Bell Atlantic Corp. v. AT&T Corp.

United States Court of Appeals Fifth Circuit F I L E D Revised August 4, 2003 July 16, 2003 IN THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT Charles R. Fulbruge III Clerk No. 02-40656 BELL ATLANTIC CORPORATION; ET AL, Plaintiffs, ROCHELLE COMMUNICATIONS INC; ADROIT MEDICAL SYSTEMS INC, a Tennessee Corporation, Plaintiffs-Appellants, versus AT&T CORPORATION; ET AL, Defendants, AT&T CORPORATION, Defendant. Appeal from the United States District Court for the Eastern District of Texas Before GARWOOD, SMITH and BARKSDALE, Circuit Judges. GARWOOD, Circuit Judge: In this Rule 23(f) interlocutory appeal, the plaintiffs–appellants, Rochelle Communications, Inc. (Rochelle) and Adroit Medical Systems, Inc. (Adroit), challenge the district court's denial of their motion to certify two classes of plaintiffs allegedly injured by the refusal of the defendant, AT&T Corp. (AT&T), to permit the passage of caller identification (caller ID) data across its long-distance telephone network. Because we conclude that the appellants cannot satisfy the predominance requirement of Rule 23(b), we affirm. Facts and Proceedings Below In 1996, Bell Atlantic Corp. (Bell) brought suit under section 4 of the Clayton Act1 against AT&T and its subsidiary Lucent Technologies, Inc., seeking treble damages and injunctive relief for alleged violations of the antitrust laws. According to Bell, AT&T attempted to monopolize the market for caller-ID service in violation of section 2 of the Sherman Act2 when, for approximately four years beginning in 1992, AT&T blocked the free passage of caller-ID data over its long-distance network. Shortly after Bell brought suit, the named class plaintiffs, Rochelle and Adroit, intervened and moved to certify two classes of plaintiffs who allegedly suffered antitrust injury because of AT&T's conduct during the class period, a period running between March 19, 1992, 1 15 U.S.C. § 15. 2 15 U.S.C. § 2. 2 and November 30, 1995.3 A. Caller ID The decision to certify a class may often necessitate a highly factual inquiry, see Alabama v. Blue Bird Body Co., Inc., 573 F.2d 309, 316 (5th Cir. 1978), and the propriety of class certification here hinges in part upon evidence introduced below concerning the nature and operation of caller ID. The record reflects that caller ID is a service marketed and provided by local telephone companies that permits the display, on a device either attached to or incorporated into the telephone of the recipient of a call, of the telephone number, and occasionally the name, of the calling party. The service operates by transmitting data containing, at a minimum, a calling party's telephone number (CPN) over the telephone networks until it ultimately reaches, and is displayed on, the call recipient's caller-ID display unit.4 3 March 19, 1992, represents that date on which AT&T began blocking the free passage of caller ID data over its long- distance telephone network. AT&T ultimately ceased blocking the free transmission of caller ID data on December 1, 1995, the effective date of the Federal Communications Commission's (FCC) regulation ordering all long-distance carriers to pass such data over their networks free of charge. See Miscellaneous Rules Relating to Common Carriers, 60 Fed. Reg. 29490, 29491 (June 5, 1995) (codified at 47 C.F.R. § 64.1604 (2002)). 4 For at least the early portion of the class period, caller ID was advertised primarily as a local service. The parties do not dispute this fact, nor do they allege that caller ID was ever advertised, during the class period and on a class- wide basis, as providing any utility with respect to long- 3 There is no dispute that access to caller ID information may be of benefit to a number of businesses and may, for certain businesses, produce substantial efficiency gains and accompanying cost savings. Businesses, for example, may use caller ID data to return calls received after hours where the caller left an incomplete message or no message at all. And because the display units also are sometimes able to record information for later recall, the caller ID display units may also be used to track call volume. In addition, businesses that maintain reverse-charge, long-distance telephone numbers benefit from caller ID by using the calling party number to screen out unwanted calls, thereby reducing long-distance calling expenses. At a more sophisticated level, the caller-ID data transmitted to a call recipient may be linked, through the use of computer telephony integration (CTI) equipment and software, to a business's computerized database. CTI equipment thus allows a business to use a caller ID signal to rapidly retrieve information related to a particular caller, permitting a business, for example, to route a call to a specific employee, or to provide faster and more efficient service to a customer, resulting in reduced telephone bills and labor costs, and in some circumstances, increased distance calls, or that the class members purchased caller ID service from their local telephone companies with the expectation that caller ID service would be available on long-distance calls. 