The hypothetical monopolist test is one of the organizing principles of the Horizontal Merger Guidelines, and it is a test increasingly applied to define markets, not just in merger cases, but throughout antitrust, and not just in the U.S., but throughout the world. While it is certainly true that the hypothetical monopolist test "advanced a new approach to market definition" whose "importance and brilliance . . . cannot be overstated,"⁽¹⁾ it is equally true that the hypothetical monopolist test is "not flawless."⁽²⁾ Indeed, three years after the 1982 Guidelines were published George Stigler commented that, "[the Guidelines' market definition test] has one, wholly decisive defect: it is completely nonoperational. No method of investigation of data is presented, and no data . . . are specified that will allow the market to be determined empirically."⁽³⁾ History has proven that Stigler's concern was overstated, as there clearly is data that will operationalize the hypothetical monopolist test. Using reliable data is critically important because without such data, courts frequently resort to the Brown Shoe factors⁽⁴⁾ such as industry recognition or the product's peculiar characteristics,⁽⁵⁾ without clear discussion of how these factors relate to the hypothetical monopolist test.

My remarks address how to render the hypothetical monopolist test — in Stigler's phrase — "operational" by using sound empirical methods that answer the questions posed by the Guidelines' market definition test.

The Relevant Question

In Section 1.11 of the 1992 Merger Guidelines, the "relevant question" is posed as follows:

If, in response to the price increase, the reduction in sales of the product would be large enough that a hypothetical monopolist would not find it profitable to impose such an increase in price, then the Agency will add to the product group the product that is the next-best substitute for the merging firm's product . . . . The price increase question is then asked for a hypothetical monopolist controlling the expanded product group. This process will continue until a group of products is identified such that a hypothetical monopolist over that group of products would profitably impose at least a "small but significant and nontransitory" increase ["SSNIP"], including the price of a product of one of the merging firms.⁽⁶⁾

The iterative process required by the hypothetical monopolist test implies two subsidiary questions: first, what volume of lost sales will make the SSNIP unprofitable; and second, what volume of sales will be lost as a result of a SSNIP. Much of the economic literature focuses on the first question — in fact, that is the question that "critical loss" attempts to answer. The second question, what volume of sales will be lost as a result of a SSNIP, is frequently not answered. As a result, a fact finder — whether an agency or a court — may have confidence that a hypothetical monopolist could not profitably impose a SSNIP if it were to lose X percent of its sales, but that fact finder may have no confidence that a hypothetical monopolist would lose more or less than X percent of its sales.

The purpose of these remarks is to provide some guidance toward answering the second question — what volume of sales would the hypothetical monopolist lose in response to a SSNIP. To that end, we will discuss four forms of evidence: historical evidence, econometric evidence, affidavit evidence, and survey evidence.

Types of Evidence

1. Customer reaction to historical price changes.

The Merger Guidelines expressly authorize the use of historical data, although they also caution it may produce misleading results.⁽⁷⁾ Past price changes will often reflect changes in cost affecting the whole industry rather than market power, and consumers' reactions to such changes do not illuminate how they would respond to a hypothetical monopolist. Cf. United States v. Eastman Kodak Co., 63 F.3d 96, 106 (2d Cir. 1995) (fact "that Kodak film sells for different prices in different parts of the world is insufficient to establish price discrimination with proof that Kodak's costs are uniform"). Moreover, price changes typically are of short duration, and thus do not satisfy the non-transitory requirement of the SSNIP test. See Gulfstream Park Racing Assoc. v. Tampa Bay Downs, Inc., 2003 U.S. Dist. LEXIS 20225 at *50-51 (M.D. Fla. Nov. 7, 2003) (rejecting historical price increase on the grounds that it lasted "for a transitory period of time, five years.")⁽⁸⁾ For this reason, the Guidelines themselves caution: "the picture of competitive conditions that develops from historical evidence may provide an incomplete answer to the forward-looking inquiry of the Guidelines."⁽⁹⁾

