Promedica Health System, Inc. v. Federal Trade Commission

RECOMMENDED FOR FULL-TEXT PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 14a0083p.06 UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT _________________ PROMEDICA HEALTH SYSTEM, INC., ┐ Petitioner, │ │ │ No. 12-3583 v. │ > │ FEDERAL TRADE COMMISSION. │ Respondent. │ ┘ On Petition for Review of a Final Order of the Federal Trade Commission No. 9346. Argued: March 7, 2013 Decided and Filed: April 22, 2014 Before: KETHLEDGE, WHITE, and STRANCH, Circuit Judges. _________________ COUNSEL ARGUED: Douglas R. Cole, ORGAN COLE + STOCK LLP, Columbus, Ohio, for Petitioner. Michele Arington, FEDERAL TRADE COMMISSION, Washington, D.C., for Respondent. ON BRIEF: Douglas R. Cole, Erik J. Clark, ORGAN COLE + STOCK LLP, Columbus, Ohio, David Marx, Jr., Stephen Y. Wu, MCDERMOTT WILL & EMERY LLP, Chicago, Illinois, for Petitioner. Michele Arington, John F. Daly, FEDERAL TRADE COMMISSION, Washington, D.C., for Respondent. Beth Heifetz, Tara Stuckey Morrissey, JONES DAY, Washington, D.C., Mark J. Botti, Hyland Hunt, AKIN GUMP STRAUSS HAUER & FELD LLP, Washington, D.C., for Amici Curiae. 1 No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 2 _________________ OPINION _________________ KETHLEDGE, Circuit Judge. This is an antitrust case involving a proposed merger between two of the four hospital systems in Lucas County, Ohio. The parties to the merger were ProMedica, by far the county’s dominant hospital provider, and St. Luke’s, an independent community hospital. The two merged in August 2010, leaving ProMedica with a market share above 50% in one relevant product market (for so-called primary and secondary services) and above 80% in another (for obstetrical services). Five months later, the Federal Trade Commission challenged the merger under § 7 of the Clayton Act, 15 U.S.C. § 18. After extensive hearings, an Administrative Law Judge and later the Commission found that the merger would adversely affect competition in violation of § 7. The Commission therefore ordered ProMedica to divest St. Luke’s. ProMedica now petitions for review of the Commission’s order, arguing that the Commission was wrong on both the law and the facts in its analysis of the merger’s competitive effects. We think the Commission was right on both counts, and deny the petition. I. A. Lucas County is located in the northwestern corner of Ohio, with approximately 440,000 residents. Toledo lies near the county’s center; more affluent suburbs lie to the southwest. Two- thirds of the county’s patients have government-provided health insurance, such as Medicare or Medicaid. Twenty-nine percent of the county’s patients have private health insurance, which pays significantly higher rates to hospitals than government-provided insurance does. (Medicare and Medicaid reimbursements generally do not cover the providers’ actual cost of services.) A relatively large proportion of the county’s privately insured patients reside in the county’s southwestern corner. This case concerns the market—or markets, depending on how one defines them—for “general acute-care” (GAC) inpatient services in Lucas County. GAC comprises four basic categories of services. The most basic are “primary services,” such as hernia surgeries, No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 3 radiology services, and most kinds of inpatient obstetrical (OB) services. “Secondary services,” such as hip replacements and bariatric surgery, require the hospital to have more specialized resources. “Tertiary services,” such as brain surgery and treatments for severe burns, require even more specialized resources. And “quaternary services,” such as major organ transplants, require the most specialized resources of all. Different hospitals offer different levels of these services. There are four hospital providers in Lucas County. The most dominant is ProMedica, with 46.8% of the GAC market in Lucas County in 2009. ProMedica operates three hospitals in the county, which together provide primary (including OB), secondary, and tertiary services. The county’s second-largest provider is Mercy Health Partners, with 28.7% of the GAC market in 2009. Mercy likewise operates three hospitals in the county, which together provide primary (including OB), secondary, and tertiary services. The University of Toledo Medical Center (UTMC) is the county’s third-largest provider, with 13% of the GAC market. UTMC operates a single teaching and research hospital, just south of downtown Toledo, and focuses on tertiary and quaternary services. It does not offer OB services. The remaining provider is St. Luke’s Hospital, which before the merger was an independent, not-for-profit hospital with 11.5% of the GAC market. St. Luke’s offers primary (including OB) and secondary services, and is located in southwest Lucas County. B. With respect to privately insured patients, hospital providers do not all receive the same rates for the same services. Far from it: each hospital negotiates its rates with private insurers (known as Managed Care Organizations, or MCOs); and the rates themselves are determined by each party’s bargaining power. The parties’ bargaining power depends on a variety of factors. An MCO’s bargaining power depends primarily on the number of patients it can offer a hospital provider. Hospitals need patients like stores need customers; and hence the greater the number of patients that an MCO can offer a provider, the greater the MCO’s leverage in negotiating the hospital’s rates. But MCOs compete with each other just as hospitals