In this interlocutory appeal1 we must decide whether the Ohio Supreme Court would recognize a market share theory of liability in DES2 litigation. We hold that the district court erred in predicting that it would.

Plaintiffs are twenty-seven women3 allegedly suffering damage to their reproductive systems due to their in utero exposure to DES. Plaintiffs claim that each of the defendants was either a manufacturer or distributor of DES. In their complaints each plaintiff asserted strict liability under products liability, negligence under products liability, breach of warranty, market share liability, concert of action and alternative liability. Under the claim headed "market share liability" each plaintiff uniformly alleged that "[t]he passage of time since DES exposure and the Defendants' methods of sale, distribution and promotion of DES has made it unreasonably difficult to determine which Defendant manufactured the exact DES which harmed the Plaintiff." Each plaintiff claimed that "[a]s a result of the difficulty of proof stated [above], the market share theory of liability should be applied, requiring each Defendant [sic] affirmatively prove it did not cause the Plaintiff's injuries."

All defendants moved for summary judgment on all claims based on market share liability on the ground that Ohio had neither adopted such a cause of action nor abandoned the traditional tort requirement that a plaintiff must show, as part of the causation element of her claim, that she was injured by the conduct of a specific defendant. Plaintiffs filed cross-motions for summary judgment on the legal issue of their right to proceed on the theory of market liability. After examining the only two Ohio cases on the subject, the district court concluded that if properly presented with the question, the Ohio Supreme Court would recognize a market share theory of liability in DES litigation. The district court noted that on the one occasion in which the Ohio Supreme Court considered the theory of market share liability in the context of asbestos litigation, the court "[l]e[ft] the issue of market share liability in DES cases open for future consideration." See Goldman v. Johns-Manville, 33 Ohio St. 3d 40, 514 N.E.2d 691 (1987)(suit against manufacturers and suppliers of asbestos products). The district court pointed out that the Goldman court did not ultimately apply the theory not because of any disapproval with the concept, but because it found that asbestos was not a fungible product. Moreover, the district court observed that the Goldman court favorably discussed the development of, and policy reasons for, judicially-created market share liability.

The district court also found persuasive Jackson v. Glidden Co., 98 Ohio App. 3d 100, 647 N.E.2d 879 (1995), wherein the Ohio Court of Appeals impliedly adopted the theory. See id. 647 N.E.2d at 884 (allowing plaintiffs, a class of persons allegedly injured by exposure to lead paint, to proceed with market share claim). The district court relied on the Jackson court's reading of Goldman to suggest that lack of fungibility was the sole barrier to allowing a market share theory of recovery in an asbestos case. Thus, according to the district court "[b]ecause the Ohio Supreme Court has not definitively ruled whether it would adopt the theory of market share liability under appropriate facts, the Jackson opinion is authoritative."

In response to defendants' argument that it is the legislature's job to create new theories of recovery, the district court pointed out that the Ohio Supreme Court had not hesitated to adopt the theory of alternative liability in Minnich v. Ashland Oil Co., 15 Ohio St. 3d 396, 473 N.E.2d 1199 (1984), so as to avoid injustice upon an innocent plaintiff. The district court also found that the theory of market share liability is entirely consistent with the causation requirement under the Ohio Products Liability Act, because as Goldman noted, market share liability does not eliminate the need for proof of proximate cause, but "merely relaxes" the requirement that the plaintiff identify which one of a group of negligent tortfeasors caused the injury to the plaintiff. See Goldman, 514 N.E.2d at 693.

The district court therefore denied defendants' motion and granted plaintiffs' motion. It then made the determinations necessary under 28 U.S.C. § 1292(b).4 This court accepted defendants' petitions for permission to appeal on October 11, 1996.

On appeal, defendants contend that the district court misconstrued the duties of a diversity court when it held that the Ohio Supreme Court would adopt a market share theory of liability in this case. They argue that the Supreme Court once declined to adopt that novel theory and has affirmed that a plaintiff must prove that the acts of a specific actor caused her harm. These common law rules were codified in a comprehensive Product Liability Act in 1988, which does not include market share liability, and is antithetical to essential elements of market share. Further, the Ohio Products Liability Act was recently amended in ways that even more clearly preclude a market share theory of liability. The district court's treatment of an intermediate appellate court decision that "impliedly" recognized a market share theory in Ohio as dispositive is wrong, because that decision is merely one datum to be considered by a court sitting in diversity.

