In Re Brand Name Prescription Drugs Antitrust Litigation. Appeals of Robert A. Huggins

POSNER, Chief Judge. We have consolidated for decision four appeals (in two of which we have jurisdiction under 28 U.S.C. § 1292(b) and in the other two under 28 U.S.C. § 1291 and Fed.R.Civ.P. 54(b)) from rulings in a huge price-fixing litigation that the Judicial Panel on Multidis-trict Litigation has consolidated in the Northern District of Illinois for pretrial proceedings. The consolidation covers hundreds of separate cases (a number of them class actions) brought under section 1 of the Sherman Act, 15 U.S.C. § 1, by retail pharmacies against manufacturers and wholesalers of prescription drugs. The pharmacies complain that the defendants have conspired among themselves to deny all pharmacies, *603 including chains and buying groups, discounts off the list price of brand-name drugs that the manufacturers sell to the wholesalers and that the wholesalers in turn resell to the pharmacies. A brief sketch of the operation of the alleged conspiracy will provide the essential background to understanding the issues presented by these appeals. While refusing to give pharmacies any discounts, the defendants give steep discounts to favored classes of customers, including hospitals, health maintenance organizations, nursing homes, and mail-order companies. The defendants maintain this differential pricing through a “chargeback” system. Under that system, the manufacturer makes a contract with the favored customer establishing a discounted price at which the customer is entitled to buy from wholesalers; the wholesaler sells to the favored customer at that price; and the manufacturer then reimburses the wholesaler for the difference between the regular wholesale price and the discounted price. So if the manufacturer’s regular price to the wholesaler for some drug is $100 and the contractually agreed upon discounted price for a favored customer is $75, the wholesaler will pay the manufacturer $100 for the drug but resell it to the favored customer at $75 and the bill the manufacturer $25. The plaintiffs claim that the purpose of the chargeback system is to make it difficult for the favored customers to engage in arbitrage, that is, to buy more than they need and resell the surplus to pharmacies at a price between the discounted price that the favored customers pay and the higher, undis-counted wholesale price that nonfavored customers pay. The chargeback system permits the wholesalers to buy cheap only when they are reselling to someone whom the manufacturer wants to be given a discount. The defendants’ differential pricing of their drugs is discriminatory in the technical economic sense — it involves charging different prices for the same goods, the differences being unrelated to savings in the costs of serving the favored customers. When the lower of two discriminatory prices covers the seller’s cost, the higher price must exceed that cost. This creates an incentive for the favored purchasers to order more of the good than they need for their own use and to sell the surplus to disfavored customers at a price somewhere in between the seller’s different prices. For example, an $80 resale by a hospital or other favored customer that had bought at $75 to a pharmacy that had bought at $100 would make both parties to the resale better off; the hospital would have a profit of $5 and the pharmacy would obtain a cost savings of $20. This is arbitrage and would erode the two-price system. The chargeback system prevents arbitrage; the wholesaler who resold to a pharmacy at a significant discount would incur a loss, since he would not be able to charge back any part of the discount to the manufacturer. Although a federal statute forbids hospitals and other providers of health care to resell to other sellers the pharmaceutical drugs that they buy, the statute does not cover all the favored customers for such drugs. 21 U.S.C. § 353(c)(3). Anyway statutes are not always fully obeyed. The chargeback system fills the gap in the statute’s coverage and does not require heavy enforcement costs. The presence of price discrimination in the economic sense is evidence of the presence of monopoly power' — the power to raise price above cost without losing so many sales as to make the price rise unsustainable. If the lower price covers the seller’s cost, the higher price must exceed it; so competition must be weak or absent, because it has failed to force price down to cost (including in “cost” a reasonable return on investment). Since monopoly power can be created by collusion among competing sellers, the existence of industry-wide price discrimination is some evidence of collusion. But it is not conclusive evidence, especially in an industry such as pharmaceuticals many of the products of which are patented. The sellers may be selling goods that although close substitutes are not perfect substitutes, with the result that each seller has some monopoly power and therefore can price discriminate unilaterally. It might want to do so to take advantage of the fact that some consumers are less able to resist high prices than others. A fully developed record might show, in accordance with contested evidence in the record compiled to date, that a pharmacy has little choice but to buy a wide range of *604 competing drugs because it cannot know in advance which drug its customers’ doctors will prescribe. An HMO, however, can (within limits) tell the doctors it employs what drugs to prescribe, and it can use that power to extract price concessions from the individual manufacturers, who naturally however do not wish to extend the concessions to captive consumers such as the pharmacies. In the extensive pretrial proceedings that have been conducted to date in this litigation, the plaintiffs have presented evidence that the defendant manufacturers agreed among themselves, and also with the defendant wholesalers, to refuse discounts to pharmacies and to make this refusal stick by adopting the chargeback system in order to prevent arbitrage. In other words, the claim is that pervasive price discrimination in the pharmaceutical market is the result not of individual decisions by manufacturers who possess some monopoly power but of an agreement to practice price discrimination. The plaintiffs’ objection is not to the discrimination as such; although there is a Robinson-Patman claim in the complaint, it is not part of the appeal. The plaintiffs’ objection is to having to pay high prices that, but for the defendants’ alleged conspiracy, would be brought down by competition. One might have supposed that if the defendants were going to collude on price, they would go the whole hog and agree not to provide discounts to the hospitals and other customers favored by the discriminatory system. But the defendants’ cartel — if that is what it is — may not be tight enough to prevent hospitals and other bulk purchasers with power to shift demand among different manufacturers’ drugs from whipsawing the members of the cartel for discounts; or maybe these purchasers could shift demand to manufacturers that are not members of the cartel. If, for whatever reason, the elasticity of demand for a cartel’s product differs among groups of purchasers, a single cartel price will not be profit-maximizing unless a discriminatory price scheme cannot be enforced at reasonable cost. The manufacturers moved for summary judgment, arguing that there wasn’t enough evidence of collusion to warrant a trial. The district judge denied the motion. The correctness of his ruling is not before