WALLACE, Circuit Judge:
In this consolidated action, the parties appeal from the district court’s judgment and orders certified for appeal which encompass a plan for distributing a fund created by the settlement of a multidistrict class action alleging a nationwide conspiracy to fix the price of cement. The appellants contend that the “offset” provision in the plan is inequitable, that certain parties should have been excluded from participation, and that other potential claimants should have been allowed to share in the fund. We have jurisdiction pursuant to 28 U.S.C. § 1291, and we affirm.
I
Beginning in 1976, 35 actions were filed in 12 district courts against a number of cement manufacturers and their trade association alleging, among other things, a nationwide conspiracy to fix prices in violation of section 1 of the Sherman Act, 15 U.S.C. § 1. Six of these actions also alleged another conspiracy among certain sellers of ready-mix concrete in Arizona (the Ready-mix Companies). Several of the lawsuits alleged other antitrust violations, including violations of state laws. On September 12, 1977, the Judicial Panel on Multidistrict Litigation transferred all pending actions to the District of Arizona for coordinated pretrial proceedings. In re Cement and Concrete Antitrust Litigation, 437 F.Supp. 750 (J.P.M.D.L. 1977).
On March 9, 1979, the district court granted a motion to certify the actions as class actions and establish four classes. The first and broadest class was the National Cement Class, made up of all those who purchased cement during the relevant time period other than the defendants and the members of the governmental class»» [1438]*1438The next two classes, the California Governmental Class and the Oregon Governmental Class, consist of the respective states and their political subdivisions that purchased cement. The fourth class, the Arizona Ready-mix Class, consists of those entities that purchased ready-mix concrete for delivery within the State of Arizona. In the fall of 1979, the defendants in this fourth class action (the Ready-mix Companies) settled with this class and a partial judgment was entered on July 8, 1980, pursuant to Fed.R.Civ.P. 54(b).
Between July 1979 and October 1981, the first three classes entered into a series of settlement agreements with seven of the major defendants. The settlements totaled approximately $32,000,000 in cash plus 135,000 shares of stock in one of the defendant companies. The settlement agreements left distribution of the settlement fund for later resolution, subject to the approval of the court. On January 6, 1984, the district court decertified the National Cement Class. The decertification order provided, however, that absent class members retained their eligibility to participate in the settlement fund. After decertification, the remaining defendants reached separate settlements with the California and Oregon Governmental Classes and with most of the remaining individual plaintiffs.
On January 21, 1985, the district court approved a plan for distributing the settlement fund. Under this plan, the fund would be distributed to the class members in proportion to the amount of their cement purchases both from manufacturers who settled with the class (settling defendants) and from manufacturers who did not settle with the class (nonsettling defendants). Any claimant, however, whose claim was based on purchases from nonsettling defendants would have to offset against his share of the fund any amount he had already received in an individual settlement with a nonsettling defendant. Furthermore, if a claimant chose to continue litigating claims against the nonsettling defendants, the claimant’s share of the settlement fund would be placed in escrow pending the outcome of that litigation. Any amount received by that claimant from continued litigation with the nonsettling defendant would then be offset against the claimant's share when released from escrow. Alternatively, a claimant could choose to waive any claims against nonsettling defendants and immediately receive a full share of the fund.
On appeal, the parties raise five objections to the district court’s allocation of the settlement fund. First, they argue that the court should have adopted a proposal for assigning “weights” to claims as a substitute for the plan’s offset provision. Second, Arizona contends that the court should have excluded the Ready-mix Companies from sharing in the fund. Finally, some argue that the district court erred in its refusal to allow payments out of the fund for claims based on: (1) an alleged concerted refusal to deal, (2) state civil penalty statutes, and (3) state indirect purchaser statutes. We consider each of these five objections in turn.
II
Several of the parties in this appeal object to the offset provision in the district court’s plan and to the procedures followed by the district court in its adoption. Rule 23(e) declares that a “class action shall not be ... compromised without the approval of the court.” Fed.R.Civ.P. 23(e). When evaluating a proposed class action settlement, a district court must determine whether it is “fundamentally fair, adequate and reasonable.” Officers for Justice v. Civil Service Commission, 688 F.2d 615, 625 (9th Cir.1982) (Officers for Justice), cert. denied, 459 U.S. 1217, 103 S.Ct. 1219, 75 L.Ed.2d 456 (1983). We review the district court’s determination for an abuse of discretion. In re Equity Funding Corporation of America Securities Litigation, 603 F.2d 1353, 1362 (9th Cir.1979) (Equity Funding).
