Antitrust Law

Professor Sandoval

Santa Clara University Law School

Submitted by

Richard B. Fox

May 19, 2008


I………. Introduction ……………………………………………………………………………………………. 1


A…….. The Relevant Market Depends On Both the Product Market and the Geographic Market Definition……………………………………………………………………………………………………… 2

B…….. The Product Market Definition…………………………………………………………………….. 3

C…….. The Geographic Market Definition……………………………………………………………….. 5


A.. The Horizontal Merger Guidelines and the Critical Loss Test………………………………. 8

B…….. Limitations of the Critical Loss Test…………………………………………………………….. 12

1……… Perfect competition: marginal cost equal average cost………………………………….. 12

2……… Use of average variable cost instead of marginal cost favors a broader market.. 13

3……… Direct evidence of elasticity is usually not available and indirect evidence may not be relevant or reliable…………………………………………………………………………………………………… 14

4……… The Cellophane Fallacy……………………………………………………………………………. 16

5……… The kinked demand curve problem……………………………………………………………. 17

C…….. Limitations of the Critical Loss Test in Inpatient Hospital Markets……………….. 18

1……… Consumers Are Not Informed as to Prices of Healthcare Services………………… 18

2……… Price inelasticity of healthcare markets invalidates critical loss analysis………….. 19

D…….. The Elzinga-Hogarty Test……………………………………………………………………………. 20

E…….. Limitations of the Elzinga-Hogarty Test in General……………………………………… 22

F……… Limitations of the Elzinga-Hogarty Test in Inpatient Hospital Markets…………. 24

G…….. Fallacies Common To Both Elzinga-Hogarty and Critical Loss Analysis……….. 25

1……… The ‘Silent Majority’ Fallacy…………………………………………………………………….. 25

2……… The Payer Fallacy……………………………………………………………………………………. 27

3……… The Acuity Fallacy………………………………………………………………………………….. 28

4……… The Complexity Fallacy…………………………………………………………………………… 31

5……… The residential zip code fallacy…………………………………………………………………. 33


A…….. Regulatory Barriers to Entry in the Pediatric ICU Market…………………………… 33

B…….. Regulatory Barriers to Pediatric ICU “Shipments”………………………………………. 34

V…….. Ineffectiveness of the Horizontal Guidelines as an Antitrust Enforcement Tool in Hospital Merger Cases………………………….. 35

A…….. In re Adventist Health Sys.……………………………………………………………………………. 37

B…….. California v. Sutter Health Sys.……………………………………………………………………… 41

VI……. SUMMARY AND CONCLUSION………………………………………………………………….. 44



The thesis of this paper is that the standard tools used to define geographic market in antitrust cases, such as the SSNIP Test, the Critical Loss Test, and the Elzinga-Hogarty Test, are invalid when applied to a highly differentiated and price inelastic product such as hospital services in general, including a highly specialized service like pediatric intensive care services in particular.  This thesis will be discussed by examining previous judicial opinions, the writings of antitrust economists, and by examining primary data publically available from California’s hospital planning agency, the California Office of Statewide Health Planning and Development.  The paper concludes that econometric tests, such as SSNIP, Critical Loss, and Elzinga-Hogarty, are not applicable to relevant geographic market definition in markets, such as health care markets, comprised of large numbers of highly differentiated products and/or services, many of which comprise relevant sub-markets in their own right, and are not substitutable for one another.  For this reason, lower courts that have used these tests will have to return to the standard set forth by the Supreme Court in Tampa Electric Co., examination of the individual and peculiar facts of each case without the benefit of a universally applicable rule that fits all cases.

Antitrust cases are governed by the Sherman Act of 1890, 15 U.S.C. § 1 et seq.[1]

  Section one of the Sherman Act makes it unlawful to conspire to restrain trade in interstate commerce:

Every contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade or commerce among the several States, or with foreign nations, is declared to be illegal.


            However, not all restraints of interstate commerce are unlawful, only those that unreasonably restrain trade.[2]  The factors to be considered under the Rule of Reason were set forth in the Chicago Board of Trade case:

The true test of legality is whether the restraint imposed is such as merely regulates and perhaps thereby promotes competition or whether it is such as may suppress or even destroy competition. To determine that question the court must ordinarily consider the facts peculiar to the business to which the restraint is applied; its condition before and after the restraint was imposed; the nature of the restraint and its effect, actual or probable. The history of the restraint, the evil believed to exist, the reason for adopting the particular remedy, the purpose or end sought to be attained, are all relevant facts.[3]


            Factors held to be important in the Chicago Board of Trade case in determining whether there has been an unlawful restraint of trade include “the facts peculiar to the business.”[4]  Here, “business” refers to the relevant market or line of commerce.  In most cases, the relevant market has both a product definition dimension and a geographic dimension.[5]  In healthcare cases, both dimensions may prove challenging for the reasons discussed below.


