contracts, sole source contracts allow hospitals the freedom to buy from others.²²⁴ Another commentator noted that GPO loyalty rebate programs allow buyers to purchase from rivals offering lower prices.²²⁵ Other panelists noted that many long-term contracts are qualified in that "almost all GPO contracts can be terminated on 60- to 90-days notice."²²⁶

Some panelists argued that, even if the GPOs were doing what their critics alleged, these contracting practices can actually increase, not decrease, consumer welfare.²²⁷ For example, one source reported that GPOs use the challenged contracting practices "as incentives for manufacturers to provide deeper discounts and for hospital members to concentrate purchasing volume to obtain better prices."²²⁸

Some researchers and industry representatives claim that providers who make purchases pursuant to GPO contracts generally save 10 to 15 percent of the price they would otherwise pay.²²⁹ Also, GPO contracts that bundle products can be "simply ways of making products more attractive, effectively cutting price, or reducing costs by disposing of excess inventory."²³⁰

One panelist asserted that programs that allow suppliers to "reward [buyers'] higher levels of compliance" can be procompetitive "because they're offering increased dividends in exchange for volume," and because they standardize the buyers' products, which "leads to lower inventory costs [and] the ability to standardize patient care, leading to better quality, better staff education and improved safety."²³¹ This same panelist explained that long-term contracts are sometimes necessary in light of the costs of "large clinical evaluations." He explained the process involved for clinically evaluating a particular product:

The evaluation took 18 months. Our direct costs were over $150,000 .... We looked at product utilization in over 8,500 surgical cases in 60 of our facilities with over 2,100 surgeons participating. At the end of that evaluation process, our owners said this was too much work to award just a three-year contract... [and] they decided to award a five-year contract.²³²

He further stated that "strong" GPO programs are needed to counter the growing market power of suppliers that have consolidated in recent years.²³³ Finally, he also questioned whether the challenged practices could really be injuring the upstream supplier market, citing evidence that the medical device market is flourishing.²³⁴

Others, however, question GPOs' claimed efficiencies.²³⁵ For example, after a pilot study, the General Accounting Office reported that "GPOs' prices were not always lower and were often higher than prices paid by hospitals negotiating with vendors directly."²³⁶

According to the GPO industry, GPOs provide additional benefits to their members, including reduced overhead costs for purchasing departments. In addition, GPOs claim to provide "assistance with product-comparison analysis and standardization of products."²³⁷ Through GPOs, members may be able to reduce their supply costs via group purchasing, rebates, and surplus dividend payments.²³⁸

As one panelist stated, GPOs can not only "eliminate wasteful administrative duplication[ , _]. they increase competition between rival GPOs, manufacturers and their member hospitals, all of which can translate into lower prices and higher quality for consumers."²³⁹ Moreover, "GPOs assist members in product selection, an activity that would otherwise use up large amounts of member staff time."²⁴⁰ One estimate suggested that hospitals would spend on average $155,000 per hospital to duplicate the administrative and other functions GPOs provide.²⁴¹

The structure and incentives of individual GPOs may play an important role in determining the level of efficiencies they obtain. For example, GPOs acting on their members' behalf may strive to achieve efficiencies for the members.²⁴² Thus, some noted that when the challenged contracting practices are arranged not by one rival manufacturer seeking to foreclose others, but by "a buyer or its agent" - i.e., the buyers or their GPO - "in order to get lower prices from the manufacture[r]," the practices are likely to be pro-competitive.²⁴³

One panelist suggested this circumstance distinguishes the challenged practices from typical tying and bundling cases.²⁴⁴ Indeed, another panelist believed that a GPO's refusal to carry a given manufacturer's product likely reflects buyers' skepticism about the manufacturer's claims about its product, not any competitive injury.²⁴⁵

By contrast, some commentators suggest that if suppliers control a GPO, the organization may lack the incentive to promote efficiencies for its members. Some panelists suggested that such GPOs may have an incentive to collude with suppliers aiming to injure rival suppliers in a bid to acquire market power over the market for providing goods and services to the buyers.²⁴⁶ Under this theory, the GPOs agree to raise barriers against rival suppliers through contract terms imposing tying, bundling, or exclusive dealing arrangements on the buyers.²⁴⁷ These terms seek to "exclude rival manufacturers from competing for hospital sales even when the rival products are better or cheaper."²⁴⁸ Although suppliers do not need GPO support to attempt to exclude their rivals from the downstream market,²⁴⁹ one panelist suggested that the GPOs can streamline the efforts to exclude.²⁵⁰

GPO members may also find it difficult to pursue other means of procuring goods for a variety of reasons. For example, member hospitals may be contractually bound to purchase certain supplies through a given GPO; the efficiencies that GPOs afford may outweigh their anticompetitive costs; member hospitals may enjoy "side-payments or special discounts" that give them private incentives to stay; a race-to-the-bottom effect may persuade a hospital to maintain its special GPO discount so that it does not suffer vis-a-vis its rivals; or agency problems that reward hospital administrators for winning short-term price cuts regardless of long-term harms may prevent hospitals from taking action against these anticompetitive practices.²⁵¹

Others counter that GPOs are unlikely to collude with suppliers in this way for long, because buyers unhappy with the anticompetitive results can always leave the GPO for other means of purchasing supplies.²⁵² One panelist noted that GPOs must compete for hospitals' business and that hospitals "are free to select GPOs that best represent their interests."²⁵³

E. Statement 7 Does Not Protect Anticompetitive Contracting Practices

Health Care Statement 7 addresses the formation of a GPO. See Box 7-1. Some have proposed altering Statement 7, citing to concerns about alleged anticompetitive contracting practices.²⁵⁴ The Agencies, however, do not believe that it is appropriate or wise to amend Statement 7, because the statement and its safety zone thresholds do not prevent and should not be appropriately read as preventing antitrust challenges to any of the alleged anticompetitive contracting practices about which panelists and others have raised concerns.