A.
The district court required class members who purchased cement from nonsettling defendants to offset their share of the settlement fund with any moneys they [1439]*1439have received or will receive in litigation against the nonsettling defendants. Without this requirement, the plan was perceived as being inequitable to class members who purchased cement only from the settling defendants. Some time before the parties agreed to settle, the defendants had entered into a judgment-sharing agreement. Because of this agreement, the settling defendants could terminate their liability in the case only by having the plaintiffs agree not to sue the nonsettling defendants based upon their joint liability for the acts of the settling defendants. A successful settlement was possible therefore only because the plaintiffs covenanted not to sue the nonsettling defendants based upon their joint liability for the plaintiffs’ purchases of price-fixed cement from the settling defendants.
Nevertheless, even with this covenant, the plaintiffs retained the right to continue litigation against nonsettling defendants if the lawsuits were based on purchases from those defendants. Therefore, a portion of the class — those who purchased only from the settling defendants — had to release all claims against the nonsettling defendants, while the remainder of the class did not. This left open the possibility that class members with claims against nonsettling defendants could secure double recovery: collect from the settlement fund for purchases made from settling defendants, and also collect against the nonsettling defendants under another settlement or a judgment. The district court adopted the offset provision to put plaintiffs who purchased cement from nonsettling defendants in the same position as plaintiffs who had to waive their joint-liability claims against those nonsettling defendants. On appeal, some parties argue that the court instead should have adopted one of two other proposed solutions.
One proposal would have allowed only class members who purchased cement from settling defendants to share in the fund. The district court properly rejected this solution. Antitrust defendants are jointly and severally liable for the acts of their coconspirators. See, e.g., In re Uranium Antitrust Litigation, 617 F.2d 1248, 1257 (7th Cir.1980). Therefore, the settling defendants were liable to all class members, not to just those members who purchased cement from them personally. Accordingly, class members who purchased cement from nonsettling defendants had valid claims, and thus were entitled to share in the fund.
The second proposal would have the court “weight” claims. Under this proposal, a weight of 1.0 would be assigned to claims based upon purchases from the settling defendants and a lower weight, possibly as low as 0.0, would be assigned to claims based upon purchases from nonsettling defendants. The proposal did not offer any criteria for such a weighting, nor did it suggest examples of what might be appropriate weights for the claims of any of the class members.
Although the district court rejected the weighting proposal, the offset provision it adopted in effect applies an equitable weighting to claims. Class members who purchased cement from nonsettling defendants will participate fully in the settlement fund except to the extent they have benefited from their separate lawsuits against the nonsettling defendants. That is, claims to the fund based upon purchases from nonsettling defendants are effectively given a lower weight. Moreover, this weight is not based upon an artificial estimate. Instead, the weight of a claim to the fund is discounted by the precise value that a class member receives from its claims against the nonsettling defendants. Those who do not receive any value from these claims receive a full share of the fund. Those who do receive a value from these claims, through either a settlement with or a judgment against these defendants, receive a proportionately smaller share of the fund. The court’s plan thus “weights” the claims to the fund in an equitable manner.
On at least two other occasions we have affirmed a district court’s approval of a class action settlement that included an offset provision. In Equity Funding, 603 F.2d 1353, the district court approved a [1440]*1440plan that required claimants to offset their share of the fund against any moneys they received in a related bankruptcy proceeding. We affirmed the district court, stating that the court’s approval of the offset provision “was a component of its duty to insure the equitable distribution of the settlement proceeds.” Id. at 1365. In Officers for Justice, 688 F.2d 615, the district court approved a class action settlement in an employment discrimination lawsuit that contained a “mitigation clause.” Under the clause, a claimant’s share of the back pay settlement was reduced if he had already received some compensation by occupying certain higher paying positions. Id. at 629. We found the mitigation clause to be “an entirely reasonable provision.” Id. at 630.
All participants in the Cement settlement wanted more. We must recognize, however, that “[a] just result is often no more than an arbitrary point between competing notions of reasonableness.” In re Corrugated Container Antitrust Litigation, 659 F.2d 1322, 1325 (5th Cir. Unit A Oct. 1981), cert. denied, 456 U.S. 998, 102 S.Ct. 2283, 73 L.Ed.2d 1294 (1982). We hold that, in adopting the offset provision, the district judge forged an equitable solution and did not abuse his discretion.