  A.  The Relevant Market Depends On Both the Product Market and the Geographic Market

The joint Horizontal Merger Guidelines of the Antitrust Division of the U.S. Department of Justice and the Federal Trade Commission generally define the relevant market as having both a product dimension and a geographic dimension.[6]  The product market is generally determined first.[7]

  B. The Product Market Definition

           The product market includes those goods or services that are reasonably interchangeable and cross-elastic with one another.[8]  In each case, the market must be defined based on the “facts peculiar to the case.[9]

            In cases involving physicians’ services, the “facts peculiar” to product definition are generally based on the medical specialty of the plaintiff physician.  In Jefferson Parish Hospital Dist. No. 2 v. Hyde, a case brought by an anesthesiologist excluded from a hospital by an exclusive contract, the Supreme Court held that the product market was “anesthesiological services.”[10] 

            In Summit Health, Ltd. v. Pinhas, a case brought by an excluded ophthalmologist, the Supreme Court accepted, based on the pleadings, that the market was “ophthalmological services.”[11] 

            In Bhan v. NME Hospitals, Inc., the issue was whether a nurse anesthetist competes in the same market as physician anesthesiologists.[12]  The Ninth Circuit held that the test was whether the nurse anesthetist’s services were reasonably interchangeable with those of M.D. anesthesiologists.  Id.  In Oltz, another case brought by an excluded nurse anesthetist, the Ninth Circuit agreed with the district court that the relevant product was “anesthesia services.”[13] 

            Thus, in these cases involving medical professional services, the question of product definition requires a determination of reasonable interchangeability and cross-elasticity of demand.  Some clinical services are clearly different product markets, like orthopedics and cancer treatment.[14]  Other are not, such as M.D. anesthesiology services and nurse anesthetist services.[15] 

            While hospital and physician services are generally classified for product definition purposes by clinical specialty, they can also be defined by acuity. That is, how urgent is the need for the service? In the Chattanooga hospital merger, the Court observed that, “Going to another city is out of the question in medical emergencies…”[16]  Saying the same thing but looking at it from the other end of the acuity spectrum, the Court in the Freeman Hospital merger held that elective procedures may constitute a separate product submarket.[17]

  C.  The Geographic Market Definition

           While the product market definition is seldom difficult in medical cases, the geographic market definition often is.  As in the product market definition, geographic market definition hinges on practical considerations.[18]  These are generally issues for the trier of fact.[19] 

            In Jefferson Parish, the allegation was that an exclusive contract between a hospital and a group of anesthesiologists damaged competition by denying patients living in Jefferson Parish the opportunity to choose an anesthesiologist who was not a member of the exclusive group. [20]  However, the Supreme Court found that “[t]here are at least 20 hospitals in the New Orleans metropolitan area and about 70 per cent of the patients living in Jefferson Parish go to hospitals other than East Jefferson [Hospital]”).  Id.  The Court also noted the District Court’s finding that, “As pointed out by plaintiff, the majority of surgeons have privileges at more than one hospital in the area. They have the option of admitting their patients to another hospital where they can select the anesthesiologist of their choice. Similarly a patient can go to another hospital if he is not satisfied with the physicians available at East Jefferson.”[21]  From this the Court concluded that, “[there is] insufficient evidence in this record to provide a basis for finding that the Roux contract, as it actually operates in the market, has unreasonably restrained competition…[since] there is no evidence that any patient who was sophisticated enough to know the difference between two anesthesiologists was not also able to go to a hospital that would provide him with the anesthesiologist of his choice.”[22] 

            By way of contrast, in the Oltz case, the Ninth Circuit found an unreasonable restraint of trade in anesthesia services in the geographic market of Helena, Montana, “one of Montana’s smaller cities” where there was only one general hospital:

The conclusion that Helena was the relevant geographic market for assessing such harm is inescapable. St. Peter’s enjoyed the overwhelming majority of the market for general surgery. As a result, an anesthesia service provider desiring to serve that market had to work at St. Peter’s. More importantly, there was no evidence that patients could effectively turn outside of St. Peter’s for alternate sources of anesthesia services. There was also no evidence of any other service that was a reasonable substitute for anesthesia services in terms of use and cross- elasticity of demand. Accordingly, we conclude that anesthesia services and the Helena area framed the appropriate product and geographic components of a relevant market in which the jury could assess injury to competition.[23]


            The Oltz case underscores that importance of looking at, from a practical point of view, what are the consumers’ alternatives?  In Oltz, the court defined the geographic and product markets practically by its conclusion that consumers could not “effectively turn outside of St. Peter’s for alternate sources of anesthesia services,” nor was there any “reasonable substitute for anesthesia services in terms of use and cross- elasticity of demand.” 

            However, in another nurse anesthetist case, Bhan II, the Court held that the plaintiff had failed to establish a restraint in a relevant market.  Bhan alleged that the geographic market was limited to the town of Manteca, California, because the hospital required its providers to live in Manteca.[24]  But the Court held that the hospital’s residency requirement did not constrain competition in the market for anesthesia providers because, given time, the hospital could recruit providers to Manteca from well beyond that area.[25]  However, just because the hospital could, given enough time, recruit other anesthesia providers to Manteca does not mean that consumers could, as a practical matter, turn to those services in a time of need.