do. And to attract patients, an MCO’s health-care plan must offer a comprehensive range of services—primary, secondary, tertiary, and quaternary—within a geographic range that patients are willing to travel for each of those No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 4 services. (The range is greater for some services than others.) These criteria in turn create leverage for hospitals to raise rates: to the extent patients view a hospital’s services as desirable or even essential—say, because of the hospital’s location or its reputation for quality—the hospital’s bargaining power increases. But another important criterion for a plan’s competitiveness is its cost. Thus, if a hospital demands rates above a certain level—the so-called “walk-away” point—the MCO will try to assemble a network without that provider. For example, rather than include all four hospital providers in its network, the MCO might include only three. If a provider becomes so dominant in a particular market that no MCO can walk away from it and remain competitive, however, then that provider can demand—and more to the point receive—monopoly rates (i.e., prices significantly higher than what the MCOs would pay in a competitive market). Here, before the merger, MCOs in Lucas County had sometimes offered networks that included all four hospital providers, but sometimes offered networks that included only three. From 2001 until 2008, for example, Lucas County’s largest MCO, Medical Mutual of Ohio, successfully marketed a network of Mercy, UTMC, and St. Luke’s. Since 2000, however, no MCO has offered a network that did not include either ProMedica or St. Luke’s—the parties to the merger here. C. The likely reason MCOs have historically found it necessary to include either ProMedica or St. Luke’s in their networks is that those providers are dominant in southwest Lucas County, where St. Luke’s is located. In that part of the county—relatively affluent, and with a high proportion of privately insured patients—ProMedica and St. Luke’s were direct competitors before the merger at issue here. Indeed, St. Luke’s viewed ProMedica as its “most significant competitor,” while ProMedica viewed St. Luke’s as a “[s]trong competitor”—strong enough, in fact, that ProMedica offered to discount its rates by 2.5% for MCOs who excluded St. Luke’s from their networks. But in this competition ProMedica had the upper hand. It is harder for an MCO to exclude the county’s most dominant hospital system than it is for the MCO to exclude a single hospital that services just one corner of the county—a corner, moreover, that the dominant system also services. And that means the MCOs’ walk-away point for the dominant system is No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 5 higher—perhaps much higher—than it is for the single hospital. Here, the record bears out that conclusion: ProMedica’s rates before the merger were among the highest in the State, while St. Luke’s rates did not even cover its cost of patient care. That was true even though St. Luke’s quality ratings on the whole were better than ProMedica’s. As a result, St. Luke’s struggled in the years before the merger, losing more than $25 million between 2007 and 2009. To improve matters, St. Luke’s hired Daniel Wakeman, a hospital-turnaround specialist, as its CEO. Wakeman implemented a three-year plan to reduce costs, increase revenues, and regain patient volume from ProMedica. Eventually St. Luke’s fortunes began to improve: by August 2010, St. Luke’s was out of the red (albeit barely), and Wakeman reported that “this positive margin confirms that we can run in the black if activity stays high.” By then, however, St. Luke’s was contemplating other options. In August 2009, Wakeman presented three options to St. Luke’s Board. The first was for St. Luke’s to “[r]emain independent” by “cut[ting] major services” until an “accepted margin is realized.” The second was for St. Luke’s to “[p]ush the [MCOs] . . . to raise St. Luke’s reimbursement rates to an acceptable margin.” Under this option, Wakeman noted, “the message [to MCOs] would be [to] pay us now (a little bit more) or pay us later (at the other hospital system contractual rates).” The third option was for St. Luke’s to join one of the three other providers in Lucas County— ProMedica, Mercy, or UTMC. Of all these options, Wakeman believed that a merger with ProMedica “ha[d] the greatest potential for higher hospital rates. A ProMedica-[St. Luke’s] partnership would have a lot of negotiating clout.” Wakeman also recognized, however, that an affiliation with ProMedica could “[h]arm the community by forcing higher hospital rates on them.” Three months later, Wakeman recommended to St. Luke’s Board that it pursue a merger with ProMedica. The Board accepted the recommendation the same day. Six months later, on May 25, 2010, ProMedica and St. Luke’s signed a merger agreement. D. In July 2010—less than two months after the agreement was signed—the FTC opened an investigation into the merger’s competitive effects. A month later, the FTC and ProMedica No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 6 entered into a “Hold Separate Agreement” that allowed ProMedica to close the deal, but that, during the pendency of the FTC investigation, barred ProMedica from terminating St. Luke’s contracts with MCOs, eliminating or transferring St. Luke’s clinical services, or terminating St. Luke’s employees without cause. With these restrictions in place, ProMedica and St. Luke’s closed the merger deal on August 31, 2010. In January 2011, the FTC filed