OPINION OF THE COURT
COWEN, Circuit Judge.The City of Philadelphia (“City”) and the Philadelphia Housing Authority (“PHA”) brought this action against manufacturers of lead pigment and their trade association to recover the costs of abating hazardous lead-based paint which plaintiffs must incur pursuant to newly promulgated federal regulations. Plaintiffs allege that for decades defendants knew their product, lead pigment, caused lead poisoning in children. Nevertheless, defendants marketed lead pigment for use in paint intended for residential buildings and refused to warn consumers of the potential harm.
We hold that the statute of limitations bars the City, but not PHA, from asserting claims for negligence, strict products liability, breach of warranty and fraud. As an agency of the Commonwealth of Pennsylvania, PHA is exempt from the statute of limitations under the doctrine of nullum tempus. Because plaintiffs concede they are unable to link a specific manufacturer to the lead pigment in any particular property, they pro*115pose three theories of collective liability— alternative liability, market share liability and enterprise liability — to establish the causation element of each of their causes of action. We hold that plaintiffs may not proceed under these theories because Pennsylvania law has not adopted any of them in products liability cases or sent an authoritative signal that it would do so. As a federal court in a diversity case, we may not significantly expand state law without a clear indication that the Pennsylvania Supreme Court would do the same. We therefore will affirm the district court’s order dismissing the plaintiffs’ amended complaint.
I. FACTUAL AND PROCEDURAL BACKGROUND
According to the Department of Health and Human Services, lead poisoning is a serious threat to the health of American children. See U.S. Dep’t of Health and Human Services, The Nature and Extent of Lead Poisoning in Children in the United States: A Report to Congress 1 (July 1988). Children may ingest lead by chewing on surfaces covered with lead-based paint or by breathing air that contains dust from crumbling paint. Exposure to lead paint occurs in homes, schools and day care centers. Inner city residents are the highest risk population segment because older city buildings, which predominate in the inner city, are most likely to have unremoved lead paint on the walls.
At low levels, lead poisoning causes IQ reduction, shortened attention span, hyperactivity, aggressive behavior, loss of appetite, vomiting and abdominal pain. Ingestion of lead in higher quantities may cause convulsions, brain damage and eventually death. Currently, fifteen percent of all preschoolers, approximately 3,000,000 children, have elevated lead levels sufficient to impair their neurological development. See U.S. Environmental Protection Agency, Strategy for Reducing Lead Exposures 1 (February 21, 1991). A study estimates that sixty-two percent of all Philadelphia children between the ages of six months and five years have lead levels in their blood capable of causing learning and central nervous disorders. See Environmental Defense Fund, Legacy of Lead: America’s Continuing Epidemic of Childhood Lead Poisoning A-5 (March 1990). The damage caused by lead is irreversible. Therefore, the sole cure for lead poisoning is prevention of exposure to lead. See Centers for Disease Control, Preventing Lead Poisoning in Young Children: A Statement by the Centers for Disease Control 39 (Oct. 1991).
Preventing future injury through lead-based paint abatement is critical, but also enormously expensive. United States Department of Housing and Urban Development (“HUD”) regulations, promulgated pursuant to the Lead-Based Paint Poisoning Prevention Act, 42 U.S.C. §§ 4821-46 (1988 & Supp. II 1990), require the City and PHA to notify all tenants of residential HUD-associated housing constructed prior to 1978 about the dangers of lead poisoning and the appropriate precautions to be taken, see 24 C.F.R. § 35.5(a) (1992), and'to cover or remove lead-based paint from HUD housing units built before 1978, see 24 C.F.R. § 35.-24(b) (1992). HUD provides no funding to aid compliance with the regulations.
To shift their large financial burden to the parties alleged to be primarily responsible for the lead poisoning public health crisis, the City and PHA brought this lawsuit.1 Plaintiffs seek damages from NL Industries, Inc., Atlantic Richfield Company, The Sherwin-Williams Company, The Glidden Company *116and Fuller-O’Brien Corporation, all of which are lead pigment manufacturers, and the Lead Industries Association (“LIA”).2 Plaintiffs seek compensatory damages in excess of $100,000,000 for the costs of (1) inspecting HUD and privately owned housing; (2) removing lead paint from public and private residential properties built or painted prior to 1950; (3) testing individuals to detect elevated lead blood levels; (4) treating city residents for exposure to lead paint; (5) educating the public about the hazards of lead paint; and (6) recovering liability imposed on plaintiffs in their capacity as property owners for personal injury arising from the ingestion of lead paint. Plaintiffs also seek punitive damages and injunctive relief to require the defendants to eliminate the hazards caused by lead paint on the walls of City properties.