In earlier antitrust cases the courts considered many different “facts peculiar to the case” in determining geographic markets.[26]  However, in 1982 the Antitrust Division of the U.S. Department of Justice set forth its Merger Guidelines and, on the same date, the Federal Trade Commission set forth its Statement Concerning Horizontal Mergers, which gave considerable weight to the Department’s Merger Guidelines.[27]   The Guidelines aimed to provide competitors with objective and uniform criteria that they could use to foretell whether particular combinations would be opposed by the federal government.  The heart of the Guidelines was the Hypothetical Monopolist Test, a test which asked a very simple question, will the proposed merger result in a price increase in the relevant market?[28]

   A.  The Horizontal Merger Guidelines and the Critical Loss Test

          Under the Horizontal Merger Guidelines, the definition of the relevant product and geographic markets is based on the Hypothetical Monopolist Test.  This test has been reduced to a more precise and mathematical test known as the “Critical Loss Test.” 

            As applied to the product market, the Critical Loss Test defines the relevant product market to be the narrowest market in which a hypothetical monopolist would be able to profitably impose a small but significant and nontransitory increase in price (“SSNIP”), typically assumed to be a 5% increase.[29]   If consumers have no practical alternative products to which they can turn, then they will have to pay the higher price and the monopolist’s profits will increase.  In this case, the Critical Loss Test holds that the relevant product market has been defined.[30]

            However, if consumers facing such a price increase could shift their purchases to other practical substitutes, then the hypothetical monopolist’s profits will fall if enough consumers turn to other products such that the lost sales more than offset the additional profits from the increased price.[31]  The amount of lost sales that suffices to make the price increase unprofitable is the “critical loss.” Id.  The actual loss that results from a price increase is compared to the critical loss.  If the actual loss is less than the critical loss, then this defines a relevant market.  Id.  If the actual loss is greater than the critical loss, then this shows that the hypothetical monopolist does not operate in a relevant market.  In this case, the Critical Loss Test holds that the product market has been drawn too narrowly and should be expanded to include the next closest alternative product or products.[32]  Then the analysis is repeated in ever expanding circles until the hypothetical monopolist in this market can profitably impose a SSNIP.[33]  This is then the relevant product market under the Critical Loss Test.[34] 

            Essentially the same methodology is applied to determine geographic market under the Critical Loss Test.[35]  The narrowest geographic market that includes the suppliers at issue is drawn and then a SSNIP is applied by all suppliers within that geographic area.  Id.  If enough consumers leave that geographic area to make their purchases outside of it such that the suppliers within that geographic area see a decline in their profits, then the market is too narrowly drawn and must be expanded.  Id. The same analysis is repeated, expanding the market as many times as necessary until a SSNIP in that market becomes profitable.[36] 

            An important assumption that underlies the Critical Loss Test is that there are alternative products and/or suppliers available in the market to impose price discipline on the other products and/or suppliers.[37]  If the alternative products and/or suppliers are too small or lack capacity to expand output in response to a SSNIP, then they will not restrain price increases by more dominant products and/or suppliers.  Id.

            In practice, the Critical Loss Test is applied in a three-stage process.[38]  First, the profit margin must be determined based, not on average profits overall, but on the profits gained or lost when a SSNIP is imposed and some customers are lost, sometimes termed the contribution margin.[39]  That is, the contribution margin is then marginal revenue less marginal costs divided by marginal revenue.  Id.  From this measure of marginal profits, the number of customers who would have to leave in order to make a SSNIP unprofitable, the “critical loss,” as a percentage of sales, can be calculated as: 

            x=[Y/(Y + CM)] *100,

where Y is defined as the percentage increase in price over the premerger level and CM is defined as the contribution margin (i.e., the extent to which price exceeds the cost, measured as a percentage of the price).  Id.


            The second step in this process is to determine the actual loss of customers due to the SSNIP.[40]  This requires a determination of the elasticity of demand, reflecting those customers who would not purchase the product at all at the higher price, or the cross-elasticity of demand, reflecting those customers who would purchase the product from another supplier. Id.   Elasticity (ε), then, is the change in quantity demanded divided by the change in price:

            ε = ΔQ/ ΔP.[41]

            The third step in the critical loss analysis is to compare the calculated critical loss from step one with the observed SSNIP-induced loss from step two.[42]  If the latter exceeds the former, then the SSNIP is unprofitable, indicating that the posited market is too narrow and should be expanded.  Id.

  B.  Limitations of the Critical Loss Test

    1.  Perfect Competition: Marginal Cost Equal Average Cost

          In conditions of perfect competition, all suppliers operate at the point where marginal revenue equals marginal cost.[43]  In this case, marginal profit is zero and the critical loss in 100%.  Id.  That is, the hypothetical monopolist would have to lose all of its sales for there to be no increase in profit by increasing price.  Id.  Thus, when markets are perfectly, or at least highly, competitive, the critical loss is very large.[44]  A large critical loss defines a narrow market.  Id.  Thus, a party arguing in favor of a merger or other restraint of trade because the market is both very competitive and also very broad may be advancing an inherently contradictory argument. Id.

    2.  Use of Average Cost Instead of Marginal Cost Favors a Broader Market

           Marginal cost data is seldom available from the usual business records.[45]  Thus, average variable cost is often used instead.  Id.  However, marginal cost is typically greater than average variable cost because the cost of inputs, such as skilled labor, usually rise with quantity produced.[46]  Furthermore, the incremental costs of selling rise since each additional sale requires more effort than the last.[47]  When average variable cost is substituted for marginal cost, the profit margin is biased upward and the critical loss is biased downward.[48]  This biases the analysis in favor of an overly broad market, thus favoring the finding of an absence of anti-competitive effect.  Id.