4 customer goodwill.5 A number of technological prerequisites, however, must be met before a call recipient can receive a caller ID signal. Chief among these is the need for each portion of the telephone network that the caller ID signal must traverse to be connected to a telephone network known as the Signaling System 7 (SS7) network.6 Thus, for any given call to carry a caller ID signal to a call recipient over the AT&T long-distance network, the local telephone exchange networks of both the caller and call recipient must have SS7 capability, and the local telephone exchange networks of both caller and call recipient must be connected to AT&T's long-distance 5 The FCC, for example, in 1994 concluded: “Service providers who respond to telephone orders, such as stock brokers or parts and equipment dealers, could use the calling party's number to direct the call immediately to the appropriate department for service. Banks could program data sources to have customer profile information available as a call is answered. With interstate delivery of calling party number, calls to national service centers could be routed automatically to local service centers closest to the calling party. Consumers making orders could have their name, address and billing information verified instantaneously. Indeed, a significant number and kind of customized national services can develop as a result of instant recognition of the calling party.” In re Rules and Policies Regarding Calling Number Identification Service—Caller ID, 9 F.C.C. Rcd. 1764 (1994) (hereinafter Rules and Policies). 6 “The term 'Signaling System 7' (SS7) refers to a carrier to carrier out-of-band signaling network used for call routing, billing and management.” 47 C.F.R. § 64.1600 (2002). The district court found, and all the parties concede, that full SS7 connectivity is a prerequisite for a call to carry CPN from the caller to the call recipient. 5 SS7 network.7 Even where there was complete SS7 connectivity, however, a number of additional factors, other than AT&T's conduct, may have prevented the unimpeded passage of a caller ID signal during the class period. Some states, concerned with the implications of caller ID for privacy rights and existing wiretapping legislation, imposed regulations that blocked the transmission of CPN on all calls, both local and long-distance. In re Rules and Policies Regarding Calling Number Identification Service—Caller ID, 9 F.C.C. Rcd. 1764 (1994). Thus AT&T alleged, and the plaintiffs do not dispute, that Texas prohibited the transmittal of CPN for a substantial portion of the class period, while California did not permit the passage of CPN at any time during the class period. Pennsylvania did not amend its statutes to permit caller ID service until December 1993. See 66 PA. CONS. STAT. § 2906(a) (providing for caller ID service and overruling Barasch v. Pennsylvania Public Utilities Commission, 576 A.2d 79 (1990), which held that caller ID violated state wiretap laws). Other states required, and still require, that telephone companies provide consumers with the option of blocking the display of their telephone numbers. See, e.g., 7 “Because transmission of the calling party number requires SS7 technology, technical feasibility exists wherever SS7 technology is used.” Rules and Policies, supra note 5. According to the record, only 65% of business-access lines were fully SS7 connected as of July 1993; only 72% of local networks had achieved full SS7 connectivity by the end of the class period in 1995. 6 CAL. PUB. UTIL. CODE § 2893(a) (West 2003) (requiring, with certain exceptions, that “every telephone call identification service . . . shall allow a caller to withhold display of the caller's telephone number”); 66 PA. CONS. STAT. § 2906(a) (same). Finally, beyond state regulations and SS7 connectivity, other technological barriers may have prevented the transmission of caller ID data during the class period. Thus, the record indicates that CPN does not accompany a call where the call is placed either from a pay phone or from a cellular phone. In addition, caller ID service may be unavailable where either the calling party or the call recipient employs a private branch exchange (PBX) telephone system, a type of telephone system widely used by businesses during the class period. Assuming, however, that none of these various barriers would have impeded the receipt of caller ID, there is no question but that caller ID was unavailable on certain calls during the class period because of AT&T's decision to block the free transmission of caller ID signals over its long-distance network. Moreover, for purposes of deciding class certification, we shall simply assume without deciding, as did the district court, that such conduct amounted to an attempt to monopolize the market for caller ID service in violation of the Sherman Act. The only question remaining before this court, therefore, is the propriety of the district court's denial of class certification. B. Class Definitions 7 The named plaintiffs, Rochelle and Adroit, initially sought to certify two classes of plaintiffs who they maintained suffered antitrust injury as a result of AT&T's blocking of caller ID data: (1) a reverse charge class comprising business and organizations who purchased AT&T's reverse-billed (“1-800") long-distance service; and (2) a call recipient class comprising businesses and organizations that were actual or potential purchasers of caller-ID service for long-distance calls.8 After a hearing during which the district court received evidence concerning the operation of caller ID service and the nature of the antitrust injury alleged, the plaintiffs moved to redefine the putative classes.9 The plaintiffs' amended motion for class certification, the denial