B.
The State of Arizona raises three procedural objections to the offset provision. Arizona first argues that class members did not have adequate notice and an opportunity to object to the offset provision. Second, Arizona argues that the offset provision constituted an impermissible unilateral modification of the proposed settlement. Finally, Arizona contends that the waiver provisions of the plan create an impermissible “spurious” class action.
Rule 23(e) of the Federal Rules of Civil Procedure states that notice of a compromise in a class action “shall be given to all members of the class in such manner as the court directs.” Although the rule gives a court broad discretion, both the form and content, of the notice must satisfy constitutional due process requirements. Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 172-77, 94 S.Ct. 2140, 2150-52, 40 L.Ed.2d 732 (1974). The due process clause requires that notice be “reasonably calculated, under all the circumstances, to apprise interested parties of the pendency of the action and afford them an opportunity to present their objections.” Mullane v. Central Hanover Bank & Trust Co., 339 U.S. 306, 314, 70 S.Ct. 652, 657, 94 L.Ed. 865 (1950). This standard does not mean that the notice must recite the language of every provision of a proposed settlement agreement. Notice is satisfactory if it “generally describes the terms of the settlement in sufficient detail to alert those with adverse viewpoints to investigate and to come forward and be heard.” Mendoza v. United States, 623 F.2d 1338, 1352 (9th Cir.1980) (Mendoza), cert. denied, 450 U.S. 912, 101 S.Ct. 1351, 67 L.Ed.2d 336 (1981). Whether notice satisfies due process is a question of law which we review de novo.
On July 29,1984, the class members in this case received a notice informing them that the parties had entered into the proposed settlement agreement creating the settlement fund. The question of how to distribute the funds was held in abeyance pending consideration by the court of “the views of all interested parties.” On November 9, 1984, the district judge announced for the first time his intention to require class members to offset their claims against the fund with any moneys received from nonsettling defendants. Thus, the parties were never directly given notice of the offset provision until after it had been imposed.
The July 29 notice, however, warned the parties that the district court would be determining “whether the plan of distribution should give preference to claims for purchases from the settled defendants ... by assigning a weighting factor in favor of such purchases as compared to purchases from all other cement manufacturers.” Class members were thus on notice that to the extent a claim was based on purchases from nonsettling defendants, the court might reduce a claimant’s share of the pro[1441]*1441ceeds. The distribution plan ultimately adopted by the district court is within the scope of this notice. Under the plan, a claim based on purchases from nonsettling defendants may be reduced through an offset for any recoveries from those defendants. As discussed earlier, the district court has in effect assigned a lower weight to claims based on purchases from nonsettling defendants. The notice sufficiently alerted those class members adverse to such a scheme “to investigate and to come forward and be heard.” Mendoza, 623 F.2d at 1352; see In re Corrugated Container Antitrust Litigation, 643 F.2d 195, 223-24 (5th Cir.1981) (holding that notice of a settlement agreement that failed to provide an estimated range of per unit recovery was nonetheless satisfactory), cert. denied, 456 U.S. 998, 102 S.Ct. 2283, 73 L.Ed.2d 1294 (1982); Equity Funding, 603 F.2d at 1361-62 (holding that notice need not give a precise valuation figure if the notice informs the class that the judge will be deciding the question at the hearing).
Arizona’s second objection is based on the proposition that a court may not unilaterally modify a settlement agreement but may only approve or reject the agreement as a whole. See Officers for Justice, 688 F.2d at 630 (dicta); Pettway v. American Cast Iron-Pipe Co., 576 F.2d 1157, 1172 (5th Cir.1978), cert. denied, 439 U.S. 1115, 99 S.Ct. 1020, 59 L.Ed.2d 74 (1979). Arizona suggests that the district court has done exactly this because the precise terms of the plan of distribution were not contained in any plan proposed by the parties but instead were provided by the court.
We need not decide whether a court may under some circumstances unilaterally modify a class action settlement because we conclude that here the district court did not make such a unilateral modification. In 1981, the district court approved, without modification, the seven settlement agreements that created the settlement fund. Each of these agreements provided that “[t]he allocation and distribution of the Settlement Fund shall be in accordance with the provisions of any agreements or plans approved by the Court.” When the court approved these settlement agreements it explicitly retained jurisdiction “for the purpose of supervising and directing the administration of the settlements and distribution of the Settlement Funds.”