an administrative complaint against ProMedica. Later that month, the FTC and the state of Ohio filed a separate complaint in federal district court in Toledo, seeking a preliminary injunction that would extend the Hold Separate Agreement pending the outcome of the FTC’s administrative complaint. The district court granted the injunction. Meanwhile, in the administrative proceeding, an ALJ held a hearing that lasted over 30 days and produced more than 8,000 pages of trial testimony and over 2,600 exhibits. In December 2011, the ALJ issued a lengthy written decision. The ALJ found that the merger would “result[] in a tremendous increase in concentration in a market that already was highly concentrated”; that the merger would eliminate competition between ProMedica and St. Luke’s, thereby increasing ProMedica’s bargaining power with MCOs; and that ProMedica would be particularly dominant in southwest Lucas County—an area with a relatively high proportion of privately insured patients. Thus, the ALJ found that the merger would allow ProMedica unilaterally to increase its prices above a competitive level. The ALJ also found that the merger did not create any efficiencies sufficient to offset its anticompetitive effects. Consequently, the ALJ concluded that the merger likely would substantially lessen competition in violation of § 7 of the Clayton Act. As a remedy, the ALJ ordered ProMedica to divest St. Luke’s. ProMedica appealed the ALJ’s decision to the Commission, which found that the merger increased ProMedica’s market share far above the threshold required to create a presumption that the merger would lessen competition. The Commission also found that a large body of other evidence—including documents and testimony from the merging parties themselves, testimony from the MCOs, and expert testimony—confirmed that the merger would have a substantial anticompetitive effect. The Commission therefore affirmed the ALJ’s decision and ordered ProMedica to divest St. Luke’s. No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 7 This petition followed. II. We review the Commission’s legal conclusions de novo, and its factual findings under the substantial-evidence standard. 15 U.S.C. § 21(c); Realcomp II, Ltd. v. FTC, 635 F.3d 815, 823 (6th Cir. 2011). Substantial evidence is evidence that “a reasonable mind might accept as adequate to support a conclusion.” Realcomp II, 635 F.3d at 824 (quoting Universal Camera Corp. v. NLRB, 340 U.S. 474, 477 (1951)). Section 7 of the Clayton Act prohibits mergers “where in any line of commerce . . . the effect of such acquisition may be substantially to lessen competition, or to tend to create a monopoly.” 15 U.S.C. § 18. As its language suggests, Section 7 deals in “probabilities, not certainties.” Brown Shoe Co. v. United States, 370 U.S. 294, 323 (1962). A. “Merger enforcement, like other areas of antitrust, is directed at market power.” FTC v. H.J. Heinz Co., 246 F.3d 708, 713 (D.C. Cir. 2001) (quoting Lawrence A. Sullivan & Warren S. Grimes, The Law of Antitrust § 9.1 at 511 (2000)). Market power is itself a term of art that the Department of Justice’s Horizontal Merger Guidelines (which we consider useful but not binding upon us here) define as the power of “one or more firms to raise price, reduce output, diminish innovation, or otherwise harm consumers as a result of diminished competitive constraints or incentives.” Horizontal Merger Guidelines (2010) (“Merger Guidelines”) § 1 at 2. Often, the first steps in analyzing a merger’s competitive effects are to define the geographic and product markets affected by it. See United States v. Gen. Dynamics Corp., 415 U.S. 486, 510 (1974). Here, the parties agree that the relevant geographic market is Lucas County. The relevant product market or markets, however, are more difficult. The first principle of market definition is substitutability: a relevant product market must “identify a set of products that are reasonably interchangeable[.]” Horizontal Merger Guidelines § 4.1. Chevrolets and Fords might be interchangeable in this sense, but Chevrolets and Lamborghinis are probably not. See 2B Phillip E. Areeda, Herbert Hovenkamp & John L. Solow, Antitrust Law ¶ 533e at 259 (3d ed. 2007). “The general question is whether two products can be used for the same purpose, No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 8 and if so, whether and to what extent purchasers are willing to substitute one for the other.” F.T.C. v. Arch Coal, Inc., 329 F. Supp. 2d 109, 119 (D.D.C. 2004) (quotations omitted). By this measure, each individual medical procedure could give rise to a separate market: “[i]f you need your hip replaced, you can’t decide to have chemotherapy instead.” United States v. Rockford Mem’l Corp., 898 F.2d 1278, 1284 (7th Cir. 1990). But nobody advocates that we analyze the effects of this merger upon hundreds if not thousands of markets for individual procedures; instead, the parties agree that we should “cluster” these markets somehow. The parties disagree, however, on the principles that should govern which services are clustered and which are not. Two theories of clustering are pertinent here. The first—which the FTC advocates and the Commission adopted—is the “administrative-convenience” theory. (A better name might be the “similar-conditions” theory.) This theory holds, in essence, that there is no need to perform separate antitrust analyses for separate product markets when competitive conditions are similar for each. See Emigra Group v. Fragomen, 612 F. Supp. 2d 330, 353 (S.D.N.Y. 2009). In Brown Shoe, for example, the Supreme Court analyzed together the markets for men’s, women’s, and children’s shoes, because the competitive conditions for each of them were similar. 