of which is before us, removed certain problematic aspects of the initial definitions, and defined 8 The plaintiffs’ first motion defined the reverse charge class as “a class of all businesses and organizations that, at any time between March 19, 1992 and November 30, 1995, have been actual purchasers from AT&T of reverse-billed (typically “800") switched-access, long-distance service, and who have been actual or potential purchasers of Caller-ID service for the processing of incoming switched-access long distance telephone calls.” The call recipient class was defined as “a class of all businesses and organizations that, at any time between March 19, 1992 and November 30, 1995, have been actual or potential purchasers of Caller-ID service for the processing of incoming switched-access long distance telephone calls.” 9 Between the plaintiffs' initial motion to certify the two classes and the subsequent hearing on that motion, Bell Atlantic settled its claim against AT&T. 8 the reverse charge class as “All businesses and organizations that, at any time between March 19, 1992 and November 30, 1995, were actual purchasers from AT&T of reverse-billed (typically “800") switched-access, long distance service, and were actual purchasers of Caller-ID service for the processing of incoming switched-access telephone calls.” The call recipient class, whose members who were not necessarily all AT&T subscribers, was defined as “All business and organizations that, at any time between March 19, 1992 and November 30, 1995 (a) were actual purchasers of Caller-ID service for the processing of incoming switched-access telephone calls; and (b) received at least one AT&T long-distance call carried over the SS7 signaling network each month during which such purchaser was a subscriber to the switched-access Caller-ID service.” In addition to defining the proposed classes, the plaintiffs' motion for certification also included a formula for calculating the amount of damages to which the plaintiffs claimed the class members were entitled. Specifically, the plaintiffs proposed to calculate damages for individual class members in both classes based upon the national “average number of seconds saved per call” [both long-distance and local] through the use of caller ID, an average wage rate for the typical employee answering and processing telephone calls, and the total number of AT&T calls to class members made during the class period.10 For the reverse charge 10 With respect to the reverse charge class, the plaintiffs' expert presented the following proposal for measuring damages: “Measure the number of AT&T switched access 800 calls that were completed to class members over the damages period. This amount, multiplied by the average net cost savings per call that is typical for users of 9 class, the plaintiffs also proposed to adjust the damages calculus, using AT&T's billing records, to include recovery of any long- distance charges assessed against class members that might have otherwise been avoided through the use of caller ID. AT&T opposed class certification arguing, inter alia, that the plaintiffs had failed to carry their burden of establishing predominance and numerosity. Specifically, AT&T contended that the plaintiffs' motion for certification failed the predominance bar on two grounds. First, AT&T maintained that the plaintiffs could not prove antitrust injury with regard to each class member absent an individualized inquiry into issues of causation. Second, even assuming the plaintiffs could establish liability on a classwide basis, AT&T argued that the plaintiffs still could not clear the predominance hurdle since the variegated nature of the class member businesses and organizations precluded a formulaic calculation of damages. The district court, after holding a second hearing on the long-distance Caller-ID, results in the net amount of the potential savings that class members would have realized if not for the alleged actions of AT&T.” With respect to the damages owed members of the call recipient class, the plaintiffs' expert testified that “it is only necessary to determine the number of inbound AT&T long-distance calls to the class during the damages period and multiply that by the average net cost savings per call that is typical for users of long-distance Caller-ID. That product reflects the net amount of the potential savings that class members would have realized if not for the alleged actions of AT&T.” 10 issue, denied class certification. Seizing on the first of AT&T's two above objections and relying on Alabama v. Blue Bird Body Co., Inc., 573 F.2d 309 (5th Cir. 1978), the district court found that difficulties in establishing that AT&T had actually caused an antitrust injury to any given class member defeated predominance.11 Following the district court's ruling, the plaintiffs petitioned for, and were granted leave to file this interlocutory appeal under Rule 23(f). Because we also conclude that the plaintiffs clearly failed to surmount the predominance hurdle of Rule 23(b)(3), albeit on different grounds from those relied upon by the district court, we affirm. Discussion A. Standard of Review 11 With respect to the reverse charge class, the court concluded that the plaintiffs could not demonstrate a viable method of establishing, through proof common to the class, that any individual class member was both connected to the SS7 network and received a telephone call from a calling party also connected