Courts have similarly cautioned against the use of historical evidence. See Gulfstream Park Racing Assoc, 2003 U.S. Dist. LEXIS 20225 at *50-51 ("The Guidelines are applied to hypothetical future situations and not to historical evidence"). Indeed, what separates the Guidelines' method of market definition from prior methods, such as Elzinga-Hogarty, is its hypothetical or forward-looking approach. See California v. Sutter Health Sys., 84 F. Supp. 2d 1057, 1072 (N.D. Cal. 2000) ("Plaintiff's E-H test results do not end the Court's analysis, in any event, as the E-H test is only a starting point in analyzing a geographic market. E-H test results reflect only current market behavior, and are insufficient to determine where patients could practically turn for acute inpatient services if faced with a future anticompetitive price increase.") United States v. Mercy Health Servs., 902 F. Supp. 968, 978 (N.D. Iowa 1995), vacated and remanded as moot by 107 F.3d 632 (8th Cir. 1997) held "[t]he court finds that the government's case rests too heavily on past health care conditions and makes invalid assumptions as to the reactions of third-party payers and patients to price changes. The government's case also fails to undergo a dynamic approach to antitrust analysis, choosing instead to look at the situation as it currently exists within a competitive market." Similarly, the court in FTC v. Freeman Hosp., 69 F. 3d 260, 268 (8th Cir. 1995), concluded: "In order to meet its burden, the FTC is required to present evidence addressing the crucial question of where consumers of acute care inpatient hospital services could practicably turn for alternative sources of the product should the Hospitals' merger be consummated and Joplin Hospital prices become anticompetitive. The FTC has failed to produce such evidence."⁽¹⁰⁾

When considering historical evidence, it is important to distinguish between evidence that customers switched in response to price changes and evidence that customers did not switch. If a significant number of customers switched from product X to product Y in response to small but significant price changes in X relative to Y, this seems to be evidence that the two belong in the same relevant market. The criticism that the price change in X may be cost justified or not sufficiently long in duration would simply suggest that even more customers would switch from X if the price increase were not cost justified, imposed by a monopolist, or of a longer duration. Alternatively, evidence that customers did not switch from product X to Y in response to a small but significant price change in X relative to Y may not be evidence that the two belong in the same relevant market, as customers would be more likely to switch if the price increase were not cost justified or of longer duration.

Thus, courts have rejected as evidence of a relevant market the fact that customers did not switch to alternatives in response to historical price changes but accepted evidence that customers either have switched or currently use both the relevant product and other substitutes. Compare Gulfstream Park Racing Assoc., 2003 U.S. Dist. LEXIS 20225 at *50-51 (rejecting evidence that customers did not switch in response to a price increase) with FTC v. Tenet Healthcare Corp., 186 F.3d 1045, 1050 (8th Cir. 1998) ("Poplar Bluffs") (evidence of current dual usage strongly suggested that a hypothetical monopolist could not profitably impose a SSNIP).

Where there are pricing anomalies caused by past anticompetitive activity or government regulation, the facts may speak for themselves and it is possible to apply the hypothetical monopolist test with a high degree of confidence. In an extreme case, if the firms selling the products in the proposed market have in the past engaged in cartel activity, there would be no reason to doubt that the market is properly defined.⁽¹¹⁾ An interesting concurring opinion in E.W. French & Sons v. General Portland, 885 F.2d 1392, 1402 (9th Cir. 1989), discussed at length the significance of past cartel activity for proving a relevant market. Following a trial in which the court incorrectly charged the jury on the legal significance of finding a conspiracy to eliminate a competitor who did not join a cartel, the defendant argued that the incorrect charge was harmless error because there was not proof of a relevant market or anticompetitive effect. Plaintiff argued that the cartel itself conclusively established a relevant antitrust market. In his concurrence, Judge Farris agreed:

As a purely logical matter, French is unquestionably correct. A price-fixing conspiracy confined to manufacturers of concrete would not have been able to succeed if concrete were not a distinct product market: when the cartel attempted to raise prices, customers would simply switch to sand, brick, gravel or some other construction material. . . . . Every price-fixing conspiracy thus identifies directly, in a real world context, a group of firm which is insulated from outside competitive pressures. That is precisely what conventional market definition evidence attempts to identify artificially, by the collection and interpretation of economic data regarding the relationship between various demand curves, by common sense assumptions about the interchangeability of similar products, and the like.