Plaintiffs’ amended complaint asserts claims for negligent product design, strict product liability, negligent failure to warn, breach of warranty, fraud and misrepresentation, indemnification, restitution and punitive damages. Because the procedural context of this appeal is a motion to dismiss, we assume that plaintiffs’ allegations are true. The amended complaint alleges the following. Defendants knew of the severe hazards of lead paint since the early 1900s, long before this information was widely circulated to the public. Defendants also were aware that non-toxic pigments, such as zinc-oxide pigment, were available as substitutes for lead pigments in paint. Despite this knowledge, defendants continued to promote their product for use in paint intended for residential interior surfaces and refused to warn potential consumers of the known hazards.
The LIA, organized by some of the defendants in 1928, collected literature addressing the toxicity of lead pigment and attempted to discredit studies documenting the hazards of lead paint. The LIA also funded its own medical research to refute scientific evidence concerning the dangers of lead. Through expensive advertising campaigns, the LIA misrepresented the safety of lead paint to the consumer public. Through vigorous lobbying, it also misled legislative bodies considering regulating or banning lead paint. All of the defendants were members of the LIA at some point, though at different times and for different lengths of time.
Plaintiffs concede they have been aware for years that lead paint is dangerous to children. In the 1940s and 1950s, plaintiffs “began to be aware of the hazards of lead paint to young children.... ” Amended Complaint ¶ 73, App. at 290-91. Philadelphia made lead poisoning a reportable disease in 1950, and throughout the 1950s and 1960s, it operated programs to determine the extent to which lead paint caused lead poisoning in children. In 1966, Philadelphia enacted an ordinance that prohibited the use of lead paint on residential walls, regulated the labeling of lead paint, and required property owners to remove lead paint from buildings posing a health hazard to children. See Philadelphia, Pa., Code § 6 — 403(2)(a), (3), (4)(b) (1966). Following the LIA’s suggested standard, the City defined lead paint as paint containing more than one percent lead. See Philadelphia Dep’t of Public Health, Regulations Relating to Labeling, Application and Removal of Lead Paint § 6^403(l)(d) (June 27, 1966).3 To conform city buildings to the standards set forth in the 1966 ordinance, the City has paid to inspect and, when appropriate, abate lead paint hazards in public and some private buildings.
In 1976, Congress amended the Lead-Based Paint Poisoning Prevention Act to define hazardous lead paint as paint containing more than .06 percent lead. See Pub.L. No. 94-317, § 204(c)(1), 90 Stat. 706 (1976). This is the current definition of lead paint under federal law. See 42 U.S.C. § 4841(3)(B)(ii). In 1977, the Consumer Products Safety Commission enacted regulations banning the sale *117in interstate commerce of paint for residential use with more than .06 percent lead. See 16 C.F.R. § 1303 (1992). The new HUD regulations, 24 C.F.R. §§ 35.1-35.70 (1992) incorporate the .06 percent standard and therefore require significantly more extensive abatement actions than Philadelphia’s 1966 ordinance.
Defendants filed a motion to dismiss the amended complaint pursuant to Federal Rules of Civil Procedure 12(b)(6) and 9(b). The district court held that the statute of limitations barred the City, but not PHA, from asserting claims for negligent product design, strict product liability, negligent failure to warn, breach of warranty, and fraud and misrepresentation. See City of Philadelphia v. Lead Indus. Ass’n, No. 90-7064, 1992 WL 23450 (E.D.Pa. Feb. 4, 1992).4 The district court noted that the doctrine of nullum tempus grants the Commonwealth and its agencies immunity from the statute of limitations absent an explicit legislative directive to the contrary. Finding that PHA is a Commonwealth agency, the court held that nul-lum tempus exempts PHA from the statute of limitations. The district court also noted that the City, a political subdivision of the Commonwealth, may invoke nullum tempus only when enforcing a duty imposed by law. Finding that defendants’ potential liability to the City in tort arose from the defendants’ voluntary decision to sell lead pigment to paint manufacturers, the court held that the City was subject to the statute of limitations.