to the SS7 network. Absent such proof, the court concluded that the plaintiffs could not establish that, but for AT&T's refusal to transmit CPN over its network, any class member would have received caller ID data with any given call. The district court reached a similar conclusion with respect to the call recipient class. According to the district court, with the exception of records covering only one year of the class period, no records were available from which the class members could determine the source of any given call during the class period. Absent records from which a call recipient could determine that he had been called on a certain date from a certain number, the court concluded that the plaintiffs could not identify the origin of a specific call, let alone whether that call originated from an area that was connected to the SS7 network. Absent that information, no given class member could establish the requisite element of causation. 11 We review the district court’s decision to certify a class for an abuse of discretion, see McManus v. Fleetwood Enterprises, Inc., 320 F.3d 545, 548 (5th Cir. 2003), and note that the district court must “conduct a ‘rigorous analysis of the Rule 23 prerequisites’ before certifying a class.” O'Sullivan v. Countrywide Home Loans, Inc., 319 F.3d 732, 738 (5th Cir. 2003). We also note that in those cases where the plaintiff seeks to certify a class under Rule 23(b)(3), the Rules “invite[] a close look at the case before it is accepted as a class action.” Amchem Products, Inc. v. Windsor, 117 S.Ct. 2231, 2246 (1997) (quoting Kaplan, Continuing Work of the Civil Committee: 1966 Amendments of the Federal Rules of Criminal Procedure (I), 81 HARV. L. REV. 356, 375 (1967)). Finally, we stress that it is the party seeking certification who bears the burden of establishing that the requirements of Rule 23 have been met. O'Sullivan, 319 F.3d at 737–738. B. Class Certification There are no “hard and fast rules . . . regarding the suitability of a particular type of antitrust case for class action treatment.” Alabama v. Blue Bird Body Co., Inc., 573 F.2d 309, 316 (5th Cir. 1978). Rather, “[t]he unique facts of each case will generally be the determining factor governing certification.” Id. Under Rule 23(a), a plaintiff seeking to certify a class must satisfy four threshold requirements: “(1) numerosity (a 'class [so large] that joinder of all members is impracticable'); (2) 12 commonality ('questions of law or fact common to the class'); (3) typicality (named parties' claims or defenses 'are typical . . . of the class'); and (4) adequacy of representation (representatives 'will fairly and adequately protect the interests of the class').” Amchem Products, 117 S.Ct. at 2245. Beyond these four prerequisites of Rule 23(a), Rule 23(b)(3) demands of a party seeking class certification yet two further requirements, namely the burden of demonstrating both (1) that questions common to the class members predominate over questions affecting only individual members, and (2) that class resolution is superior to alternative methods for adjudication of the controversy. Id. at 2246. By inquiring into predominance, Rule 23(b)(3) thus tests “whether the proposed classes are sufficiently cohesive to warrant adjudication by representation.” Id. at 2249. The standard for certification imposed by Rule 23(b)(3) is also more demanding than the commonality requirement of Rule 23(a), and as such, mandates caution, particularly where “individual stakes are high and disparities among class members great.” Id. at 2250; see also FED. R. CIV. P. 23 advisory committee's note (“In the situations to which this subdivision [Rule 23(b)(3)] relates, class-action treatment is not as clearly called for as in those described above, but it may nevertheless be convenient and desirable depending upon the particular facts.”) (emphasis added). Determining whether the plaintiffs can clear the predominance 13 hurdle set by Rule 23(b)(3) also requires us to consider “how a trial on the merits would be conducted if a class were certified.” Sandwich Chef of Texas, Inc. v. Reliance Nat'l Ins. Indem. Co., 319 F.3d 205, 218 (5th Cir. 2003); Castano v. Am. Tobacco Co., 84 F.3d 734, 740 (5th Cir. 1996). This, in turn, “entails identifying the substantive issues that will control the outcome, assessing which issues will predominate, and then determining whether the issues are common to the class,” a process that ultimately “prevents the class from degenerating into a series of individual trials.” O'Sullivan, 319 F.3d at 738. C. The Antitrust Violation The offense of attempted monopolization in violation of section 2 of the Sherman Act has three elements, namely: (1) that the defendant engaged in predatory or exclusionary conduct, (2) that the defendant possessed the specific intent to monopolize, and (3) that there was a dangerous probability that the defendant would succeed in his attempt. Taylor Pub. Co. v. Jostens, Inc., 216 F.3d 465, 474 (5th Cir. 2000). Proof of these elements will necessarily be identical for the members of both proposed classes, and under the facts of the instant case, these issues, therefore, create no bar to class certification. Moreover, as indicated above, we assume, for purposes of addressing the issue of class certification, that AT&T's alleged conduct constituted a violation of section 2. 