3.  Direct Evidence of Elasticity Is Usually Not Available and Indirect Evidence May Not Be Relevant of Reliable

            Elasticity of demand is based on the change in quantity demanded divided by the change in price:

            ε = ΔQ/ ΔP.[49]

            There are two schools of thought as to how to get elasticity data, the direct and the indirect approaches.[50]  In the direct approach, the best source of elasticity information is deemed to actual experience, such as when a competitor increases or decreases price and holds that change for a non-transitory period.[51]  The change in quantity sold as a result can then be used to determine elasticity.  Id. 

            The second approach to estimating elasticity is the infer elasticity from firm pricing policy.[52]  In this approach, elasticity is assumed to be inversely rated to profit margins.[53]  This relationship is represented by the Lerner Equation.[54]  This equation posits that, if a product is priced so as to maximize the profits from that product, then ε is equal to the inverse of gross margin (m):  ε = 1/m.  Id.  In plain language, if a firm chooses a high margin on its product, it must think that the demand for the product is not very elastic.  Id.

            If ε can be thus determined, can actual loss be calculated and compared to critical loss under this method?  There are two schools of thought.  The Revealed Preference approach holds that the Lerner equation is valid because the firm is in the best position to know the demand elasticity for its own product and sets it price accordingly.[55]  The Direct Evidence approach holds that actual pricing decisions reflect many factors other than perceived elasticity, including spillover to sales of complementary or substitute products, customer loyalty, reputation, learning-curve effects (building market share), network effects, and, in oligopolistic markets, competitor responses.[56]  The first approach permits a much simpler critical loss analysis but only by ignoring the real world factors raised in the second, raising serious issues of relevance and reliability.[57]

    4.  The ‘Cellophane Fallacy’

           The market defined by critical loss analysis is sensitive to market power at the beginning point of the analysis.[58]  For example, in the case of a monopolist maintaining price at the profit-maximizing level, any further SSNIP will be unprofitable.  Id.  Thus, the critical loss is zero, implying a broader relevant market when, in fact, the monopolist is already operating in a relevant market.  Id.  This illustrates that critical loss analysis is only valid where the SSNIP is added to a competitive price, not a price reflecting market power.  Id.  The possibility of erroneous results in the presence of market power is called the Cellophane fallacy since it refers to the failure of the Supreme Court to find a monopoly in the Cellophane case when there actually was one.[59]

    5.  The Kinked Demand Curve Problem

           In the usual critical loss analysis, the SSNIP is assumed to be 5%, although other values can be used.[60]  When only one value for SSNIP is used it is possible to reach an erroneous conclusion if the demand elasticity curve is nonlinear.[61]  For example, it could be the case that 10% of the purchasers in the market are price sensitive and would shift their purchases to other suppliers in the face of a 5% SSNIP.  Id.  If the marginal profit rate is more than 45%, then the SSNIP will be unprofitable.  Id.  But if the other 90% are price insensitive, then a larger SSNIP, such as 10%, will not cause any other purchasers to shift their purchases to other suppliers and will be profitable.  Id.

  C. Limitations of the Critical Loss Test In Inpatient Healthcare Markets

    1.  Consumers Are Not Informed As To Prices of Healthcare Services

The Critical Loss Test assumes that consumers will re-direct their purchases in response to a SSNIP.  Id.  This, in turn, assumes that consumers know what prices they are paying and that they are the ones paying them.[62]  However, this assumption breaks down in healthcare markets, such as hospital and physician services where third party payers pay most of the charges.  Id.   Pricing of hospital services  is complicated by several factors, including the fact that hospitals have many different prices for different purchasers and the service definitions tend to vary between hospitals.[63]

    2.  Price Inelasticity of Healthcare Markets Invalidates Critical Loss Analysis

Markets for hospital and physician services tend to be price inelastic because, as discussed above, consumers are uninformed about price and because, as discussed below, the costs are born, not directly by the consumers, but by third party payers (see the “Payer Fallacy” below).  As shown above, where elasticity is quite low or non-existent, there will never be an actual loss of profit due to a SSNIP, the actual loss can never exceed the critical loss, and, thus, the critical loss test is of no utility in defining relevant market.  The implication, then, is that these markets are not defined by price but by other factors.[64]

D.  The Elzinga-Hogarty Test

The Elzinga-Hogarty Test is another market definition test, first introduced in 1973,[65] that predates the Guidelines and its Hypothetical Monopolist Test, first introduced in 1982.[66]  Elzinga-Hogarty analysis of a geographic market produces two statistics, LIFO (little in from outside) and LOFI (little out from inside).[67]  Both require that the relevant product market be identified first.  Id.  LIFO represents the absence of importation of the product into the hypothetical market, meaning that there is little external supply to constrain price within that market.  Id.  LOFI means that there is little external demand, abeyance of which in response to rising prices could also constrain price within the market.  Id.  Thus, in such a market, price is not constrained by external forces.  Id.