The City’s claims accrued, for purposes of statutes of limitations analysis, when it knew or should have known of its right to file a lawsuit. The district court concluded that the City’s claims accrued no later than 1976, when Congress expanded the definition of hazardous lead paint to include paint with a lead content of .06 percent. At that time, plaintiffs were put on notice that the prior application of paint containing that amount of lead had inflicted injury. The district court therefore held that the City’s claims were time-barred under all applicable statutes of limitations.
In a later opinion, the district court dismissed the entire amended complaint with prejudice. See City of Philadelphia v. Lead Indus. Ass’n, No. 90-7064, 1992 WL 98482 (E.D.Pa. April 23, 1992). The design defect claims were dismissed because although the end product, lead paint, could be made safer, lead pigment could not. The district court dismissed plaintiffs’ breach of warranty claims because the defendants did not sell the dangerous end product, lead paint. The fraud and misrepresentation claims were dismissed on the merits for failure to adequately plead justifiable reliance. Specifically, plaintiffs did not aver that they were unaware of the medical information in the public domain concerning the hazards of lead paint at the time they purchased lead paint. The claim for indemnification was dismissed because plaintiffs failed to allege that a judgment of liability was entered pursuant to which it paid an award of damages.
Finally, the district court dismissed the remaining restitution and failure to warn claims for insufficient pleading of proximate causation. Plaintiffs conceded that they could not prove that the lead pigment of a particular manufacturer was applied to the walls of a given housing unit. They thus advanced five theories of collective liability to hold defendants jointly and severally liable: (1) alternative liability, (2) market share liability, (3) enterprise liability, (4) concert of action, and (5) conspiracy. The district court rejected each collective liability theory suggested by plaintiffs.5 Characterizing the collective liability theories as radical and noting that the Pennsylvania Supreme Court has not embraced any of them in the context of toxic torts, the district court concluded that a federal court sitting in diversity was not the appropriate tribunal to drastically expand state tort law.
*118The City and PHA filed a motion for reconsideration, which was denied. They appeal the February 4, 1992 and the April 23, 1992 decisions of the district court as well as the denial of their motion for a rehearing. The defendants filed a cross-appeal challenging the district court’s ruling that PHA’s claims were not time-barred, and the court’s failure to dismiss the restitution claim on grounds other than causation.
II. DISCUSSION
The district court had jurisdiction pursuant to 28 U.S.C. § 1332 (1988) and this court has appellate jurisdiction pursuant to 28 U.S.C. § 1291 (1988). We exercise plenary review over an order dismissing a complaint for failure to state a claim. Marshall-Silver Constr. Co. v. Mendel, 894 F.2d 593, 595 (3d Cir.1990). In the procedural context of a motion to dismiss, we accept the factual allegations contained in the amended complaint as true and plaintiff receives the benefit of all reasonable inferences to be drawn therefrom. See Angelastro v. Prudential-Bache Sec., Inc., 764 F.2d 939, 944 (3d Cir.), cert. denied, 474 U.S. 935, 106 S.Ct. 267, 88 L.Ed.2d 274 (1985). We may not affirm the dismissal of the complaint unless plaintiffs can prove no set of facts that would entitle them to relief. Conley v. Gibson, 355 U.S. 41, 45-46, 78 S.Ct. 99, 102, 2 L.Ed.2d 80 (1957).
A. STATUTE OF LIMITATIONS
In its opinion and order dated February 4, 1992, the district court held that PHA is an agency of the Commonwealth of Pennsylvania and therefore immune from a statute of limitations defense under the doctrine of nul-lum tempus occurit regí (“time does not run against the king”). The court, however, held that the City is not entitled to nullum tem-pus immunity, and that its claims in Counts I to V of the complaint all accrued by 1976. The court therefore dismissed those claims as time-barred because the applicable limitations periods had expired before this action was filed in 1990. The City appeals the dismissal of its claims on timeliness grounds. Defendants, in turn, cross-appeal the district court’s refusal to dismiss PHA’s claims as time-barred.