14 The plaintiffs' task, however, is not limited to establishing the elements of a completed offense under section 2 of the Sherman Act. Rather, to establish civil liability under the Clayton Act, a plaintiff must also establish that he has been injured in his “business or property by reason of anything forbidden in the antitrust laws.” 15 U.S.C. § 15. Thus, where a plaintiff seeks a private civil remedy and treble damages for a violation of section 2, he must not only make out a violation of the antitrust laws, but also (1) establish that it was the defendant's conduct that actually caused injury to his business or property,12 and (2) provide “some indication of the amount of damage.” Blue Bird Body Co., 573 F.2d at 317. Establishing causation, or “fact of damage”, requires the plaintiff to demonstrate a causal connection between the specific antitrust violation at issue and an injury to the business or property of the antitrust plaintiff. Id. This requirement is in no way lessened by reason of being raised in the context of a class action. Rather, this court has held that the issue of fact of damage “is a question unique to each particular plaintiff and one that must be proved with certainty.” Id. at 327. Accordingly, we have repeatedly held that where fact of damage cannot be established for every class member through proof common to the 12 This requirement of causation is often referred to as “impact” or “fact of damage.” Blue Bird Body Co., 573 F.2d at 317 n.18. 15 class, the need to establish antitrust liability for individual class members defeats Rule 23(b)(3) predominance. See Nichols v. Mobile Board of Realtors, Inc., 675 F.2d 671 (5th Cir. Unit B 1982); Alabama v. Blue Bird Body Co., 573 F.2d 309 (5th Cir. 1978); Schumate & Co. v. National Ass'n of Secs. Dealers, 509 F.2d 147 (5th Cir. 1975). In addition to establishing fact of damage, section 4 of the Clayton Act also requires a plaintiff to show “some indication of the amount of damage” suffered. See Blue Bird Body, 573 F.2d at 317. We have recognized, however, that the nature of an antitrust claim means that “some plaintiffs can only hypothesize about what the state of their affairs would have been absent the wrong,” H&B Equipment Co. v. International Harvester Co., 577 F.2d 239, 246 (5th Cir. 1978), and we have, therefore, declined to hold antitrust plaintiffs to the same burden of proof of damages as demanded of plaintiffs in other civil cases. See Eleven Line, Inc. v. North Texas State Soccer Ass'n, 213 F.3d 198, 206–207 (5th Cir. 2000). Such leniency notwithstanding, an antitrust plaintiff may not merely rely on “guesswork or speculation” to establish damages. Id. Rather, our cases indicate that the plaintiff must provide a “just and reasonable estimate of the damage based on relevant data.” Id. (quoting Bigelow v. RKO Radio Pictures, 66 S.Ct. 574, 580 (1946); see, e.g., Kestenbaum v. Falstaff Brewing Co., 575 F.2d 564, 569 (5th Cir. 1978)(“When asserting injury from the imposition 16 of price ceilings, the plaintiff must show when prices would have been raised, by what amount, and approximately what sales would have been at the higher price.”). And we have accordingly rejected claims where the plaintiff's proposed method of calculating damages failed to reasonably approximate actual economic losses. See Eleven Line, Inc., 213 F.3d at 208–209; Keener v. Sizzler Family Steak Houses, 597 F.2d 453, 457 (5th Cir. 1979). D. Damages and Predominance Having thus identified fact of damage and the amount of damages as the two elements of the plaintiffs' claim that would be at issue at trial were the two proposed classes to be certified, we now address whether, in light of those elements, individual issues would predominate at trial. The district court found that the plaintiffs failed to demonstrate that they could establish antitrust liability through common proof, and that individual issues concerning fact of damage, therefore, defeated predominance. The plaintiffs assign this finding as error and ask us to reverse. We demur, since even if we were to conclude that the district court's decision as to fact of damage was in error, we find that the plaintiffs' motion for certification nevertheless founders on the issue of the amount of damages.13 13 AT&T argued, both before this court and before the district court, that the issue of damages precluded class certification. Specifically, AT&T maintained that the 17 As discussed above, see Part I(B) supra, the plaintiffs proposed to calculate damages for the members of both classes according to a formula that utilized a nationwide average cost of labor and a nationwide average amount of time that the class members would have saved per call had caller ID been available on long-distance calls during the class period. Upon reviewing the plaintiffs' damages formula was inadequate to determine the class members' damages, and that the need for individual inquiries into damages defeated rule 23(b)(3) predominance. The record, consequently, is fully developed on this point, and, although the district court did not rule on this basis, we see no bar to basing our decision on this alternative ground. AT&T also asserted, both on appeal and before the district court, that the plaintiffs had failed to allege a cognizable antitrust injury. See Bell v. Dow Chemical, 847 F.2d 1179, 1183 (5th Cir. 1988) (holding that a “plaintiff's injury must be the type that the antitrust laws were intended to prevent”). The plaintiffs' asserted theory of antitrust injury is admittedly a novel one, consisting of the claim that