Box 4-1: Health Care Statement 7. This statement provides in part: "The Agencies will not challenge, absent extraordinary circumstances, any joint purchasing arrangement among health care providers where two conditions are present: (1) the purchases account for less than 35 percent of the total sales of the purchased product or service in the relevant market; and (2) the cost of the products and services purchased jointly accounts for less than 20 percent of the total revenues from all products or services sold by each competing participant in the joint purchasing arrangement."

Statement 7 and its safety zone thresholds aim to address monopsony and oligopoly concerns with the formation of a GPO.²⁵⁵ This statement reflects concerns that a particular GPO could (1) create monopsony power, injuring competition in the supplier market or (2) facilitate collusion in the sale of hospital products or services, injuring competition in the downstream market.

Statement 7 does not address all potential issues that GPOs may raise. For example, it is silent on alleged exclusive dealing, tying, and bundling concerns that many panelists discussed in the Hearings. It is also silent on other potential competitive concerns, such as price-fixing, market allocation, mergers, etc. No statement is likely to cover every issue that could arise. The Agencies believe amending the statement to address some, but not all potential issues, is likely to be counterproductive. For example, some might argue that because certain issues were discussed, Statement 7 implicitly endorses as legal whatever conduct is not specifically addressed. If a supplier coordinates with the buyers, or with GPOs that have turned on their buyers, to exclude rival suppliers, Statement 7 would not protect such conduct from antitrust challenge.

In sum, Statement 7 governs Agency actions examining monopsony and oligopoly issues in connection with a GPO's formation. It does not preclude Agency action challenging anticompetitive conduct - such as anticompetitive contracting practices - that happens to occur in connection with GPOs. The Agencies will examine, on a case-by-case basis, the facts of any alleged anticompetitive contracting practices to determine whether the practice violates the antitrust laws.

VIII. TIERING AND PAY-FOR- PERFORMANCE

Commentators and panelists noted that some providers have resisted tiering and pay-for-performance programs, and refused to provide information regarding the quality of care they provide.²⁵⁶ When providers collectively refuse to enter into such arrangements or provide information to purchasers, the Agencies will carefully examine such conduct. As appropriate, the Agencies will bring cases against providers who collusively refuse to enter into such arrangements or provide such information. The Agencies also will challenge unilateral conduct or bundled contracting practices, where appropriate.

FOOTNOTES

¹Am. Med. Int'l v. FTC, 104F.T.C. 1 (1984), as modified by 104 F.T.C. 617 (1984) and 107 F.T.C. 310 (1986). The Commission decision held that a for-profit hospital chain's acquisition of a competing hospital in the city and county of San Luis, Obispo, California, violated § 7 of the Clayton Act and § 5 of the FTC Act. The Commission found that the acquisition lessened both price and nonprice competition, and ordered divestiture of the acquired hospital.

²U.S. Dep't of Justice & Federal Trade Comm'n, Antitrust Enforcement Policy Statements in the Health Care Area § 1 (1996) [hereinafter Health Care Statements], available at http://www.ftc.gov/reports/hlth3s.pdf. Agency review of most proposed hospital mergers is typically completed in less than a month. Id. § 1. See also J. Jacobs 3/28 at 69.

³Health Care Statements, supra note 2, § 1. The safety zone encompasses mergers between two general acute-care hospitals "where one of the hospitals (1) has an average of fewer than 100 licensed beds over the three most recent years, and (2) has an average daily inpatient census of fewer than 40 patients over the three most recent years, absent extraordinary circumstances." Id. This safety zone does not necessarily apply if one of the hospitals is less than five years old. Transactions that fall outside the safety zone are not necessarily anticompetitive and may be pro-competitive.

⁴The Agencies challenge relatively few mergers overall. In 2001, the Agencies were notified of 2,376 total mergers (the FTC challenged 23 and DOJ challenged 32) and a few of those were below the thresholds for notification. Federal Trade Comm'n Staff, U.S. Department of Justice, Antitrust Division, Annual Report to Congress, Fiscal Year 2002 (2003), available at http://www.ftc.gov/os/2003/08/hsrannualreport.pdf.

⁵See Health Care Services & Products Division, Federal Trade Comm'n, FTC Antitrust Actions in Health Care Services and Products (2003), available at http://www.ftc.gov/bc/hcupdate 031024.pdf; U.S. Dep't of Justice Antitrust Division, Health Care Task Force: Recent Enforcement Actions, at http://www.usdoj.gov/atr/public/ health_care/2044.htm; U.S. Dep't of Justice Antitrust Division Summary of Antitrust Division Health Care Cases Since August 25, 1983, athttp://www.usdoj. gov/atr/public/ health_care/0000.pdf.

⁶Martin Gaynor & William B. Vogt, Competition Among Hospitals, 34 Rand J. Econ. 764, 764 (2003).