us. And whether it was correct or not, the reader should bear in mind that the manufacturers have not been found to have violated the Sherman Act; the only determination is that there is enough evidence of a violation to require that the case be allowed to proceed to trial. The judge granted summary judgment to one of the manufacturers, however, DuPont Merck Pharmaceutical Company. The plaintiffs’ appeal from that ruling is one of the four appeals before us. The judge also granted summary judgment to the wholesaler defendants because he thought there was insufficient evidence of their participation in the manufacturers’ conspiracy to warrant a trial. That is another ruling appealed from. Another is the judge’s refusal to dismiss indirect-purchaser claims by pharmacies that paid overcharges as a consequence of the alleged manufacturers’ conspiracy. The manufacturers argued unsuccessfully that only the first tier of purchasers (“direct purchasers”), composed of the wholesalers and others who purchased drugs directly from the manufacturers, and not the second tier, composed of pharmacies that purchased the manufacturers’ drugs from the wholesalers (“indirect purchasers”), are permitted to bring a suit for overcharges under the Sherman Act. In the last ruling that has been appealed to us, the judge refused to remand a class action that alleges violations not of the Sherman Act but of Alabama’s antitrust statute, which expressly authorizes suits by indirect purchasers. The indirect-purchaser issue (with which we begin) is separate from the issue of the wholesalers’ participation in the manufacturers’ alleged conspiracy. It is true that if we reversed the judge’s ruling on the latter issue and so reinstated the wholesalers as defendants, and if the plaintiffs went on to obtain a judgment against the wholesalers and manufacturers, any indirect-purchaser defense would go by the board, since the pharmacies would then be direct purchasers from the conspirators. Fontana Aviation, Inc. v. Cessna Aircraft Co., 617 F.2d 478, 481 (7th Cir.1980); Arizona v. Shamrock Foods *605 Co., 729 F.2d 1208, 1212-13 (9th Cir.1984); see also In re Beef Industry Antitrust Litigation, 600 F.2d 1148, 1163 (5th Cir.1979) (requiring that the direct sellers, here the wholesalers, be joined as defendants — but that requirement is satisfied). But even if we do reinstate the wholesalers as defendants, an issue discussed later in this opinion, the plaintiffs may fail at trial to establish their liability, in which event the indirect-purchaser issue will be decisive. So, the issue being fully briefed and argued in this court, we should decide it; and the fact that it may in the end not prove decisive does not show that the district judge and we were wrong to certify his ruling on the issue under 28 U.S.C. § 1292(b) (interlocutory appeal of a ruling on a controlling question) for an immediate appeal. Sokaogon Gaming Enterprise Corp. v. Tushie-Montgomery Associates, Inc., 86 F.3d 656, 658-59 (7th Cir.1996); Johnson v. Burken, 930 F.2d 1202, 1205 (7th Cir.1991); Katz v. Carte Blanche Corp., 496 F.2d 747, 755 (3d Cir.1974); 16 Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure § 3930, pp. 426-27 (2d ed. 1996). A brief review of the evolution of the indirect-purchaser doctrine in the Supreme Court will point us toward a resolution of the issue. In Hanover Shoe, Inc. v. United Shoe Machinery Corp., 392 U.S. 481, 88 S.Ct. 2224, 20 L.Ed.2d 1231 (1968), the defendant in a Sherman Act suit, a manufacturer of machinery for making shoes, defended on the ground that the plaintiff, a shoe manufacturer that had bought the defendant’s machinery, had passed on any monopoly overcharge to its own customers, the wholesale purchasers of its shoes, and hence had not been injured. A firm hit with an increase in the cost of one of its inputs will try so far as competition allows to pass that cost on to its customers in the form of a higher price for its product. The Supreme Court held, however, that an antitrust defendant would not be permitted to defend against a damages suit on the ground that the plaintiff had shifted the cost of the defendant’s wrongdoing to the plaintiffs customers. Such a defense would complicate antitrust enforcement by requiring an apportionment of damages between different tiers of purchasers of the defendant’s product. Tracing a price hike through successive resales is an example of what is called “incidence analysis,” and is famously difficult. The Court took the next step in Illinois Brick Co. v. Illinois, 431 U.S. 720, 97 S.Ct. 2061, 52 L.Ed.2d 707 (1977), and held that the second or subsequent tiers, the indirect purchasers from the antitrust violators, couldn’t sue; only the first tier could. This was a logical corollary of the rejection of the passing-on defense in Hanover Shoe, since to determine the damages suffered by subsequent tiers of purchasers would require the very apportionment of damages that the Court had rejected in the earlier case. Illinois Brick left unclear whether there might be exceptions for cases in which the amount of the overcharge that was passed on to a lower tier of purchasers could be determined simply and with mechanical precision. A plausible example, we thought, would be a case in which the first tier of purchasers consisted of public utilities that as a consequence of government regulation passed on any cost increase dollar for dollar to their customers. Illinois v. Panhandle Eastern Pipe Line Co., 852 F.2d 891 (7th Cir.1988) (en banc). Shortly afterward, in a similar case, the Supreme Court held that such cases are not within any exception to the Illinois Brick doctrine, Kansas v. Utilicorp United, Inc., 497 U.S. 199, 110 S.Ct. 2807, 111 L.Ed.2d 169 (1990), and we duly overruled our Panhandle opinion. Illinois v. Panhandle Eastern Pipe Line Co., 935 F.2d 1469 (7th Cir.1991). UtiliCorp implies that the only exceptions to the Illinois Brick doctrine are those stated in Illinois Brick itself — “where the direct purchaser is owned or controlled by its customer,” 431 U.S. at 736 n. 16, 97 S.Ct. at 2070 or, we suppose, vice versa. The first exception (ownership) is conceded to be inapplicable here; the wholesalers are not corporate affiliates of the manufacturers. The second (control) is inapplicable as well. The manufacturers do not control the wholesalers through interlocking directorates, minority stock ownership, loan agreements that subject the wholesalers to the manufacturers’ operating control, trust agreements, or other modes of control sepa *606 rate from ownership of a majority of the wholesalers’ common stock. Jewish Hospital Ass’n v. Stewart Mechanical Enterprises, Inc., 628 F.2d 971, 975 (6th Cir.1980); cf. Gould v. Ruefenacht, 471 U.S. 701, 705, 105 S.Ct. 2308, 2310, 85 L.Ed.2d 708 (1985). The district judge held, however, primarily on the basis of the chargeback system, that the wholesalers are really nothing more than “glorified warehouses” of the manufacturers. In so ruling, the judge gave undue weight to the chargeback system. The favored customers, the ones who had contracts with the manufacturers though they took delivery from the wholesalers, are not parties to this litigation. They certainly are not complaining about the system of discriminatory pricing. They were not overcharged, and their right if any to recover overcharges in a suit against the manufacturers is not in issue. The plaintiffs are the disfavored customers. They did not have contracts with the manufacturers, they did not receive discounts, and the wholesalers did not receive chargebacks on sales