by stripping calls of caller ID data, AT&T injured the class members by “degrading” the value of their caller ID service. Whether such a general, unquantifiable degradation in the value of a product, unconnected to any objective market indicators, amounts to an antitrust injury, however, is unclear. See H&B Equipment Co. v. International Harvester Co., 577 F.2d 239, 247 (5th Cir. 1978) (noting that where a plaintiff relies on lost sales to show fact of damage, the plaintiff will have to “show a specific monetary loss”); Kestenbaum v. Falstaff Brewing Co., 575 F.2d 564, 569 (5th Cir. 1978) (holding, in the context of a price fixing claim, that “the plaintiff must show when prices would have been raised, by what amount, and approximately what sales would have been at the higher price”); Midwestern Waffles Inc. v. Waffle House Inc., 734 F.2d 705, 723 n.3 (11th Cir. 1984) (noting that a showing of injury to business or property within the meaning of the Clayton Act requires a plaintiff to “be able to demonstrate that it suffered economic damages which are quantifiable”). And although AT&T's argument may have some merit, because we hold that the plaintiffs cannot establish predominance under Rule 23(b)(3), it is not necessary to pass on that argument for the purposes of resolving the instant appeal. 18 record, however, we are not convinced that this proposed damages calculus represents an adequate approximation of any single class member's damages, let alone a just and reasonable estimate of the damages of every class member included in the two putative classes. The record indicates that rather than merely examining lost time and average labor costs, any adequate estimation of actual damages suffered would require consideration of the variegated nature of the businesses included in both the proposed classes, together with the range of uses, depending on the size and technological sophistication of any given business, to which caller ID could be applied. In light of the need for such individualized inquiries, we cannot conclude that the plaintiffs have established that the requirements of Rule 23(b)(3) can be satisfied in the present case. In Eleven Line, Inc. v. North Texas State Soccer Ass'n, 213 F.3d 198, 208–209 (5th Cir. 2000), we found inadequate an antitrust plaintiff's damages formula that, based on an average of the rates of return of similar businesses, failed to account, among other things, for differences in location and size between the various businesses used to calculate the average. We also rejected the argument that to disallow the plaintiff's damages formula would be to let anticompetitive conduct go unpunished because of “mere uncertainty in the amount of loss inflicted,” noting instead that the lenient standard for proving the amount of damages under the 19 antitrust laws should not be stretched so far as to permit recovery where there was no evidence that competition had actually been eroded. Id at 209. The plaintiffs' formula in the case sub judice suffers from the same flaws that proved fatal to the plaintiff's formula in Eleven Line, Inc. v. North Texas State Soccer Ass'n. Numerous factors that would have affected the amount of damages, if any, suffered by any given class member denied caller ID are not accounted for in the proposed formula. It is not contested that certain class member businesses were denied substantial efficiency gains because of their inability to receive caller ID with every call. Not every business included in the class, however, would have achieved the same per call savings, a fact that the plaintiffs conceded when during the second hearing on class certification, counsel for the class representatives agreed that while some class members may have a claim for substantial damages, others “may not even get a dollar's worth of damages . . . . They may get nothing.” Indeed, we fail to see, and the plaintiffs have failed to demonstrate, how for some businesses within each class, the absence of caller ID on a handful of calls would have had any effect whatever on those businesses' bottom lines. As AT&T has repeatedly pointed out, both before this court and before the district court, it is not difficult to conceive of a business that would fall within the definition of the call 20 recipient class, but that would nevertheless not have suffered any economic injury by reason of being denied caller ID on one long- distance call each month while a caller-ID subscriber during the class period. Thus AT&T posits the local pizza-delivery business, that serves a local customer base, and that receives one long- distance call each month. Such a business falls within the definition of the reverse charge class, yet it is unlikely that such a business would ever suffer any actual economic injury by being denied caller ID data on a single long-distance call each month. Even assuming that such a pizza-delivery business utilized a non-local supplier, before concluding that the business suffered any actual economic injury from the absence of caller ID information on a call from the supplier, one would have to also make both of the following dubious and highly speculative assumptions about each such call, namely that had caller ID been present, the employee answering the call would have saved some amount of time and that the employee’s time was otherwise fully utilized. The conclusion that such a local business would be actually economically injured from the absence of caller ID