⁷Id. at 764. The seven cases were: California v. Sutter Health Sys., 84 F. Supp. 2d 1057 (N.D. Cal.), aff'd mem., 2000-1 Trade Cas. (CCH) U 87,665 (9th Cir. 2000), revised, 130 F. Supp. 2d 1109 (N.D. Cal. 2001); FTC v. Tenet Healthcare Corp., 17 F. Supp. 2d 937 (E.D. Mo. 1998), rev 'd 186 F.3d 1045 (8th Cir. 1999); United States v. Long Island Jewish Med. Ctr., 983 F. Supp. 121 (E.D.N.Y. 1997); FTC v. Butterworth Health Corp., 946 F. Supp. 1285, 1300-1301 (W.D. Mich. 1996), aff'd, 1997-2 Trade Cas. (CCH) U 71,863, 71,867-68 (6th Cir. 1997); United States v. Mercy Health Services, 902 F. Supp. 968 (N.D. Iowa 1995), vacated as moot, 107 F.3d 632 (8th Cir. 1997); FTC v. Freeman Hosp., 911 F. Supp. 1213 (W.D. Mo.), aff'd, 69 F.3d 260 (8th Cir. 1995); In re Adventist Health Sys., 117 F.T.C. 224 (1994). One of the seven cases was brought by state antitrust enforcers without either Agency's involvement. See Sutter Health Sys., 84 F. Supp. 2d 1057.

⁸See Thomas L. Greaney, Night Landings on an Aircraft Carrier: Hospital Mergers and Antitrust Law, 23 Am. J.L. & Med. 191 (1997). As Professor Greaney notes, inFreeman Hospital, the FTC produced patient-origin data that showed a high percentage of patients stayed in the government's proposed geographic market, as well as forward looking testimony of market participants, including competitors, buyers, and consumers. The Court placed the Commission in a "Catch 22: hard evidence like historical patient-origin data was unacceptable because it did not address future contingencies, and managed care testimony was inadequate, although it addressed future contingencies, because it lacked the specificity of hard evidence." Id. at 207-08. Similarly, Professor Greaney noted that in Mercy Health Systems, the courts ignored most of DOJ's subjective and objective evidence designed to provide a dynamic analysis of the market and discounted opinion testimony of the most knowledgeable market participants, including third party payors and physicians. Id. at 209-212.

See also Peter Hammer & William Sage, Critical Issues in Hospital Antitrust Law, 22 Health Affairs 88, 90 (Nov./Dec. 2003) (noting merging hospitals have persuaded some courts "that nonprofit hospitals will not raise prices in the same manner as would for-profits or businesses outside of health care with comparable market share" and that relevant geographic markets include hospitals 70 to 100 miles away); William Sage et al., Why Competition Law Matters to Health Care Quality, 22 Health Affairs 31, 41-42 (Mar./Apr., 2003) (some courts presume nonprofit health facilities act in the public interest, and that increased revenues will be spent on quality improvements).

As the current Chairman of the Federal Trade Commission recently observed, "In hospital merger cases, the government is zero for the last seven. I don't know the specifics of every case, but what's striking is the zero. I can certainly accept the idea that the government should not have won them all. But it seems very unlikely the government should have lost them all." William M . Sage, Protecting Competition and Consumers: A Conversation With Timothy J. Muris, 22 Health Affairs 101, 103 (Nov./Dec. 2003).

⁹U.S. Dep't of Justice & Federal Trade Comm'n, Horizontal Merger Guidelines § 0.1 (1992 rev. 1997, efficiencies section only) [hereinafter Merger Guidelines], available at http://www.ftc.gov/bc/docs/horizmer.htm.

¹⁰Merger Guidelines, supra note 9, § 0.1.

¹¹Id. §0.1n.6.

¹²Id. § 0.2. The last factor is sometimes referred to as the "failing firm defense." As the guidelines explain:

A merger is not likely to create or enhance market power or facilitate its exercise if the following circumstances are met: 1) the allegedly failing firm would be unable to meet its financial obligations in the near future; 2) it would not be able to reorganize successfully under Chapter 11 of the Bankruptcy Act [11 U.S.C. §§1101-1174 (1988)]; 3) it has made unsuccessful good-faith efforts to elicit reasonable alternative offers of acquisition of the assets of the failing firm that would both keep its tangible and intangible assets in the relevant market and pose a less severe danger to competition than does the proposed merger; and 4) absent the acquisition, the assets of the failing firm would exit the relevant market. Id. § 5.1.

¹³e.g., In re Schering-Plough Corp., No. 9297 at 16-17 (Dec. 18, 2003) (discussing FTC v. Indiana Fed'n of Dentists, 476 U.S. 447, 460-61 (1986)), available at http://www.ftc.gov/os/adjpro/ d9297/031218 commissionopinion.pdf.

¹⁴Indiana Fed'n of Dentists, 476 U.S. at 460-61.

¹⁵See, e.g., Todd v. Exxon Corp., 275 F.3d 191, 206 (2d Cir. 2001) (evidence of "an actual adverse effect on competition ... arguably is more direct evidence of market power than calculations of elusive market share figures"); Toys R' Us v. FTC, 221 F.3d 928, 937 (7th Cir. 2000) (market power can be proved "through direct evidence of anticompetitive effects"); United States v. Baker Hughes Inc., 908 F.2d 981, 992 (D.C. Cir. 1990) ("'Market share is just away of estimating market power, which is the ultimate consideration,' and ... '[w]hen there are better ways to estimate market power, the court should use them'" (quoting Ball Mem'l Hosp. v. Mutual Hosp. Ins., 784 F.2d 1325, 1336 (7th Cir. 1986)).).