After studying in detail the regional market for steam coal as purchased by electric utility companies in the continental United States, Elzinga and Hogarty posited that an average of these two numbers equaling 90% delineated a “strong” market for this product by these customers and that a average of 75% delineated a “weak” market.[68]  However, this determination rested, not on price data, but on shipment data.  Id.  Indeed, that was felt to be an advantage of the method since shipment data was felt to be more readily available than price data.[69]

  E.  Limitations of the Elzinga-Hogarty Test in General

           The Elzinga-Hogarty analysis purports to identify “strong” and “weak” markets and to distinguish them from non-markets using the LIFO and LOFI criteria, where 75% identifies a “weak” market and 90% identifies a “strong” market.55  These thresholds were developed for analysis of the coal industry, an industry with a very uniform and fungible commodity-type product. Id.  However, Elzinga and Hogarty recognized, in their initial description of the method, that these thresholds were arbitrary and might have to be varied depending on the circumstances of the case.[70]  However, the courts, primarily the district courts, have generally accepted these thresholds with little or no analysis and applied them as a matter of law.[71]  However, this mechanistic application of the test has been criticized as having no economic basis when applied in this market.[72]

This mechanical application of an econometric test may, moreover, conflict with Supreme Court rulings in this area, which stress that the proper test of expansive market definitions is whether these distant suppliers are ones “to which the purchaser can practicably turn for supplies.”[73]  The lower court decisions do not reflect consideration of the possibility that, under the test of where the purchaser can practically turn for these services, a geographic market definition threshold of 90% LOFI/LIFO for steam coal for electric power plants, which can be stockpiled for months in advance, may not be the same definition applicable to the immediate treatment of life-threatening medical conditions.  That is, just because 10% of patients can and do seek care outside the proposed market for their own elective medical needs does not necessarily mean that the other 90% could, as a practical matter, do so for their urgent or emergent medical needs.  This limitation of the method is discussed below as the “Silent Majority” fallacy.

One case in which the court did consider the practical aspects of the matter of whether traveling patients constrained the pricing faced by non-traveling patients was Hospital Corporation of America v. F.T.C., in which Judge Posner applied the test of feasibility to find that prices are not so constrained because patients do not practically travel far for medical care, especially in emergencies, unless it is not available locally.[74] 

Another problem with the arbitrary use of bright-line rules like the 90% rule is that they can lead to absurd results, as in the Tenet merger case in which the defendant’s expert was unable to define a relevant market smaller than the entire state of Missouri for a merger of two small hospitals in Poplar Bluffs, Missouri, population 17,000.[75]

  F.  Limitations of the Elzinga-Hogarty Test In Inpatient Hospital Markets

The Elzinga-Hogarty test requires a well defined product, such as coal or beer.[76]  Where the product is an amalgam of products or services, such as “commercial banking,” the test is less applicable.  Id. 

In addition to this product definitional problem, Elzinga-Hogarty anaylsis in healthcare markets suffers from two fallacies, as seen in the recent Evanston Northwestern Hospital merger[77] where the FTC presented the testimony of Prof. Elzinga as to why the Elzinga-Hogarty Test is not applicable to hospital merger cases.  He reached this conclusion based primarily on his conclusion that there are two fallacies in applying Elzinga-Hogarty analysis to hospital merger cases, the Silent Majority Fallacy and the Payer Fallacy.[78]  Because both fallacies also apply to the Critical Loss test, they are discussed below.

  G.  Fallacies Common To Both Elzinga-Hogarty and Critical Loss Analysis

    1.  The ‘Silent Majority’ Fallacy 

Use of the Critical Loss Test or the Elzinga-Hogarty Test may erroneously indicate that a geographic market is larger than it actually is because they draw inferences about the purchasing preferences of the majority of patients based on the purchasing preferences of a small minority, as little as ten percent.[79]  It posits that the existence of traveling consumers may not limit seller market power with respect to non-traveling consumers.  Professor Elzinga himself acknowledges that, in the case of hospital markets, the use of patient flow data is misleading because of the Silent Majority Fallacy.[80]  Professor Elzinga also testified that he felt that patient travel to more distant hospitals was not so much a matter of price but was due to non-availability of the necessary services in their own community or for personal reasons.[81]  Essentially, Prof. Elzinga finds that the decision to travel or not travel for medical care is not generally based on price and that the Elzinga-Hogarty test is simply inapplicable to that market analysis because the traveling patients do not constrain the prices paid by the non-traveling patients.

    2.  The Payer Fallacy

           The Payer Fallacy refers to the problem that geographic market analyses based on patient flow data can be misleading because patients do not necessarily travel based on price, since these costs are, for the most part, paid for by a third party, such as a health insurance provider.[82]  Furthermore, Medicare requires Medicare-contracted hospitals to provide emergency care for all patients who present themselves to the hospital, regardless of ability to pay.[83]  Indeed, Prof. Elzinga has concluded that the most important variable in patients’ decisions as to which hospitals they will utilize is where their own doctor directs them.[84]  Judge Posner reached the same conclusion in Hospital Corp. of America v. F.T.C..[85]

    3.  The Acuity Fallacy

           The Acuity Fallacy arises from the fact that some medical care is more urgent than other medical care that can be electively scheduled, and that the patient travel that constrains prices in the latter market does not necessarily constrain price in the former market.[86]  Analyses based on patient flow data that fail to account for these different submarkets may be flawed, as seen in the following data.

