SUPREME COURT OF THE UNITED STATES
27 MARCH 2013
(delivered the opinion of the Court)
The District Court and the Court of Appeals approved certification of a class of more than 2 million current and former Comcast subscribers who seek damages for alleged violations of the federal anti-trust laws. We consider whether certification was appropriate under Federal Rule of Civil Procedure 23(b)(3).
Comcast Corporation and its subsidiaries, petitioners here, provide cable-television services to residential and commercial customers. From 1998 to 2007, petitioners engaged in a series of transactions that the parties have described as “clustering,” a strategy of concentrating operations within a particular region. The region at issue here, which the parties have referred to as the Philadelphia “cluster” or the Philadelphia “Designated Market Area” (DMA), includes 16 counties located in Pennsylvania, Delaware, and New Jersey. Petitioners pursued their clustering strategy by acquiring competitor cable providers in the region and swapping their own systems outside the region for competitor systems located in the region. For instance, in 2001, petitioners obtained Adelphia Communications’ cable systems in the Philadelphia DMA, along with its 464,000 subscribers; in exchange, petitioners sold to Adelphia their systems in Palm Beach, Florida, and Los Angeles, California. As a result of nine clustering transactions, petitioners’ share of subscribers in the region allegedly increased from 23.9 percent in 1998 to 69.5 percent in 2007. See 264 F. R. D. 150, 156, n. 8, 160 (ED Pa. 2010).
The named plaintiffs, respondents here, are subscribers to Comcast’s cable-television services. They filed a class-action anti-trust suit against petitioners, claiming that petitioners entered into unlawful swap agreements, in violation of §1 of the Sherman Act, and monopolized or attempted to monopolize services in the cluster, in violation of §2. Ch. 647, 26Stat. 209, as amended, 15 U. S. C. §§1, 2. Petitioners’ clustering scheme, respondents contended, harmed subscribers in the Philadelphia cluster by eliminating competition and holding prices for cable services above competitive levels.
Respondents sought to certify a class under Federal Rule of Civil Procedure 23(b)(3). That provision permits certification only if “the court finds that the questions of law or fact common to class members predominate over any questions affecting only individual members.” The District Court held, and it is uncontested here, that to meet the predominance requirement respondents had to show
that the existence of individual injury resulting from the alleged anti-trust violation (referred to as “anti-trust impact”) was “capable of proof at trial through evidence that [was] common to the class rather than individual to its members”; and
that the damages resulting from that injury were measurable “on a class-wide basis” through use of a “common methodology.” 264 F. R. D., at 154.
Respondents proposed four theories of anti-trust impact:
First, Comcast’s clustering made it profitable for Comcast to withhold local sports programming from its competitors, resulting in decreased market penetration by direct broadcast satellite providers.
Second, Comcast’s activities reduced the level of competition from “overbuilders,” companies that build competing cable networks in areas where an incumbent cable company already operates.
Third, Comcast reduced the level of “benchmark” competition on which cable customers rely to compare prices.
Fourth, clustering increased Comcast’s bargaining power relative to content providers. Each of these forms of impact, respondents alleged, increased cable subscription rates throughout the Philadelphia DMA.
The District Court accepted the overbuilder theory of anti-trust impact as capable of classwide proof and rejected the rest. Id., at 165, 174, 178, 181. Accordingly, in its certification order, the District Court limited respondents’ “proof of anti-trust impact” to “the theory that Comcast engaged in anti-competitive clustering conduct, the effect of which was to deter the entry of overbuilders in the Philadelphia DMA.” App. to Pet. for Cert. 192a–193a.
The District Court further found that the damages resulting from overbuilder-deterrence impact could be calculated on a classwide basis. To establish such damages, respondents had relied solely on the testimony of Dr. James McClave. Dr. McClave designed a regression model comparing actual cable prices in the Philadelphia DMA with hypothetical prices that would have prevailed but for petitioners’ allegedly anti-competitive activities. The model calculated damages of $875,576,662 for the entire class. App. 1388a (sealed). As Dr. McClave acknowledged, however, the model did not isolate damages resulting from any one theory of anti-trust impact. Id., at 189a–190a. The District Court nevertheless certified the class.