370 U.S. at 327-28. The competitive conditions for hospital services include the barriers to entry for a particular service—e.g., how difficult it might be for a new competitor to buy the equipment and sign up the professionals necessary to offer the service—as well as the hospitals’ respective market shares for the service and the geographic market for the service. See Jonathan B. Baker, The Antitrust Analysis of Hospital Mergers and the Transformation of the Hospital Industry, Law & Contemp. Probs., Spring 1988, at 93, 138; United States v. Long Island Jewish Med. Ctr., 983 F. Supp. 121, 142-43 (E.D.N.Y. 1997). If these conditions are similar for a range of services, then the antitrust analysis should be similar for each of them. Long Island, 983 F.Supp. at 142-43. Thus, if the competitive conditions for, say, secondary inpatient procedures are all reasonably similar, then we can cluster those services when analyzing a merger’s competitive effects. No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 9 Here, the Commission applied this theory to cluster both primary services (but excluding OB, for reasons discussed below) and secondary services for purposes of analyzing the merger’s competitive effects. Substantial evidence supports that demarcation. The respective market shares for each of Lucas County’s four hospital systems (ProMedica, Mercy, UTMC, St. Luke’s) are similar across the range of primary and secondary services. A hospital’s market share for shoulder surgery, for example, is similar to its market share for knee replacements. Barriers to entry are likewise similar across primary and secondary services. So are the services’ respective geographic markets. Thus, the competitive conditions across the markets for primary and secondary services are similar enough to justify clustering those markets when analyzing the merger’s competitive effects. See Emigra Group, 612 F. Supp. 2d at 353. But the same is not true for OB services, whose competitive conditions differ in at least two respects from those for other services. First, before the merger, ProMedica’s market share for OB services (71.2%) was more than half-again greater than its market share for primary and secondary services (46.8%). And the merger would drive ProMedica’s share for OB services even higher, to 80.5%—no small number in this area of the law. Second, and relatedly, before the merger there were only three hospital systems that provided OB services in Lucas County (ProMedica, Mercy, St. Luke’s) rather than four; after the merger, there would be only two. (One might also suspect that the geographic market for OB services is smaller than it is for other primary services—one can drive only so far when the baby is on the way—but the record is not clear on that point.) The Commission therefore flagged OB as a separate relevant market for purposes of analyzing the merger’s competitive effects. For the reasons just stated, substantial evidence supports that decision. Finally, the Commission excluded tertiary services from its analysis of the merger’s competitive effects. The competitive conditions for tertiary services differ from those for primary and secondary services, in part because patients are willing to travel farther for tertiary services (e.g., a liver transplant) than they are for primary or secondary services (e.g., hernia surgery). Indeed, UTMC’s representative testified that, “[f]or the tertiary . . . services, we compete with . . . institutions such as the University of Michigan, the Cleveland Clinic, University Hospital in Cleveland, and the Ohio State University.” The geographic market for tertiary services is therefore larger than the geographic market for primary and secondary No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 10 services. Moreover, the hospitals’ respective market shares for these services are different than their respective shares for primary or secondary services; St. Luke’s market share for tertiary services, for example, is nearly zero. Thus, the competitive conditions for tertiary services differ from those for primary and secondary services. (The same is undisputedly true for quaternary services, which the Commission likewise excluded from its analysis.) To all this ProMedica offers two responses. The first concerns the 2010 Horizontal Merger Guidelines. Section 4 of the Guidelines provides that “[m]arket definition focuses solely on demand substitution factors”—that is, the extent to which consumers regard one product as a substitute for another. And ProMedica points out that the Commission’s use of the administrative-convenience theory (to cluster the markets for primary and secondary services) focuses on market shares and entry conditions—both of which, ProMedica correctly observes, are “supply-side” considerations. (Entry conditions, for example, concern the ease with which new competitors can enter the relevant market and thus augment the supply for a particular product.) Thus, ProMedica concludes, the Commission’s clustering methodology contradicts the Horizontal Merger Guidelines. But ProMedica’s conclusion does not follow. The reference to demand-side considerations in § 4 of the Guidelines concerns the manner in which one defines a relevant market, not the conditions under which one can cluster admittedly different markets when analyzing a merger’s competitive effects. The administrative-convenience theory asks