            Data reporting over one hundred demographic and clinical details on all hospital discharges from non-federal hospitals in the state of California is publicly available from the California Office of Statewide Health Planning and Development (“OSHPD”).[87]  Among other things, this data can be analyzed to show whether the hospital admission was: 1) elective, in that it was scheduled at least twenty four hours in advance, 2) urgent, in that it was not scheduled twenty four hours in advance but the patient did not require emergency room services at the time of admission, or 3) emergent, in that the patient required emergency room services at the time of admission. 

            Review of the OSHPD data shows that there were 3,442,395 patients admitted to California non-federal hospitals for the year 2005.  See Table I.  Of these, only 24% were admissions scheduled at least twenty four hours in advance and 76% were unscheduled.  Id.  Of the total number of admissions, 47% of these patients were admitted to the hospital as emergency cases that came in through the emergency room.  Id.  Another 29% were admitted as urgent cases.  Id.  As between these different groups, it seems quite likely that the opportunity for consumers to price shop for hospital care is probably limited, for the most part, to the 24% of admissions that are elective, that is, scheduled more than twenty four hours in advance.[88]  Where the admission is emergent or urgent, comprising 76% of the market, the opportunity for price shopping is, therefore, necessarily limited, and price elasticity much less than in the electively admitted group.[89]

            The predominance of emergent and urgent modes of admission is even more pronounced in the pediatric age range.  Among children admitted to non-federal California hospitals in 2005, only 14% were scheduled admissions and 86% were unscheduled, 43.1% being admitted from the emergency department and another 42.7% being admitted urgently but not from the emergency department.  Table I. 

            When the acuity of the medical condition is higher, it would be expected that patients would chose closer hospitals for treatment than when the acuity is lower and they can travel farther.[90]  The OSHPD data support this effect.  Out-Of-County patients[91] were 44% more likely to be electively admitted, 32.5%, as compared to In-County patients, 22.5%.  Pediatric Out-Of-County patients were 92% more likely to be electively admitted, 23.2%, versus pediatric In-County patients, 12.1%.  Table I. 

            From this data it can be seen that the hospital inpatient market is actually comprised of different submarkets based on acuity.  Although there are no studies of the elasticity of these submarkets, it is likely that elasticity differs significantly among them.  Thus, if the patients who are scheduled admissions are more likely to travel for their care than those who are emergently admitted, then the travel of the former does not necessarily reflect the likelihood that the latter group will also travel in response to price.  Thus, the travel of the former would not constrain the pricing faced by the latter.  This would be an instance of the Silent Majority Fallacy.

            The acuity fallacy is also important because of the statutory limitations imposed by the federal Emergency Medical Transfer and Active Labor Act, “EMTALA,”[92] that limits the ability of emergency room patients to transfer to other hospitals, as discussed below.  Elzinga and Hogarty include statutory and regulatory constraints on product “shipments” as important factors in relevant market definition.[93]

    4.  The Complexity Fallacy

           Some medical conditions are simple to treat, others are highly complex.  Just because some consumers will travel great distances to seek elective treatment for expensive and complex conditions, such as open-heart surgery or cancer, does not necessarily mean that they would travel great distances for simple conditions, like a nosebleed or sore throat.  Yet, many cases involving studies of patient flows lump all of these together and assume that, if the former group will travel, so will the latter.[94]  In fact, these are different submarkets markets with different elasticities.[95]  Complex procedures should be excluded from the analysis of community hospitals.[96]  Furthermore, complex treatments are simply not available in some rural hospitals and healthcare markets, making these separate submarkets.[97]

The data to determine these submarkets is theoretically available since Medicare and California require all hospitals to classify patients into one of 543 Diagnostically Related Groups.[98]  Where patients must travel due to complexity it is erroneous to assume that they are travelling because of price.[99]  Indeed, the destination hospital may be even more expensive than the origin hospital since it probably enjoys more market power due to the uniqueness of its services.

    5.  The Residential Zipcode Fallacy

Reported cases using patient travel to define relevant market have all used residential zip codes as the point of origin of the “shipment” and the hospital as the destination.[100]  However, this may be an unwarranted assumption in all cases, especially in emergencies, such as patients who enter hospitals while traveling away from home or patients stricken by illness or injury while at work.  As shown above, non-scheduled admissions comprise the majority of hospital admissions.


  A.  Regulatory Barriers to Entry In the Pediatric ICU Market

          There are a number of regulatory barriers to entry into the pediatric intensive care services market.  Since, by definition, this is a hospital service, one must first, in California, have a license to operate a hospital.[101]  Hospitals that wish to provide intensive care services in California must also meet specific requirements for facilities and staffing.[102]  There must be a licensed physician with training in critical care medicine appointed to provide administrative oversight of a critical care unit.[103] 

            There are no specific licensing requirements for pediatric intensive care services as compared to intensive care services in general for any age group.  Physicians who wish to practice pediatric intensive care in California must have a license to practice medicine from the California Medical Board.[104]  Although not statutorily required, the hospital medical staff may require a physician wishing to practice in a pediatric specialty to have at least specialty training and certification in the specialty of pediatrics from the American Board of Pediatrics.[105]

            Thus, there are substantial barriers to entry into the pediatric critical care market.