Id. at 1402.

United States v. Archer-Daniels-Midland Co., 866 F.2d 242 (8th Cir. 1988), presented a somewhat analogous situation. The issue was whether government had correctly defined a market for high fructose corn sweetener ("HFCS"), the closest substitute for sugar in carbonated sodas. Government price supports for sugar set a floor for its price that was 10 percent above the price of HFCS, and, as a result of its cost advantage, HFCS had gradually replaced sugar in soft drinks. The Eighth Circuit observed that the price floor for sugar provided a direct answer to the hypothetical monopolist question: "As long as an effective price support program is in existence, a monopolist of HFCS will be able to raise the price of HFCS to just below the supported price of sugar before being constrained by the competitive forces of sugar. In other words, the [hypothetical] HFCS monopolist is able to exercise excess market power." Id. at 246.

Finally, FTC v. Staples, 970 F. Supp. 1066 (D.D.C. 1997), could be viewed as using a variation on a natural experiment. Staples is sometimes discussed as a case that skipped the market definition step altogether by asking directly whether a merger from three to two or two to one office superstores would lessen competition.⁽¹²⁾ But, Staples can also be described as testing whether there is a relevant market for office superstore competition for the sale of consumable office supplies in geographic markets where such competition exists. As a rule, cross-market comparisons are likely to engender controversy because it is difficult to control for all of the factors that affect pricing in different markets. For instance, three office superstore geographic markets are likely to be more densely populated with larger economies of scale and more intense retail competition of all types. Still, FTC's various price studies — which on the whole the Court found persuasive — tended to show that a hypothetical office superstore monopolist could impose a SSNIP on the sale of consumable office supplies, because where there was an actual office superstore monopolist that is precisely what it did. If, as in Staples, the empirical problems of cross-market comparison can be solved to the satisfaction of the trier-of-fact, such studies provide a method of applying the Guidelines' test to many retail and other markets.

2. Econometric Evidence

Econometrics uses historical evidence in a controlled and scientific manner to answer the hypothetical monopolist test. As Judge Easterbrook in Menasha Corp. v. News America, 354 F.3d 661, 2004 U.S. App. LEXIS 241, at *9 (7th Cir. 2004) (criticizing the plaintiff for failing to introduce "econometric evidence of any kind" in defining the relevant market) recently recognized, econometric evidence can provide highly reliable evidence of whether a sufficient number of customers will switch to defeat a SSNIP.

There are many types of econometric analysis that can be used to answer the hypothetical monopolist test. One type is the sort used in Staples, which, in effect allowed for a controlled natural experiment. This type of analysis requires information on pricing, number of competitors, and factors that may influence the price of goods at retail. A second type of analysis directly estimates market elasticities. To be done properly, this type of analysis requires time-series consumer level information on price and quantity.

While there is general agreement that in appropriate circumstances econometric estimation of elasticities can provide perhaps the best evidence on market definition, there are significant issues with respect to its use. One issue involves the availability of data — it is quite unusual for private parties to have access to price and quantity information from all market participants. This is, of course, less of an issue for the government, but it does raise important issues about transparency, as the government may be unable to share this data with the merging parties. A second issue involves the quality of the data — retail scanner data frequently does not give sufficient information on coupons or other discounts. A third issue involves the appropriateness of the data — retail scanner data, for example, may not be the appropriate type of data with which to analyze mergers between manufacturers.

Courts have frequently used very high standards in evaluating econometric evidence. See, e.g., FTC v. Swedish Match, 131 F. Supp. 2d at 161 (rejecting econometric estimation of demand because the model was not robust and because the results were significant only at an 85 percent level of certainty). That being said, econometric evidence has been used by courts in market definition both expressly, Kraft Gen. Foods, Inc., 926 F. Supp. at 333 (crediting plaintiff's econometric estimation of cross-elasticities as evidence of a ready-to-eat cereal market), and implicitly, Staples, 970 F. Supp. at 1077 (expressing some doubt at econometric evidence but crediting it because it was consistent with other evidence).