a different question (whether the competitive conditions for two markets are similar enough to analyze them together) than the one answered by § 4 of the Guidelines (how one defines an individual market in the first place). To analogize to a different area of law: ProMedica’s argument is like saying that a district court should not certify a particular class because it includes different plaintiffs. ProMedica’s second response is to offer an altogether different approach to clustering, which in some quarters is known as the “transactional-complements” theory. (Per Orwell’s admonition to use concrete terms instead of vague ones, see Orwell, Politics and the English Language (1946), we call this the “package-deal” theory instead.) The package-deal theory holds that, if “most customers would be willing to pay monopoly prices for the convenience” of No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 11 receiving certain products as a package, then the relevant market for those products is the market for the package as a whole. 2B Areeda, Antitrust Law, ¶ 565c at 408. That is true even though the individual products in the package are not substitutes for each other. Id. For example, in United States v. Grinell Corp., 384 U.S. 563, 572 (1966), the Supreme Court found that the relevant market for a package of centrally monitored alarm services (burglar and fire) was the market for the package as a whole. ProMedica argues that the package-deal theory applies here because MCOs typically bargain for all of a hospital’s services in a single negotiation. That is true enough; but the specific “package” that ProMedica advocates is one comprising not only primary (excluding OB) and secondary services—which everyone agrees should be clustered when analyzing the merger’s competitive effects—but also tertiary and OB services. And that makes the question presented by ProMedica’s argument much narrower. To wit: whether the MCOs are willing to pay a premium to have a package of services that includes tertiary and OB delivered by a single provider. If so, the relevant market is the market for the package as a whole. See 2B Areeda ¶ 565c at 408. But the record makes plain that the MCOs do not demand from each hospital a package of services that includes tertiary and OB. For example, St. Luke’s offers virtually no tertiary services, and yet the MCOs still contract for the services that St. Luke’s does offer. Likewise, UTMC does not offer OB services, and yet the MCOs still contract with UTMC. And as for the hospital systems that do provide all those services—i.e., ProMedica and Mercy—there is no evidence that MCOs are willing to pay a premium to have all of those services delivered by either of those providers in a single package. It is true that MCOs must offer their members (i.e., patients) a network that provides a complete package of hospital services. But the record shows that the MCOs do not need to obtain all of those services from a single provider. There are no market forces that bind primary, secondary, tertiary, and OB services together like a single plywood sheet. In summary, even ProMedica conceded in its answer to the FTC’s complaint that the “more sophisticated and specialized tertiary and quaternary services, such as major surgeries and organ transplants, also are properly excluded from the relevant market[.]” Answer ¶ 13. No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 12 ProMedica was correct to make that concession then, and incorrect to seek to retract it now. The relevant markets, for purposes of analyzing the merger’s competitive effects, are what the Commission says they are: (1) a cluster market of primary (but not OB) and secondary inpatient services (hereafter, the “GAC market”), and (2) a separate market for OB services. B. ProMedica’s next argument is that the Commission relied too heavily on market- concentration data to establish a presumption of anticompetitive harm. Agencies typically use the Herfindahl-Hirschman Index (HHI) to measure market concentration. “The HHI is calculated by summing the squares of the individual firms’ market shares, and thus gives proportionately greater weight to the larger market shares.” Merger Guidelines § 5.3 at 18. Agencies use HHI data to classify markets into three types: “unconcentrated markets,” which have an HHI below 1500; “moderately concentrated markets,” which have an HHI between 1500 and 2500; and “highly concentrated markets,” which have an HHI above 2500. Id. at 19. The Guidelines further provide that “[m]ergers resulting in highly concentrated markets that involve an increase in the HHI of more than 200 points will be presumed to be likely to enhance market power.” Thus, as a general matter, a merger that increases HHI by more than 200 points, to a total number exceeding 2500, is presumptively anticompetitive. Id. § 5.3 at 19; see also, e.g., Heinz, 246 F.3d at 716 (merger that would have increased HHI by 510 points to 5,285 created presumption of anticompetitive effects by a “wide margin”); United States v. H & R Block, Inc., 833 F. Supp. 2d 36, 72 (D.D.C. 2011) (merger that would have increased HHI by approximately 400 points to 4,691 created presumption of anticompetitive effects). The merger here blew through those barriers in spectacular fashion. In the GAC market, the merger would increase the HHI by 1,078 (more than five times the increase necessary to trigger the presumption of illegality) to a total number of 4,391 (almost double the 2,500 threshold for a highly concentrated market). The OB numbers are even worse: the merger would