information in such a situation, is therefore equally speculative.14 14 It is also possible to conceive of a member of the reverse charge class that might not have suffered any actual economic injury from the denial of caller ID information on one long distance call each month. Thus, the proposed reverse charge class would include a local business owner who purchased switched-access, long-distance service from AT&T, but who only received calls at that number from an out-of-state relative. It 21 According to the record, most labor and time savings would only have been realized by those businesses that possessed both CTI equipment and software that would have enabled them to utilize caller ID in conjunction with a customer database. The class definitions, however, do not seek to distinguish between those businesses that did, and those that did not, possess CTI equipment. Moreover, even among those businesses that did possess CTI equipment, the amount of labor savings realized, if any, would have varied greatly. The record reveals that a wide variety of CTI equipment, ranging in expense and performance, was available during the class period, the effectiveness of which apparently depended, to at least some extent, on the type of database software with which it was used, a factor that also varied among businesses. Not only does the plaintiffs' proposed damages formula thus fail to account for disparities in potential labor savings among class members due to variations in, or the absence of, CTI equipment, but it also fails to account for other differences among class members that would have affected the amount, if any, of actual economic damages suffered. The damages formula does not account for those businesses, included in the definitions of both classes, that employed PBX telephone systems (or received long- distance calls only or primarily from other business that employed is difficult to conclude that the denial of caller ID information on those personal calls would have had any effect whatever on that business owner’s bottom line. 22 PBX systems) and that, therefore, could never have received caller ID data regardless of AT&T's conduct. Nor does the formula reflect that those businesses that served high volumes of repeat customers stood to gain more from caller ID, both in terms of reduced labor costs and increased customer satisfaction, than businesses that served ever-changing customer bases.15 Finally, neither the class definitions nor the damages formula purports to adjust for the reduced level of damages that would be due those businesses that may have served customers primarily residing in states that required that callers be given the option of blocking caller ID signals from accompanying their calls. Where an antitrust plaintiff seeks to project lost profits by comparing like businesses, it is the plaintiff who must “demonstrate the reasonable similarity of the business whose earning experience he would borrow.” Eleven Line, Inc., 213 F.3d at 208. Similarly, where the injury alleged is measured in terms of an average of lost labor savings and an average amount of time saved per phone call, it is the plaintiff who must demonstrate the reasonable similarity of the businesses used to calculate those 15 This difference arises from the fact that CTI equipment is most useful where data concerning the caller is already stored in the call recipient's database. If a business serves repeat customers, information concerning most of that business's customers will be stored in that business's database. If, however, a business deals primarily with one-time customers, there will be no data stored in a database for most callers. Caller ID and CTI, therefore, provide less utility for, and its deprivation less harm to businesses in the latter group. 23 averages. This the plaintiffs have failed to do. The plaintiffs' proposed damages formula instead attempts to project a measure of damages, for all the class members, that in no way accounts for the vast differences among those class members. Any reasonable approximation of the damages actually suffered by the various class members would instead require a much tighter inquiry into the nature of the class member businesses. Given the need for such individualized damages inquiries, we conclude that individual issues concerning damages predominate over questions common to the proposed classes. We realize that relatively few motions to certify a class fail because of disparities in the damages suffered by the class members. Even wide disparity among class members as to the amount of damages suffered does not necessarily mean that class certification is inappropriate, see Gold Strike Stamp Co. v. Christensen, 436 F.2d 791, 798 (10th Cir. 1970),16 and courts, 16 The courts' ability to sever the damages portion of a class action suit from the liability portion is the principal reason why variations among class members in the amount of damages suffered frequently does not defeat predominance. See, e.g., Bogosian v. Gulf Oil Corp., 561 F.2d 434, 456 (3d Cir. 1977) (noting that “[i]f for any reason the district court were to conclude that there would be problems involved in proving damages which would outweigh the advantages of class certification, it should give appropriate consideration to certification of a class limited to the determination of liability.”); 7B CHARLES ALAN WRIGHT, ARTHUR R. MILLER & MARY KAY KANE, FEDERAL PRACTICE AND PROCEDURE § 1781 (2d ed. 1986) (“[T]he question of damages can be severed from that of liability and tried on an individual basis.”). The plaintiffs here, however, never proposed such a bifurcated trial. 