¹⁶Merger Guidelines, supra note 9, § 1.0. This test further assumes that the hypothetical profit- maximizing firm is not subject to price regulation and that the terms of sale of all other products are held constant. Id.

¹⁷Id. § 1.0.

¹⁸Seth Sacher & Louis Silvia, Antitrust Issues in Defining the Product Market for Hospital Services, 5 Int'l J. Econ. BUS. 181, 182-83 (1998) at http://www.ftc.gov/ogc/healthcarehearings/docs/030326sethbsacher.pdf.

¹⁹See also supra Chapter 1.

²⁰See, e.g., Guerin-Calvert 3/26 at 125, 130 (suggests using the merger guidelines and the hypothetical monopolist test; "although there is a great deal that is unique and specific about health care and hospitals in particular, [the best approach for analyzing hospital industry competition and transactions is] the same kinds of principles and the same kinds of fact-intensive analysis that is used in all other industries"); Margaret E. Guerin- Calvert, Defining Geographic Markets for Hospitals 6-11 (3/26) (slides) [hereinafter Guerin-Calvert Presentation], at http://www.ftc.gov/ogc/health carehearings/docs/030326guerincalvert.pdf; Vistnes 3/26 at 147-148 (stating the geographic market definition "should be driven, principally if not exclusively, by the Merger Guidelines;" the key test is whether a plan could divert enough patients to a different hospital in a different region to make the price increase unprofitable); Gregory Vistnes, Geographic Markets and Hospital Competition 5 (3/26) (slides) [hereinafter Vistnes Presentation], at http://www.ftc.gov/ogc/healthcarehearings/docs/vistnes.pdf; Werden 3/26 at 201 (noting the merger guidelines' hypothetical monopolist paradigm is the right approach); Gregory Werden, Hospital Mergers and the Hypothetical Monopolist Test 2 (3/26) (slides) [hereinafter Werden Presentation], at http://www.ftc.gov/ogc/healthcarehearings/docs/werden.pdf; David Argue 3/28 at 41-42.

²¹Hammer & Sage, supra note 8, at 90, citing to United States v. Rockford Mem'l Corp., 898 F.2d 1278, 1285 (7th Cir. 1990).

²²See, e.g., Michael Vita & Seth Sacher, The Competitive Effects of Not-For-Profit Hospital Mergers: A Case Study, 49 J. Indus. Econ. 63 (2001) (using a control group methodology to assess competitive effects). Here, the competitive effect of the transaction is identified by comparing the change in price at the merging hospitals to the change in price (measured over the same time period) at a set of "control" hospitals. The control hospitals are hospitals in other geographic areas that are otherwise similar to the merging hospitals. Note, however, that a price increase by itself may not be sufficient to prove anticompetitive effects.

²³The Elzinga-Hogarty test is named for the two economists who first proposed this particular analysis. See Kenneth Elzinga & Thomas Hogarty, The Problem of Geographic Market Delineation in Antitrust Suits, 18 Antitrust Bull. 45 (1973) [hereinafter Elzinga & Hogarty, The Problem]; Kenneth Elzinga & Thomas Hogarty, The Problem of Geographic Market Delineation Revisited: The Case of Coal, 23 Antitrust Bull. 1 (1978) [hereinafter Elzinga & Hogarty, The Problem Revisited].

²⁴The term "critical loss analysis" was first used in an article: Barry Harris & Joseph Simons, Focusing Market Definition: How Much Substitution Is Necessary? 12 Res. IN L. & Econ. 207 (1989).

²⁵See FTC v. Tenet Healthcare Corp., 17 F. Supp. 2d 937 (E.D. Mo. 1998), rev 'd 186 F.3d 1045 (8th Cir. 1999). In this case, the Eighth Circuit relied on both an Elzinga-Hogarty test and a critical loss analysis to conclude that a broad geographic market was appropriate. Similarly, in United States v. Mercy Health Services, 902 F. Supp. 968 (N.D. Iowa 1995), vacated as moot, 107 F.3d 632 (8th Cir. 1997), the District Court relied on patient migration patterns, regional hospitals' outreach clinics, and the lack of evidence that patients' loyalty to their physicians would prevent them from defeating a price increase to find a broad geographic market. See also J. Jacobs 3/28 at 72-74 (noting DOJ lost the Mercy Health case on the geographic market definition for all of these reasons, but believes that the government could address successfully some of these issues today); California v. Sutter Health Sys., 84 F. Supp.2d 1057 (N.D. Cal. 2000) (insufficient evidence of a relevant geographic market); FTCv. Freeman Hosp., 911 F. Supp. 1213 (W.D. Mo.), aff'd, 69 F.3d 260 (8th Cir. 1995) (holding the Commission had failed to identify a relevant geographic market).

²⁶See Hammer & Sage, supra note 8, at 90; Frech 3/26 at 189-191; Greaney 2/27 at 141-42; Greaney , supra note 8.

²⁷See Werden 3/26 at 248-50 (the data may provide descriptive information, but you cannot draw strong conclusions); Guerin-Calvert 3/26 at 139; Guerin-Calvert Presentation, supra note 20, at 17; Frech 3/26 at 190-91 (noting that patient flow data and the Elzinga-Hogarty ratios are useful background, but make no sense when used as a bright line to define the geographic market); Vistnes 3/26 at 251-52; Argue 3/28 at 44 (Elzinga-Hogarty is a static analysis and does not address the dynamic nature of markets).