A divided panel of the Court of Appeals affirmed. On appeal, petitioners contended the class was improperly certified because the model, among other shortcomings, failed to attribute damages resulting from overbuilder deterrence, the only theory of injury remaining in the case. The court refused to consider the argument because, in its view, such an “attac[k] on the merits of the methodology [had] no place in the class certification inquiry.” 655 F. 3d 182, 207 (CA3 2011). The court emphasized that, “[a]t the class certification stage,” respondents were not required to “tie each theory of anti-trust impact to an exact calculation of damages.” Id., at 206. According to the court, it had “not reached the stage of determining on the merits whether the methodology is a just and reasonable inference or speculative.” Ibid. Rather, the court said, respondents must “assure us that if they can prove anti-trust impact, the resulting damages are capable of measurement and will not require labyrinthine individual calculations.” Ibid. In the court’s view, that burden was met because respondents’ model calculated “supra-competitive prices regardless of the type of anti-competitive conduct.” Id., at 205.
We granted certiorari. 567 U. S. ___ (2012).
The class action is “an exception to the usual rule that litigation is conducted by and on behalf of the individual named parties only.” Califano v Yamasaki, 442 U. S. 682–701 (1979). To come within the exception, a party seeking to maintain a class action “must affirmatively demonstrate his compliance” with Rule 23. Wal-Mart Stores, Inc. v Dukes, 564 U. S. ___, ___ (2011) (slip op., at 10). The Rule “does not set forth a mere pleading standard.” Ibid. Rather, a party must not only “be prepared to prove that there are in fact sufficiently numerous parties, common questions of law or fact,” typicality of claims or defences, and adequacy of representation, as required by Rule 23(a). Ibid. The party must also satisfy through evidentiary proof at least one of the provisions of Rule 23(b). The provision at issue here is Rule 23(b)(3), which requires a court to find that “the questions of law or fact common to class members predominate over any questions affecting only individual members.”
Repeatedly, we have emphasized that it “‘may be necessary for the court to probe behind the pleadings before coming to rest on the certification question,’ and that certification is proper only if ‘the trial court is satisfied, after a rigorous analysis, that the prerequisites of Rule 23(a) have been satisfied.’” Ibid. (quoting General Telephone Co. of Southwest v Falcon, 457 U. S. 147–161 (1982)). Such an analysis will frequently entail “overlap with the merits of the plaintiff’s underlying claim.” 564 U. S., at ___ (slip op., at 10). That is so because the “ ‘class determination generally involves considerations that are enmeshed in the factual and legal issues comprising the plaintiff’s cause of action.’” Ibid. (quoting Falcon, supra, at 160).
The same analytical principles govern Rule 23(b). If anything, Rule 23(b)(3)’s predominance criterion is even more demanding than Rule 23(a). Amchem Products, Inc. v Windsor, 521 U. S. 591–624 (1997). Rule 23(b)(3), as an “‘adventuresome innovation,’” is designed for situations “‘in which “class-action treatment is not as clearly called for.”’” Wal-Mart, supra, at ___ (slip op., at 22) (quoting Amchem, 521 U. S., at 614–615). That explains Congress’s addition of procedural safeguards for (b)(3) class members beyond those provided for (b)(1) or (b)(2) class members (e.g., an opportunity to opt out), and the court’s duty to take a “‘close look’” at whether common questions predominate over individual ones. Id., at 615.
Respondents’ class action was improperly certified under Rule 23(b)(3). By refusing to entertain arguments against respondents’ damages model that bore on the propriety of class certification, simply because those arguments would also be pertinent to the merits determination, the Court of Appeals ran afoul of our precedents requiring precisely that inquiry. And it is clear that, under the proper standard for evaluating certification, respondents’ model falls far short of establishing that damages are capable of measurement on a classwide basis. Without presenting another methodology, respondents cannot show Rule 23(b)(3) predominance: Questions of individual damage calculations will inevitably overwhelm questions common to the class. This case thus turns on the straightforward application of class-certification principles; it provides no occasion for the dissent’s extended discussion, post, at 5–11 (Ginsburg and Breyer, JJ., dissenting), of substantive anti-trust law.