    B.  Regulatory Barriers to Pediatric ICU “Shipments”

           The Emergency Medical Transfer and Active Labor Act, hereafter the “EMTALA” Act, 42 U.S.C. § 1395dd, requires all Medicare-participating hospitals to evaluate and stabilize any patients of any age who seek services in their emergency rooms.[106]  Hospitals are required to provide necessary services, including physician services, regardless of the patient’s ability to pay.  Id.  Furthermore, the hospital may not transfer the patient to another hospital before that patient is “stabilized” unless: (1) the patient or other responsible person acting on behalf of the patient requests a transfer in writing, (2) a physician certifies that the benefits of transfer outweigh the risks, and (3) the transfer to the receiving hospital is appropriate, meaning that the receiving hospital has the necessary staff and facilities and has agreed to accept the transfer.[107]  Given these requirements, most of which very few patients would be aware of, it is not very practical for an unstable patient already in one hospital to price shop other hospitals, even if they would wish to focus on price under these circumstances, and the experience is that most do not.[108]


The Guidelines were revised to include the SSNIP test in 1982.[109]  Between 1994 and 2000, federal and state antitrust enforcers lost all seven of the hospital merger cases they litigated.[110]  Indeed, the FTC has not litigated a hospital merger case in a federal court since losing these seven cases.  While critical loss analysis is the mainstay of the Horizontal Merger Guidelines, a number of commentators have questioned its usefulness as an antitrust tool, in particular in hospital merger cases.[111],[112] 

In fairness, these failures cannot be directly laid at the feet of the Critical Loss Test because it was not formally used in any of them.[113]  In most cases, the court simply held that the plaintiff (the antitrust enforcer) had failed to meet its burden of showing the restricted geographic market necessary for the court to find market power in that market.  Id.   Indeed, it may well be impossible to use Critical Loss or Elzinga-Hogarty analysis to prove up hospital markets.  Id.  The following two cases illustrate some of these difficulties.

  A.  In re Adventis Health System (1994)

In this case, the Federal Trade Commission opposed the acquisition of the assets of the Ukiah General Hospital by the Adventist Health System, operator of a general hospital in Ukiah, California, as a violation of Section 7 of the Clayton Act, 15 U.S.C. § 18.  The Administrative Law Judge, hereafter the “ALJ”, found that the acquisition led to a sharp increase in the hospital market concentration in the southeastern Mendocino/Lake County area.[114]  Adventist Health defended by denying that the F.T.C. had jurisdiction over not-for-profit entities such as Adventist Health.[115]  The Commission agreed and, in its final order, dismissed the complaint.  Id.

However, in its initial order in the case, the ALJ considered the geographic market.  The complaint counsel applied the Elzinga-Hogarty analysis and found the following:[116]


Geographic Area LOFI LIFO Average
Ukiah 71% 82% 76%
Ukiah + Willits 85% 81% 83%
Ukiah + Willits + Lakeport 91% 75% 83%
All Mendocino Co. + all Lake Co. + Santa Rosa 83% 78% 81%
All Mendocino Co. + all Lake Co. + Santa Rosa + Healdsburg + Sebastopol + St. Helena 89% 83% 86%


The ALJ accepted the conclusion of Elzinga and Hogarty that non-migration of 75% of consumers reflected a “weak” market and non-migration of 90% reflected a “strong” market.[117]

Complaint counsel then went on to look at the out-migrating population, inquiring as to whether they out-migrated simply because the services they needed were not offered in Ukiah because they were too complex.[118]  In that case, they would not be relevant to market determination.  The problem was that the DRG data upon which counsel relied was not detailed enough to show, in each case, whether the service needed was or could be provided in Ukiah.[119]  This is hardly surprising since the DRG classification was not designed to define different product or service classes but was, instead, designed to classify groups of services that would be similarly priced under Medicare’s prospective payment system.[120]  The most that counsel could show was that the out-migrating patients received more complex services than those who did not out-migrate but there was significant overlap in complexity between the two groups.[121],[122]   Due to substantial questions as to the methodology applied by complaint counsel, the ALJ rejected complaint counsel’s geographic market definition.[123]  Despite all this, the ALJ also held that, “The Elzinga-Hogarty test offers an easily-applied numerical standard for geographic market analysis…”[124]

After rejecting the complaint counsel’s Elzinga-Hogarty definition of the relevant geographic market, the ALJ considered other evidence proffered, such as strategic planning documents of the hospitals in Ukiah as well as the testimony of physicians and insurance companies.[125]  Taking all of this into account, the ALJ concluded that the relevant geographic market included Ukiah, Willits (20 miles from Ukiah), and Santa Rosa (68 miles from Ukiah).  Id.