²⁸Guerin-Calvert 3/26 at 139; Guerin-Calvert Presentation, supra note 20, at 17.

²⁹See, e.g., Harris 3/26 at 171-78, 222-24; Guerin-Calvert 3/26 at 125, 130-31; Werden 3/26 at 201-205, 212-20, 248-50; Frech 3/26 at 189-90; Daniel O'Brien & Abraham Wickelgren, A Critical Analysis of Critical Loss Analysis, 71 Antitrust L.J. 161, 161-62 (2003); Michael Katz & Carl Shapiro, Critical Loss: Let's Tell the Whole Story, 17 Antitrust, Spring 2003, at 49-50; James Langenfeld & Wenqing Li, Critical Loss Analysis in Evaluating Mergers, 46 Antitrust Bull. 299, 299-301 (2001); Kenneth L. Danger & H.E. Frech III, Critical Thinking About 'Critical Loss' in Antitrust, 46 Antitrust Bull. 339, 340-42 (2001); David Scheffman & Joseph Simons, The State of Critical Loss Analysis: Let's Make Sure We Understand the Whole Story, 3 The Antitrust Source, Nov. 2003, at http://www.abanet.org/antitrust/source/nov03/ scheffman.pdf.

³⁰Vistnes 3/26 at 144, 147-49; Vistnes Presentation, supra note 20, at 4-5, 11-18; Guerin-Calvert 3/26 at 131-33; Guerin-Calvert Presentation, supra note 20, at 4, 12. See also Leibenluft 3/28 at 8-9 ("[On] geographic market, it's sort of a Catch 22. The courts require ~ and, I think, rightfully so ~ that the analysis be dynamic. What will happen if the hospitals merge? As a result of that, the plaintiff is faced with a difficult task. What they have is traditional hard evidence which relates to, for example, patient flow data, which reflects historical patient patterns, and is historical conduct. But that doesn't reflect what might happen in the future. But when the Government tries to find what may or look to what may suggest what will happen dynamically, then that evidence could be attacked as being speculative or anecdotal."); Feller 9/24 at 66 (discussing geographic markets for physician services and also noting that "zip code analysis, however, only presents a static and limited view of the relevant geographic market").

³¹Elzinga & Hogarty, The Problem, supra note 23; see also Elzinga & Hogarty, The Problem Revisited, supra note 23.

³²Elzinga & Hogarty, The Problem, supra note 23, at 52-64.

³³Id. at 72-75 & n.75 ("Where the appropriate product market is a set of heterogeneous goods, or where there is product differentiation, or where there are important physical differences among units within the product market, adding together physical units will be difficult if not impossible. In such cases, measuring output in sales instead of physical units might be necessary.").

³⁴See, e.g., Zwanziger 3/26 at 92 (The Elzinga-Hogarty approach "is poorly suited to hospital mergers" because it does not recognize the underlying heterogeneity on the supply or demand side of hospital services.); Jack Zwanziger, Defining Hospital Markets 2 (3/26) (slides) [hereinafter Zwanziger Presentation], at http://www.ftc.gov/ogc/ healthcare hearings/ docs/ zwanziger.pdf.

³⁵See Gregory Vistnes, Hospitals, Mergers, and Two-Stage Competition, 67 ANTITRUST L.J. 671, 689 (2000); Sacher & Silvia, supra note 18, at 192-93.

³⁶See Vistnes, supra note 35, at 689; Elzinga & Hogarty, The Problem, supra note 23, at 72-76; Elzinga & Hogarty, The Problem Revisited, supra note 23, at 2-3.

³⁷See Elzinga & Hogarty, The Problem, supra note 23, at 73-75; Elzinga & Hogarty, The Problem Revisited, supra note 23, at 2.

If the LIFO and LOFI are both 10 percent or less, then the geographic market satisfies the "strong" Elzinga-Hogarty test. If the LIFO and LOFI are both 25 percent or less then the geographic market satisfies the "weak" Elzinga-Hogarty test. Elzinga & Hogarty, The Problem Revisited, supra note 23, at 2.

³⁸Frech 3/26 at 190-97; Greaney 2/27 at 141-42 (noting that the courts naively interpret Elzinga-Hogarty in health care cases, and that because hospitals offer heterogeneous services and patients have highly diverse preferences, this results in "thoroughly wrong-headed precedents and subdoctrines").

³⁹Frech 3/26 at 190-95.

⁴⁰Id. at 195.

⁴¹Zwanziger 3/26 at 232-33. See also id. at 97-99 (noting that large markets based on patient flow data and Elzinga-Hogarty are incompatible with research knowledge: travel distance is the most important criteria for a patient in deciding which hospital to use).

⁴²Cory Capps et al., The Silent Majority Fallacy of the Elzinga-Hogarty Criteria: A Critique and New Approach to Analyzing Hospital Mergers 1 (Nat'l Bureau of Econ. Research, Working Paper No. w8216, 2001) [hereinafter Capps et al., Silent Majority]. See also Cory Capps et al., Geographic Market Definition in Hospital Merger Cases 4 (4/16) [hereinafter Capps et al. (stmt)], at http://www.ftc.gov/ogc/healthcare hearings/docs/030410capps2.pdf; Cory Capps, For-Profit and Non-Profit Pricing: The Empirical Evidence (4/10) (slides), at http://www.ftc.gov/opp/hc/030410corycapps.pdf [hereinafter Capps Presentation]. See also Cory Capps et al., Antitrust Policy and Hospital Mergers: Recommendations for a New Approach, 47 ANTITRUST Bull. 677, 713-14 (2002) [hereinafter Capps et al., Antitrust Policy].