to them. The plaintiffs’ complaint is that they were overcharged because the wholesalers passed on to them the overcharge that the wholesalers had to pay the manufacturers by virtue of the price-fixing conspiracy. This is just the kind of complaint that Illinois Brick bars. The only entities permitted to complain about the manufacturers’ overcharging the wholesalers are the wholesalers themselves, the direct purchasers, even if every cent of the overcharge was promptly and fully passed on to the pharmacies in the form of a higher wholesale price. Some wholesalers were plaintiffs in this litigation; they settled. Had they not done so, and had the case proceeded to trial and the pharmacies been permitted to seek damages for the amount of the overcharge passed on to them, the court would have had to apportion the overcharge between the wholesalers and the pharmacies. That’s just what the Supreme Court in Hanover Shoe, Illinois Brick, and Utilicorp told the federal courts not to do. We can imagine the present case reconfigured in a way that might take it out of the orbit of these decisions; it would not be a matter of carving a further exception. A number of pharmacies have tried to improve their bargaining position vis-á-vis the drug manufacturers by forming buying groups. The bigger a buyer is, the more likely it is to be able to obtain a discount from a member of a cartel, since the volume of its purchases may compensate the member for endangering the cartel by granting a discount. George J. Stigler, “A Theory of Oligopoly,” in Stigler, The Organization of Industry 39, 43-44 (1968). That is one motive for forming a buying group. The manufacturers have been steadfast in refusing to grant discounts to such groups. If this refusal, taking as it does the form of a refusal to enter into direct contractual relations with certain retailers, such as the manufacturers have with their favored customers, were successfully challenged as a boycott, see FTC v. Superior Court Trial Lawyers Ass’n, 493 U.S. 411, 428, 110 S.Ct. 768, 777, 107 L.Ed.2d 851 (1990); FTC v. Indiana Federation of Dentists, 476 U.S. 447, 458-59, 106 S.Ct. 2009, 2017-18, 90 L.Ed.2d 445 (1986); Collins v. Associated Pathologists, Ltd., 844 F.2d 473, 479 (7th Cir.1988), the Illinois Brick rule, which is a rule concerning overcharges, would fall away. The plaintiffs would be permitted to prove up whatever damages they could show had flowed from the boycott, Mid-West Paper Products Co. v. Continental Group, Inc., 596 F.2d 573, 585 n. 47 (3d Cir.1979), provided they weren’t seeking to recover overcharges, for that would entail the very incidence analysis that Illinois Brick bars. Merican, Inc. v. Caterpillar Tractor Co., 713 F.2d 958, 966-68 and n. 21 (3d Cir.1983). But that is precisely what they are seeking. The certified class is of pharmacies that paid overcharges, and the certification was based on the uniformity of the harm. It would be more difficult to justify class treatment of a boycott of buying groups. Compare White Industries, Inc. v. Cessna Aircraft Co., 845 F.2d 1497, 1502-03 (8th Cir.1988), with Bogosian v. Gulf Oil Corp., 561 F.2d 434, 455 (3d Cir.1977). That may be why the plaintiffs have not cast their case in the boycott mold. We need not decide whether it is still open to them to do so in the district court. *607 To conclude our discussion of the drug-manufacturers’ federal antitrust liability to the indirect purchasers, the federal class actions should have been dismissed unless the wholesalers should not have been dropped as defendants, an issue we take up later. The Alabama class suit, which the district judge refused to remand, also involves the indirect-purchaser question; so let us turn to that suit. It was actually the second prescription-drug price-fixing suit brought in the Alabama state courts. The first had been removed to federal district court under the diversity jurisdiction and then transferred by the multi-district panel to the Northern District of Illinois for consolidation with the other prescription-drug price-fixing suits. The district court had denied a motion to remand, so the suit remains in that court. Then our Alabama suit was filed, and like the first suit it was removed to federal district court and transferred to the Northern District of Illinois. It is a class suit on behalf of consumers in several states, not only Alabama, and it names as defendants a large number of drug manufacturers none of which either is a citizen of Alabama or sells exclusively to that state’s residents. The suit is based, or at least purports to be based, on an Alabama statute that is modeled on the Sherman Act but that contains a provision which expressly authorizes indirect-purchaser claims — the very type of claim that Illinois Brick bars in suits under the Sherman Act. Ala.Code § 6-5-60(a). The defendants argued, and the district court agreed, that the suit is removable to federal court under both the diversity statute and, by virtue of the doctrine of “artful pleading,” the federal-question statute as well. 28 U.S.C. §§ 1331,1332. There is complete diversity of citizenship among the parties; the question, so far as the issue of diversity jurisdiction is concerned, is only whether the minimum amount in controversy required to maintain a diversity suit in federal court ($50,000 at the time the suit was filed) is present. The court cannot just add up the damages sought by each member of the class. Snyder v. Harris, 394 U.S. 332, 89 S.Ct. 1053, 22 L.Ed.2d 319 (1969); Zahn v. Int’l Paper Co., 414 U.S. 291, 301, 94 S.Ct. 505, 511, 38 L.Ed.2d 511 (1973); In re Corestates Trust Fee Litigation, 39 F.3d 61, 64 (3d Cir.1994). At least one named plaintiff must satisfy the jurisdictional minimum. If he does, the other named plaintiffs and the unnamed class members can, by virtue of the supplemental jurisdiction conferred on the federal district courts by 28 U.S.C. § 1367, piggyback on that plaintiffs claim. That is, they remain plaintiffs, or unnamed members of the class, as the case may be, even though their own claims are for less than the jurisdictional minimum amount. So the Fifth Circuit held in In re Abbott Laboratories, 51 F.3d 524, 527-29 (5th Cir.1995), and we signified our agreement with that holding in Stromberg Metal Works, Inc. v. Press Mechanical, Inc., 77 F.3d 928, 930-33 (7th Cir.1996), and repeat it today. The plaintiffs in this case, however, because they did not want their case removed to federal court, were careful to plead that the damages sought by each did not exceed $50,000. This is plausible — you would have to buy an awful lot of expensive drugs to run up a bill the overcharge portion of which alone was more than that amount. And plausible or not, a plaintiff can always stay under the minimum amount in controversy by waiving his right to more, In re Amino Acid Lysine Antitrust Litigation, 918 F.Supp. 1181, 1185-86 (N.D.Ill.1996), though these plaintiffs have not established that they did mean to waive their right. Compensatory damages, which we have just seen are not likely to exceed $50,-000 for any of the named plaintiffs, are not the only form of monetary relief sought, however. The antitrust statute on which the Alabama class action is based authorizes the court to award up to $500 for each “instance of ... injury or damages” as a statutory penalty, in addition to any compensatory damages. Ala.Code § 6-5-60(a). But the defendants cannot simply wave the statute in our faces. They have the burden of establishing federal jurisdiction when they seek to remove a ease from state to federal