One could construe these provisions to mean that the district court merely retained the authority to approve or disapprove a plan of distribution. Such a construction, however, accords neither with the course of proceedings nor with the understanding of the parties. The parties did not reach an agreement on a unified distribution plan and then submit it to the district court for approval. Instead, the parties submitted several issues as to distribution of the funds to the court for its resolution. Moreover, correspondence among the parties indicates that they understood that “all plaintiffs are free to engage in any ‘donnybrook’ that may be appropriate to divide up the ‘spoils’ ” and that failure to achieve agreement among plaintiffs “will necessitate resolution by the court.” We conclude that the parties intended to be bound by the district court’s resolution of these distribution issues in the same way that parties to a contract might agree to be bound by the findings of an appraiser or an arbitrator. In short, the parties consented to the district court’s reasonable resolution of conflicts that might arise when distributing the settlement fund. Therefore, the court did not “unilaterally” modify the settlement agreement. Cf. Equity Funding, 603 F.2d at 1365-66 (affirming class action settlement that gave the district court authority to place a value on certain securities and to determine whether to offset recoveries as part of the distribution plan); Curtiss-Wright Corp. v. Helfand, 687 F.2d 171, 173 (7th Cir.1982) (holding that a district court may unilaterally reduce a class member’s share of a settlement fund if the court reserved decision on the issue before approving the settlement).
Arizona’s final contention is that the waiver provision of the plan in effect requires those who purchased cement from nonsettling defendants to “opt-in” to the settlement class after the settlement was [1442]*1442already approved. Such an opt-in provision, Arizona argues, violates the intent and purpose of rule 23 by creating what is known as a “spurious class action.” This argument mischaracterizes the nature of the election that these class members faced. The class members who purchased cement from the nonsettling defendants were, by definition, already members of the class when the settlement was approved. As members, they are entitled to share in the settlement fund. The only election they were required to make was not whether to opt-in, but instead whether to waive their rights against the nonsettling defendants. The waiver provision does not create a “spurious class action.”
Ill
Arizona argues that the Ready-mix Companies should not be allowed to share in the settlement fund because they were not part of the National Cement Class. The district court erred, Arizona contends, by ruling in the final order distributing the funds that the Ready-mix Companies were included in the class when the notice of settlement was approved on October 20, 1980.
No one doubts that the Ready-mix Companies were “purchasers of cement” under the definition of the National Cement Class. Arizona argues, however, that the Ready-mix Companies fell within a provision in the class definition excluding “defendants” from the plaintiff class. Arizona contends that the Ready-mix Companies were defendants in the cement conspiracy because they were coconspirators with the cement manufacturers. Arizona also points out that, in any case, the Ready-mix Companies were certainly defendants in the separate count that alleged an Arizona-wide conspiracy among the Ready-mix Companies to fix the price of ready-mix concrete.
We need not address this issue directly. In 1981, the parties stipulated that the Ready-mix Companies were part of the National Cement Class. The court approved this stipulation and entered an order on April 14, 1981, to reflect the parties’ agreement on this issue. None of the parties moved the district court for relief from the stipulation. Indeed, even on appeal Arizona does not argue that the stipulation was not binding or valid. Because the validity of the stipulation was questioned neither in the district court nor on appeal, the parties are bound by it.. The stipulation specifically encompasses the question of the Ready-mix Companies’ participation; therefore, we conclude that they are part of the National Cement Class and entitled to participate in the settlement.
Arizona makes a final argument with respect to one of the Ready-mix Companies, New Pueblo Constructors, Inc. (New Pueblo). Arizona contends that New Pueblo must be excluded from participation in the settlement fund because New Pueblo stipulated to the dismissal of the separate action it brought against the defendants. This argument is plainly without merit, however, because that stipulation expressly stated that it did not preclude New Pueblo from asserting any rights it may have to the settlement fund.
IV
Appellants Smith & Green Corp. and Arizona Slump Block, Inc. (collectively Smith & Green) argue that the district court should have allowed them to share in the settlement fund based on their claims alleging that the defendants conspired to refuse to deal with them in violation of section 1 of the Sherman Act, 15 U.S.C. § 1. In an order filed November 23, 1984, the district court ruled that Smith & Green “are not members of the plaintiff class of direct purchasers” and therefore could not share in the settlement fund. The court later clarified this ruling in an order filed January 23,1985, by stating that Smith & Green could make claims based upon direct purchases of cement but that “[c]laims based upon a refusal to deal will not be allowed.”