I. Nullum Tempus
a. Plaintiff PHA
Under the doctrine of nullum tempus, statutes of limitations are not applicable to actions brought by the Commonwealth or its agencies unless a statute expressly so provides. Commonwealth, Dep’t of Transp. v. Rockland Constr. Co., 498 Pa. 531, 448 A.2d 1047, 1047 (1982); Commonwealth, Dep’t of Transp. v. J.W. Bishop & Co., 497 Pa. 58, 439 A.2d 101, 101 (1981). The rationale of this rule is that the Commonwealth, as a plaintiff, seeks the vindication of public rights and the protection of public property. J.W. Bishop & Co., 439 A.2d at 104. None of the statutes of limitations applicable in this case expressly apply to the Commonwealth. Therefore, if PHA is an agency of the Commonwealth, none of its claims is time-barred.
PHA was created under the Housing Authorities Law, which provides, in pertinent part: “An Authority shall constitute a public body, corporate and politic, exercising public powers of the Commonwealth as an agency thereof-” Pa.Stat.Ann. tit. 35, § 1550 (1977) (emphasis added). By its clear language, this provision of the statute designates PHA an agency of the Commonwealth.
Defendants, however, rely on T & R Painting Co., Inc. v. Philadelphia Housing Authority, 466 Pa. 493, 353 A.2d 800, 800 (1976), which held that PHA is not an agency of the Commonwealth and therefore not within the Commonwealth Court’s jurisdiction. The court rejected the argument that the “exercising public powers of the Commonwealth as an agency thereof’ language rendered PHA an agency of the Commonwealth because at least eight other provisions of the Housing Authorities Law indicated that PHA is not a Commonwealth agency, but rather a local agency operating within a limited area. 353 A.2d at 801. Concluding that the language of the Housing Authorities Law did not specifically resolve the issue of whether PHA was a local or Commonwealth agency, the T & R Painting Co. court reached the issue of legislative intent, albeit *119in the context of determining whether PHA was a local or a Commonwealth agency for purposes of the Appellate Court Jurisdiction Act. Id. 353 A.2d at 801-02.
PHA argues that T & R Painting Co. was implicitly overruled by Marshall v. Port Authority of Allegheny County, 524 Pa. 1, 568 A.2d 931 (1990). In that case, the defendant Port Authority (“PAT”) claimed that it was an agency of the Commonwealth and therefore immune from tort liability under the doctrine of sovereign immunity. The court cited one provision of the enabling statute which created the Authority: “There are hereby created bodies corporate and politic in counties of the second class, to be known as Port Authority of (insert name of county), which shall constitute public bodies corporate and politic; exercising the public powers of the Commonwealth as an agency thereof.” Id. 568 A.2d at 933-34 (quoting Pa.Stat.Ann. tit. 55, § 553(a)) (emphasis added by Marshall court). The court concluded that “[i]n view of this plain statutory language, it would be impossible to conclude that PAT is anything other than an agency of the Commonwealth.” Id. 568 A.2d at 934. The court refused to consider evidence of contrary legislative intent, reasoning that “[t]he best indi-cium of legislative intent regarding the scope of sovereign immunity is the plain language employed in the statute, and we are constrained to give effect to that language.” Id. Notably, the court did not refer to, much less discuss, any other provisions of the statute, including those which, as in T & R Painting Co., cut the other way.
We agree with the district court that Marshall is controlling. The enabling statutes considered in Marshall and T & R Painting Co. contained virtually identical language. In T & R Painting Co., the court went beyond that language and considered the statute as a whole, found it ambiguous, and therefore examined other indicia of legislative intent. See 353 A.2d at 801-02. This approach was rejected in Marshall, where the court relied exclusively on the plain language of one provision without considering other statutory provisions. See 568 A.2d at 933-34. Thus, if there is language in an enabling statute which clearly states that a party is a Commonwealth agency, that statutory language is controlling.
Relying on Marshall, intermediate appellate courts recently have held that housing authorities are agencies of the Commonwealth for purposes of sovereign immunity. See Battle v. Philadelphia Housing Auth., 406 Pa.Super. 578, 594 A.2d 769, 771 (1991) (PHA is a Commonwealth agency); Crosby v. Kotch, 135 Pa.Cmwlth. 470, 580 A.2d 1191, 1193 (1990) (Luzerne County Housing Authority is a Commonwealth agency). Though these cases involved sovereign immunity rather than nullum tempus, the legislature’s designation of which entities are Commonwealth parties is dispositive for all governmental privileges. Smith v. Mognet, 618 A.2d 1215, 1217 (Pa.Commw.Ct.1992); Northampton County Area Community College v. Dow Chem, U.S.A., 389 Pa.Super. 11, 566 A.2d 591, 597-98 (1989), aff'd, 528 Pa. 502, 598 A.2d 1288 (1991). Because Pa.Stat. Ann. tit. 35, § 1550 unambiguously states that a housing authority “exercis[es] public powers of the Commonwealth as an agency thereof,” and because, as a federal court sitting in diversity, we are bound to follow the pronouncement of a state’s highest court on an issue of state law, we hold that PHA is an agency of the Commonwealth which can invoke the doctrine of nullum tempus. None of PHA’s claims is time-barred.