court, and so they must present evidence of federal jurisdiction once the existence of that jurisdiction is fairly cast into doubt. Chase v. Shop ’N Save Warehouse Foods, Inc., 110 F.3d 424, 427 (7th Cir.1997); Wellness Com *608 munity-National v. Wellness House, 70 F.3d 46, 49 (7th Cir.1995); Selcke v. New England Ins. Co., 2 F.3d 790, 792 (7th Cir.1993). It was east into doubt here by the complaint itself, which does not allege stakes in excess of $50,000 or facts from which such stakes can readily be inferred. Yet the defendants presented no evidence that, even with the statutory penalty added to the compensatory damages sought, any of the named plaintiffs is asking for more than $50,000. The defendants point out that it is possible that at least one of the plaintiffs had more than $50,000 in damages and penalties. A hundred purchases within the four-year period covered by the complaint would carry a purchaser over the threshold, even if the overcharge on each purchase was tiny, because each purchase, constituting we assume a separate “instance of ... injury or damage,” would entitle the purchaser to the $500 statutory penalty. But the defendants put in no evidence that any of the named plaintiffs in fact made this many purchases. Instead they argue that under Alabama law the entire statutory penalties awarded in a case are the indivisible penalty for a defendant’s misconduct and so are the stakes in each of the plaintiffs’ claims. If this is correct, and the plaintiffs have not waived a claim for total damages (compensatory damages plus the penalty) per plaintiff of more than $50,000, then the defendants had no need to present any evidence on the jurisdictional issue. In arguing their interpretation of the Alabama statute, with the support of Tapscott v. MS Dealer Service Corp., 77 F.3d 1353, 1359 (11th Cir.1996), and less directly of Allen v. R & H Oil & Gas Co., 63 F.3d 1326, 1334 (5th Cir.1995), but in opposition to Gilman v. BHC Securities, Inc., 104 F.3d 1418, 1428-31 (2d Cir.1997), the defendants are gesturing toward the Supreme Court’s statement in Snyder v. Harris, supra, 394 U.S. at 355, 89 S.Ct. at 1066 that when “two or more plaintiffs unite to enforce a single title or right in which they have a common and undivided interest,” the amount in controversy is the aggregate in which they each have their undivided share. An example is an action by the heirs of an intestate estate against the estate’s administrator. A successful prosecution of the action would result in making the estate larger, and each heir would have an undivided interest in the larger, as in the original, estate. Shields v. Thomas, 58 U.S. (17 How.) 3, 15 L.Ed. 93 (1855). Other examples are set forth in Gilman v. BHC Securities, Inc., supra, 104 F.3d at 1423. This is not such a case. The penalty prescribed by the Alabama statute is presumably per violation, that is, per sale at an unlawful price; and it is awarded to the victim of the particular violation, the direct or indirect buyer, rather than to the victims of the price-fixing conspiracy as a group or to a representative member of the group. If one plaintiff disclaimed the penalty awarded him under the statute, or settled with the defendant for an amount that included no penalty, the penalty thus forsworn would not go to another plaintiff; it would be subtracted from the total amount of penalties assessed against the defendant. Indeed, if the court had awarded the maximum penalty to each victim, it would be impossible for the court to shift the disclaimed penalty to another of the victims; to do so would pierce the ceiling. But we take it that even if one victim had received $300 rather than $500, the court would not give him another $200 if another victim had disclaimed his own $300 penalty. It is possible we suppose that the judge could fix some amount that represented in his mind the proper punishment for the defendant’s misconduct; divide that amount by the number of plaintiffs; and if the result of the division was greater than $500, cut down the aggregate accordingly. But even if, in acting so, the judge would be complying with the spirit as well as the letter of the statute, the resulting fund would not be a piece of property to which the plaintiffs had undivided rights. None of the victims would have an undivided right in a common fund or res such that if one claimant fell out the others’ shares would grow. Sellers v. O’Connell, 701 F.2d 575, 579 (6th Cir.1983); Eagle Star Ins. Co. v. Maltes, 313 F.2d 778, 781 (5th Cir.1963). A plaintiffs award of punitive damages is not limited by awards made to previous plaintiffs complaining of the same *609 act of the defendant. E.g., Allen v. R & H Oil & Gas Co., supra, 63 F.3d at 1334; Dunn v. HOVIC, 1 F.3d 1371, 1385-86 (3d Cir.1993); Roginsky v. Richardson-Merrell, Inc., 378 F.2d 832, 839-41 (2d Cir.1967) (Friendly, J.). This rule has been criticized (as by Judge Friendly in Roginsky), but whether it is a good rule or a bad rale it shows that the right to punitive damages is a right of the individual plaintiff, rather than a collective entitlement of the victims of the defendant’s misconduct. Gilman v. BHC Securities, Inc., supra, 104 F.3d at 1428-31. The rale may have to be qualified now that the Supreme Court has held that excessive awards of punitive damages violate the due process clause. BMW of North America, Inc. v. Gore, — U.S. —, 116 S.Ct. 1589, 134 L.Ed.2d 809 (1996). For it could be argued that a piling on of awards by different courts for the same act might result in excessive punishment for that act. We need not decide whether this argument would ever succeed; it is unlikely to succeed to the point of converting entitlements to punitive damages from individual to collective entitlements. That the defendants have failed to show that the plaintiffs are seeking more than $50,000 apiece against each defendant cannot be the end of our analysis of diversity jurisdiction. The complaint seeks an injunction against the alleged conspiracy as well as damages and the penalty, and the defendants argue that it will cost them more than $50,-000 to comply with the injunction even though the only plausible form of injunctive relief in a case like this would be to order the defendants to stop fixing prices. There are four ways in which a request for an injunction might be thought to carry a case over the amount in controversy threshold. The first way — plainly one valid way, e.g., Hunt v. Washington State Apple Advertising Comm’n, 432 U.S. 333, 347, 97 S.Ct. 2434, 2443, 53 L.Ed.2d 383 (1977); Gould v. Artisoft, Inc., 1 F.3d 544, 548 n. 4 (7th Cir.1993); Justice v. Atchison, Topeka & Santa Fe Ry., 927 F.2d 503, 505 (10th Cir.1991); Smith v. Washington, 593 F.2d 1097, 1099 (D.C.Cir. 1978), and some courts think the only valid way, Kheel v. Port of New York Authority, 457 F.2d 46, 49 (2d Cir.1972); Bernard v. Gerber Food Products Co., 938 F.Supp. 218, 220-22 (S.D.N.Y.1996) — is if the value of the injunction to the plaintiff exceeds the statutory minimum. So we could look to the present value of the future cost savings that each plaintiff anticipated from the cessation of each defendant’s price fixing. No effort to quantify this value or array of values in even the roughest terms has been made, however, so we put it to one side. Although one of our cases adopts the “plaintiff only” position, Freeman v. Sports Car Club of America, Inc., 51 F.3d 1358, 1362 (7th Cir.1995), it overlooked a decision in which we had squarely rejected that position in favor of the “either viewpoint” (plaintiffs or defendant’s) approach, McCarty