The Federal Trade Commission and the Pennsylvania Office of Attorney General recently challenged a proposed hospital system merger in the Philadelphia, PA area as a violation of Section 7 of the Clayton Act. The FTC lost this challenge in large part because it failed to convince the Court of the two alleged relevant geographic markets for inpatient general acute care despite apparent agreement between the two sides economists about plausibility of the hypothetical monopolist market definition test constructed by statistical analysis. It appears the Court required that purchaser fact-witnesses “bless” the market boundaries asserted by the Government. Although court insistence on this guiding principle has obvious merit and serves as a check on plaintiff’s(s’) expert’s quantitative findings, the Opinion references multiple “facts” that seem irrelevant to proving an antitrust market or demonstration of likely adverse unilateral effects. This article reviews the publicly available evidence in this matter, and discusses both geographic and other aspects of the case.
By David Eisenstadt & James Langenfeld1
I. INTRODUCTION
On February 27, 2020 the Government (the Federal Trade Commission and the Pennsylvania Office of Attorney General) challenged the proposed merger of Thomas Jefferson University (“TJU”) and Albert Einstein Healthcare Network (“AEHN”) (the “merging parties”) as a violation of Section 7 of the Clayton Act.2 On the same date, the Government filed a motion for Preliminary Injunction (“PI”) with the District Court for the Eastern District of Pennsylvania to temporarily enjoin the merger pending a full-trial on the merits. Six days of PI hearings occurred during October 2020. On December 8, 2020 the Court denied the Government’s motion, finding Plaintiffs had failed to meet their evidentiary burden that they would likely prevail on the merits of their market definition claims during a full trial on the merits. Two weeks later, an Emergency Motion filed by the Government to the 3rd Circuit for a temporary injunction was denied without comment. On January 6, 2021, the FTC voted to remove the matter from Adjudication, including withdrawal of its Part 3 Administrative Complaint.
The FTC lost this challenge in large part because it failed to convince the Court of the two alleged relevant geographic markets for inpatient general acute care (“GAC”) despite apparent agreement between the two sides economists about plausibility of the hypothetical monopolist(s) (“HM”) market definition test constructed with the use of statistical and other analysis. Both alleged markets were rejected on grounds the Government had failed to reconcile statistical evidence of “diversion ratios”3 constructed from patients’ hospital usage patterns with the abilities of commercial health plans to resist and defeat a price increase by contracting with facilities located outside either market. This disconnect caused the Court to question whether the purported HM in each of the two markets would have the ability to successfully leverage its ownership or control of the composite hospitals to profitably impose at least a small, but significant non-transitory increase in price (“SSNIP”) at the merging party hospital around which the HM was constructed.
The Opinion raises questions that may well be the subject of market definition in future merger litigations. Specifically, what exactly is commercial reality, how does it fit into market definition, and in what way does commercial reality relate to quantitative market definition analysis performed by economic experts?
II. FTC’S APPROACH TO MARKET DEFINITION
The Government’s market definition allegations were both similar to and different from those advanced in other recent hospital merger litigations. As with previous challenges, the proposed merger allegedly would reduce competition in the inpatient GAC market. However, for the first time the Government also alleged harm in an inpatient rehabilitation facility (“IRF”) market. As an additional wrinkle, in an apparent effort to comport more closely with the 2010 Horizontal Merger Guidelines (“HMGs”), the Government alleged separate GAC geographic markets around the merging parties’ individual hospitals allegedly most susceptible to a post-transaction price increase.4 Diversion ratio analysis, strategic documents, and payer testimony indicated that Jefferson was a closer competitor to Einstein (i.e. prevented Einstein from unilaterally raising price profitably) than vice-versa. Given this, the Government deemed the different Einstein hospitals were more susceptible to a post-transaction price increase and it constructed individual geographic markets around each.5
The HM around each Einstein hospital included (1) the Jefferson hospital with the highest diversion percentage from that Einstein facility; (2) any non-merging-party hospital for which the diversion percentage from that Einstein hospital equaled or exceeded the diversion percentage to the Jefferson hospital identified in step (1); (3) any non-merging-party hospital with a drive-time from the Einstein hospital that was shorter than the drive-time to any non-merging-party hospital included in step (2); and (4) the aggregate diversion percentage from the Einstein hospital to all the hospitals identified in steps (1) – (3) had to be large enough for the imposition of a profitable SSNIP at the Einstein hospital.6 The criteria employed in steps (2) and (3) necessarily excluded some hospitals from the market simply because they were deemed to be worse substitutes to the target Einstein hospital than the Jefferson hospital that accounted for the largest percentage diversion.