We start with an unremarkable premise. If respondents prevail on their claims, they would be entitled only to damages resulting from reduced overbuilder competition, since that is the only theory of anti-trust impact accepted for class-action treatment by the District Court. It follows that a model purporting to serve as evidence of damages in this class action must measure only those damages attributable to that theory. If the model does not even attempt to do that, it cannot possibly establish that damages are susceptible of measurement across the entire class for purposes of Rule 23(b)(3). Calculations need not be exact, see Story Parchment Co. v Paterson Parchment Paper Co., 282 U. S. 555, 563 (1931), but at the class-certification stage (as at trial), any model supporting a “plaintiff’s damages case must be consistent with its liability case, particularly with respect to the alleged anti-competitive effect of the violation.” ABA Section of Anti-trust Law, Proving Anti-trust Damages: Legal and Economic Issues 57, 62 (2d ed. 2010); see, e.g., Image Tech. Servs. v Eastman Kodak Co., 125 F. 3d 1195, 1224 (CA9 1997). And for purposes of Rule 23, courts must conduct a “‘rigorous analysis’” to determine whether that is so. Wal-Mart, supra, at ___ (slip op., at 10).
The District Court and the Court of Appeals saw no need for respondents to “tie each theory of anti-trust impact” to a calculation of damages. 655 F. 3d, at 206. That, they said, would involve consideration of the “merits” having “no place in the class certification inquiry.” Id., at 206–207. That reasoning flatly contradicts our cases requiring a determination that Rule 23 is satisfied, even when that requires inquiry into the merits of the claim. Wal-Mart, supra, at ___, and n. 6 (slip op., at 10–11, and n. 6). The Court of Appeals simply concluded that respondents “provided a method to measure and quantify damages on a classwide basis,” finding it unnecessary to decide “whether the methodology [was] a just and reasonable inference or speculative.” 655 F. 3d, at 206. Under that logic, at the class-certification stage any method of measurement is acceptable so long as it can be applied classwide, no matter how arbitrary the measurements may be. Such a proposition would reduce Rule 23(b)(3)’s predominance requirement to a nullity.
There is no question that the model failed to measure damages resulting from the particular anti-trust injury on which petitioners’ liability in this action is premised. The scheme devised by respondents’ expert, Dr. McClave, sought to establish a “but for” baseline – a figure that would show what the competitive prices would have been if there had been no anti-trust violations. Damages would then be determined by comparing to that baseline what the actual prices were during the charged period. The “but for” figure was calculated, however, by assuming a market that contained none of the four distortions that respondents attributed to petitioners’ actions. In other words, the model assumed the validity of all four theories of anti-trust impact initially advanced by respondents: decreased penetration by satellite providers, overbuilder deterrence, lack of benchmark competition, and increased bargaining power. At the evidentiary hearing, Dr. McClave expressly admitted that the model calculated damages resulting from “the alleged anti-competitive conduct as a whole” and did not attribute damages to any one particular theory of anti-competitive impact. App. 189a–190a, 208a.
This methodology might have been sound, and might have produced commonality of damages, if all four of those alleged distortions remained in the case. But as Judge Jordan’s partial dissent pointed out (655 F. 3d, at 216):
[B]ecause the only surviving theory of anti-trust impact is that clustering reduced overbuilding, for Dr. McClave’s comparison to be relevant, his benchmark counties must reflect the conditions that would have prevailed in the Philadelphia DMA but for the alleged reduction in overbuilding. In all respects unrelated to reduced overbuilding, the benchmark counties should reflect the actual conditions in the Philadelphia DMA, or else the model will identify ‘damages’ that are not the result of reduced overbuilding, or, in other words, that are not the certain result of the wrong.
[internal quotation marks omitted]
The majority’s only response to this was that “[a]t the class certification stage we do not require that Plaintiffs tie each theory of anti-trust impact to an exact calculation of damages, but instead that they assure us that if they can prove anti-trust impact, the resulting damages are capable of measurement and will not require labyrinthine individual calculations.” Id., at 206. But such assurance is not provided by a methodology that identifies damages that are not the result of the wrong. For all we know, cable subscribers in Gloucester County may have been overcharged because of petitioners’ alleged elimination of satellite competition (a theory of liability that is not capable of classwide proof); while subscribers in Camden County may have paid elevated prices because of petitioners’ increased bargaining power vis-à-vis content providers (another theory that is not capable of classwide proof); while yet other subscribers in Montgomery County may have paid rates produced by the combined effects of multiple forms of alleged anti-trust harm; and so on. The permutations involving four theories of liability and 2 million subscribers located in 16 counties are nearly endless.