v. Amoco Pipeline Co., 595 F.2d 389 (7th Cir.1979). Looked at from the defendants’ standpoint, the minimum amount in controversy would be present if the injunction sought by the plaintiffs would require some alteration in the defendant’s method of doing business that would cost the defendant at least the statutory minimum amount. See, e.g., id. at 391. This ground is not argued either. Often it will be equivalent to the previous ground, the value of the injunction to the plaintiff. The defendant would be willing to pay the plaintiff up to a shade less than the cost that the injunction would impose on the defendant to induce the plaintiff to abandon his quest for injunctive relief. In that way the cost to the defendant would be transmuted into an equivalent value to the plaintiff. If, however, there are multiple plaintiffs, actual or potential, the defendant will not be willing to pay each one as much as he would if there were only one possible plaintiff. It may seem paradoxical to defeat removal in the multiplaintiff setting on this basis. But it is implicit in the rale that forbids aggregation of class members’ separate claims that it will sometimes be more difficult for a defendant desiring to remove a diversity case to federal court to establish the minimum amount of controversy in a multiplaintiff case than in a much smaller single-plaintiff case. Compare a class action in which one million class members each has a claim worth $1 with a case in which a single plaintiff has a claim worth $100,000. There *610 is diversity jurisdiction in the second case but not (because of the nonaggregation rule in class actions, the rule of Snyder and Zahn) the first. Concern has been expressed that if the cost to the defendant may be used to establish the minimum amount in controversy in an injunction case, it may be used for this purpose in a damages case, and then the nonaggregation rule will be circumvented. E.g., Packard v. Provident Nat’l Bank, 994 F.2d 1039, 1050 (3d Cir.1993). The concern is misplaced. Whatever the form of relief sought, each plaintiffs claim must be held separate from each other plaintiffs claim from both the plaintiffs and the defendant’s standpoint. The defendant in such a ease is deemed to face multiple claims for injunctive relief, each of which must be separately evaluated. Snow v. Ford Motor Co., 561 F.2d 787, 790 (9th Cir.1977). The question then becomes, as with the penalty statute, whether each plaintiff is asserting an individual right or, rather, a right to an undivided interest in something. In this case it is the former. Each plaintiff has a right to be free from the indirect effects of collusive pricing. Moreover, the grant of an injunction in favor of a single plaintiff would be unlikely to impose a heavy cost on any of the defendants; each defendant could continue in its own way of pricing with respect to all other plaintiffs. The test, we repeat, is the cost to each defendant of an injunction running in favor of one plaintiff; otherwise the non-aggregation rule would be violated. Still another way in which the requirement of the statutory minimum amount in controversy can be satisfied in an injunctive case is by showing that the injunction would force the defendant to forgo a benefit to him that is worth more than the threshold amount specified in the diversity statute, e.g., Grotzke v. Kurz, 887 F.Supp. 53 (D.R.I.1995), as where the suit asks that the defendant be enjoined from completing a lucrative transaction. That is not argued here either. The reason may be that while an injunction against price fixing might prevent a defendant from engaging in lucrative unlawful transactions, it would not deprive the defendant of a legally protected interest. It would not be like the case in which the defendant, in order to extirpate the effects of its unlawful act, is forced to restructure its operations at a cost that may greatly exceed any profit it made from the act. Structural relief is frequently decreed in merger cases under section 1 of the Sherman Act or section 7 of the Clayton Act or in monopolization cases under section 2 of the Sherman Act, but very rarely in a price-fixing case, such as we have here. The last way of satisfying the requirement of the minimum amount in controversy in an injunction case, the way principally argued by the defendants, is that a defendant’s clerical or ministerial costs of compliance might cany a case across the threshold. Even if an injunction doesn’t require the defendant to restructure its business or give up a lucrative lawful business opportunity, but merely tells it to stop doing something illegal, such as conspiring to fix prices, there will be lawful costs of compliance. Just the cost of duplicating an injunction in a case such as this and distributing the copies to all the relevant personnel might exceed $50,000 for each defendant, and, if so, this would argue for allowing removal to federal court. The argument would be the same as before— given the possibility of a settlement, a suit is worth as much to the plaintiff in the form of an expected value of settling it as it is costly to the defendant, at least in the single-plaintiff case. But if the argument were accepted, then every case, however trivial, against a large company would cross the threshold, whether the threshold was $50,000 or as it now is $75,000, even if the plaintiff were asking for an injunction against disclosing his unlisted telephone number. It would be an invitation to file state-law nuisance suits in federal court. We needn’t bite this bullet. The defendants have made no effort to show that what is conceivable is also probable by quantifying the internal cost of compliance to each of them and then adding it to a plaintiffs compensatory damages and penalty entitlement. The alternative basis on which the district court permitted the removal of the Alabama suit to the federal district court was the “artful pleading” doctrine. The doctrine *611 is usually taken to mean that if federal law has so far occupied a field of disputes as to extinguish any basis in state law for seeking a resolution of the dispute, a plaintiff cannot prevent removal by casting his claim as one under state law — it must actually be a claim under federal law because only federal law could supply a ground for relief. Caterpillar Inc. v. Williams, 482 U.S. 386, 393-94, 107 S.Ct. 2425, 2429-30, 96 L.Ed.2d 318 (1987); Avco Corp. v. Aero Lodge No. 735, 390 U.S. 557, 88 S.Ct. 1235, 20 L.Ed.2d 126 (1968); Kaucky v. Southwest Airlines Co., 109 F.3d 349, 351 (7th Cir.1997). And as such it can be removed to federal court even if it is not within the diversity jurisdiction, and, by virtue of 28 U.S.C. § 1441(e) (added in 1986), even if the state court could not have exercised jurisdiction over the case because it is a type of case that is within the exclusive jurisdiction of the federal courts, as well as being a case in which only federal law can supply the rule of decision. It may seem odd to allow removal and retention in such cases, rather than to trust the state court to dismiss a suit that is frivolous because it is based on nonexistent (because preempted) state law, especially since a defense of preemption is normally not a basis for removal and is therefore decided by the state court. Metropolitan Life Ins. Co. v. Taylor, 481 U.S. 58, 63, 107 S.Ct. 1542, 1546, 95 L.Ed.2d 55 (1987); Franchise Tax Board v. Laborers Vacation Trust, 463 U.S. 1, 24-27, 103 S.Ct. 2841, 2854-55, 77 L.Ed.2d 420 (1983). The usual explanation is that if the suit must be based on federal law because that is the only law that such a suit can be based on (the standard example is a suit to enforce a collective bargaining agreement, which can be litigated only under federal law), the defendant is entitled to remove and his entitlement should not be defeated by the plaintiff’s evasive drafting of the complaint. E.g., Bartholet v. Reishauer A.G. (Zurich), 953 F.2d 1073, 1075 (7th Cir.1992). It’s true that the defendant should be able to defeat this maneuver in state court by moving to dismiss the suit as frivolous; if the plaintiff countered by coming out of his state-law closet and acknowledging that he was trying to plead a federal case, the defendant could then remove. 