⁴³Capps et al., Silent Majority, supra note 42, at 1-2.

⁴⁴Id.

⁴⁵Frech 3/26 at 195 ("[A]s you expand the area to get to a high enough percentage to call it a service area, you keep picking up more hospitals, and that keeps making it more difficult" to reach a cut-off.). Professor Frech noted that even at the 75 percent level, the defendant's expert could not find a cut-off for the Poplar Bluff geographic market area in the Tenet case. Id. at 195.

⁴⁶Frech 3/26 at 192 ("[R]anking zip codes by the number of patients usually gives the largest market areas.").

⁴⁷Id. at 192-93. See also H.E. Frech, III et al., Elzinga-Hogarty Tests and Alternative Approaches for Market Share Calculations in Hospital Markets, 71 ANTITRUST L.J. 921, 928-29, 941-47 (2004).

⁴⁸Frech 3/26 at 192-93.

⁴⁹e.g., Scheffman & Simons, supra note 27, 61 at 2-3; Harris 3/26 at 170-75; Werden 3/26 at 201-04.

⁵⁰See supra note 29.

⁵¹One also can ask how much of a reduction in its sales the hypothetical monopolist would be willing to tolerate to sustain a given price increase. Only asking this alternative calculation actually implements the Horizontal Merger Guidelines' hypothetical monopolist test, but the analysis described in the text yields roughly the same result under plausible conditions. Werden 3/26 at 202-04; Werden Presentation, supra note 20, at 4-5.

⁵²Harris 3/26 at 170-75. The formula for the critical loss for an x% price increase is x/(x + m), where m is the margin, expressed as a percentage price. For example, if the margin is 60 percent, the critical loss for a 5 percent price increase is 5/(5 + 60) = .077, or 7.7 percent.

⁵³Id. at 174-75.

⁵⁴Scheffman & Simons, supra note 29, at 2-3 (outlining a three-step process for conducting a critical loss analysis); see also Katz & Shapiro, supra note 29, at 49-50; O'Brien & Wickelgren, supra note 29, at 161.

⁵⁵Werden 3/26 at 204-05; Werden Presentation, supra note 20, at 8.

⁵⁶Werden 3/26 at 204-05; Werden Presentation, supra note 20, at 8, 11, 14.

⁵⁷Werden 3/26 at 209-17; Werden Presentation, supra note 20, at 15-19.

⁵⁸Werden 3/26 at 204-05; Werden Presentation, supra note 20, at 14.

⁵⁹Werden 3/26 at 219-20 (noting that it is important to properly calculate the margin, and that in hospital mergers it is possible that not all patients contribute the same margin - depending on which patients are likely to leave if faced with a price increase, the margin, and therefore, the critical loss, may differ). For similar critiques, see Danger & Frech, supra note 29; Langenfeld & Li, supra note 29.

⁶⁰Alternatively, these zip codes are identified as "at risk" or "overlapping." Harris 3/26 at 177-78; Frech 3/26 at 189-190.

⁶¹See Capps et al., Silent Majority, supra note 42; Capps et al., Antitrust Policy, supra note 42, at 679- 82, 690-92, 694-704.

⁶²Frech 3/26 at 189-90 (noting that the predicted actual loss is an important part of how critical loss analysis is implemented, and as typically implemented, critical loss analysis leads to implausibly large geographic areas).

⁶³Danger & Frech, supra note 29 at 349-51; Katz & Shapiro, supra note 29, at 49-50, 52-53; Langenfeld & Li, supra note 29, at 302-03, 307-08; O'Brien & Wickelgren, supra note 29, at 161-63.

⁶⁴See Katz & Shapiro, supra note 29, at 50; O'Brien & Wickelgren, supra note 29, at 161-62; Scheffman & Simons, supra note 29, at 4.

⁶⁵See Katz & Shapiro, supra note 29, at 50-51; O'Brien & Wickelgren, supra note 29, at 162; Danger & Frech, supra note 29, at 349-51; Langenfeld & Li, supra note 29, at 308-09, 323; Michael Katz & Carl Shapiro, Further Thoughts on Critical Loss, 3 The Antitrust Source, Mar. 2004, at http://www.abanet.org/antitrust/source/march04 /katzshapiro.pdf; Daniel O'Brien & Abraham Wickelgren, The State of Critical Loss Analysis: Reply to Scheffman and Simons, 3 The Antitrust Source, Mar. 2004, at http://www.abanet.org/ antitrust/source/march04/obrienwickel.pdf. But see Scheffman & Simons, supra note 29, at 5 (disagreeing with critiques that attempt "to infer, with greater specificity, a value of AL [actual loss] from incremental margins and (too simple an) economic theory").

⁶⁶One panelist defended critical loss at the Hearings as an appropriate mechanism for analyzing proposed hospital geographic markets. Harris 3/26 at 167, 173-74. This panelist recommended that the parties and court closely examine documents, data, and testimony to determine the elasticity of demand and how many patients are likely to leave if faced with an anticompetitive price increase. Harris 3/26 at 222-24. He did not, however, address the argument that the premerger margin itself contains substantial information about the likely switching behavior of consumers.