Defendant Dart Industries, f/k/a Rexall Drug Company (Rexall), has filed a separate brief, arguing that the district court erred in assuming that all DES products are fungible.

This court reviews the district court's determination of state law de novo. Salve Regina College v. Russell, 499 U.S. 225, 111 S. Ct. 1217, 113 L. Ed. 2d 190 (1991). In Kirk v. Hanes Corp., 16 F.3d 705 (6th Cir.1994), we described the duty of a federal diversity court:

The task of this court, sitting in diversity, is to apply the same law as would be applied by the [Ohio] state courts. See Erie R.R. v. Tompkins, 304 U.S. 64, 78, 58 S. Ct. 817, 822, 82 L. Ed. 1188 (1938). Where a state's highest court has spoken to an issue, we are bound by that decision unless we are convinced that the high court would overrule it if confronted with facts similar to those before us. Bernhardt v. Polygraphic Co. of Am., 350 U.S. 198, 205, 76 S. Ct. 273, 277, 100 L. Ed. 199 (1956). Moreover, where a state appellate court has resolved an issue to which the high court has not spoken, "we will normally treat [those] decisions ... as authoritative absent a strong showing that the state's highest court would decide the issue differently." Garrett v. Akron-Cleveland Auto Rental, Inc. (In re Akron-Cleveland Auto Rental, Inc.), 921 F.2d 659, 662 (6th Cir.1990)(emphasis added).

Id. at 707. More recently we explained that an intermediate appellate decision,

while lacking the controlling force of a decision of a state court of last resort, does serve as "a datum for ascertaining state law which is not to be disregarded by a federal court unless it is convinced by other persuasive data that the highest court of the state would decide otherwise." We may refuse to follow intermediate appellate court decisions where we are persuaded that they fail to reflect state law correctly, but "we should not reject a state rule just because it was not announced by the highest court of the state," even if we believe that the rule is "unsound."

Ziebart Int'l Corp. v. CNA Ins. Cos., 78 F.3d 245, 250 (6th Cir.1996)(internal citations omitted). "Other persuasive data" include the state's supreme court dicta, restatements of law, law review commentaries, and the majority rule among other states. Garden City Osteopathic Hosp. v. HBE Corp., 55 F.3d 1126, 1130 (6th Cir.1995).

"A federal court in a diversity case is not free to engraft onto ... state rules exceptions or modifications which may commend themselves to the federal court, but which have not commended themselves to the State in which the federal court sits." Day & Zimmermann, Inc. v. Challoner, 423 U.S. 3, 4, 96 S. Ct. 167, 168, 46 L. Ed. 2d 3 (1975). See also Bailey v. V&O Press Co., 770 F.2d 601, 605 (6th Cir.1985)(stating that "[i]n our evaluation of the content of state law under Erie, we are not commissioned to take a position regarding the advisability or fairness of the state rule to be applied, but are to determine the issue as would the highest court of the state").

Determining how the Ohio Supreme Court would rule requires us to examine the interplay between the Goldman and Jackson decisions and the Ohio Products Liability Act. We approach the question chronologically. In Goldman the Ohio Supreme Court was asked to decide if the theory of alternative liability, as recognized by that court in Minnich,5 or market share liability theory, as developed by the California Supreme Court in Sindell v. Abbott Labs., 26 Cal. 3d 588, 163 Cal. Rptr. 132, 607 P.2d 924 (1980),6 applied in a suit against asbestos manufacturers. At the outset the Goldman court explained that:

both alternative liability and market-share liability are exceptions to the general rule that a plaintiff has to prove an injury was caused by the negligence of a particular defendant. Both theories are judicially created, and shift the burden of proof to each defendant to show that the plaintiff was not injured by that defendant's negligence.

Goldman, 514 N.E.2d at 693.

Pointing out that "[t]he key point of alternative liability, then is that the plaintiff must still prove that all the defendants acted tortiously", id. at 696,7 the Goldman court rejected the theory in the asbestos context, because the plaintiff was unable to identify any manufacturers who supplied any products to the place where he worked. Id. 514 N.E.2d at 699.