24 therefore, have certified classes even in light of the need for individualized calculations of damages.17 Class treatment, however, may not be suitable where the calculation of damages is not susceptible to a mathematical or formulaic calculation, or where the formula by which the parties propose to calculate individual damages is clearly inadequate. Thus the Fourth Circuit held in Windham v. American Brands, Inc. that where the issue of damages “does not lend itself to . . . mechanical calculation, but requires 'separate “mini-trial[s]”’ of an overwhelmingly large number of individual claims,” the need to calculate individual damages will defeat predominance, 565 F.2d 59, 68 (4th Cir. 1977) (internal quotations and alterations omitted), a holding reiterated in Broussard v. Meineke Discount Muffler Shops, Inc., 155 F.3d 331, 342–343 (4th Cir. 1998), in which the Fourth Circuit again noted that where the claims of the members of a putative class of 17 “[I]t uniformly has been held that differences among the members [of a class] as to the amount of damages incurred does not mean that a class action would be inappropriate.” 7B CHARLES ALAN WRIGHT, ARTHUR R. MILLER & MARY KAY KANE, FEDERAL PRACTICE AND PROCEDURE § 1781 (2d ed. 1986) (collecting cases); see Bogosian, 561 F.2d at 456 (noting that where proving damages is a mechanical task, “the necessity for calculation of damages on an individual basis should not preclude class determination when the common issues which determine liability predominate”); Brown v. Pro Football, Inc., 146 F.R.D. 1 (D.D.C. 1992) (finding a proposed damages formula, based on individual contracts entered into by various plaintiff, adequate to calculate antitrust damages); In re Polypropylene Carpet Antitrust Litig., 996 F.Supp. 18, (N.D. Ga. 1997) (permitting class certification in an antitrust price- fixing case where the plaintiffs proposed to use regression analysis to estimate class members' damages). 25 antitrust plaintiffs are “inherently individualized,” “the need for individual proof of damages” will bar class certification. Although we have not, before the instant case, directly addressed the relationship between proof of antitrust damages and Rule 23(b)(3) predominance, this court quoted Windham v. American Brands with approval in Alabama v. Blue Bird Body Co., when we expressed “serious reservations about the manageability of a class” in an antitrust action where the determination of individual damages cannot be made by mathematical or formulaic computation. 573 F.2d 309, 329 (5th Cir. 1978). More recently, we held in the context of a claim brought under Texas law, that “[w]here the plaintiffs' damage claims 'focus almost entirely on facts and issues specific to individuals rather than the class as a whole,' the potential . . . that the class action may 'degenerate in practice into multiple lawsuits separately tried,” renders class treatment inappropriate. See O'Sullivan v. Countrywide Home Loans, 319 F.3d 732, 744 (5th Cir. 2003). Upon reviewing the record in light of O'Sullivan, Blue Bird Body Co., and Windham, we cannot conclude that the plaintiffs' two proposed classes are appropriate for certification. As discussed above, there are vast differences among the numerous businesses that compose the two classes. The businesses that fall within those classes range from sole proprietorships, with customer bases comprising almost exclusively local consumers, that received a 26 minimum of long-distance calls and for whom caller ID would have provided minimal, if any, labor savings, to large, interstate catalogue companies that maintained large call centers, possessed CTI equipment, and for whom caller ID might have produced substantial benefits in both labor savings and customer satisfaction. When applied to all the members included in the two proposed classes, however, the plaintiffs' damages formula—a formula based on nationwide averages that makes no effort to adjust for the variegated nature of the businesses included in the classes—cannot reasonably approximate the actual damages suffered by the class members by reason of AT&T's blocking of caller ID signals. In light of the requirements of Rule 23(b)(3), therefore, we hold that the plaintiffs have clearly failed to demonstrate that common issues of fact predominate over those individual issues of fact that are plainly necessary for any just estimate of the antitrust damages suffered by the class members. Conclusion Because we conclude that the issue of damages defeats predominance, we decline to address the district court's holding that the plaintiffs failed to demonstrate that they would be able to establish antitrust impact, through common proof, for either proposed class. Even assuming that the plaintiffs can establish antitrust liability with respect to all the class members, we conclude that the plaintiffs, having had a fully adequate 27 opportunity to address the issue of damages below, clearly failed to demonstrate that the calculation of individualized actual economic damages, if any, suffered by the class members can be performed in accordance with the predominance requirement of Rule 23(b)(3). We conclude, therefore, that class certification is not appropriate; the district court's order denying certification is therefore AFFIRMED. 28