b. Plaintiff City of Philadelphia
The City is a political subdivision of the Commonwealth. Absent an express statutory provision, the doctrine of nulhim tem-pus is unavailable to political subdivisions except in very limited circumstances:
statutes of limitations cannot be pleaded' against such political subdivisions when they are seeking to enforce strictly public rights, that is, when the cause of action accrues to them in their governmental capacity and the suit is brought to enforce an obligation imposed by law as distinguished from-one arising out of an agreement voluntarily entered into by the defendant.
City of Philadelphia v. Holmes Elec. Protective Co., 335 Pa. 273, 6 A.2d 884, 887 (1939). The district court read this language as establishing a two-part test for the availability *120of nullum tempus. The plaintiffs claims must (1) accrue to it in its governmental capacity and (2) seek enforcement of an obligation imposed on the defendant by law rather than a voluntary agreement. The district court assumed arguendo that the first prong was satisfied because the City was suing to remove lead-based paint from HUD-associated housing units, as required by the Lead-Based Paint Poisoning Prevention Act. The court concluded, however, that the City could not invoke nullum tempus because its claims arose out of agreements voluntarily entered into by the defendants. Though Pennsylvania case law provides no clear answer to the question before us, we believe that the Supreme Court of Pennsylvania would have reached the same result as the district court.
It is clear that nullum tempus applies when a government entity sues to enforce a statutory obligation imposed upon a defendant, and the claim is one which by its nature can accrue only to the government. For example, in In re Envy’s Estate, 337 Pa. 542, 12 A.2d 333, 335 (1940), milium tempus applied to the City of Philadelphia’s suit against the parents of an insane man for reimbursement of amounts spent to maintain him in a public institution, where the parents’ obligation to pay was imposed by a statute. In City of Philadelphia v. Litvin, 211 Pa.Super. 204, 235 A.2d 157, 158 (1967), the court held that the City was not subject to a statute of limitations in an action to collect taxes. In Pocono Township v. Hall, 127 Pa.Cmwlth. 116, 561 A.2d 53 (1989), the township enacted an ordinance which required developers to provide a bond covering the cost of paving roads in the areas they develop. The township’s action against a developer to enforce the “road bond” was not subject to a statute of limitations. Id. 561 A.2d at 56.
In contrast, nullum tempus clearly does not apply to claims arising out a contract voluntarily entered into by both a government entity and the defendant. In Holmes, an ordinance authorized, but did not require, the City of Philadelphia to permit the defendant power company to lay wires underground in exchange for a share of the profits. The City’s action to collect amounts due was held to be time-barred. Id. 6 A.2d at 887-88. The court reasoned that the City was enforcing a contractual obligation in the nature of an annual rental, rather than “a tax or license fee,” which is the kind of claim subject to nullum tempus. Id. 6 A.2d at 887. In Pennsylvania Turnpike Commission v. Atlantic Richfield Co., 482 Pa. 615, 394 A.2d 491, 494 (1978), the doctrine of nullum tem-pus did not apply to the Commission’s claim for unpaid rent arising from a lease voluntarily entered into by the parties.
In Borough of West Fairview v. Hess, 130 Pa.Cmwlth. 385, 568 A.2d 709 (1989), a borough sued its former elected auditors for negligently failing to discover deficiencies in its accounts. The court held that the borough’s tort claim is subject to a statute of limitations, as “the borough is not acting in a role which is exclusively governmental, such as tax collector, but rather is seeking a judgment against allegedly negligent public servants, just as any private litigant having standing could do.” Id. 568 A.2d at 713.