increase HHI by 1,323 points (almost seven times the increase necessary for the presumption of illegality) to a total number of 6,854 (almost triple the threshold for a highly concentrated market). The Commission therefore found the merger to be presumptively illegal. No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 13 ProMedica responds that this sort of analysis—measuring HHI to apply a presumption of illegality—applies only in “coordinated-effects” cases, rather than in “unilateral-effects” ones. And the FTC admittedly challenges the merger only on unilateral-effects grounds here. The two theories are different: the idea behind coordinated effects is that, “where rivals are few, firms will be able to coordinate their behavior, either by overt collusion or implicit understanding in order to restrict output and achieve profits above competitive levels.” H&R Block, 833 F. Supp.2d at 77. A simple example might be parallel pricing by two gas stations located across the street from each other in a remote small town. Unilateral-effects theory, on the other hand, holds that “[t]he elimination of competition between two firms that results from their merger may alone constitute a substantial lessening of competition.” Merger Guidelines § 6 at 20. The most obvious example of this phenomenon is a “merger to monopoly”—e.g., where a market has only two firms, which then merge into one—but unilateral effects “are by no means limited to that case.” Id. The Guidelines also distinguish between unilateral effects for “homogeneous products” and for “differentiated products.” Homogeneous products are indistinguishable from each other—oil, corn, coal—whereas differentiated products are similar enough to compete in a relevant market, but different enough that some customers prefer one product over another. The market for cola products is an example. Here, the relevant markets involve differentiated products: hospitals have different doctors, facilities, and (perhaps above all) locations, which means that some patients prefer certain hospitals over others. “The extent of direct competition between the products sold by the merging parties is central to the evaluation of unilateral effects.” Id. § 6.1. “Direct competition,” in this sense, does not mean merely that products are within a relevant market; instead, it refers to the extent to which consumers regard the products as close substitutes. Thus, unilateral-effects analysis examines whether differentiated products are not merely substitutes for one another, but close substitutes for some fraction of consumers. In the market for upscale sedans, for example, Audi and Jaguar might be closer substitutes for some consumers than Audi and Lincoln are. (For other consumers in the same market–say, consumers who prefer domestic brands—Lincoln and Cadillac might be closer substitutes.) These hierarchies of consumer preference, which are themselves iridescent from consumer to consumer, are critical to unilateral-effects analysis. For No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 14 “[u]nilateral price effects are greater, the more the buyers of products sold by one merging firm consider products sold by the other merging firm to be their next choice.” Id. For a merger to raise concerns about unilateral effects, however, not every consumer in the relevant market must regard the products of the merging firms as her top two choices. Instead, “[s]ubstantial unilateral price elevation post-merger for a product sold by one of the merging firms normally requires that a significant fraction of the customers purchasing that product view products formerly sold by the other merging firm as their next-best choice.” Id. at 20-21. That “significant fraction,” moreover, “need not approach a majority.” Id. at 21. But none of this, in ProMedica’s view, has much to do with market concentration per se. Thus, what the Commission should have focused on, ProMedica says, is the extent to which consumers regard ProMedica as their next-best choice after St. Luke’s, or vice-versa. And ProMedica therefore argues that the Commission was wrong to presume the merger illegal based upon HHI data alone. The argument is one to be taken seriously. The Guidelines themselves state that “[a]gencies rely much more on the value of diverted sales [i.e., in rough terms, the extent to which the products of the merging firms are close substitutes] than on the level of HHI for diagnosing unilateral price effects in markets with differentiated products.” Id. But this case is exceptional in two respects. First, even without conducting a substitutability analysis, the record already shows a strong correlation between ProMedica’s prices—i.e., its ability to impose unilateral price increases—and its market share. Before the merger, ProMedica’s share of the GAC market was 46.8%, followed by Mercy with 28.7%, UTMC with 13%, and St. Luke’s with 11.5%. And ProMedica’s prices were on average 32% higher than Mercy’s, 51% higher than UTMC’s, and 74% higher than St. Luke’s. Thus, in this market, the higher a provider’s market share, the higher its prices. In ProMedica’s case, that fact is not explained by the quality of ProMedica’s services or by its underlying costs. Instead, ProMedica’s prices—already among the highest in the State—are explained by bargaining power. As the Commission explained: “the hospital provider’s bargaining leverage will depend upon how the MCO would fare if its network did not include the hospital provider (and therefore became less attractive to potential members who prefer that provider’s