The Goldman court then turned to a discussion of market-share liability, which it described as "fundamentally, a theory for assessing liability ... advanced to provide redress where alternative liability was inapplicable." Id. The court recited the policy reasons supporting such a theory,8 but ultimately rejected the theory in principal part based on lack of fungibility,9 a factor which it felt was not shared in DES cases.10 "This fundamental difference between DES and asbestos--indeed, asbestos tape alone--is enough to undercut the Sindell justification for market-share theory in this case." Id. 514 N.E.2d at 701. Other factors which led to the rejection of a market-share theory of liability were the difficulty of defining the market, id. at 700-01; and the absence of a major portion of the market (there Johns-Mansville) which could defeat the market share prerequisite that a "substantial share" of the manufacturers be before the court. Id. at 701.

Significantly, the Ohio Supreme Court added that "[w]e can conceive of no problem more in need of a legislative solution," 514 N.E.2d at 701, borrowing the philosophy of the Iowa Supreme Court in a case that actually rejected market share liability theory in DES cases:

Although Mulcahy [v. Eli Lilly, 386 N.W.2d 67 (Iowa 1986) ] was a DES case wherein the court rejected market-share liability theory [which according to the Goldman court involved "a situation with fewer complexities than those surrounding asbestos exposure"], the same reasoning is at least as applicable to asbestos:

"We reject the market share liability theory on a broad policy basis. We acknowledge that a plaintiff in a DES case with an unidentified product manufacturer presents an appealing claim for relief. Endeavoring to provide relief, courts have developed theories which in one way or another provided plaintiffs recovery of loss by a kind of court-constructed insurance plan. The result is that manufacturers are required to pay or contribute to payment for injuries which their product may not have caused.

"This may or may not be a desirable result. We believe, however, that awarding damages to an admitted innocent party by means of a court-constructed device that places liability on manufacturers who were not proved to have caused the injury involves social engineering more appropriately within the legislative domain....

" ...

" ... Plaintiffs request that we make a substantial departure from our fundamental negligence requirement of proving causation, without previous warning or guidelines. The imposition of liability upon a manufacturer for harm that it may not have caused is the very legal legerdemain, at least by our long held traditional standards, that we believe the courts should avoid unless prior warnings remain unheeded. It is an act more closely identified as a function assigned to the legislature under its power to enact laws." Id. at 75-76.

Goldman, 514 N.E.2d at 701-02 (quoting Mulcahy v. Eli Lilly & Co., 386 N.W.2d 67, 75-76 (Iowa 1986)). The Goldman court concluded by saying that "[e]ven if we were to recognize market-share liability as a viable theory of recovery, this is not the case in which to do so." 514 N.E.2d at 702.

Although Goldman stopped short of a wholesale rejection of market-share liability in other contexts and presumed in dicta that DES was fungible, at the same time it recognized the difficulties accompanying an alteration to the fundamental principle of tort law. Most importantly, the Goldman court acknowledged that such a task is better suited to the legislature, and in fact invited such a response.

In 1988, one year after Goldman, the Ohio Legislature enacted the Ohio Products Liability Act, Ohio Rev.Code § 2307.71-2307.80 (1988)(the "Act"). The Act codified and replaced pre-existing common law causes of actions for harms caused by an allegedly defective product. See Ohio Rev.Code § 2307.72(A) ("[a]ny recovery of compensatory damages based on a products liability claim is subject to §§ 2307.71-2307.79 of the Revised Code") and § 2307.71(M)(defining "product liability claim" as "a claim that is asserted in a civil action and that seeks to recover compensatory damages from a manufacturer or supplier for death, physical injury to person, emotional distress, or physical damage to property other than the product in question.... "). Cf. § 2307.72(C)(specifically preserving common law only for nonproduct liability claims for economic loss). See also Sikorski v. Link Elec. & Safety Control Co., No. 70187, 1997 WL 15274, at * 6 (Ohio.Ct.App. Jan.16, 1997); Long v. Tokai Bank, No. 15748, 1996 WL 814056 (Ohio.Ct.App. Sept.20, 1996)("legislature's intent in passing the Ohio Product Liability Act was to codify, to the extent possible, the common law of products liability"); LaPuma v. Collinwood Concrete, No. 64882, 1994 WL 86211, at * 2 (Ohio.Ct.App. March 17, 1994)(stating that "[i]n 1988, the state legislature codified products liability causes of action and such actions no longer exist at common law"), rev'd on other grounds, 75 Ohio St. 3d 64, 661 N.E.2d 714, 715-16 (1996)(ruling that claims for economic loss were not to be construed under the Act because § 2307.72(C) of Act explicitly preserved common law remedies for economic loss); McAuliffe v. Western States Import Co., No. 65297, 1993 WL 527880, at * 4 (Ohio.Ct.App. Dec.16, 1993)("State legislature codified product liability causes of action and such actions no longer exist at common law"); rev'd on other grounds, 72 Ohio St. 3d 534, 651 N.E.2d 957 (1995).