The case before us more closely resembles this second line of cases (Holmes, Atlantic Richfield and Hess), which held that nullum tempus does not apply to common law contract and tort claims arising out of voluntary agreements. The defendants voluntarily entered into contracts to sell lead pigment to manufacturers of lead-based paint, and the City voluntarily contracted to purchase the paint. The City’s claims all arise out of these voluntary agreements. The claims are for negligent product design, negligent failure to warn, strict products liability, breach of warranty, and fraud. These are common law tort and contract claims which could be filed against the defendants by any private litigant. These claims do not, by their nature, accrue only to government entities, and the defendants do not have any obligation imposed upon them by statute. Thus, the City is not suing in a “governmental capacity ... to enforce an obligation imposed by law,” Holmes, 6 A.2d at 887, and cannot invoke nullum tempus.
The City relies heavily on recent cases which held that school districts are not subject to statutes of limitations when suing to recover damages to school buildings caused by asbestos products, Mt. Lebanon School *121Dist. v. W.R. Grace & Co., 414 Pa.Super. 455, 607 A.2d 756, 762 (1992), and by inadequate design and construction, Stroudsburg Area School Dist. v. R.K.R. Assoc./Architects, 417 Pa.Super. 85, 611 A.2d 1276, 1279-80 (1992). These cases are distinguishable. In Mt. Lebanon, the plaintiff school district, as an agency of the legislature, was compelled by the Constitution of Pennsylvania and by statute to construct and maintain school buildings. 607 A.2d at 760-62. To fulfill this duty, the school district must contract with private parties who can provide suitable facilities. Therefore, any claim arising from those contractual relations accrues to the plaintiff in a governmental capacity, and the suit is brought to enforce public rights and an obligation imposed by law. Nullum tempus applies to such claims. Id. 607 A.2d at 762. This reasoning was followed in Stroudsburg, 611 A.2d at 1279-80.
In this case, the City has not alleged that it was required by law, as an agency of the legislature, to contract for the 'purchase of lead-based paint or to construct the buildings in which that paint was used. The district court properly held that the City’s claims in Counts I to V are subject to statutes of limitations.
2. Accrual of Claims
The City argues that even if it cannot invoke nullum tempus, its claims are timely because they did not accrue until April of 1990, when federal regulations were enacted which required the City to remove lead-based paint from federally funded housing.6 The district court rejected this argument, and so do we.
Limitations periods are computed from the time the cause of action accrued. 42 Pa.Cons.Stat.Ann. § 5502(a) (1981). Under Pennsylvania law, a cause of action accrues at “the time when the plaintiff could have first maintained the action to a successful conclusion.” Kapil v. Association of Pa. State College and Univ. Faculties, 504 Pa. 92, 470 A.2d 482, 485 (1983); see Foley v. Pittsburgh-Des Moines Co., 363 Pa. 1, 68 A.2d 517, 535 (1949). However, under Pennsylvania’s “discovery rule,” a statute of limitations does not run while the plaintiff, despite the exercise of due diligence, is unaware of his injury. Pocono Int’l Raceway, Inc. v. Pocono Produce, Inc., 503 Pa. 80, 468 A.2d 468, 471 (1983); Schaffer v. Larzelere, 410 Pa. 402, 189 A.2d 267, 270 (1963).
The City argues that it was injured in 1990 when federal law required it to abate lead-based paint. Though the cost of removing the lead possibly could be the measure of damages, see Miller v. C.P. Centers, Inc., 334 Pa.Super. 623, 483 A.2d 912, 915 (1984), it is not the injury. The City and its citizens-were injured when the buildings were rendered unsafe by the use of lead-based paint, not when the City, years later, was forced by federal regulations to take remedial action. The City’s claims in Counts I to V — negligent product design, negligent failure to warn, strict products liability, breach of warranty, and fraud — are all state tort and contract claims which do not depend upon any provision of federal law for their success. The City could have filed the tort claims7 when the paint was applied to its buildings, and those claims therefore normally would have accrued at that time. However, under the “discovery rule,” they did not accrue until the City reasonably could have known that it was injured by the lead-based paint.