The Government respectively labeled the separate HMs around EMCP, EMCM and EMCEP as the North Philadelphia, Montgomery area, and Philadelphia area markets. The North Philadelphia market centered around EMCP included eleven GAC hospitals, two Einstein facilities (EMCP and EMCEP), two Jefferson hospitals (Jefferson-Abington and Jefferson-Frankford), two Temple University facilities, individual hospitals owned by Prime and Tower Health, and three individual specialty facilities, one pediatric and two that specialized in cancer treatment. In that market, the Government alleged that consummation of the merger would cause the Herfindahl Hirschman Index (“HHI”) to increase by 1,200 points to 4,500.
The Montgomery area market included nine hospitals, EMCM, two Jefferson (Abington and Abington-Lansdale) facilities, two Tower hospitals, two Prime hospitals, and two others. In this market, the Government alleged an HHI increase of 700 points to 3,500. The Philadelphia area (“IRF”) market centered around EMCEP included two Einstein facilities, three Jefferson, an IRF facility owned by the University of Pennsylvania, and another non-party IRF situated at Trinity’s Nazareth Hospital. In this alleged market, the HHI was projected to rise by 2,500 points to 5,900.
In all three markets, the alleged post-merger HHI and increase therein exceeded the standards set forth in the HMGs for a merger to likely enhance market power – a post-merger HHI of 2,500 or greater and an HHI increase of at least 200 points.7
Both parties’ economic experts agreed that the HMs for each of the two GAC markets would be able to profitably impose a SSNIP at the Einstein facility around which the market was constructed.8
III. ANALYSIS OF THE COURT’S OPINION
At first read, the District Court’s decision about geographic market definition appears to be on more solid footing than several prior District Court hospital merger opinions that denied the Government’s Motion for a Preliminary Injunction. The portions of expert and fact witness testimony that pertained to market definition were discussed in greater detail, with the Court making specific findings about the credibility of different fact-witness testimony, all of which suggests its decision would have been difficult to reverse on factual grounds as “clearly erroneous.” Further, both parties conceded the importance of reconciling expert and fact-witness testimony with the Court finding that the Government failed do this.
In particular, the alleged markets were rejected on grounds the Government failed to reconcile with commercial realities its statistical evidence of diversion ratios and the construction of each HM.9 The nature of the evidentiary inconsistencies varied between the two alleged product markets. The GAC geographic markets were rejected because health plan payers that provided “supporting” geographic market definition testimony failed to confirm the alleged market boundaries to the Court’s satisfaction. In rejecting each GAC market, the Court seemed to emphasize that United, Cigna, and IBC, in that order, were insufficiently versed with the contours of either market to validate their existence. In addition, the Court found the veracity of one health plan witness that offered supporting market definition testimony on behalf of the Government was likely influenced by economic concerns unrelated to market definition.
The Court was particularly critical of the live market definition testimony that was offered by two health insurers, Independent Blue Cross (“IBC”) and Cigna. In rejecting IBC’s concerns, the Court noted IBC could not identify how dependent its enrollees were on the merging parties’ hospitals in the Montgomery area market for obstetrics, notwithstanding the same witness testified a contract with the merging parties’ hospitals in this market, for this service, was important. But the Court went beyond the weakness of IBC’s market definition testimony. It enumerated several other confounding problems including IBC’s status as a health plan competitor to the merging parties, its prior contracting behavior with Einstein and Jefferson, its position as the Philadelphia area’s leading health plan, and its professed goal to offer networks with greatest range of provider choice.
Cigna’s concerns were dismissed largely on grounds its witness could not easily embrace the boundaries of the Government’s two GAC markets and the Government’s unilateral effects theory of harm.
The Court was also disturbed by the lack of corroborating testimony from the two other leading health plans, Aetna and United, reducing the weight which it assigned to the testimonies of the other two. Specifically, Aetna had no concerns about the transaction. United did not appear as a hearing witness on behalf of the Government because its witness’s deposition testimony apparently did not affirm the boundaries or the two alleged GAC markets, that it would agree to pay higher rates to the merged firm post-transaction, or that its hospital network would be unmarketable to the residents of either GAC market if Jefferson and Einstein were excluded from its hospital network. In addition, an ordinary course, United analysis mentioned certain non-party hospitals located outside either GAC market as substitutes for EMCP and/or EMCM.
The IRF market was rejected because the payer witnesses that testified lacked experience in negotiating IRF rates, the Government did not prove local access to IRFs was important to employers (it is an infrequently utilized service), or that access to an IRF network was an important factor in their selection of a health plan.