In light of the model’s inability to bridge the differences between supra-competitive prices in general and supra-competitive prices attributable to the deterrence of overbuilding, Rule 23(b)(3) cannot authorize treating subscribers within the Philadelphia cluster as members of a single class. Prices whose level above what an expert deems “competitive” has been caused by factors unrelated to an accepted theory of anti-trust harm are not “anti-competitive” in any sense relevant here. “The first step in a damages study is the translation of the legal theory of the harmful event into an analysis of the economic impact of that event.” Federal Judicial Center, Reference Manual on Scientific Evidence 432 (3d ed. 2011) (emphasis added). The District Court and the Court of Appeals ignored that first step entirely.
The judgment of the Court of Appeals for the Third Circuit is reversed.
It is so ordered.
Justice Ginsburg and Justice Breyer
(with whom Justice Sotomayor and Justice Kagan join, dissenting)
Today the Court reaches out to decide a case hardly fit for our consideration. On both procedural and substantive grounds, we dissent.
This case comes to the Court infected by our misguided reformulation of the question presented. For that reason alone, we would dismiss the writ of certiorari as improvidently granted.
Comcast sought review of the following question: “[W]hether a district court may certify a class action without resolving ‘merits arguments’ that bear on [Federal Rule of Civil Procedure] 23’s prerequisites for certification, including whether purportedly common issues predominate over individual ones under Rule 23(b)(3).” Pet. for Cert. i. We granted review of a different question (567 U. S. ___ (2012)):
Whether a district court may certify a class action without resolving whether the plaintiff class has introduced admissible evidence, including expert testimony, to show that the case is susceptible to awarding damages on a class-wide basis.
Our rephrasing shifted the focus of the dispute from the District Court’s Rule 23(b)(3) analysis to its attention (or lack thereof) to the admissibility of expert testimony. The parties, responsively, devoted much of their briefing to the question whether the standards for admissibility of expert evidence set out in Federal Rule of Evidence 702 and Daubert v Merrell Dow Pharmaceuticals, Inc., 509 U. S. 579 (1993), apply in class certification proceedings. See Brief for Petitioners 35–49; Brief for Respondents 24–37. Indeed, respondents confirmed at oral argument that they understood our rewritten question to center on admissibility, not Rule 23(b)(3). See, e.g., Tr. of Oral Arg. 25.
As it turns out, our reformulated question was inapt. To preserve a claim of error in the admission of evidence, a party must timely object to or move to strike the evidence. Fed. Rule Evid. 103(a)(1). In the months preceding the District Court’s class certification order, Comcast did not object to the admission of Dr. McClave’s damages model under Rule 702 or Daubert. Nor did Comcast move to strike his testimony and expert report. Consequently, Comcast forfeited any objection to the admission of Dr. McClave’s model at the certification stage. At this late date, Comcast may no longer argue that respondents’ damages evidence was inadmissible.
Comcast’s forfeiture of the question on which we granted review is reason enough to dismiss the writ as improvidently granted. See Rogers v United States, 522 U. S. 252, 259 (1998) (O’Connor, J., concurring in result) (“[W]e ought not to decide the question if it has not been cleanly presented.”); The Monrosa v Carbon Black Export, Inc., 359 U. S. 180, 183 (1959) (dismissal appropriate in light of “circumstances .... not fully apprehended at the time certiorari was granted” (internal quotation marks omitted)). The Court, however, elects to evaluate whether respondents “failed to show that the case is susceptible to awarding damages on a class-wide basis.” Ante, at 5, n. 4 (internal quotation marks omitted). To justify this second revision of the question presented, the Court observes that Comcast “argued below, and continue[s] to argue here, that certification was improper because respondents had failed to establish that damages could be measured on a classwide basis.” Ibid. And so Comcast did, in addition to endeavouring to address the question on which we granted review. By treating the first part of our reformulated question as though it did not exist, the Court is hardly fair to respondents.
Abandoning the question we instructed the parties to brief does “not reflect well on the processes of the Court.” Redrup v New York, 386 U. S. 767, 772 (1967) (Harlan, J., dissenting). Taking their cue from our order, respondents did not train their energies on defending the District Court’s finding of predominance in their briefing or at oral argument. The Court’s newly revised question, focused on predominance, phrased only after briefing was done, left respondents without an unclouded opportunity to air the issue the Court today decides against them. And by resolving a complex and fact-intensive question without the benefit of full briefing, the Court invites the error into which it has fallen. See infra, at 5–11.