We must determine initially the scope of our review when examining a district court’s interpretation of a class definition. If this were a district court’s initial order either denying or certifying a class, [1443]*1443we would review for an abuse of discretion. See Moore v. Hughes Helicopters, Inc., 708 F.2d 475, 479 (9th Cir.1983). It may be that this same standard of review should also be used when determining whether a district court properly modified the scope of a class by altering the class definition. When, however, a court does not modify a class description, but instead interprets an existing class definition, the court does not exercise its discretion. Instead, the district court construes the language of the class definition in much the same way that a court might construe the language of a contract or a statute. We therefore conclude that we should review de novo the district court’s analysis of the language of the class definition. Cf. Miller v. Safeco Title Insurance Co., 758 F.2d 364, 367 (9th Cir.1985) (holding that a district court’s analysis of contract language is reviewed de novo).
The National Cement Class consisted of entities “that have purchased for delivery within the United States cement from any cement manufacturer.” Thus, the plain language of the class definition encompasses entities harmed by purchasing cement, not those harmed because they were not permitted to purchase cement. Moreover, language in the district court’s order certifying this class indicates that the class was intended to cover only those harmed by some form of price-fixing conspiracy. The court found that the question common to the claims of the class was “whether any of the defendants conspired to fix prices, allocate customers or standardize credit terms.” The court affirmed the nationwide scope of the class specifically because of the evidence of nationwide price-fixing. Furthermore, when concluding that the impact of the defendants’ acts was not individualized, the court stated that “an illegal price fixing scheme presumptively impacts upon all purchasers of a price fixed product.” Finally, the plaintiffs recognized in their motion for class certification that “only the purchase of cement or ready-mix at elevated price levels ... is germane to this litigation.” Nowhere in the class certification proceedings did the district court or the parties refer to claims based upon a concerted refusal to deal. We therefore hold that the district court properly concluded that refusal to deal claims were not within the scope of the class action.
V
The district court also ruled that “[cjlaims will not be allowed for any State civil penalties.” California, Alabama, and Minnesota (the States) argue that, because they expressly released in the settlement agreement their state civil penalty claims, they are entitled to participate in the settlement fund as a matter of contract law. This argument misses the mark, however, because the critical issue is not whether the States gave consideration in the settlement agreement, and therefore have an enforceable contract. Instead, the issue is whether those bound by the settlement agreement have waived the right to challenge the validity of the States’ civil penalty claims. We review de novo the interpretation of contract language but defer to the district court’s findings of fact with respect to extrinsic evidence necessary to interpret a contract. See Miller v. Safeco Title Insurance Co., 758 F.2d 364, 367 (9th Cir.1985).
We conclude that the parties to the settlement agreement retained the right to challenge the state civil penalty claims. In approving the settlement agreements, the district court did not determine “any of the disputed issues of fact or law involved in the litigation ” (emphasis in original). As stated previously, the court explicitly retained jurisdiction “for the purpose of supervising and directing the administration of the settlements and distribution of the Settlement Funds.” Furthermore, correspondence between counsel for the parties shows that they understood that the settlement agreements left unresolved the validity of the various state law claims. For example, correspondence between the parties relating to the settlement agreement declares that “[a]ll questions relating to allocation of the settlement proceeds have been preserved to a later date, without any [1444]*1444waiver and/or commitment by anyone with regard to the formula which will be. adopted.” Thus, although the settlement agreements gave the parties the right to argue that the state civil penalty claims are valid, the agreements did not resolve that or any other issue pertaining to which claims would be compensated from the fund. Accordingly, the States’ contention that they have a contractual right to participate in the settlement fund through their state civil penalty claims is without merit.
Turning then to the rejection, the district court did not express a reason for disallowing civil penalty claims. Nevertheless, we need not decide whether the court should have included a “reasoned response” to the objections of civil penalty claimants, see Mandujano v. Basic Vegetable Products, Inc., 541 F.2d 832, 836 (9th Cir.1976), “(bjecause the parties neither raise the issue nor contend that it affects our jurisdiction.” Spaulding v. University of Washington, 740 F.2d 686, 694 n. 2 (9th Cir.), cert. denied, 469 U.S. 1036, 105 S.Ct. 511, 83 L.Ed.2d 401 (1984). The States have not raised any other objections to the district court’s ruling on this issue.