⁶⁷Katz & Shapiro advocate focusing on what they term the "aggregate diversion ratio" to indicate whether the elasticity of demand for the candidate market is sufficiently lower than the firm-level demand elasticities so that the candidate market is, in fact, a market. Suppose there are three products in the candidate market, A, B, and C, and the price of A is increased by five percent. The aggregate diversion ratio is the percentage of sales lost by A that is recaptured by B and C. Katz and Shapiro argue that the actual loss is less than the critical loss if and only if the aggregate diversion ratio exceeds the critical loss. Katz & Shapiro, supra note 29, at 53-54. See also O'Brien & Wickelgren, supra note 29, at 184 ("We have shown that the inference typically drawn from critical loss analysis - that high margins make a merger less likely to be anticompetitive - is often inconsistent with economic theory .... In our opinion, critical loss analysis has led to enormous confusion about the economic factors that govern firms' pricing incentives. The technique has been misused so frequently that arguments that are inconsistent with basic economic theory have almost gained a measure of legitimacy in antitrust cases.").

⁶⁸Frech 3/26 at 189, citing to Danger & Frech, supra note 29. See also Langenfeld & Li, supra note 29, at 301, 313, 323-333; O'Brien & Wickelgren, supra note 29, at 162, 168-73, 177-84; Katz & Shapiro, supra note 29, at 50-51,54-55.

⁶⁹Langenfeld & Li, supra note 29, at 323-24, 332-33. Many of these same problems have been identified by other researchers. See, e.g., Danger & Frech, supra note 29, at 341-42; O'Brien & Wickelgren, supra note 29, at 162, 184; Katz & Shapiro, supra note 29, at 52-55.

⁷⁰Langenfeld & Li, supra note 29, at 332-333. The formula for critical loss is x/(x + m), where x is the percentage price change of interest (e.g., 5%) and m is the premerger price cost margin ((p-c)/p), expressed as a percentage. In equilibrium, m = 1/,, where , is the elasticity of demand. If , is small and premerger margins are therefore high, it will also be true (by definition of elasticity) that a given price increase will induce only small changes in quantity. See O'Brien & Wickelgren, supra note 29, at 167-68; Katz & Shapiro, supra note 29, at 50-53; Danger & Frech, supra note 29, at 342-50; Langenfeld & Li, supra note 29, at 303-05, 334-337; But see Scheffman & Simons, supra note 29, at 5-8 (arguing that critiques of critical loss analysis that use the formula (m = 1/,), or the Lerner Equation, use "the simplest economic model of pricing" to infer that actual loss would be equal or close to critical loss in equilibrium and thereby inappropriately shift the burden of proof to defendants).

⁷¹Vistnes 3/26 at 145-146; Vistnes Presentation, supra note 20, at 2, 4; Vistnes, supra note 35, at 671-692.

⁷²Vistnes 3/26 at 148; Vistnes Presentation, supra note 20, at 5; Vistnes, supra note 35, at 674-81, 692. See also Town 4/9 at 60-67 (discussing simulation study that showed significant post-merger price increases to HMOs even though an Elzinga-Hogarty analysis suggested little, if any competitive harm; this suggests that it is important to focus on the price negotiations between hospitals and payors and the ability of a payor to exclude a particular hospital if they cannot reach a price agreement).

⁷³Vistnes 3/26 at 157-60; Vistnes Presentation, supra note 20, at 11-14; Vistnes, supra note 35, 671-74, 681-84, 688-92. See also Frech 3/26 at 196-98 (agreeing that with managed care, there are now two stages of competition, and that patient flow data is static and only reflects competition at the consumer or second-stage level, but not at the payor or first-stage level, because changes in payors' hospital networks move too slowly to be captured in the patient flow data).

⁷⁴Vistnes 3/26 at 160; Vistnes Presentation, supra note 20, at 13-14; Vistnes, supra note 35, at 681-84.

⁷⁵Vistnes 3/26 at 146-47; Vistnes Presentation, supra note 20, at 13-14; Vistnes, supra note 35, at 672-7 4, 688.

⁷⁶Vistnes 3/26 at 160; Vistnes Presentation, supra note 20, at 14; Vistnes, supra note 35, at 672-73.

⁷⁷Guerin-Calvert 3/26 at 230.

⁷⁸Id.at 230-31. But see Vistnes 3/26 at 243 (arguing that even if all hospitals are in a plan's network today, as long as the plan can credibly threaten to exclude the hospital, that possibility of exclusion is a constraint on pricing).

⁷⁹Capps et al. (stmt), supra note 42, at 5.

⁸⁰Id. at 5-6.

⁸¹Id. at 6. The authors refer readers to another paper (Cory Capps et al., Competition and Market Power in Option Demand Markets (April 2003) (unpublished manuscript)), in which they "provide a step by step derivation and empirical implementation of a market power measure that correctly incorporates the ex-ante nature of hospital pricing." Id. at 6-7. These authors also published another article outlining option demand analysis, as well as two other analyses. The authors suggest that the other two analytical techniques are not as accurate as the formal option demand analysis, but they are useful in defining hospital geographic markets. See Capps et al., Antitrust Policy, supra note 42, at 681.

⁸²Guerin-Calvert 3/26 at 141-43, 226, 237-39; Harris 3/26 at 223.

⁸³See, e.g., Guerin-Calvert 3/26 at 141 (stating that documents show who the hospitals see as their competitors and strategic plans of hospitals competing with merging hospitals often show strategies for taking patients from another hospital); Guerin-Calvert Presentation, supra note 20, at 12.