While the Court’s decision to review the merits of the District Court’s certification order is both unwise and unfair to respondents, the opinion breaks no new ground on the standard for certifying a class action under Federal Rule of Civil Procedure 23(b)(3). In particular, the decision should not be read to require, as a prerequisite to certification, that damages attributable to a classwide injury be measurable “‘on a class-wide basis.’” See ante, at 2–3 (acknowledging Court’s dependence on the absence of contest on the matter in this case); Tr. of Oral Arg. 41.
To gain class-action certification under Rule 23(b)(3), the named plaintiff must demonstrate, and the District Court must find, “that the questions of law or fact common to class members predominate over any questions affecting only individual members.” This predominance requirement is meant to “tes[t] whether proposed classes are sufficiently cohesive to warrant adjudication by representation,” Amchem Products, Inc. v Windsor, 521 U. S. 591, 623 (1997), but it scarcely demands commonality as to all questions. See 7AA C. Wright, A. Miller, & M. Kane, Federal Practice and Procedure §1778, p. 121 (3d ed. 2005) (hereinafter Wright, Miller, & Kane). In particular, when adjudication of questions of liability common to the class will achieve economies of time and expense, the predominance standard is generally satisfied even if damages are not provable in the aggregate. See Advisory Committee’s 1966 Notes on Fed. Rule Civ Proc. 23, 28 U. S. C. App., p. 141 (“[A] fraud perpetrated on numerous persons by the use of similar misrepresentations may be an appealing situation for a class action, and it may remain so despite the need, if liability is found, for separate determination of the damages suffered by individuals within the class.”); 7AA Wright, Miller, & Kane §1781, at 235–237.
Recognition that individual damages calculations do not preclude class certification under Rule 23(b)(3) is well nigh universal. See 2 W. Rubenstein, Newberg on Class Actions §4:54, p. 205 (5th ed. 2012) (ordinarily, “individual damage[s] calculations should not scuttle class certification under Rule 23(b)(3)”). Legions of appellate decisions across a range of substantive claims are illustrative. See, e.g.,
Tardiff v Knox County, 365 F. 3d 1, 6 (CA1 2004) (Fourth Amendment);
Chiang v Veneman, 385 F. 3d 256, 273 (CA3 2004) (Equal Credit Opportunity Act);
Bertulli v Independent Assn. of Continental Pilots, 242 F. 3d 290, 298 (CA5 2001) (Labor-Management Reporting and Disclosure Act and Railway Labor Act);
Beattie v CenturyTel, Inc., 511 F. 3d 554, 564–566 (CA6 2007) (Federal Communications Act);
Arreola v Godinez, 546 F. 3d 788, 801 (CA7 2008) (Eighth Amendment).
Anti-trust cases, which typically involve common allegations of anti-trust violation, anti-trust impact, and the fact of damages, are classic examples. See In re Visa Check/MasterMoney Anti-trust Litigation, 280 F. 3d 124, 139–140 (CA2 2001). See also 2A P. Areeda, H. Hovenkamp, R. Blair, & C. Durrance, Anti-trust Law ¶331, p. 56 (3d ed. 2007) (hereinafter Areeda & Hovenkamp); 6 A. Conte & H. Newberg, Newberg on Class Actions §18:27, p. 91 (4th ed. 2002). As this Court has rightly observed, “[p]redominance is a test readily met” in actions alleging “violations of the anti-trust laws.” Amchem, 521 U. S., at 625.
The oddity of this case, in which the need to prove damages on a classwide basis through a common methodology was never challenged by respondents, see Brief for Plaintiffs-Appellees in No. 10–2865 (CA3), pp. 39–40, is a further reason to dismiss the writ as improvidently granted. The Court’s ruling is good for this day and case only. In the mine run of cases, it remains the “black letter rule” that a class may obtain certification under Rule 23(b)(3) when liability questions common to the class predominate over damages questions unique to class members. 2 Rubenstein, supra, §4:54, at 208.
Incautiously entering the fray at this interlocutory stage, the Court sets forth a profoundly mistaken view of anti-trust law. And in doing so, it relies on its own version of the facts, a version inconsistent with factual findings made by the District Court and affirmed by the Court of Appeals.