3. Customer interviews and affidavits.

If you want to know what somebody would do in a hypothetical situation, one solution is to ask him. This may explain the practical appeal of the Guidelines' 5 to 10 percent price increase formulation of the SSNIP/hypothetical monopoly test; it gives lawyers and economists a concrete question to ask customers in interviews and depositions. It also has led to the frequent use of customer affidavits in recent merger litigation.⁽¹³⁾

There are significant issues with respect to the use of customer affidavits and Areeda and Turner go so far as to suggest that customer affidavits are the "least reliable" evidence of whether customers would switch in response to a SSNIP.⁽¹⁴⁾ Despite this criticism, the government and litigants in merger cases frequently use affidavits. A review of cases involving the use of affidavits reveal two frequent errors: (i) affidavits frequently ignore the presence of switching costs, and (ii) affidavits frequently do not cover a representative sample of customers.

As to the first question, if customers would not switch between the products being grouped together for the purpose of applying the hypothetical monopolist test, it is not meaningful to ask whether they would switch to a product outside the proposed market following a SSNIP. For example, United States v. Engelhard Corp., 126 F.3d 1302 (11th Cir. 1997), involved a merger between Engelhard and Floridan, two sellers of Gellant Quality Attapulgite ("GQA"). The DOJ argued that the fact that current GQA customers would not switch in response to a SSNIP in GQA was evidence that there existed a relevant market for GQA. The Eleventh Circuit rejected the DOJ's argument, in part, because Engelhard's GQA customers would not switch to Floridan GQA in response to a 5-10% increase in the price of Engelhard's GQA.⁽¹⁵⁾

The Guidelines acknowledge a variation of this point when they speak of adjusting the size of a SSNIP to account for situations, such as a tariff or commission, where the relevant product constitutes a small percentage of the overall cost of the good or service. See Horizontal Merger Guidelines at § 1.1 ("the price for which an increase will be postulated will be whatever is considered to be the price of the product at the stage of the industry being examined") & n.11. But see NaBanco v. VISA USA, 779 F.2d 592 (11th Cir. 1986) (finding a lack of market power with respect to the fees for credit card charges because credit cards compete with other means of payment). Other solutions may include making some allowance for switching costs before applying the SSNIP, or making the SNIPP inquiry with respect to just new customers or customers who otherwise are not locked-in by an initially investment in one firm's product or service.

Affidavits also must cover an adequate sample of customers. The Eleventh Circuit's decision in Englehard is instructive. GQA had a wide variety of different users. The government offered witnesses who testified that they would not switch away from GQA in response to a 5 to 10 percent price increase. But, the government did not classify its witnesses by type of user or present evidence concerning the number of each type of user. As a result, the Eleventh Circuit held that the government failed to prove a relevant market: "it is possible for only a few customers who switch to alternatives to make the price increase unprofitable, thereby protecting a larger number of customers who would have acquiesced in higher GQA prices. To evaluate such possibilities, the Government should have ascertained the size of the GQA market in each of its end-use applications . . . . No matter how many customers in each end-use industry the government may have interviewed, those results cannot be predictive of the entire market if those customers are not representative of the market." Englehard, 126 F.3d at 1306.

Sungard Data Sys., 172 F. Supp. 2d at 183, applied similar analysis to much the same result. In light of the "the striking heterogeneity" of the customers, the Court found that

[t]he government has failed . . . to show whether [the group of captive customers] is substantial enough that a hypothetical monopolist would find it profitable to impose such an increase in price. The sampling of customer statements before the Court is minuscule when compared with the entire universe of defendants' shared hotsite customers. Although the government has submitted approximately 50 statements from customers stating that they either would not or could not switch from shared hotsites, there are more than 7,500 customers that currently use defendants' shared hotsites. Without more information, the Court simply cannot determine whether these 50 declarations are representative of the shared hotsite client base.