            The Adventist case shows how, even in a simple case of two hospitals in one rural community, the Elzinga-Hogarty test is very hard to apply.  The primary cause of this difficulty is the diversity of the services provided, in other words, the product market definition.  Whereas coal and beer, for which the Elzinga-Hogarty test was first reported, are fairly uniform and fungible commodities, medical care is not.  Medicare divides hospital care into twenty five Major Diagnostic Categories[126] and five hundred and forty three Diagnostically Related Groups, each DRG being assigned to one of the MDC’s.[127]  These groupings are determined by a computer program, “Grouper,” based, among other things, on specific diagnosis codes, of which there are approximately fourteen thousand, and procedural codes (codes for services provided), of which there are approximately four thousand.[128]  Furthermore, there are other submarkets, such as those based on health insurance provider,[129] acuity, as discussed above, and young age.[130]   All of these submarkets and differentiated products complicate the geographic market analysis, often beyond any possibility of achieving a definitive result using quantitative measures such as Elzinga-Hogarty.[131]  This limitation of the method in complex, poorly differentiated markets was recognized by Elzinga and Hogarty when they first described the method.[132]

  B. California v. Sutter Health System (2000)

This action was brought by the Attorney General of California to obtain a preliminary injunction against the merger of two general acute care hospitals in the Oakland, California area.[133]  The Court adopted the Horizontal Merger Guidelines as a non-binding standard of merger analysis, citing Tenet and Mercy Health Services.[134]  In determining geographic market, then court first applied the Elzinga-Hogarty test, considering the results of both the plaintiff’s and defendant’s analyses under that test.[135]  The principal difference between those two analyses was that the plaintiff adopted an 85% threshold for the average of LIFO and LOFI[136] whereas the defendant used the 90% threshold (“strong market”)[137] to define the relevant geographic market.  The Court chose to adopt the 90% threshold, Elzinga and Hogarty’s standard for “strong” market definition in the coal industry.[138]  The Court also chose to include, as a competitor, the Kaiser hospital in Walnut  Creek,[139] on the basis that, in the long run, even though not in the short run, patients could be treated at the Kaiser hospitals if they changed to the Kaiser health plan.[140]

The Court then held that Elzinga-Hogarty analysis was only a starting point and that it was necessary to look, not just at where patients are already going, but at where patients could go if faced with a price increase by defendants, a question not answered by the Elzinga-Hogarty approach.[141]  The Court found that there were three factors that were relevant to this question: (1) service area overlap between hospitals inside the test market and hospitals outside the test market; (2) geography and travel times, and (3) the perceptions of market participants.[142]

As to overlap of service areas, the Court found that some of defendant’s patients came from the fringes of the defendant’s service area, such as Contra Costa county, and who thus could, practically, have gone to hospitals in Contra Costa county.[143]  Thus, Contra Costa county was counted in defendant’s service area.  Id.  The Court also found that some of the patients from defendant’s immediate service area could go to hospitals across the San Francisco Bay in the city of San Francisco.[144]  Thus, San Francisco hospitals were deemed to be practical alternatives to defendant’s hospitals.  Id.

In this question of service area overlap, the Court made the error of the Silent Majority Fallacy when it held that, just because a few patients from defendants’ service area go to hospitals outside plaintiff’s posited geographical market area, such as San Francisco or Contra Costa county, does not mean that most could, or more importantly, would, turn to those other distant hospitals in the event of a five percent increase in defendants’ prices.


           If the Hypothetical Monopolist and Elzinga-Hogarty tests are either inapplicable or impractical of application in differentiated product markets, the utility of the present Horizontal Merger Guidelines in defining geographic markets may be limited in many highly differentiated markets such as hospital services, including pediatric intensive care services.  The same limitations would apply where courts borrow these tools to apply to section one claims under the Sherman Act.

This paper concludes that econometric tests, such as SSNIP, Critical Loss, and Elzinga-Hogarty, are not applicable to define relevant geographic market in markets, such as health care markets, comprised of large numbers of highly differentiated products and/or services, many of which comprise relevant sub-markets in their own right, and are not substitutable for one another.  Pediatric intensive care services are one of these submarkets because of the high level of acuity of these patients and the statutory limitations on “shipping” them to other hospitals.         For these reasons, definition of relevant markets in the pediatric intensive care market should use the standard set forth by the Supreme Court in Tampa Electric Co., examination of the individual and peculiar facts of each case.









Scheduled Admissions           Emergency Department          Urgent Admissions[146]             Total Admissions



All Patients                 827,146 (24.0%)                     1,627,862 (47.3%)                  987,387 (28.7%)                     3,442,395 (100.0%)


Out-of-County                        173,526 (32.5%)                        180,903 (33.9%)                  178,783 (33.5%)                        533,212 (100.0%)


In-County                   653,620 (22.5%)                     1,446,959 (49.7%)                  808,604 (27.8%)                     2,909,183 (100.0%)


AGE < 17

Scheduled Admissions           Emergency Department          Urgent Admissions                 Total Admissions



All Patients                 34,842 (14.3%)                       105,326 (43.1%)                     104,295 (42.7%)                     244,463 (100.0%)


Out-of-County            11,030 (23.2%)                         14,946 (31.4%)                       21,669 (45.5%)                       47,645 (100.0%)



In-County                   23,812 (12.1%)                         90,380 (45.9%)                       82,626 (42.0%)                     196,818 (100.0%)