services).” Op. 36. Here, the record makes clear that a No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 15 network which does not include a hospital provider that services almost half the county’s patients in one relevant market, and more than 70% of the county’s patients in another relevant market, would be unattractive to a huge swath of potential members. Thus, the Commission had every reason to conclude that, as ProMedica’s dominance in the relevant markets increases, so does the need for MCOs to include ProMedica in their networks—and thus so too does ProMedica’s leverage in demanding higher rates. The second respect in which this case is exceptional is simply the HHI numbers themselves. Even in unilateral-effects cases, at some point the Commission is entitled to take seriously the alarm sounded by a merger’s HHI data. And here the numbers are in every respect multiples of the numbers necessary for the presumption of illegality. Before the merger, ProMedica already held dominant market shares in the relevant markets, which were themselves already highly concentrated. The merger would drive those numbers even higher—ProMedica’s share of the OB market would top 80%—which makes it extremely likely, as matter of simple mathematics, that a “significant fraction” of St. Luke’s patients viewed ProMedica as a close substitute for services in the relevant markets. On this record, the Commission was entitled to put significant weight upon the market-concentration data standing alone. These two aspects of this case—the strong correlation between market share and price, and the degree to which this merger would further concentrate markets that are already highly concentrated—converge in a manner that fully supports the Commission’s application of a presumption of illegality. What ProMedica overlooks is that the “ultimate inquiry in merger analysis” is not substitutability, but “‘whether the merger is likely to create or enhance market power or facilitate its exercise.’” Carl Shapiro, The 2010 Horizontal Merger Guidelines: From Hedgehog to Fox in Forty Years, 77 Antitrust L.J. 49, 57 (2010) (emphasis added) (quoting U.S. Dep’t of Justice & Fed. Trade Comm’n, Commentary on the Horizontal Merger Guidelines (2006)). Here, as shown above, the correlation between market share and price reflects a correlation between market share and market power; and the HHI data strongly suggest that this merger would enhance ProMedica’s market power even more, to levels rarely tolerated in antitrust law. In the context of this record, therefore, the HHI data speak to our “ultimate inquiry” as directly as an analysis of substitutability would. The Commission was correct to presume the merger substantially anticompetitive. No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 16 C. The remaining question is whether ProMedica has rebutted that presumption. ProMedica argues on several grounds that it has; but more remarkable is what ProMedica does not argue. By way of background, the goal of antitrust law is to enhance consumer welfare. See, e.g., Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 221 (1993); 2B Areeda ¶ 100 at 4 (“the principal objective of antitrust policy is to maximize consumer welfare by encouraging firms to behave competitively”) (cited in Kirtsaeng v. John Wiley & Sons, Inc, 133 S. Ct. 1351, 1363 (2013)); cf. Reiter v. Sonotone Corp., 442 U.S. 330, 343 (1979) (“Congress designed the Sherman Act as a ‘consumer welfare prescription’”) (quoting Bork, The Antitrust Paradox 66 (1978)). And the Merger Guidelines themselves recognize that “a primary benefit of mergers to the economy is their potential to generate significant efficiencies and thus enhance the merged firm’s ability and incentive to compete, which may result in lower prices, improved quality, enhanced service, or new products.” Merger Guidelines § 10 at 29; see also Shapiro, supra at 80 (“Efficiencies generate downward pricing pressure that may outweigh the upward pricing pressure”). Thus, the parties to a merger often seek to overcome a presumption of illegality by arguing that the merger would create efficiencies that enhance consumer welfare. See, e.g., FTC v. Univ. Health, Inc., 938 F.2d 1206, 1222 (11th Cir. 1991). But ProMedica did not even attempt to argue before the Commission, and does not attempt to argue here, that this merger would benefit consumers (as opposed to only the merging parties themselves) in any way. To the contrary, St. Luke’s CEO admitted that a merger with ProMedica might “[h]arm the community by forcing higher rates on them.” The record with respect to the merger’s effect on consumer welfare, therefore, only diminishes ProMedica’s prospects here. That the Commission did not merely rest upon the presumption, but instead discussed a wide range of evidence that buttresses it, makes ProMedica’s task more difficult still. On that score the Commission’s best witnesses were the merging parties themselves. Those witnesses established that ProMedica and St. Luke’s are direct competitors: St. Luke’s CEO testified that ProMedica was St. Luke’s “most significant competitor,” while a ProMedica witness testified that ProMedica viewed St. Luke’s as a “[s]trong competitor”—strong enough that ProMedica offered at least one MCO a 2.5% discount off its rates if the MCO excluded St. Luke’s from its network. St. Luke’s management was also candid about the merger’s potential impacts on its No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 17 prices: its CEO stated that a merger with ProMedica “has the greatest potential for higher