VI
The district court disallowed claims arising from indirect purchases of cement — purchases of cement or products containing cement from a middle level distributor rather than directly from a manufacturer. The court concluded that state statutes purporting to authorize claims by indirect purchasers are preempted by federal law because “[s]uch statutes are clear attempts to frustrate the purposes and objectives of Congress, as interpreted by the Supreme Court in Illinois Brick [Co. v. Illinois, 431 U.S. 720, 97 S.Ct. 2061, 52 L.Ed.2d 707 (1977)].” The question whether federal law preempts state law is an issue of law which we review de novo.
The supremacy clause mandates that “[t]his Constitution, and the laws of the United States which shall be made in pursuance thereof ... shall be the supreme law of the land; ... any thing in the constitution or laws of any state to the contrary notwithstanding.” U.S. Const. art. VI. The Supreme Court has identified two general ways in which federal law, through the supremacy clause, preempts state law. First, “[i]f Congress evidences an intent to occupy a given field, any state law falling within that field is pre-empted.” Silkwood v. Kerr-McGee Corp., 464 U.S. 238, 248, 104 S.Ct. 615, 621, 78 L.Ed.2d 443 (1984) (Silkwood). Second,
[i]f Congress has not entirely displaced state regulation over the matter in question, state law is still preempted to the extent it actually conflicts with federal law, that is, when it is impossible to comply with both state and federal law, ... or where the state law stands as an obstacle to the accomplishment of the full purposes and objectives of Congress.
Id. (citation omitted) (emphasis added).
No one argues that Congress intended federal law to occupy entirely the field of antitrust law, nor do we need to reach that issue. We therefore decide the question before us by applying the second, more limited, test for preemption, and examining the relationship of the state indirect purchaser statutes to the purposes and objectives of Congress with respect to the federal antitrust laws.
The Supreme Court described the purpose of federal antitrust laws in the course of two cases. First, in Hanover Shoe, Inc. v. United Shoe Machinery Corp., 392 U.S. 481, 88 S.Ct. 2224, 20 L.Ed.2d 1231 (1968) (Hanover Shoe), the Supreme Court held that a defendant in an antitrust suit cannot assert as a defense that the plaintiff did not suffer damages to the extent that the plaintiff passed on any of the defendant’s illegal overcharges to the plaintiff's customers (the indirect purchasers). The Court rejected this defense because proving the amount of any passed-on overcharge “would often require additional long and complicated proceedings involving massive evidence and complicated theories.” Id. at 493, 88 S.Ct. at 2231. In addition, splitting up a potential damage [1445]*1445award among direct and indirect purchasers would reduce the incentives for any one person to bring an action. Id. at 494, 88 S.Ct. at 2232. As a result, “[t]reble-damage actions ... would be substantially reduced in effectiveness,” thus interfering with Congress’s goal of encouraging private enforcement of the antitrust laws. Id.
In a second case, Illinois Brick, 431 U.S. 720, 97 S.Ct. at 2061, the Supreme Court held that indirect purchasers could not, except under special circumstances, assert claims under section 4 of the Clayton Act, 15 U.S.C. § 4. The Court reached this conclusion in two. steps. The Court reaffirmed the holding of Hanover Shoe, again emphasizing that a “pass-on” defense would “add whole new dimensions of complexity to treble-damages suits,” Illinois Brick, 431 U.S. at 737, 97 S.Ct. at 2070, and would reduce “the effectiveness of those suits if brought by indirect purchasers with a smaller stake in the outcome than that of direct purchasers suing for the full amount of the overcharge,” id. at 745, 97 S.Ct. at 2074. Given Hanover Shoe, the Court then considered whether the indirect purchase theory could be used by plaintiffs rather than defendants. The Court observed that if every plaintiff affected by an overcharge could sue only for the amount it absorbed, treble-damages actions would be subject to the same complications found objectionable in Hanover Shoe. Id. at 731-35, 97 S.Ct. at 2067-69. Moreover, the Court recognized that if direct purchasers could bring claims for the full overcharge, any additional lawsuit brought by an indirect purchaser “would create a serious risk of multiple liability for defendants.” Id. at 730, 97 S.Ct. at 2067. The Court declined to “ ‘open the door to duplicative recoveries.’ ” Id. at 731, 97 S.Ct. at 2067, quoting Hawaii v. Standard Oil Co., 405 U.S. 251, 264, 92 S.Ct. 885, 892, 31 L.Ed.2d 184 (1972).