A. Commercial Realities and Market Definition
The Opinion raises at least three sets of questions that may well be the future subject of market definition in merger litigations: First, what exactly does commercial reality mean in the context of market definition analysis, and how should commercial realities overlay with a statistical analysis of market definition performed by the plaintiff’s expert? Second, should supporting testimony rendered by a complaining buyer (i.e. a health plan or third-party payer) be rejected or discounted when the merging parties are direct competitors to the buyer? Third, does the Government’s theory market definition (or unilateral effects) require payers’ unanimous endorsement.10
To an economist, the questions about commercial reality that are pertinent to geographic market definition are pretty straightforward. They are: do buyers (e.g. health plans) agree that access to locations under the control of the putative HM are necessary for the commercial success of a product sold to customers (employers) with significant concentrations of members (employees and/or dependents) that reside in the affected area (e.g. North Philadelphia)? Second and relatedly, does common control of the locations which comprise the HM make the merging-party-hospital around which the market is centered vulnerable to a profitable price increase of at least a SSNIP? If not, as a practical matter, what additional locations should be added to the proposed HM for the assumed price increase at the target-party hospital to be profitable?
The Court does not cite to health plan-witness testimony demonstrating that either GAC HM contained too few competing firms or locations to profitably impose at least a SSNIP at the pertinent Einstein facility. The Court also did not cite payer testimony which explained how an attempted SSNIP by the HM at either target Einstein hospital would be easily defeated, i.e. how the payer would substitute to hospitals not included in the HM to defeat the proposed price increase. The Court also cited no testimony that revealed a GAC hospital network which excluded the facility locations hypothetically owned by either HM could be as effectively marketed as a network that included them.
In rejecting each GAC HM, the commercial realities that the Court did mention seem to pertain more to each merging party’s competitors and not the adequacy or lack thereof of the proposed HM at issue. For example, the Court mentioned a United internal analysis that listed a number of alternative providers to EMCP or EMCM that were not part of the either GAC HM. What the Court did not address is whether these outside hospitals could be substituted for each relevant Einstein facility if either was hypothetically under the control of the (non-Einstein) hospitals that comprised each respective HM. When the HM can successfully condition participation of all facilities under its ownership on the inclusion of each Einstein hospital, substitution away from the Einstein hospital at issue would be impossible because it would mean losing access to all of the facilities under the HM’s control. Finally, while the Court observed United never explicitly recognized the legitimacy of either GAC market, its witness representative apparently never offered testimony that disproved their existence.11
Some of the Court’s grounds for rejecting IBC’s testimony relate to the lack of pre-merger unilateral market power possessed by each merging party. These comments fail to shed light on the ability of each HM to profitably impose at least a SSNIP at the target-Einstein hospital or the merged firm’s ability to raise price after the merger. For example, the Opinion states that IBC leveraged each merging party pre-merger by threatening to exclude it from one or more of its product networks. While this certainly suggests neither party possessed unilateral monopoly power or “must have” status, it does not demonstrate profitable price increases of at least a SSNIP by each HM at the relevant Einstein hospital would be unprofitable.12
The Court also referenced the contemplated marketing of a network by IBC’s parent that excluded both Jefferson facilities and two Main Line hospitals (Paoli and Bryn Mawr) as evidence that a merged Jefferson-Einstein would gain no unilateral market power. At best, the analogy is relevant only to unilateral market power or the lack thereof that would result from other combinations with Jefferson. No discussion is provided about how the combined GAC share of Jefferson-Main Line in their relevant area compares to the combined share of Jefferson-Einstein in either of the two alleged GAC markets. At a minimum, appropriately applying the outcome in the former geography to the latter demands this type of comparison. Relatedly, the comment also does not address the legitimacy of either proposed GAC HM and its ability to impose at least a SSNIP at the targeted-Einstein hospital around which the market was centered.
The Court cited other dynamics to underscore its suspicions about IBC’s true motive for testifying that it was vulnerable to a post-merger price increase. In this regard, the Opinion references two specific “market” facts: IBC’s status as a dominant buyer and its desire to offer broad networks.13 The first fact is used to suggest IBC would be protected against merger-related hospital input price increases because as a buyer it has more bargaining leverage than Jefferson-Einstein would possess as a seller. The second fact implies that even if IBC were susceptible to a post-merger price increase, it could simply exclude the merged firm from its different product networks and avoid adverse impact.
IBC’s apparent position as a dominant buyer does not establish it would be immune from a post-merger price increase. What matters is the change in the relative bargaining positions of IBC and Jefferson and Einstein that results from their merger. Further, IBC’s choice to differentiate its products by offering greater provider choice allows it to sell larger networks (all else equal), potentially at a higher price. From IBC’s perspective, a merger that induces it to shrink its hospital provider network to avoid paying higher prices to Jefferson-Einstein is the equivalent to a quality-adjusted price increase for its products.