hospital rates” and would bring “a lot of negotiating clout.” The parties’ own statements, therefore, tend to confirm the presumption rather than rebut it. The same is true of testimony from the MCO witnesses. Those witnesses testified that a network comprising only Mercy and UTMC—the only other providers who would remain after the merger—would not be commercially viable because it would leave them with a “hole” in the suburbs of southwest Lucas County. (That no MCO has offered such a network during the past decade corroborates the point.) Consequently, the MCO witnesses explained, they would have no walk-away option in post-merger negotiations with ProMedica—and thus little ability to resist ProMedica’s demands for even higher rates. ProMedica responds that this testimony is self- serving, which might well be true (though one might construe ProMedica’s response as an implicit admission of the MCOs’ point). But ProMedica otherwise offers no reason to think the MCOs’ predictions are wrong—and the record offers plenty of reason to think they are right. ProMedica’s task, then, is to overcome not merely the presumption of anticompetitive effects, but also the statements of the merging parties themselves, and the MCOs’ testimony, and ProMedica’s failure to cite any efficiencies that would result from this merger. To that end, ProMedica argues that Mercy, rather than St. Luke’s, is ProMedica’s closest substitute—because Mercy, like ProMedica, offers tertiary services, whereas St. Luke’s does not. But any argument about substitutes must begin with a definition of the relevant market; and ProMedica’s argument is based upon a market definition that we have already rejected. That Mercy offers tertiary services, and St. Luke’s for the most part does not, matters only if the relevant market is one for a primary, secondary, and tertiary services wrapped together in a single package. That is not the relevant market here. See supra at 12-14. Instead, the relevant markets are those for GAC services and OB services, respectively—markets in which the merging parties’ own statements show that ProMedica and St. Luke’s are direct competitors. ProMedica’s argument is meritless. ProMedica also argues that MCOs, rather than patients, are the relevant consumers here, and that the Commission therefore erred by “assess[ing] substitutability from the patients’ perspective.” But this is an argument about semantics. MCOs assemble networks based primarily upon patients’ preferences, not their own; and thus the extent to which an MCO No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 18 regards ProMedica and St. Luke’s as close substitutes depends upon the extent to which the MCO’s members do. Finally, ProMedica argues that St. Luke’s was in such dire financial straits before the merger that it “was not a meaningful competitive constraint on ProMedica.” This argument is known as a “weakened competitor” one, and is itself “probably the weakest ground of all for justifying a merger.” Kaiser Aluminum & Chem. Corp. v. FTC, 652 F.2d 1324, 1339 (7th Cir. 1981). Courts “credit such a defense only in rare cases, when the [acquiring firm] makes a substantial showing that the acquired firm’s weakness, which cannot be resolved by any competitive means, would cause that firm’s market share to reduce to a level that would undermine the government’s prima facie case.” Univ. Health, 938 F.2d at 1221. In other words, this argument is the Hail-Mary pass of presumptively doomed mergers—in this case thrown from ProMedica’s own end zone. The record demonstrates that St. Luke’s market share was increasing prior to the merger; that St. Luke’s had sufficient cash reserves to pay all of its obligations and meet its capital needs without any additional borrowing; and that, according to St. Luke’s CEO, “we can run in the black if activity stays high.” St. Luke’s difficulties before the merger provide no basis to reject the Commission’s findings about the merger’s anticompetitive effects. ProMedica has failed to rebut the presumption that its merger with St. Luke’s would reduce competition in violation of the Clayton Act. We therefore need not address ProMedica’s remaining criticisms of various other evidence that merely buttressed that presumption. D. ProMedica argues that the Commission erred in ordering divestiture as a remedy. We review the Commission’s choice of remedy for abuse of discretion. Jacob Siegel Co. v. FTC, 327 U.S. 608, 611-12 (1946). In doing so, we resolve “all doubts” in the Commission’s favor. United States v. E.I. du Pont de Nemours & Co., 366 U.S. 316, 334 (1961). Once a merger is found illegal, “an undoing of the acquisition is a natural remedy.” Id. at 329. Here, the Commission found that divestiture would be the best means to preserve competition in the relevant markets. The Commission also found that ProMedica’s suggested “conduct remedy”—which would establish, among other things, separate negotiation teams for No. 12- 3583 ProMedica Health Sys., Inc. v. Fed. Trade Comm’n Page 19 ProMedica and St. Luke’s—was disfavored because “there are usually greater long term costs associated with monitoring the efficacy of a conduct remedy than with imposing a structural solution.” And the Commission found no circumstances warranting such a remedy here. We have no basis to dispute any of those findings. The Commission did not abuse its discretion in choosing divestiture as a remedy. * * * The Commission’s analysis of this merger was comprehensive, carefully reasoned, and supported by substantial evidence in the record. The petition is denied.