Id. at 191-92.⁽¹⁶⁾

The Court indicated it may have afforded the customer affidavits more weight if the government were able to show they represented an adequate distribution of customers. "Instead of fine-tuning its presentation to account for significant differences among defendants' customers, the government lumped all customers together." Id. at 192

Thus, small samples drawn from a large number of customers are not likely to be convincing unless they reflect a representative sample. But, even large numbers of affidavits, however, are not sufficient if they are all from one class of customers, which does not consist of customers whose response will determine whether a SSNIP will be profitable. FTC v. Cardinal Health, 12 F. Supp. 2d 34, 48-49 (D.D.C. 1998), sustained the FTC's narrow market definition for the wholesale distribution of prescription drugs against the parties' claims that other methods of distribution, most prominently self-supplied warehousing and distribution, should also be included in the market. The merging parties presented significant evidence that chain pharmacies would turn to self-supply if prices increased. However, the court was persuaded by the FTC showing that other classes of customers, "namely hospitals and independent pharmacies — would have no reasonable substitutes." Id. at 47. Significantly, the Court found that "[d]ispensers who self-warehouse are overwhelmingly limited to a small segment of retail chain pharmacies. Thus, this Court finds that the majority of Defendants' customers cannot replicate the wholesalers' services themselves nor obtain them from any other source or supplier." Id. at 48-49. Where the plaintiff is able to show that defendant's affidavits are drawn from a class of customers with uniquely elastic demand, they will not defeat other evidence tending to show that a hypothetical monopolist could increase price to other customers.

In some instances, a few customers may be able to speak to the overall demand elasticity of an industry. In Swedish Match, 131 F. Supp. 2d at 162, the FTC may have avoided the sampling problem when it used affidavits of loose leaf chewing tobacco distributors as a proxy for the views of a broad spectrum of customers. Likewise, the insurance companies and other payors in Poplar Bluffs who submitted affidavits saying they could not shift patient demand to other hospitals spoke to the question of overall demand. But, although they appear to have been some of the only evidence in that litigation directly addressing the hypothetical monopolist question as opposed to current market conditions, the court afforded them little weight. 186 F.2d at 1054 (calling the insurers' affidavits "suspect" and relying on other evidence tending to show that they could resist a price increase). If the number of customer who say they can switch exceeds the critical loss, there can be no concern whether the sample of affidavits is representative. For example, if, absent an allegation of price discrimination, the critical loss is only eight percent, and customers representing 10 percent of sales swear they would respond to a SNIPP by switching to products outside the market, the proposed market is not correctly defined.

Frequently, the merging parties and the government will get contradictory affidavits from the same customers. In this regard, the government's subpoena power can be used to test the conviction of a defendant's affiant. Defendants, thus, may not want to get affidavits unless the customer's view is strongly held and the defendant has an understanding of the antitrust basis for the customer's position. From a government perspective, customers may simply be complaining about changes or disruptions in the market that will cause the customer inconvenience. Such concerns do not necessarily speak to an overall lessening of competition. It is not fatal to defendants, however, if the government flips a pro-merger customer. If the customer's equivocation obscures the issue, the government may fail to carry its burden of proof. Cf. Sungard, 172 F. Supp. at 183, 189.

4. Surveys.

A frequent criticism of customer affidavits is that they are form affidavits, which simply give yes/no responses to a series of questions, including whether customers would switch in response to a SSNIP. While it is clear that the government is unlikely to give form affidavits much weight, it appears that courts frequently give these affidavits at least some weight.

If the goal of a client is simply to obtain as many affidavits as possible, without taking the time to create detailed customer specific affidavits, it may be advisable to conduct a survey instead. Indeed, if conducted properly by an expert trained in acceptable survey methodology, surveys should carry more weight than even an allegedly representative sample of affidavits. For example, surveys may have more methods to control leading questions, selection bias, and randomness. Surveys may also make it easier to organize customers by key competitive traits reflecting the probable elasticity of their demand.