The Opinion characterizes Cigna’s witness as being both contradictory and uncertain about exact market boundaries. Its witness testified EMCM, Jefferson-Abington, and Jefferson-Abington Lansdale would alone constitute a valid HM (in which case the Government’s Montgomery Area HM would more than satisfy a SSNIP test), but he was less certain about the boundaries of the North Philadelphia market centered around EMCP. On the other hand, Cigna apparently offered no testimony that indicated it would easily defeat an attempted SSNIP at EMCP or EMCM by either GAC HM by selling a network that included only outside hospitals.
The Court determined Cigna’s sale of health plan services to large employers throughout the Philadelphia area provided probative evidence of hospital geographic market boundaries. The employed work forces of these employers reside throughout the five-county Philadelphia PA area, prompting Cigna to contract with a large number of health systems situated therein to satisfy that wide-area demand. From this observation, the Court’s seemed to conclude the entire metropolitan area is a more appropriate geographic market for hospital services than either one of the two alleged GAC markets.
Cigna’s strategy to offer a hospital network of which contains multiple hospitals and different hospital-access points underscores the different contexts which are used to define markets. From a business and marketing perspective, Cigna’s need to offer a hospital network that spans a wide geography is attributable to the spatial heterogeneity of the employees of its different employer-client accounts. That business imperative does not mean all hospitals in Cigna’s network are equally good substitutes. In fact, Cigna’s need to offer a large panel of hospitals actually demonstrates the opposite; otherwise, employers would be content to purchase a network with many fewer hospital-access points. By contrast, antitrust market definition focuses on the hospital substitution patterns of a defined enrollee/inpatient population, and considers whether employers with, say, a significant concentration of employees residing in the North Philadelphia area would readily switch to a hospital network that excludes the facilities that comprise the putative (North Philadelphia) HM if it attempted to implement a SSNIP at the relevant Einstein hospital.
B. Commercial Realities and Likely Competitive (Adverse Unilateral) Effects
Unilateral effects relate to the merged firm’s ability to profitably raise price at one or more facilities under its ownership.14] Since the Court’s denial of the Government’s Motion seems at least partly based on a lack of showing of adverse unilateral effects, we discuss those commercial realities that would run counter to the Government’s theory of unilateral harm and then discuss several of the Court’s references to this issue.
Commercial realities that would indicate a lack of unilateral effects include credible testimony that buyers would be able to “as effectively” sell a hospital network to those commercial customers whose employees exhibit significant usage of the targeted Einstein facilities if (1) the hospital locations owned by the merged firm which are part of the HM were excluded from the payer’s network; (2) non-merging party hospitals that are part of the HM were included; and (3) additional non-merging party hospitals not part of the HM (“outside hospitals”) were also included. Such evidence would demonstrate a payer could form a substitute hospital network without inclusion of the relevant merging parties’ facilities, which in terms of both enrollment and price, would be as commercially attractive as a network including those merged firm’s facilities.15
This fact pattern cannot be demonstrated simply by evidence the same buyer offered or contemplated offer of a network that excluded the relevant locations of both merging parties. Although this narrower network might well be commercially salable, it could also be inferior to one that included the merging parties’ locations, would have to be sold at a discount to be attractive, and even then, might be purchased by only a subset of the customers that previously bought a network which included them. The key issue is whether these same customers would agree to pay a higher price for network that includes both merging parties’ relevant hospitals than one that excludes them.16
Although the Court certainly recognized the significance of these questions to unilateral effects issues, the testimony referenced in the Opinion does not provide needed answers.17 In rejecting IBC and/or Cigna testimony, the Court made three observations: (1) health plans only need to contract with one of the two merging hospitals or systems; (2) health plan products that excluded the hospitals of both systems were either sold or contemplated for sale; (3) Jefferson (TJUH, Abington and Abington-Lansdale) were not particularly competitively constrained by Einstein, and were more constrained by the two University of Pennsylvania hospitals located in Philadelphia.18
Of these three observations, only the second is likely to be somewhat relevant to unilateral effects, but without more does not demonstrate the equivalence of a network which included both merging parties and one that excluded them. The Court’s first observation, that a successful network could consist of one but not the other merging firm is not on-point to whether a network could be equally successful if both were excluded. The third issue pertains to the apparent asymmetry of unilateral effects, not their absence.