28 U.S.C. § 1446(b). But should the defendant be put to the bother? If as a matter of fact the plaintiff is really intending to bring a federal suit though failing to cite federal law, it can be argued that his intentions should be taken as the reality and so the defendant allowed to remove what is functionally though not formally a federal suit. The problem comes in setting limits to the doctrine. There are countless cases in which a suit under state law could be thought to be a federal suit in state clothing. Antitrust law, for example, with an isolated exception, Flood v. Kuhn, 407 U.S. 258, 284-85, 92 S.Ct. 2099, 2112-13, 32 L.Ed.2d 728 (1972), is a field in which Congress has not sought to replace state with federal law. California v. ARC America Corp., 490 U.S. 93, 101-02, 109 S.Ct. 1661, 1665-66, 104 L.Ed.2d 86 (1989). The states are free to enact their own antitrust laws, reaching the same conduct as the federal laws except insofar as the states’ power to regulate economic activities in other states is limited by the commerce and due process clauses of the federal Constitution. See Herbert Hovenkamp, “State Antitrust in the Federal Scheme,” 58 Ind. L.J. 375 (1983). This is a potentially significant qualification, as we shall see; but on the view taken by the defendants in this case, any time an antitrust plaintiff brings a suit in state court under a state antitrust statute that contains substantive provisions similar to that of a federal antitrust statute, the defendant can remove on the ground that the plaintiff is trying to bring a federal antitrust suit yet to insulate it from removal to a federal court. This surprising possibility gets a boost from a footnote in Federated Department Stores, Inc. v. Moitie, 452 U.S. 394, 101 S.Ct. 2424, 69 L.Ed.2d 103 (1981). The plaintiffs in that ease brought a class suit in a state court under state fraud law and state unfair competition law. The suit was removed to federal district court, properly in the Supreme Court’s judgment because the district court had found as a fact that the plaintiffs “had attempted to avoid removal jurisdiction by ‘artful[ly]’ casting their ‘essentially federal law claims’ as state-law claims.” Id. at 397 n. 2, 101 S.Ct. at 2427 n. 2. The suit had been filed after the district court had dismissed an *612 earlier version, explicitly premised on federal antitrust law, on the basis of a federal defense that, like the “passing on” defense of Illinois Brick, the plaintiffs hoped would not be recognized by state law. It is not easy to see why this is “artful pleading” in some invidious, evasive sense. Once the federal defense was held to block the plaintiffs’ federal antitrust claim, their only hope was to proceed under state law. They had little motive to conceal a federal claim in state clothing, for their federal claim was dead. See In re Application of County Collector, 96 F.3d 890, 897 (7th Cir.1996). It is the same here. The only motive the plaintiffs in our Alabama case could have for filing a case under the Alabama statute was to avoid the federal passing-on defense of Illinois Brick, a defense they could avoid only if they pressed their claim exclusively under state law — -and if they did that the case would belong in state court because, as we have seen, it is not within the diversity jurisdiction and so is not removable to federal court on that basis. The Supreme Court went on to hold in Moitie that the “artfully pleaded” (hence federal) claims that had been removed to federal court were barred by res judicata. The suit had been refiled in state court after final judgment had been entered against the plaintiffs in federal court. We can now see how the refiling of these suits in state court under state law could be thought “artful pleading” in an invidious sense; and the Court did not say it was artful pleading — only that it would not question the district court’s finding that it was. The plaintiffs had been trying to dodge a federal court’s judgment. The defendants could have set up the judgment as res judicata in the state court in which the suits were refiled. But if the sole basis for filing a state suit is to get around, however temporarily and hopelessly, a federal judgment, it can be argued that the new “state law” suit is really the old federal suit in a transparent guise and that the federal court ought to say so in order to get rid of it quickly and thus protect the federal judgment against the possibility that the state court might abet the plaintiffs effort to get around a dispositive defense or other fatal flaw in his federal case. Doe v. Allied-Signal, Inc., 985 F.2d 908, 911-12 (7th Cir. 1993); Rivet v. Regions Bank of Louisiana, F.S.B., 108 F.3d 576, 586 (5th Cir.1997); Ultramar America Ltd. v. Dwelle, 900 F.2d 1412 (9th Cir.1990). Furthermore, any state claim in Moitie had been extinguished by the federal judgment, by operation of the doctrine of merger. Recall that the Court held the claim barred by res judicata. The reason was that the claim could have been joined to the plaintiffs’ federal claim and arose from the same cluster of facts. In these circumstances, since it was not joined, it merged into the federal judgment and disappeared, leaving nothing on which to base a suit in state court. There is no federal judgment here. Neither when the Alabama suit was filed nor when the motion to remand was filed was there any ruling by the district court, let alone a judgment, barring the suit on Illinois Brick (or any other federal) grounds. On the contrary, the district court thought Illinois Brick not a bar to a federal antitrust suit by indirect purchasers. It is true that the judge had refused to certify the first Alabama suit removed to the district court as a class action, but the denial of class certification is not a final judgment, terminating the underlying suit; the suit continues, only as an individual action rather than as a class action. The plaintiffs may well be stretching the Alabama statute to the breaking point in seeking damages for nonresident plaintiffs from nonresident defendants who sell primarily in other states. If it were clear that the plaintiffs could get no significant relief under Alabama law, this would strengthen the inference that they were merely recap-tioning their federal suit as one under state law. But it is not clear, even though the defendants are able to cite Alabama cases which say that Alabama’s antitrust statute is indeed limited to intrastate commerce and it is doubtful that any of the price-fixed sales attacked in the suit took place in intrastate rather than interstate commerce. The cases on which the defendants rely, for example Georgia Fruit Exchange v. Turnipseed, 9 Ala.App. 123, 62 So. 542, 546 (1913), date *613 from a period in which, interstate commerce being narrowly defined, see, e.g. Hadley-Dean Plate Glass Co. v. Highland Glass Co., 143 Fed. 242, 244 (8th