Although some cases have used surveys in defining the relevant market, see e.g., Kraft Gen. Foods, 926 F. Supp. at 327 ("Only about one-third of all adults responding in a 1991 survey said they would not eat all types of cereals"), few cases have involved directly asking customers whether they would switch to alternatives in response to a SSNIP. One case in which a survey was used was Tenet Health, 17 F. Supp. 2d 937, 945, where the district court gave the defendant's survey no weight because of a number of technical and substantive failings.⁽¹⁷⁾ Although this decision was later reversed on appeal, the Eighth Circuit did not discuss the survey evidence.

Because surveys could be one of few methods of addressing the hypothetical monopolist question without resorting to incomplete and fragmentary evidence that does not address overall demand characteristics, litigants should be able to devise methods to overcome any technical criticisms that can be levied against them.

Conclusion

For the hypothetical monopolist test to serve as something more than an instructive theoretical paradigm, litigants, agencies, and courts should look to empirical evidence that can directly answer the question: how many customers will switch in response to a SSNIP. While historical evidence, affidavits, and survey evidence can all be subject to criticism, there can be little question that they are better than mere conjecture on what is a critical question on market definition.

FOOTNOTES

1. David Scheffman, Malcolm Coate, & Louis Silvia, 20 Years of Merger Guidelines Enforcement at the FTC: An Economic Perspective 6-7, at http://www.usdoj.gov/atr/hmerger/12881.htm (last visited Feb. 5, 2004).

2. Id. at 6.

3. George Stigler & Robert Sherwin, The Extent of the Market, 28 J. L. & Econ. 555, 582 (1985).

4. See Brown Shoe Co. v. United States, 370 U.S. 294 (1962) (markets may "be determined by examining such practical indicia as industry or public recognition of the submarket as a separate economic entity, the product's peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors").

5. See, e.g., Olin Corp. v. FTC, 986 F.2d 1295, 1298-1305 (9th Cir. 1993) (which includes an extensive discussion of both Brown Shoe factors and the Merger Guidelines), United States v. Sungard Data Sys., 172 F. Supp. 2d 172, 189 (D.D.C. 2001) ("Thus, the Brown Shoe factors — especially industry recognition and the peculiar characteristics and uses of the product — support a finding that internal hotsites fall within the same product market as shared hotsite services."); FTC v. Swedish Match N. Am., Inc., 131 F. Supp. 2d 151, 163-66 (using Merger Guidelines as a tool to answer Brown Shoe's cross-elasticity question, but then going to extensively discuss other factors such as industry recognition); FTC v. Staples, Inc., 970 F. Supp. 1066, 1075 (D.D.C. 1997) ("The Commission discussed several of the Brown Shoe 'practical indicia' in its case, such as industry recognition, and the special characteristics of superstores which make them different from other sellers of office supplies, including distinct formats, customers, and prices."); New York v. Kraft Gen. Foods, 926 F. Supp. 321, 359 (S.D.N.Y. 1995) (citing practical indicia as a factor but not relying upon them); Bon-Ton Stores v. May Depot Stores, 881 F. Supp. 860, 872 (W.D.N.Y. 1994) ("That this evidence supports a finding of a traditional department store submarket is made even more apparent when one considers the factors set forth in Brown Shoe."); and FTC v. Owens-Illinois, Inc., 681 F. Supp. 27, 35 (D.D.C. 1988) vacated by 850 F. 2d 694 (D.C. Cir. 1988) (citing Brown Shoe's practical indicia, but then attempting to implement the hypothetical monopolist test).

6. Department of Justice & Federal Trade Commission, Horizontal Merger Guidelines, reprinted in 4 Trade. Reg. Rep. ¶ 13,104, at § 1.11 (1992).

7. "In considering the likely reaction of buyers to a price increase, the Agency will take into account all relevant evidence, including, but not limited to, the following: . . .
evidence that buyers have shifted or have considered shifting purchases between different geographic locations in response to relative changes in price or other competitive variables." Id.

8. Although the Guidelines do not define a non-transitory period of time, one would certainly think that 5 years would be sufficient.