The 1988 Act embodies the general common law principle that a plaintiff has to prove an injury proximately caused by a particular defendant.11 The Act states that a manufacturer is subject to liability only if the plaintiff establishes, by a preponderance of the evidence, a product defect and that the product defect proximately caused the claimants' harm. Ohio Rev.Code §§ 2307.73(A)(1)-(2)(emphasis added). "Proximate cause" as used in the Act, is in essence the "substantial factor" test of legal cause stated in the Restatement (Second) of Torts, Sections 431, 433 (1965), and comment a, which focuses on the defendant's conduct in relation to the plaintiff. See Horton v. Harwick Chem. Corp., 73 Ohio St. 3d 679, 653 N.E.2d 1196 (1995)(adopted); Pang v. Minch, 53 Ohio St. 3d 186, 559 N.E.2d 1313 (1990)(discussed with approval); see also Jeffers v. Olexo, 43 Ohio St. 3d 140, 539 N.E.2d 614, 616-17 (1989)(rule of proximate cause "requires that the injury sustained shall be the natural and probable consequence of the negligence alleged").

Further, product defect is defined by reference to statutory provisions that set out how a product may be found defective in manufacture (§ 2307.74); design (§ 2307.75); due to inadequate warning (§ 2307.76); or due to breach of express representation of material fitness (§ 2307.77). These defined elements of "defect" in product liability claims all arise from the acts of the particular manufacturer of the product. Claims of defective manufacture or design, §§ 2307.74 and 2307.75(A)(1), are based on the condition of the product when "it left the control of its manufacturer." §§ 2307.74, 2307.75(A)(1)(emphasis added). Claims based on inadequate warning or instruction depend on the state of knowledge of its manufacturer at the time "it left the control of its manufacturer," § 2307.76(A)(1)(a)(emphasis added), and depend on the specific warnings or instructions that its manufacturer supplied. § 2307.76(A)(1)(b). A claimed defect due to breach of an express warranty depends on the words of the representation and on the condition of the product "when it left the control of its manufacturer." § 2307.77 (emphasis added). See generally, § 2307.71(M) (defining "product liability claim" as claim asserted in a civil action that seeks to recover compensatory damages "from a manufacturer or supplier" arising from the product in question).

Noticeably absent from this scheme are exceptions based on alternative liability as applied in Minnich, or market-share liability as discussed in Goldman. As a matter of statutory construction we must presume that the legislature was aware of both Goldman and Minnich when it drafted the Products Liability Act. Cf. Seeley v. Expert, Inc., 26 Ohio St. 2d 61, 269 N.E.2d 121, 129 (1971)(holding that in interpreting meaning of legislative intent, it is not unimportant that the General Assembly failed to amend legislation subsequent to interpretation by Ohio Supreme Court); Spitzer v. Stillings, 109 Ohio St. 297, 142 N.E. 365, 367 (1924)(noting that "[b]y the rules of construction of statutes, if a statute is amended in certain particulars, after the same has been interpreted and defined by the courts, without change in other respects, it will be presumed that the Legislature was satisfied with the court's interpretation"). See International Union, United Auto., Aerospace & Agric. Implement Workers, 72 F.3d 1243, 1248 (6th Cir.), cert. denied, --- U.S. ----, 117 S. Ct. 63, 136 L. Ed. 2d 24 (1996)("It is a settled principle of statutory construction that when Congress drafts a statute, courts presume that it does so with full knowledge of the existing law.") This silence, coupled with the exclusivity provision of the Act, points to the logical conclusion that the Ohio General Assembly did not wish to create a market share theory of liability. Thus, it is more than a fair guess to say that the Ohio Supreme Court would not adopt a market share theory of liability today.

Recent amendments to the Act reinforce the fundamental theory of liability originally codified in 1988. See Amended Act § 2307.73(A).12 In particular, one amendment provides that a manufacturer is subject to liability under the Act only if the claimant establishes, inter alia, that "the manufacturer designed, formulated, produced, created, made, constructed, assembled, or built the product." Id. (Emphasis added.)