Cir.1906), and federal power to regulate such commerce being deemed exclusive, id.; United States v. E.C. Knight Co., 156 U.S. 1, 11, 15 S.Ct. 249, 253, 39 L.Ed. 325 (1895), a state statute limited to intrastate commerce would have some, albeit a strictly limited, scope and could not have a greater scope no matter how much the state wanted it to. The eases thus were not interpreting the statute; they were interpreting the Constitution as placing upper and lower bounds on the reach of the statute, and the Constitution has since been reinterpreted. If the statute is limited today as it once was to commerce that is not within the regulatory power of Congress under the commerce clause, it is a dead letter because there are virtually no sales, in Alabama or anywhere else in the United States, that are intrastate in that sense. United States v. Lopez, 514 U.S. 549, 558, 115 S.Ct. 1624, 1630, 131 L.Ed.2d 626 (1995); Wickard v. Filburn, 317 U.S. 111, 63 S.Ct. 82, 87 L.Ed. 122 (1942); United States v. Hicks, 106 F.3d 187, 189-90 (7th Cir.1997). Other states read their antitrust statutes to reach what is now understood to be interstate commerce. E.g., R.E. Spriggs v. Adolph Coors Co., 37 Cal.App.3d 653, 112 Cal.Rptr. 585 (1974); Heath Consultants, Inc. v. Precision Instruments, Inc., 247 Neb. 267, 527 N.W.2d 596, 607 (1995) (citing cases). The reading is constitutionally permissible, Clay v. Sun Ins. Office, Ltd., 377 U.S. 179, 84 S.Ct. 1197, 12 L.Ed.2d 229 (1964), and we are given no reason to suppose that Alabama would buck this trend and by doing so kill its statute. A state’s power to regulate interstate commerce is limited, however, by the provisions of the federal Constitution that limit the extraterritorial powers of state government. A state cannot regulate sales that take place wholly outside it. K-S Pharmacies, Inc. v. American Home Products Corp., 962 F.2d 728, 730 (7th Cir.1992). State A cannot use its antitrust law to make a seller in State B charge a lower price to a buyer in C. Insofar as the Alabama suit challenges sales from plants or offices in other states to pharmacies in other states, it exceeds the constitutional scope of the Alabama antitrust law. But insofar as it challenges sales from other states to pharmacies in Alabama, it is within the intended and permissible scope of the statute, and, since there may well be a nontrivial number of such sales, the suit has enough potential merit as an Alabama antitrust suit to defeat the application of the “artful pleading” doctrine. The twist that Moitie gave to the doctrine is (very uncharacteristically for its author, Justice, now Chief Justice, Rehnquist) based on distrust of state courts, and, especially since it appears only in a footnote, should be narrowly construed in the interest of maintaining comity between the federal government and the states and keeping federal jurisdiction within the limits prescribed by Congress. But the plaintiffs are wrong to argue that if their suit, if reconceived as a federal suit, is so plainly barred by Illinois Brick as to be frivolous, this would mean that it could not be removed to federal court because federal courts lack jurisdiction over frivolous federal claims. It is quite true that a case can be so utterly lacking in merit that the proper disposition of it is dismissal under Rule 12(b)(1) of the civil rules (lack of subject-matter jurisdiction) rather than under Rule 12(b)(6) (failure to state a claim). See, e.g., Hagans v. Lavine, 415 U.S. 528, 536-37, 94 S.Ct. 1372, 1378-79, 39 L.Ed.2d 577 (1974); Korzen v. Local Union 705, 75 F.3d 285, 289 (7th Cir.1996). But it would be a considerable paradox if, the less merit a claim had, the more opportunity the plaintiff would have to restart the suit in another court. Moitie bars plaintiffs in hopeless federal cases from staving off the evil day of dismissal by shifting the case into a state court that may be confused about or even indifferent to the lack of merit of the case. Although the issue must be considered a close one because of persisting uncertainty about the estimation of the amount in controversy in injunction cases and about the scope of the doctrine of artful pleading after Moitie’s footnote, we conclude that the motion to remand the Alabama suit should have been granted, and we move on to the question whether the wholesalers should have *614 been dropped as defendants. Pretrial discovery included the taking of a thousand depositions and the production of fifty million pages of documents, and from this indigestible mass the plaintiffs have plucked a number of tasty morsels to garnish their briefs. We shall not extend this opinion with quotations. Suffice it to say that the record discloses a number of instances in which officers of the defendant wholesalers urge manufacturers to hold the line against discounting to pharmacies and their buying groups, and pledge to adhere to the charge-back system. The defendants argue that each of these “smoking guns” is susceptible of an innocent interpretation. But the issue before us is not whether the wholesalers were in fact participants in the price-fixing conspiracy; it is whether there is sufficient evidence of this to create a jury issue. In deciding this question we must construe the evidence as favorably to the plaintiffs as the record permits, not as favorably to the defendants as it permits. The defendants’ interpretations may be correct; they are not inevitable. But they argue, pointing to Matsushita and other decisions by the Supreme Court and this court, that summary judgment for a defendant is proper, even if there is some evidence of an antitrust violation, if the plaintiffs theory of violation makes no economic sense. Matsushita Electric Industrial Co. v. Zenith Radio Corp., 475 U.S. 574, 587, 106 S.Ct. 1348, 1356, 89 L.Ed.2d 538 (1986); Eastman Kodak Co. v. Image Technical Services, Inc., 504 U.S. 451, 467-69, 112 S.Ct. 2072, 2082-83, 119 L.Ed.2d 265 (1992); Reserve Supply Corp. v. Owens-Corning Fiberglas Corp., 971 F.2d 37, 49 (7th Cir.1992); Illinois Corporate Travel, Inc. v. American Airlines, Inc., 806 F.2d 722, 726 (7th Cir. 1986). This has to be the right rule, given the potential for jury confusion in litigation as enormous and esoteric as a billion-dollar antitrust damages action. The wholesalers argue that it would have been contrary to them economic self-interest for them to have joined a conspiracy that prevents them from selling at discounted prices to the pharmacies. The lower the price at which they sell to the pharmacies, the larger their volume of sales, and if their markup is unaffected this will translate into larger gross and probably net revenues. But this misconceives the plaintiffs’ theory of the wholesalers’ violation. The theory is that the wholesalers were the manufacturers’ cats-paws. There is nothing new about the idea that a cartel might “hire” a customer to help police the cartel. See Elizabeth Granitz & Benjamin Klein, “Monopolization by ‘Raising Rivals’ Costs’: The Standard Oil Case,” 39 J.Law & Econ. 1 (1996). The theory is especially plausible in the circumstances of the present case. (That doesn’t mean it’s correct; that’s not the issue.) Drag wholesalers appear to be an endangered commercial species. Before the chargeback system was adopted, the manufacturers would often sell directly to hospitals, HMOs, and other favored customers, bypassing the wholesalers, since by selling directly they could monitor each customer’s purchases and so try to identify instances in which a customer was purchasing for purposes of arbitrage rather than for its own