9. Horizontal Merger Guidelines, 4 Trade Reg. Rep. ¶ 13,104, at § 0.

10. See also Delaware Health Care, Inc. v. MCD Holding Co., 957 F. Supp 535, 542 (D. Del. 1997) ("Defendants argue that the [Elzinga-Hogarty] test is insufficient as a tool of market definition. They note critics have opined that the usefulness of the [Elzinga Hogarty] test is limited to determining when a market is not correctly defined . . . . [T]he antitrust law treatise with which Dr. Solow is associated indicates reliance solely on historical tests like the [Elzinga Hogarty] test 'can misstate the size of the market.'") (quoting Phillip E. Areeda & Herbet Hovenkamp 2A Antitrust Law ¶ 550 (1995)); Bathke v. Casey's Gen. Stores, Inc. 64 F.3d 340, 346 (9th Cir. 1995) ("using discovery to obtain the addresses of a seller's customers seldom provides us with useful information about geographic market. What we really need to know is the extent to which people from the immediate area can readily turn to alternative sellers.") (quoting Herbert Hovenkamp Federal Antitrust Policy ¶ 3.6d (1994).)

11. The same conclusion may be warranted if a similar market in another jurisdiction had been subject to cartel activity. Other instances where prices are not competitive due to regulation may be similarly instructive.

12. Jonathan B. Baker, Econometric Analysis in FTC v. Staples, Address before the American Bar Association's Antitrust Section Economics Committee (July 18, 1997), available at http://www.ftc.gov/speeches/other/stspch.htm (last visited Feb. 5, 2004).

13. While it may be the best available evidence of how customers would respond to a SSNIP, customer views have not always carried the day. Compare FTC v. Butterworth Health Corp., 946 F. Supp. 1285, 1292-93 (W.D. Mich. 1996), aff'd by 121 F.3d 708 (6th Cir. 1997) ("The most persuasive of these are the views of consumers.") and Bathke v. Casey's Gen. Stores, Inc., 64 F.3d 340, 346-47 (8th Cir. 1995) (most important dynamic evidence is that indicating where consumers could practicably turn for alternative services to avoid doing business with the antitrust defendant) with Sutter Health Sys., 84 F. Supp. 2d at 1080 ("As stated by the court in Freeman Hosp., 'while such non-empirical data may have some probative value as a starting point to evaluate this market, such data will not carry the [plaintiff's] burden. Informal, off-the-cuff remarks and anecdotal evidence concerning the marketplace are no substitute for solid economic evidence.'") (quoting Freeman Hosp. 911 F. Supp. at 1220).

14. Philip E. Areeda & Donald Turner Antitrust Law (1980) ¶538b, at 239 ("Though not irrelevant, such statements are often unreliable.").

15. Areeda criticizes Englehard on the grounds that the fact that GQA was such as small and insubstantial part of the end products ultimately purchased by consumers was a reason why it would be easier to exercise monopoly power. Id. at 2A ¶564.

16. See also Owens-Illinois, 681 F. Supp. at 37 ("analysis serves to confirm the conclusion that those few end use segments proven to be inelastic are not significant enough, in and of themselves, to constitute a relevant product market and are not representative of the glass container market as a whole") (eleven inelastic end-uses of glass containers accounted for only about 25 percent of all glass container usage); United States v. Gillette Co., 828 F. Supp. 78, 83 (D.D.C. 1993) ("since this limited subset of fountain pen devotees does not encompass the entire universe of consumers . . . a larger market must be defined. The record indicates that there is a much larger subset of fountain pen consumers who will substitute other modes of writing for fountain pens; for these customers, fountain pens therefore are in direct competition with these other modes."); Virginia Vermiculite Ltd. v. W.R. Grace & Co., 108 F. Supp. 2d 549, 587 (W.D. Va. 2000) ("The defense experts conducted a basic cross-elasticity analysis of these substitutes, and concluded that even if these substitutes do not cover all of the potential end-uses for vermiculite, 'The existence of a variety of substitutes in a variety of applications indicates that even a hypothetical monopolist supplier of vermiculite would be unable to profitably increase price.'").