Yet, the Court relied on these facts to conclude there would be adequate hospital competition in the Philadelphia area after the merger. Although the Court’s factual findings fairly demonstrate that the addition of Einstein to Jefferson would not create an absolute GAC monopoly in any market, this high threshold is not the test for showing likely adverse unilateral effects.19
C. Commercial Realities and the IRF Market
It is unclear whether the opposing experts agreed about the composition of the IRF HM. The merging parties contended the Government’s product market inappropriately excluded high-end nursing home facilities that provided rehabilitation care that was comparable to the services offered by the IRF facilities that were part of the HM. On the other hand, it appears that no disagreement existed about the identity of the HM constructed using diversion ratios between different IRFs, or that with a diversion percentage from EMCEP of 18.5 percent, Jefferson-Magee was its most important IRF competitor.20 Less apparent is whether evidence was presented that demonstrated IRF diversion ratios from EMCEP to nursing homes were even larger justifying their inclusion in the relevant market under step (3) of the HM build-up.21
In the alleged IRF market, the Court implied IRF services were rarely used by employees or their dependents, suggesting that employers would be satisfied with networks that included some IRF facilities located anywhere, including rehabilitation hospitals situated outside the locations of the alleged HM. While reference to a market with “thin” demand is germane to measuring potential consumer dollar harm from a proposed transaction, a “size of market” test by itself has no impact on the accuracy or mechanics of product or geographic market definition.22
The Opinion referenced no specific payer testimony that when IRF services were required, employers with significant employee concentrations in the Philadelphia area would be indifferent to an IRF network that included the locations of the IRF HM, one consisting of different facilities. There was also was apparently no testimony that commercial customers would be indifferent to an IRF network that excluded the merging parties’ IRF facilities, and one that included the non-merging party facilities that were part of the HM plus “outside” (not part of the HM) IRF facilities.
D. Dual-Role Complaining Buyers and Lack of Buyer Unanimity
More nuanced considerations apply to the Court’s several observations about the mixed-motives of complaining buyers that compete with the merging firms (“dual-role” purchasers or buyers), and the weight that should be assigned to complaining purchaser testimony when (all) buyers’ opinions about the transaction differ.
-
- Dual Role Buyers
Dual role purchasers that also compete directly with the merging parties might object to a proposed merger either in their position as would-be disadvantaged buyers or disgruntled competitors. In Jefferson-Einstein, the Court noted that IBC’s opposition to all hospital mergers, its significant prior objections to (a) UPMC’s attempted acquisition of Einstein, (b) Jefferson’s proposed Accountable Care Organization with Main Line Health, and (c) Jefferson’s co-ownership with Einstein of a Medicare/Medicaid Advantage Plan that competed with IBC, warranted the classification of disgruntled competitor.23
All of the above could or would have created health plan competition to IBC. UPMC, a health system based in Western Pennsylvania owns no Philadelphia-area hospitals but does operate a health plan which is the most significant competitor to Highmark Blue Cross-Blue Shield (aka Blue Cross-Blue Shield of Western Pennsylvania). Plausibly, IBC feared that UPMC would use its acquisition of Einstein as a toe-hold for its health plan entry in the Philadelphia area. The Court fairly concluded that IBC considers Jefferson a potential competitive threat as an insurer, an eventuality which becomes more likely as Jefferson adds additional owned-facility access points to its current southeast PA-NJ health system.
Ultimately, the Court afforded little weight to IBC’s views about the merger because of Jefferson’s stance as a health plan competitor. Interestingly, neither the testimonies of United nor Cigna were discredited on grounds Jefferson would emerge as a stronger health plan competitor to them. This suggests the Court did not conclude all complaining health plans objected out of concern the merged firm would offer greater competition to them. It is also unclear what specific conclusions the Court drew from Aetna’s lack of any opposition.24
The Court’s dismissal of IBC’s testimony creates an interesting juxtaposition: whether the strong views of complaining “dual-role” buyers are inherently so suspect that their complaints about a proposed hospital merger should be discounted. Accordingly, in future litigations the Government will likely need to flesh out in greater detail the underlying incentives of complaining dual-role health buyers. In that regard, we can think of at least two potential evidence tie-breakers: one simple, yet potentially unreliable, and one quantitatively more precise, but also more difficult to construct.
For some time, economists have noted that the customers who purchase from dual-role complaining purchasers are in a superior position to objectively comment on whether the merger would harm competition.25 In this instance, those customers would be employers that purchase health plan coverage from IBC, especially ones which self-insure and incur the full-pass-through expense of higher hospital claim costs that result from a merger which creates additional negotiating leverage for the merged firm vis-à-vis health plans. Knowledgeable, credible employers that perceive a greater benefit from more health plan competition would support the merger, while those that anticipate higher claims costs because of the merger would oppose it. As far as we can discern, no evidence was offered by employer-customers of IBC which speaks to whether they believed that on balance the proposed transaction would be to their financial benefit or detriment.