To understand the anti-trust problem, some (simplified) background discussion is necessary. Plaintiffs below, respondents here, alleged that Comcast violated §§1 and 2 of the Sherman Act. See 15 U. S. C. §§1, 2. For present purposes, the §2 claim provides the better illustration. A firm is guilty of monopolization under §2 if the plaintiff proves (1) “the possession of monopoly power in the relevant market” and (2) “the willful acquisition or maintenance of that power[,] as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident.” United States v Grinnell Corp., 384 U. S. 563–571 (1966). A private plaintiff seeking damages must also show that (3) the monopolization caused “injur[y].” 15 U. S. C. §15. We have said that anti-trust injuries must be “of the type the anti-trust laws were intended to prevent and that flo[w] from that which makes defendants’ acts unlawful.” Atlantic Richfield Co. v USA Petroleum Co., 495 U. S. 328, 334 (1990) (quoting Brunswick Corp. v Pueblo Bowl-O-Mat, Inc., 429 U. S. 477, 489 (1977)). See 2A Areeda & Hovenkamp ¶391a, at 320 (To prove anti-trust injury, “[a] private plaintiff must identify the economic rationale for a business practice’s illegality under the anti-trust laws and show that its harm flows from whatever it is that makes the practice unlawful.”).
As plaintiffs below, respondents attempted to meet these requirements by showing that
Comcast obtained a 60% or greater share of the Philadelphia market, and that its share provides it with monopoly power;
Comcast acquired its share through exclusionary conduct consisting of a series of mergers with competitors and “swaps” of customers and locations; and
Comcast consequently injured respondents by charging them supra-competitive prices.
If, as respondents contend, Philadelphia is a separate well-defined market, and the alleged exclusionary conduct permitted Comcast to obtain a market share of at least 60%, then proving the §2 violation may not be arduous. As a point of comparison, the government considers a market shared by four firms, each of which has 25% market share, to be “highly concentrated.” Dept. of Justice & Federal Trade Commission, Horizontal Merger Guidelines §5.3, p. 19 (2010). A market, such as the one alleged by respondents, where one firm controls 60% is far worse. See id., §5.3, at 18–19, and n. 9 (using a concentration index that determines a market’s concentration level by summing the squares of each firm’s market share, one firm with 100% yielding 10,000, five firms with 20% each yielding 2000, while a market where one firm accounts for 60% yields an index number of at least 3,600). The Guidelines, and any standard anti-trust treatise, explain why firms in highly concentrated markets normally have the power to raise prices significantly above competitive levels. See, e.g., 2B Areeda & Hovenkamp ¶503, at 115.
So far there is agreement. But consider the last matter respondents must prove: Can they show that Comcast injured them by charging higher prices? After all, a firm with monopoly power will not necessarily exercise that power by charging higher prices. It could instead act less competitively in other ways, such as by leading the quiet life. See J. Hicks, Annual Survey of Economic Theory: The Theory of Monopoly, 3 Econometrica 1, 8 (1935) (“The best of all monopoly profits is a quiet life.”).
It is at this point that Dr. McClave’s model enters the scene. His model first selects a group of comparable outside-Philadelphia “benchmark” counties, where Comcast enjoyed a lower market share (and where satellite broadcasting accounted for more of the local business). Using multiple regression analysis, McClave’s model measures the effect of the anti-competitive conduct by comparing the class counties to the benchmark counties. The model concludes that the prices Philadelphia area consumers would have paid had the Philadelphia counties shared the properties of the benchmark counties (including a diminished Comcast market share), would have been 13.1% lower than those they actually paid. Thus, the model provides evidence that Comcast’s anti-competitive conduct, which led to a 60% market share, caused the class to suffer injuriously higher prices.
The special anti-trust-related difficulty present here stems from the manner in which respondents attempted to prove their anti-trust injuries. They proffered four “non-exclusive mechanisms” that allegedly “cause[d] the high prices” in the Philadelphia area. App. 403a. Those four theories posit that (1) due to Comcast’s acquisitions of competitors, customers found it more difficult to compare prices; (2) one set of potential competitors, namely Direct Broadcast Satellite companies, found it more difficult to obtain access to local sports broadcasts and consequently decided not to enter the Philadelphia market; (3) Comcast’s ability to obtain programming material at lower prices permitted it to raise prices; and (4) a number of potential competitors (called “overbuilders”), whose presence in the market would have limited Comcast’s power to raise prices, were ready to enter some parts of the market but decided not to do so in light of Comcast’s anti-competitive conduct. 264 F. R. D. 150, 161–162 (ED Pa. 2010).