Both sides point to § 2307.791 of the Amended Act as supporting their respective positions. Section 2307.791 provides:

Sec. 2307.791. A manufacturer shall not be held liable for damages based on a product liability claim that asserts any of the following theories:

(A) Industrywide or enterprise liability, including, but not limited to, any claim that seeks to impose liability on an entire industry or enterprise, on any members of an entire industry or enterprise, or on any trade association of an entire industry or enterprise that alleges both of the following:

(1) Joint awareness of product risks;

(2) Joint development of product safety standards or delegation of safety function to an industry trade association or another entity;

(B) Alternative liability, except when all possible tortfeasors are named and subject to the jurisdiction of the court.

Amended Act § 2307.791.

Defendants argue that § 2307.791(A) forbids any claim based on a market share theory, because subsection (A) "includes but is not limited to" industry-wide or enterprise liability. "Enterprise liability" was discussed by the Ohio Court of Appeals in Jackson. Jackson, 647 N.E.2d at 882. As the Jackson court noted, the theory was recognized in Hall v. E.I.DuPont De Nemours & Co., 345 F. Supp. 353 (E.D.N.Y.1972), a suit against manufacturers of blasting caps. The Hall court held that to establish enterprise liability, the plaintiffs would have to show that the defendants were jointly aware of the risks at issue and their joint capacity to reduce or affect those risks. Defendants argue that the words "industrywide" and "including but not limited to" in § 2307.791(A) would be unnecessary if the "enterprise liability" theory of Hall was all that the statute banned.

Plaintiffs read § 2307.791 as addressing only two theories of liability: industrywide or enterprise liability, and alternative liability. Subsection (A) sets forth the elements of enterprise liability, subsection (B) discusses alternative liability, but nowhere are the elements of market share mentioned.

Even if we accept plaintiffs' reading of § 2307.791(A), they still lose, because the critical factor in our analysis is the absence of market-share liability. As discussed, we believe that the Ohio courts would perceive the legislative silence as an implicit, but nonetheless clear, pronouncement that the Ohio legislature does not favor market-share theory of recovery. This is even more true when one considers that the Amended Act now allows for alternative liability.13 Thus, even if defendants' argument that § 2307.791(A) explicitly rules out market share liability as a theory of recovery is incorrect,14 plaintiffs' still lose. In any event, the Products Liability Act is clear: it does not by its express terms provide for market share liability and it is by its express terms exclusive. Thus, the Ohio Supreme Court would be precluded by the Act from adopting a new legal cause of action.

The district court erred in its belief that the Ohio Products Liability Code was no impediment to the Ohio Supreme Court. In stating that the preference for legislative enactment "does not prevent Ohio courts from adopting a new theory of recovery when such action is necessary in the interest of justice," the court looked to Minnich, which adopted alternative liability. The flaw in this reasoning is that Minnich predates the Ohio Products Liability Code. Thus, the Minnich court did not need to consider a legislative pronouncement when it ruled on alternative liability. An Ohio court facing the question presented today would be absolutely required to apply, let alone consider, the Ohio Products Liability Act.

Lastly, the district court's conclusion that Jackson is authoritative absent a state supreme court decision is also wrong. What Kirk and other cases from this circuit actually hold is that we must usually follow the intermediate appellate decision "absent a strong showing" that the highest court would decide otherwise. The enactment of the Ohio Products Liability Code is most certainly persuasive data that would affect the Ohio Supreme Court's determination of whether to follow the Jackson court's position. Further, although Goldman contained dicta that could arguably support plaintiffs' position, it also contained a request by the Ohio Supreme Court to the Ohio Legislature to rule on the issue. The Ohio Legislature provided that response with an integrated statutory scheme that is silent on the issue of market-share liability.

Because the Ohio Legislature's pronouncements provide the best datum on the question presented, we see no reason to look to other sources.

Rexall argues separately that the district court erred in determining that all DES products are fungible, the fundamental premise of market share liability theory. Given our conclusion above, this issue is moot.

In sum, we hold that if directly presented with the issue, the Ohio Supreme Court would not adopt a market-share theory of liability in DES cases. We therefore REVERSE the decision of the district court and REMAND for further proceedings not inconsistent with this opinion.