The obvious problem with this simple test is employer-customers often do not have first-hand knowledge of merged firm’s position as a would-be health plan competitor. Further, because they purchase (i.e. “rent”) access to hospital networks, often with the assistance of third-party benefit firms, they do not negotiate directly with the merging hospitals and are not intimately familiar with the mechanics of hospital-health plan negotiations. Simply, identifying knowledgeable employers, especially those with significant employee usage of the merging parties’ hospitals that are part of a proposed HM can be a challenge.
The second test would formally compare the profit loss to the health plan from higher hospital input costs (due to the merger) with its profit decrement from more aggressive health plan competition that would be offered by the merged firm. This type of calculation could be performed either by experts or business analysts employed by the complaining health plan. A result that indicates a larger anticipated profit-loss from higher input costs would add credibility to a complaining payer’s position that its opposition was rooted in anticipation of higher input prices.
-
- Lack of Uniformity of Opinion Among Buyers
This has always been a thorny antitrust issue, with critics suggesting that product markets should be defined up-front to include only those purchasers likely to be harmed by the horizontal hospital market power created by the proposed merger.26 In this case, the relevant product market would have included less than all the commercial volume of all four major health plans, or at least their volume minus Aetna since Aetna did not apparently object to the transaction. The advantage to this approach is it limits the Court’s ability to pit the testimony of complaining purchasers against that of non-complaining ones. It also allows the Government to more adequately explain why only some buyers are likely to be adversely affected.
This approach also suffers from limitations. First, the HMGs note that either for reasons of convenience or data reliability antitrust markets are often defined for groups of customers. Second, health plans likely to be harmed by a proposed transaction may choose to not complain for a number of reasons. As such, limiting market definition to only known-complaining buyers is likely to understate the real magnitude of consumer harm.
Preserving the status quo of defining hospital product markets according to large groups of customers (e.g. all commercial health plans), not all of which either object to the transaction or testify, leaves the Court in the delicate position of weighing different customer reactions. Perhaps the Court’s ruling in Jefferson-Einstein will induce the Government in future litigations to at least better explain to the Court why its product market definition is legitimate even though only some customers complain out of fear higher hospital prices.
IV. CONCLUSION
In sum, the details and logic of the Court’s reasoning provide a less-than satisfactory economic basis for its ultimate determination that the Government failed to prove a market. In the Court’s view, purchaser fact-witnesses must “bless” the market boundaries that were established by the Government through quantitative means. Although implementation of this guiding principle has merit and serves as a check on an expert’s quantitative findings, the Opinion also references many multiple “facts” that are irrelevant to proof of an antitrust market or demonstration of likely adverse unilateral effects.27
Witness testimony and documents referenced by the Court as part of its basis for rejecting the findings of the Government’s economic expert on market definition includes (1) conflating facts relevant to unilateral effects with principles of market definition; (2) confusing pre-merger contracting behavior with the purported lack of unilateral effects; (3) failure to consider the potential anticompetitive effects of mergers which occur in product markets populated by buyers (health plans) selling differentiated products (networks); (4) not accounting for the different contexts and purposes (business, antitrust, regulatory, etc.) for which markets are defined; and (5) applying a de minimis threshold of commerce to the process of market definition.28
On the other hand, The Court’s overriding and legitimate observation that fact-testimony should accurately overlay expert statistical analysis, and its admonition that a failure to reconcile the two exposes the Government to judicial determination it did not meet its burden of proof, justifiably highlights the need for thoroughness in the market definition process.
1 The authors are Senior Managing Directors at Ankura Consulting Inc. and former employees of the U.S. Department of Justice Antitrust Division and the Federal Trade Commission. The opinions expressed are ones held by the authors and not those of any organization.
2 Collectively, the two sets of litigants are referred to as “the parties.” TJU owned eleven GAC hospitals in the Philadelphia metropolitan area (Southeastern PA and Southeast NJ), three of which were alleged to be significant and direct competitors to two Einstein GAC hospitals, Einstein Medical Center Philadelphia (“EMCP”) and Einstein Medical Center Montgomery (“EMCM”). The three Jefferson facilities were Jefferson-Abington and Jefferson-Frankford, alleged to be direct competitors to EMCP, and Jefferson-Abington and Jefferson Abington-Lansdale, alleged to be direct competitors to EMCM. The parties also owned directly competing inpatient rehabilitation facilities (“IRF”), most notably the MossRehab facility owned by and located at Einstein Medical Center Elkins Park (“EMCEP”) and Jefferson Magee. MossRehab at EMCEP allegedly significantly competed with Jefferson-Magee, a freestanding IRF in Philadelphia as well as Jefferson-Abington’s IRF unit. MossRehab had additionally located IRF beds at EMCP, and at Jefferson-Frankford and Jefferson-Bucks, two Jefferson-owned hospitals that had contracted with MossRehab prior to their acquisition by Jefferson in 2016. EMCEP also provided GAC services.