For reasons not here relevant, the District Court found the first three theories inapplicable and limited the liability-phase proof to the “overbuilder” theory. See App. to Pet. for Cert. 192a–193a. It then asked the parties to brief whether doing so had any impact on the viability of McClave’s model as a measure of classwide damages. See 264 F. R. D., at 190. After considering the parties’ arguments, the District Court found that striking the three theories “does not impeach Dr. McClave’s damages model” because “[a]ny anti-competitive conduct is reflected in the [higher Philadelphia] price [which Dr. McClave’s model determines], not in the [the model’s] selection of the comparison counties, [i.e., the lower-price ‘benchmark counties’ with which the Philadelphia area prices were compared].” Id., at 190–191. The court explained that “whether or not we accepted all [four] .... theories .... is inapposite to Dr. McClave’s methods of choosing benchmarks.” Ibid. On appeal, the Third Circuit held that this finding was not an abuse of discretion. 655 F. 3d 182, 207 (2011).
The Court, however, concludes that “the model failed to measure damages resulting from the particular anti-trust injury on which petitioners’ liability in this action is premised.” Ante, at 8. To reach this conclusion the Court must consider fact-based matters, namely what this econometric multiple-regression model is about, what it proves, and how it does so. And it must overturn two lower courts’ related factual findings to the contrary.
We are normally “reluctant to disturb findings of fact in which two courts below have concurred.” United States v Doe, 465 U. S. 605, 614 (1984). See also United States v Virginia, 518 U. S. 515, n. 5 (1996) (Scalia, J., dissenting) (noting “our well-settled rule that we will not ‘undertake to review concurrent findings of fact by two courts below in the absence of a very obvious and exceptional showing of error’” (quoting Graver Tank & Mfg. Co. v Linde Air Products Co., 336 U. S. 271, 275 (1949))). Here, the District Court found McClave’s econometric model capable of measuring damages on a classwide basis, even after striking three of the injury theories. 264 F. R. D., at 190–191. Contrary to the Court’s characterization, see ante, at 8–9, n. 5, this was not a legal conclusion about what the model proved; it was a factual finding about how the model worked. Under our typical practice, we should leave that finding alone.
In any event, as far as we can tell, the lower courts were right. On the basis of the record as we understand it, the District Court did not abuse its discretion in finding that McClave’s model could measure damages suffered by the class – even if the damages were limited to those caused by deterred overbuilding. That is because respondents alleged that Comcast’s anti-competitive conduct increased Comcast’s market share (and market power) by deterring potential entrants, in particular, overbuilders, from entering the Philadelphia area market. See App. 43a–66a. By showing that this was so, respondents’ proof tends to show the same in respect to other entrants. The overbuilders’ failure to enter deprives the market of the price discipline that their entry would have provided in other parts via threat of the overbuilders’ expansion or that of others potentially led on by their example. Indeed, in the District Court, Comcast argued that the three other theories, i.e., the three rejected theories, had no impact on prices. See 264 F. R. D., at 166, 176, 180–181. If Comcast was right, then the damages McClave’s model found must have stemmed exclusively from conduct that deterred new entry, say from “overbuilders.” Not surprisingly, the Court offers no support at all for its contrary conclusion, namely, that the District Court’s finding was “‘obvious[ly] and exceptional[ly]’ erroneous.” Ante, at 8–9, n. 5 (quoting Virginia, 518 U. S., at 589, n. 5 (Scalia, J., dissenting)).
We are particularly concerned about the matter because the Court, in reaching its contrary conclusion, makes broad statements about anti-trust law that it could not mean to apply in other cases. The Court begins with what it calls an “unremarkable premise” that respondents could be “entitled only to damages resulting from reduced overbuilder competition.” Ante, at 7. In most §2 cases, however, the Court’s starting place would seem remarkable, not “unremarkable.”