⁸⁴FTC v. Tenet Health Care Corp., 186 F.3d 1045, 1054 & n.14 (8th Cir. 1999).

⁸⁵Tenet Health Care, 186 F.3d at 1054 & n.14. See also Greaney 2/27 at 142 (finding it inexplicable that two circuits have "adopted an evidentiary rule of thumb that discounts the credibility of the testimony of third party payers on facts that are really central to their business ... when [the testimony is] unimpeached, not impeached by a showing of bias or other defects").

⁸⁶See, e.g., Leibenluft 3/28 at 15-16; Vistnes 3/26 at 147-57.

⁸⁷See, e.g., Vistnes 3/26 at 148-50; Eisenstadt 3/28 at 60-61.

⁸⁸See, e.g., Guerin-Calvert 3/26 at 140-43 (suggesting looking not only at what payors say about which hospitals are critical to their networks, but at what payors have done in the past to respond to different market behaviors, such as price increases or quality decreases); Guerin-Calvert Presentation, supra note 20, at 13, 16, 18; see also Singer 3/28 at 37-38; Toby Singer, Issues in Litigating Hospital Mergers 2-5 (3/28) ("In particular, the courts have not been willing to believe the testimony of health plans and others when it is contradicted by other evidence, such as statistical evidence on market definition," citing to California v. Sutter Health System, 84 F. Supp. 2d 1057 (N.D. Cal.), aff'dmem., 2000-1 Trade Cas. (CCH) 1(87,665 (9th Cir. 2000), revised, 130 F. Supp. 2d 1109 (N.D. Cal. 2001)); United States v. Long Island Jewish Medical Center, 983 F. Supp. 121 (E.D.N.Y. 1997); Adventist Health System/Est, 114 F.T.C. 458 (1991), at http://www.ftc.gov/ogc/healthcarehearings/docs/030328singertoby.pdf; Argue 3/28 at 49-51.

To be sure, a court will wish to assess the consistency of a witness's testimony with its documents and evidence of its previous actions. With respect to payor testimony, however, some judicial skepticism appears to be based, at least in part, on patient flow data. For the reasons discussed supra, patient flow data does not provide reliable information about what payors could do if faced with hospital price increases.

⁸⁹Guerin-Calvert 3/26 at 138-39.

⁹⁰Id. at 138-39. Some believe that the recent increases in insurance premiums are, at least in part, due to these demands for more choice and broader provider networks. See supra Chapter 1 and infra Chapter 5.

⁹¹Guerin-Calvert 3/26 at 134, 141 (referring to cases where payors were able to move marginal patients); Vistnes 3/26 at 152-56 (listing possible strategies payors could use to divert patients: dropping a hospital from the network; adding hospitals to the network to "dilute" the patient base; creating incentives for patients to switch hospitals; creating incentives for physicians to admit elsewhere; and changing the physician panel); Vistnes Presentation, supra note 20, at 8; Harris 3/26 at 180 (stating payors use various mechanisms to shift patient choices, including different copays and deductibles, tiered plans, and cafeteria plans).

⁹²See supra Chapter 3.

⁹³See, e.g., Guerin-Calvert 3/26 at 140-43; Frech 3/26 at 186-88.

⁹⁴Frech 3/26 at 186-88.

⁹⁵Id. See also H. E. Frech III & Lee Rivers Mobley, Managed Care, Distance Traveled and Hospital Market Definition, 37 INQUIRY 369-384 (2000).

⁹⁶Frech 3/26 at 186-88; Zwanziger 3/26 at 98-99 (describing research that suggests that travel distance is the most important criteria for a patient in deciding which hospital to use, and in California, where managed care penetration went from 20 percent to 90 percent over a specific period of time, the average travel distance changed very little over that same period).

⁹⁷Guerin-Calvert 3/26 at 134, 137, 141. But see Frech 3/26 at 197 (noting that turn-over among the hospitals included in a plan is sufficiently infrequent that patient flow data will often not capture the dynamics of first-stage competition).

⁹⁸Guerin-Calvert 3/26 at 140-41, 143; see also Guerin-Calvert 3/26 at 252 (describing documents in some markets that have included letters from plans to physicians to use one hospital more than another, and patient flow data subsequently showed the shift of enrollees from one hospital to another).

⁹⁹See Guerin-Calvert 3/26 at 141-43 (noting that it is rare to find a compelling coordinated-effects story in hospital markets and that the Chattanooga case is the one exception where the court accepted a coordinated effects theory of harm, referring to the Seventh Circuit opinion in Hospital Corp. of America v. FTC, 807 F.2d 1381 (7th Cir. 1986)); Guerin-Calvert Presentation, supra note 20, at 18.

¹⁰⁰Vistnes 3/26 at 150-60.

¹⁰¹Id. at 154-56.

¹⁰²Vistnes 3/26 at 156; Vistnes Presentation, supra note 20, at 9. But see FTC v. Tenet Healthcare Corp., 186 F.3d 1045, 1054 & n. 14 (8th Cir. 1999), rev 'g finding for plaintiff in FTC v. Tenet Health Care Corp., 17 F.Supp. 2d 937 (E.D. Mo. 1998) (finding that district court erred in rejecting more distant hospitals that were more costly because in doing so it "underestimated the impact of nonprice competitive factors, such as quality").

¹⁰³See, e.g., Zwanziger 3/26 at 97-99; Zwanziger Presentation, supra no