3 Section 6.1 of the 2010 Department of Justice-FTC Horizontal Merger Guidelines defines a diversion ratio as “the fraction of unit sales lost by the first product due to an increase in its price that would be diverted to the second product.” In practice, diversion ratios between two products are computed with information of triggering events other than a price increase such as a quality decline/increase or closure/unavailability of the first product.
4 HMG’s §4.1. In recent prior litigated hospital mergers involving multi-hospital systems situated in highly urban areas a single geographic market was defined for affected facilities of the two merging parties. See, for example, FTC v. Advocate Health Care Network, 841 F.3d 460,475 (7th Cir. 2016).
5 In its Complaint for a PI, the Government cites to Einstein facility-to-Jefferson diversion ratios and vice versa. The diversion percentages from EMCP or EMCM to Jefferson were at least several times larger than the diversion percentages from Jefferson-Abington or Jefferson-Abington-Lansdale to Einstein. In the alleged IRF market, the Government’s diversion ratio analysis indicated that Jefferson was roughly twice the competitive constraint on Einstein Moss-Rehab than vice-versa.
6 A SSNIP is typically considered to be a profitable price increase of at least 5 percent imposed by the HM at the hospital location of the merging party around which the market is “centered” (although the HMGs formally apply the test to include a SSNIP at one or more of the merging firms’ locations that comprise the HM; HMGs §4.2.1). If the candidate HM does not satisfy the SSNIP test, of the non-merging-party hospitals not part of the putative HM, the one that functions as the next best substitute to the hospital around which the market is defined is added and so on. Assuming linear demand, a contribution margin of 50 percent, and application of a generalized upward pricing pressure (“GUPPI”) formula which measures first-order price effects of a horizontal merger, an aggregate diversion ratio from the target hospital of 20 percent would be sufficient for a SSNIP of 5 percent (=20 percent x 50 percent x 0.5). The Government alleged that the aggregate diversion percentage in each alleged market was at least several times larger than that necessary for the threshold 5 percent SSNIP, arguably providing statistical confirmation each HM had been properly constructed.
7 HMGs §5.3.
8 The Court did not describe other areas of agreement or disagreements between the parties’ experts on the mechanics used to construct each GAC HM.
9 The Court also criticized the Government’s economic expert for failing to reconcile the two. In response, the Government claimed the reconciliation process had in fact been performed. In conducting this task, the Government’s expert relied in part upon the content of declarations and deposition testimony rendered by payer representatives. However, the Court placed greater weight on live-payer testimony, which apparently was more equivocal, leading the Court to conclude the expert had insufficient basis to claim payer testimony supported his findings.
10 Since the issue of “complaining buyers” that compete with the merging parties in the buyer’s market is also pertinent to purchaser concerns about unilateral effects, this issue is addressed below in sub-section D.
11 This point is relevant because buyers may not be sufficiently facile with how a proposed HM is constructed to form an accurate business judgment about its legitimacy.
12 Along those same lines, the Court made reference to Jefferson documents that identify hospitals other than Einstein facilities as closer competitors, a not surprising fact given local perceptions of the two merging parties and the nature of hospital competition in the Philadelphia area. This fact is pertinent to Einstein’s apparent lesser ability to constrain Jefferson pre-merger, it does not necessarily speak to payers’ abilities to use outside hospitals as leverage to successfully discipline either HM if it attempted to impose at least a SSNIP at the relevant Einstein hospital.
13 The Court also referenced IBC’s apparent dislike of all hospital mergers.
14 While some case law precedent supports issuance of a preliminary injunction only with likely proof of a relevant market and the showing of a significant concentration increase therein, internal Agency review criteria and evidence presented by the Government in recent litigated hospital merger go beyond this threshold. The market definition process contained in the HMGs is used by the FTC and DOJ as an initial screen to identify mergers likely to raise competitive concerns. The second step of the internal review process is a more detailed consideration of competitive effects, including the extent to which the merged firm would gain unilateral market power manifesting in a price increase or quality decrease at one or more of its locations.