Suppose in a different case a plaintiff were to prove that Widget, Inc. has obtained, through anti-competitive means, a 90% share of the California widget market. Suppose the plaintiff also proves that the two small remaining firms – one in Ukiah, the other in San Diego – lack the capacity to expand their widget output to the point where that possibility could deter Widget, Inc. from raising its prices. Suppose further that the plaintiff introduces a model that shows California widget prices are now twice those in every other State, which, the model concludes is (after accounting for other possible reasons) the result of lack of competition in the California widget market. Why would a court hearing that case restrict damages solely to customers in the vicinity of Ukiah and San Diego?
Like the model in this example, Dr. McClave’s model does not purport to show precisely how Comcast’s conduct led to higher prices in the Philadelphia area. It simply shows that Comcast’s conduct brought about higher prices. And it measures the amount of subsequent harm.
* * *
Because the parties did not fully argue the question the Court now answers, all Members of the Court may lack a complete understanding of the model or the meaning of related statements in the record. The need for focused argument is particularly strong here where, as we have said, the underlying considerations are detailed, technical, and fact-based. The Court departs from our ordinary practice, risks inaccurate judicial decisionmaking, and is unfair to respondents and the courts below. For these reasons, we would not disturb the Court of Appeals’ judgment and, instead, would dismiss the writ as improvidently granted.
 A “Designated Market Area” is a term used by Nielsen Media Research to define a broadcast-television market. Strictly speaking, the Philadelphia DMA comprises 18 counties, not 16.
 Respondents sought certification for the following class: “All cable television customers who subscribe or subscribed at any times since December 1, 1999, to the present to video programming services (other than solely to basic cable services) from Comcast, or any of its subsidiaries or affiliates in Comcast’s Philadelphia cluster.” App. 35a.
 The District Court did not hold that the three alternative theories of liability failed to establish anti-trust impact, but merely that those theories could not be determined in a manner common to all the class plaintiffs. The other theories of liability may well be available forthe plaintiffs to pursue as individual actions. Any contention that the plaintiffs should be allowed to recover damages attributable to all four theories in this class action would erroneously suggest one of two things – either that the plaintiffs may also recover such damages in individual actions or that they are precluded from asserting those theories in individual actions.
 The question presented reads: “Whether a district court may certify a class action without resolving whether the plaintiff class had introduced admissible evidence, including expert testimony, to show that the case is susceptible to awarding damages on a class-wide basis.” 567 U. S., at ___. Respondents contend that petitioners forfeited their ability to answer this question in the negative because they did not make an objection to the admission of Dr. McClave’s testimony under the Federal Rules of Evidence. See Daubert v Merrell Dow Pharmaceuticals, Inc., 509 U. S. 579 (1993). Such a forfeit would make it impossible for petitioners to argue that Dr. McClave’s testimony was not “admissible evidence” under the Rules; but it does not make it impossible for them to argue that the evidence failed “to show that the case is susceptible to awarding damages on a class-wide basis.” Petitioners argued below, and continue to argue here, that certification was improper because respondents had failed to establish that damages could be measured on a classwide basis. That is the question we address here.
 The dissent is of the view that what an econometric model proves is a “question of fact” on which we will not “undertake to review concurrent findings .... by two courts below in the absence of a very obvious and exceptional showing of error.” Post, at 9 (quoting United States v Virginia, 518 U. S. 515, n. 5 (1996) (Scalia, J., dissenting) (internal quotation marks omitted)). To begin with, neither of the courts below found that the model established damages attributable to overbuilding alone. Second, while the data contained within an econometric model may well be “questions of fact” in the relevant sense, what those data prove is no more a question of fact than what our opinions hold. And finally, even if it were a question of fact, concluding that the model here established damages attributable to overbuilding alone would be “obvious[ly] and exceptional[ly]” erroneous.
 We might add that even if the model had identified subscribers who paid more solely because of the deterrence of overbuilding, it still would not have established the requisite commonality of damages unless it plausibly showed that the extent of overbuilding (absent deterrence) would have been the same in all counties, or that the extent is irrelevant to effect upon ability to charge supra-competitive prices.
 A class may be divided into subclasses for adjudication of damages. Fed. Rule Civ. Proc. 23(c)(4)–(5). Or, at the outset, a class may be certified for liability purposes only, leaving individual damages calculations to subsequent proceedings. See 2 W. Rubenstein, Newberg on Class Actions §4:54, pp. 206–208 (5th ed. 2012). Further, a certification order may be altered or amended as the case unfolds. Rule 23(c)(1)(C).
all rights reserved