Together Hanover Shoe and Illinois Brick define three purposes or objectives of federal antitrust law in this context: (1) avoiding unnecessarily complicated litigation; (2) providing incentives to direct purchasers to bring private damages actions; and (3) avoiding multiple liability for defendants. With the contours of federal policy outlined by the Supreme Court in these two cases, we can proceed to determine whether the state indirect purchaser claims made in this case stand as an obstacle to accomplishing fully these objectives.
The state indirect purchaser statutes can be construed two different ways. Under one interpretation, they restrict a plaintiff to suing only for the amount of any illegal overcharge that it has absorbed. With this construction, the state laws directly conflict with federal law. Illinois Brick holds that a direct purchaser has a right, under federal law, to claim damages for the entire amount of any overcharge, whether or not he actually absorbed that overcharge. The states are not free to complicate treble damages actions and reduce the incentives for direct purchasers to bring such lawsuits by curtailing a direct purchasers’ right to sue for the full overcharge.
The alternative interpretation of the indirect purchaser statutes is that, although they do not affect the rights of direct purchasers, they permit indirect purchasers to bring claims for damages in addition to the claims brought by direct purchasers. We conclude, however, that even with this construction the indirect purchase claims asserted here would impermissibly interfere with the three policy goals outlined in Hanover Shoe and Illinois Brick.
First, under the latter interpretation, allowing the indirect purchaser claims asserted here would complicate antitrust enforcement. We recognize that the state law indirect purchaser claims could have been brought separately from the federal direct purchaser actions. Thus, indirect purchaser claims need not necessarily inject excessive complications into direct purchase actions that are based on federal law. Nevertheless, as this case illustrates, given the broad scope of joinder in modem civil actions, the claims of direct and indirect purchasers may often end up in the same [1446]*1446lawsuit. Indeed, res judicata principles may require a plaintiff with both direct and indirect claims against a single defendant to bring them together in one lawsuit. Cf Restatement (Second) of Judgments § 24 (1982) (the doctrines of merger and bar extinguish “all rights of the plaintiff to remedies against the defendant with respect to all or any part of the transaction, or series of connected transactions, out of which the action arose”). If so, complications are unavoidable.
Second, permitting the state law indirect purchaser claims asserted here could also limit the recoveries of direct purchasers and thus reduce their incentives to bring antitrust actions. A defendant’s ability and willingness to settle its claims with direct purchasers depends in part on the magnitude of the defendant’s other obligations, including its potential obligations to indirect purchaser claimants. To the extent that indirect purchaser claims have been brought or are threatened, a defendant will reduce any offer to compromise with direct purchasers. Furthermore, if antitrust treble damages judgments exhausted a defendant’s net assets, then the claims of direct purchasers would have to share the defendant’s estate in bankruptcy with the claims of indirect purchasers. These possibilities reduce the expected recoveries of direct purchasers and thus reduce the incentives for direct purchasers to bring private damages actions.
Third, the state indirect purchase claims asserted here conflict most directly with federal policy because they create the risk of multiple liability for defendants. Recognizing these claims would allow both direct and indirect purchasers to collect for the same damages. The States respond, however, that a state cause of action is not preempted simply because it imposes a harsher remedy than federal law for conduct that is a violation of both federal and state law. This proposition may be true as a general matter. Nevertheless, it does not apply where, as here, an express federal policy condemns multiple liability for antitrust defendants. The Supreme Court has not attempted to define the basis of this policy, nor need we. The Supreme Court has held unequivocally that the policy against multiple liability exists and has not hesitated to make repeated reference to it. See, e.g., Associated General Contractors of California, Inc. v. California State Council of Carpenters, 459 U.S. 519, 544, 103 S.Ct. 897, 911, 74 L.Ed.2d 723 (1983); Blue Shield of Virginia v. McCready, 457 U.S. 465, 474-75, 102 S.Ct. 2540, 2545-46, 73 L.Ed.2d 149 (1982); see also Illinois Brick, 431 U.S. at 731 n. 11, 97 S.Ct. at 2067 n. 11 (“a little slopover on the shoulders of the wrongdoers” is unacceptable). We are bound by the Supreme Court’s pronouncements on this issue.