The threshold price increase which would justify a concern about unilateral effects is less than a SSNIP, although to be sure, many antitrust practitioners suspect both federal antitrust agencies exercise prosecutorial discretion and will challenge only proposed mergers likely to raise price by at least 5 percent at one or more locations owned by the merged entity. In the absence of offsetting efficiencies (of which in Jefferson-Einstein there either were none offered as an affirmative defense or they were disregarded for purpose of the PI) the threshold for a tolerable merger-related price increase is less than one-tenth of one percent (specifically, 5%/7,500: x/200; x = 0.06 percent(7,500 is the change in HHI of a merger to monopoly in a highly concentrated market with an HHI of 2500; 200 is the threshold change in the HHI for a presumptively anticompetitive merger under the HMGs).
15 Consider a proposed HM containing five hospitals A, B, C, D and E, where hospitals A and B are respectively owned by each merging party. The relevant thought experiment is whether a network containing C, D, E, and other non-merging hospitals F, G, and H that is deemed to be the next-best-alternative network to A-E would be as commercially successful if it sold at the same price. If so, an equally attractive network can be constructed without the inclusion of A and B. The greater the number of “equivalent” (in terms of enrollment) networks to A-E that can be constructed by excluding the merging parties’ hospitals and sold for the same price as A-E, the less likely the merger creates any unilateral market power. Said differently, to achieve the same level of enrollment, if network C-H referenced above must be sold at a 20 percent discount relative to A-E, then F, G and H combined are not equivalent substitutes for A and B.
In addition, assume that two networks are also available pre-merger [A, C-H] and [B, C-H]. Each would achieve same enrollment as network A-E provided each was discounted by 10 percent relative to the price of a potential A-E network. A merger of A and B, coupled with its practice of all or nothing contracting post-merger, and normalizing the price of the [A, B, C, D, E] network to “one,” would cause network prices to increase by 11 percent (=1.0/0.9 – 1).
16 An alternative way to frame the issue is to ask how much less enrollment (if any) would be realized by a network which excluded the merging parties’ relevant hospitals and was sold at the same price as one which included them?
17 The Court recognized that, unlike other geographies with embedded natural experiments that measure measurement or assessment of these different issues, none of the episodic historical hospital-payer contracting behavior in the Philadelphia area provided any meaningful guidance.
18 A Bucks County hospital owned by Trinity (St. Mary’s) and a Montgomery County hospital owned by Holy Redeemer.
19 The Court found the Government offered no employer testimony that it would be difficult to market a health plan to employees which excluded Jefferson and Einstein. This demonstrates the difficulty of identifying knowledgeable employers to opine about unilateral effects when a proposed merger creates “some” additional bargaining leverage, but not nearly enough to confer monopoly power or “must have” status on the merged firm.
While interesting, the Court’s reference to an employer located in a geography outside either proposed GAC market (the lower Merion School District in Montgomery County) which indicated it would be “fine” with a health plan that excluded Jefferson and Einstein seems a bit forced. More pertinent would have been evidence of employers with significant (by percentage) usage of EMCM or EMCP by their employees that would have been indifferent to a health plan that excluded both Einstein and Jefferson. The Court did not cite to such evidence.
20 Of additional note is the reference in the Complaint to MossRehab as an important IRF competitor to Jefferson.
21 It is also unclear whether formal measurement of rehabilitation diversion percentages from EMCEP to nursing homes was possible given available patient-origin data.
22 In its Emergency Appeal to the 3rd Circuit, the Government asserted that the infrequency with which IRF services are demanded did not justify its dismissal by the Court as a relevant product.
23 In a two-step process, Jefferson would acquire Einstein’s ownership share in Health Plan Partners Inc. as part of their proposed merger and then purchase the remaining 50 percent interest which is owned by Temple University.
24 When hospitals vertically integrate into the health plan market, their entry often manifests as one or more joint ventures with incumbent health plans. In that regard, the partnering health plans may choose not to oppose the proposed merger. While we have no idea whether this scenario applies to Aetna’s lack of stated opposition, in other markets Aetna has been known to joint venture with hospital systems to create managed care products which would compete against with offered by the Blue Cross-Blue Shield plan that operates in the same area.
25 David T. Scheffman & Richard S. Higgins, Vertical Mergers: Theory and Policy, George Mason Law Review, 2004, pp. 967-977, e.g. 975.
26 This is discussed in the HMGs §4.1.4 Product Market Definition with Targeted Customers.
27 It is also possible of course that the Court’s misapplication of these different facts to market definition or unilateral effects reflects its confusion resulting from the Government’s failure to carefully explain how different fact-witness statements or testimony were relevant or irrelevant to these different issues.
28 This last point was raised by the Government in its Emergency Appeal to the 3rd Circuit.