use. The pharmacies were trying to get into the act by forming buying groups. Buying groups frequently act as their members’ wholesaler, buying directly from the manufacturer and thus cutting out independent wholesalers. Desiring a piece of the action with the favored customers, who were proliferating, the wholesalers agreed to implement a chargeback system that would shore up the manufacturers’ system of price discrimination, an integral component of the price-fixing conspiracy. Amd desiring to discourage buying groups they joined with the manufacturers to hold the line against granting any discounts to such groups and so discourage them formation by reducing the advantages of membership. The picture that we have just sketched may not be true, but there is enough evidence supporting it to preclude summary judgment; and our main point for the present is merely that the defendants are wrong to argue that it would make no sense for the wholesalers to conspire with them to fix the prices of pharmaceutical drags. It would make perfectly good sense, and so the “smoking gun” evidence cannot be dismissed as being obviously misunderstood, empty boasting, or idle corporate gossip. *615 The wholesalers point to their wafer-thin profit margins. The margins might be even thinner if the wholesalers had refused to play their appointed role as agents of a manufacturers’ cartel — in fact they might be out of business. And absence of monopoly profits is not inconsistent with monopoly (collusive or single-firm), since firms may transform monopoly profits into costs in their efforts to engross a larger share of them. The wholesalers point to instances in which they did engage in arbitrage, sought permission to give discounts to pharmacies, and even helped to organize buying groups of pharmacies. This evidence does not erase the factual question of whether the wholesalers joined the conspiracy. It is just evidence to be weighed in the balance by the trier of fact. There are inherent strains in a cartel. A member can do better by undercutting the cartel slightly and obtaining enormously increased volume at a slight sacrifice of unit profit than by honoring the cartel price and suffering an erosion of sales because of cheating by less scrupulous members. George J. Stigler, “A Theory of Oligopoly,” in Stigler, The Organization of Industry 39 (1968). That is why cartels tend to collapse of their own weight. And if as the plaintiffs argue the wholesalers were tools of the manufacturers — reluctant accomplices, yet not the less liable for that, Albrecht v. Herald Co., 390 U.S. 145, 150 n. 6, 88 S.Ct. 869, 871, 19 L.Ed.2d 998 (1968); United States v. Parke, Davis & Co., 362 U.S. 29, 45, 80 S.Ct. 503, 512, 4 L.Ed.2d 505 (1960); MCM Partners, Inc. v. Andrews-Bartlett & Associates, Inc., 62 F.3d 967, 973 (7th Cir.1995); Isaksen v. Vermont Castings, Inc., 825 F.2d 1158, 1163 (7th Cir.1987), rather than principals— naturally they would be restive. As for the wholesalers’ sponsorship of buying groups, it did not begin until after this litigation commenced, and may be strategic. And no significance can be attached to the fact that some of the wholesalers sued the manufacturers. Illinois Brick entitles them to do so. One virtue of the rule of that case is that it creates an incentive for middlemen to break out of a cartel and sue the supplier members; it sows dishonor among thieves; they still may be thieves. The last issue is whether the district judge was right to carve DuPont Merck out of the manufacturers’ conspiracy. A joint venture of DuPont and Merck, DuPont Merck was formed in 1991, two years after the beginning of the alleged conspiracy (or at least the earliest date within the statute of limitations), to take over DuPont’s pharmaceuticals division, DuPont Pharma. Upon its formation, DuPont Merck announced that it was adopting a “single price” policy for DuPont Pharma’s drugs, the drugs involved in this suit; it was withdrawing its discounts to hospitals and other favored customers and so abandoning its participation in the charge-back system. This demarche may seem irrelevant to whether DuPont Merck should be dismissed from the case. It is conceded to be the successor to DuPont Pharma, so that if DuPont Pharma was violating the Sherman Act between 1989 and 1991, DuPont Merck is liable under standard principles of successor liability even if it cleaned up its predecessor’s act upon taking over. Chaveriat v. Williams Pipe Line Co., 11 F.3d 1420, 1424-25 (7th Cir.1993). Moreover, the adoption of a single-price policy by terminating discounts is not the termination of the antitrust violation. The violation is not the discrimination. The discrimination is merely evidence of the violation. A cartel so powerful that it did not have to grant discounts to any customer would not be exonerated from antitrust liability. All that the withdrawal of discounts would do in such a case would be to create an additional class of plaintiffs. The significance of the single-price policy lies elsewhere — in DuPont Merck’s extraordinary but not improper argument that it thumbed its nose at the manufacturers’ cartel because it had sufficient monopoly power on its own to obtain higher profits by a unilateral pricing policy, namely that of giving no discounts to anyone. The proprietary drugs at issue in this ease that DuPont Merck makes are only five in number and they include the famous anticoagulant Coum-aden, which although its patent has expired is said to have no competition because doctors refuse to prescribe a generic or other substitute. The other four drugs are sufficiently comparable to Coumaden in point of uniqueness, according to DuPont Merck’s *616 submission, that it can make more money selling them all without any discounts even though it must lose some sales to the formerly favored customers. This is not an absurd argument; it may for all we know be entirely sound; it is backed by evidence. But there is enough contrary evidence to preclude summary judgment. Before 1991, but within the period of the statute of limitations, DuPont Pharma had a two-price policy and a chargeback system to implement it, and it participated in the trade association meetings in which, if the plaintiffs’ “smoking gun” evidence is credited — as it must be, in the present posture of the case — the conspiracy was hatched or nurtured. The withdrawal of the discounts is evidence that DuPont Merck believed that it had enough unilateral monopoly power to go its own way. But it is not conclusive evidence, and even if it were, it would be consistent with DuPont Pharma’s not having shared the belief. We said that DuPont Merck is liable for its predecessor’s antitrust violations and here we add that if DuPont Pharma is found to have participated in the conspiracy, DuPont Merck could not avoid liability even for the post-1991 conduct of the conspiracy, a conspiracy in which it was not (or so a jury might find) involved. A mere change of policy, a mere cessation of involvement, is not effective withdrawal from a conspiracy. To terminate one’s liability for the continuing illegal acts of a conspiracy that one had joined, a withdrawing member must either report the conspiracy to the authorities or announce his withdrawal to his coconspirators. United States v. United States Gypsum Co., 438 U.S. 422, 463-65, 98 S.Ct. 2864, 2886-87, 57 L.Ed.2d 854 (1978); United States v. Patel, 879 F.2d 292, 294 (7th Cir.1989); United States v. Puma, 937 F.2d 151, 158 (5th Cir.1991). So far as appears, DuPont Merck did neither. The four rulings appealed from are thus REVERSED.