The States also make the related argument that the Supreme Court has only determined that federal law does not impose multiple liability: state legislatures are free, they contend, to impose multiple liability as a matter of state law. But it is clear that the state law claims asserted here, if generally allowed, would create the possibility that federal law would impose multiple liability. If a state indirect purchaser claim is adjudicated or settled before a federal claim by the related direct purchaser, then one of two consequences must follow: either (1) the federal direct purchaser claim is barred, or (2) the direct purchaser can also collect trebled damages based on the full amount of the overcharge, thereby imposing multiple liability on the defendant as a matter of federal law. Either consequence would conflict with express federal policies.
Finally, appellant Minnesota argues that its statute avoids any conflict with federal law because express language in the statute requires it to be applied so that defendants are not subject to duplicative liability. See Minn.Stat. § 325D.57 (1984). It may be that some indirect purchaser claims made under state law would not conflict with federal law. For example, indirect purchaser claims may be valid if, under state law, they accrue only after the statute of limitations has run for direct purchase claims. In that situation, arguably, indirect purchaser actions would not pose a risk of complicating direct purchaser actions or inflicting multiple liability on de[1447]*1447fendants. We need not address that issue, however, because the indirect purchaser claims asserted here, including those arising under the Minnesota law, would have interfered with the claims of the direct purchasers. This lawsuit involves a class action brought on behalf of and that tolled the limitations period for all direct purchasers. Moreover, the settlement fund in dispute was created out of a settlement with the class of all direct purchasers. This is not a case where an indirect purchaser has filed a lawsuit after his seller (the direct purchaser) failed to bring an action within the limitations period. Therefore, the claims asserted here created “a serious risk of multiple liability.” Illinois Brick, 431 U.S. at 730, 97 S.Ct. at 2067.
The primary cases cited by the States in support of their argument that federal law does not preempt the indirect purchaser claims asserted here are distinguishable. In Silkwood, 464 U.S. 238, 104 S.Ct. at 615, the Court held that federal law regulating nuclear energy facilities does not preempt an award of punitive damages under state tort law for conduct related to radiation hazards. Id. at 256, 104 S.Ct. at 625. Silkwood, however, did not present the same conflict present here because no federal law grants a private cause of action for radiation-related injuries. See id. at 254, 104 S.Ct. at 624 (Congress rejected suggestion that it adopt a federal tort to replace existing state law). Therefore, in Silk-wood, state law had no opportunity to conflict with federal policies embodied in a federal law authorizing private damage actions.
In Cancellier v. Federated Department Stores, 672 F.2d 1312 (9th Cir.) (Cancellier), cert. denied, 459 U.S. 859, 103 S.Ct. 131, 74 L.Ed.2d 113 (1982), we held that the Age Discrimination in Employment Act (ADEA), 29 U.S.C. §§ 621-634, does not preempt an award of tort damages on pendent state claims. Cancellier, 672 F.2d at 1318. We expressly found, however, that the state contract and implied covenant claims at issue in that case “did not duplicate” the ADEA claims. Id. By contrast, indirect purchaser claims are aimed at recovering for the identical antitrust violations and the identical damages claimed by the direct purchasers.
We hold that the state law claims in this case based on indirect purchases of cement that do not fall within any exception to the rule of Illinois Brick are preempted because they stand “as an obstacle to the accomplishment of the full purposes and objectives” of federal antitrust law. Silkwood, 464 U.S. at 248, 104 S.Ct. at 621.
C.
Finally, the States reiterate their argument that, because they expressly released in the settlement agreement their indirect purchaser claims, they are entitled to participate in the settlement fund as a matter of contract law. They bolster this contention by citing authority for the proposition that a party to a contract who compromises an unliquidated claim has given good consideration irrespective of whether the claim was in fact valid. As pointed out in section V of this opinion, however, it is clear from the record that, like the state civil penalty claims, the parties to the settlement agreement did not waive their right to challenge the indirect purchase claims. To cite but one of several examples, in a letter accompanying California’s executed copy of one of the settlements, California expressly acknowledged that the parties disagreed about whether claims would be allowed based upon indirect purchases. The letter concludes that “[fjailure to achieve an agreement on this issue among plaintiffs will necessitate resolution by the court.” For the same reasons we rejected the States’ civil penalty claims, we reject their contention that they have a contractual right to participate in the settlement fund based on their release of the indirect purchase claims.
AFFIRMED.