The district court held that all of the City’s claims accrued by 1976. In that year, Congress amended the Lead-Based Paint Poisoning Prevention Act to require the Secretary of HUD to develop procedures to eliminate paint with a lead content of over .06 percent from HUD-associated housing constructed or substantially rehabilitated before *1221978. See 42 U.S.C. §§ 4822(a), 4841(3)(B)(ii); see also Pub.L. No. 94-317, § 204(c)(1), 90 Stat. 706 (1976) (amending definition of lead-based paint). We agree with the district court that the 1976 amendment to section 4841(3)(B)(ii) put the City on notice that .06 percent lead paint was a health hazard. At that time, the City reasonably should have known that it was injured by the use of such paint in its buildings. The tort claims therefore accrued by 1976.8
In 1976, the statute of limitations applicable to negligent product design, negligent failure to warn, and strict products liability was six years. See Pa.Stat.Ann. tit. 12, § 31 (repealed 1978). In 1978, it was shortened to two years, 42 Pa.Cons.Stat.Ann. § 5524 (1981), though claims which had not yet expired could run for one more year or until the end of the original six-year period, whichever was less, see Act of July 9, 1976, P.L. 586, No. 142, § 25(a). The applicable statute of limitations therefore expired in 1979.
The statute of limitations for breach of warranty is four years, measured from when tender of delivery is made. 13 Pa.Cons.Stat. Ann. § 2725(a)-(b) (1984). In 1977, the City amended Philadelphia Code § 6-403(l)(a) to prohibit the use and sale of paint containing lead in excess of the amount allowed by federal law (.06 percent) on the surface of any dwelling. Because delivery of the paint could not lawfully have been tendered to the City after 1977, the statute of limitations ' expired no later than 1981.
In 1976, the statute of limitations for fraud and misrepresentation was six years. See Pa.Stat.Ann. tit. 12, § 31 (repealed 1978); Home Life Ins. Co. of America v. Greenspan, 360 Pa. 542, 63 A.2d 72, 75 (1949). The current period is the same. 42 Pa.Cons.Stat. Ann. § 5527 (Supp.1992). The statute of limitations for fraud and misrepresentation therefore expired by 1982.
The City filed its complaint in 1990. By 1982, all of the applicable limitations periods had run. The City’s claims in Counts I to V therefore are untimely.
B. PROXIMATE CAUSATION
Because the City’s restitution and indemnification claims and all of PHA’s claims are not barred by the statute of limitations, we must address whether plaintiffs may utilize a collective liability theory to establish the requisite causation for their claims.9 Plaintiffs concede that they cannot prove which lead pigment was applied to any specific property that they own, manage, or will be required to remediate. None of their claims can survive a motion to dismiss, therefore, unless plaintiffs assert a viable theory of nonidentification liability. On appeal, plaintiffs propose three nonidentification theories of liability: (1) market share liability, (2) alternative liability, and (3) enterprise liability.
Before evaluating plaintiffs’ collective liability theories, we first must consider the constraints that shape our interpretation of Pennsylvania law. As a basic premise, federal courts presiding over diversity cases must apply the substantive law of the state whose laws govern the action. Klaxon Co. v. Stentor Elec. Mfg. Co., 313 U.S. 487, 496-97, 61 S.Ct. 1020, 1021-22, 85 L.Ed. 1477 (1941); Erie R.R. Co. v. Tompkins, 304 U.S. 64, 78, 58 S.Ct. 817, 822, 82 L.Ed. 1188 (1938). *123When the state’s highest court has not addressed the precise question presented, a federal court must predict how the state’s highest court would resolve the issue, Borman v. Raymark Indus., Inc., 960 F.2d 327, 331 (3d Cir.1992); Pennsylvania Glass Sand Corp. v. Caterpillar Tractor Co., 652 F.2d 1165, 1167 (3d Cir.1981), and is not free to shape state common law as it sees fit, Adams v. Madison Realty & Dev., Inc., 853 F.2d 163, 168 (3d Cir.1988). “A federal court in a diversity case is not free to engraft onto those 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).
A federal court may act as a judicial pioneer when interpreting the United States Constitution and federal law. In a diversity case, however, federal courts may not engage in judicial activism. Federalism concerns require that we permit state courts to decide whether and to what extent they will expand state common law. See Wisniewski v. Johns-Manville Corp., 759 F.2d 271, 274 (3d Cir.1985) (“We leave to ... the state legislatures and, where relevant, to the state courts the task of expanding or restricting liability for asbestos production.”); Bruffett v. Warner Communications, Inc., 692 F.2d 910, 920 (3d Cir.1982). Our role is to apply the current law of the appropriate jurisdiction, and leave it undisturbed. As the Court of Appeals of the District of Columbia Circuit stated, when it declined to permit a plaintiff to utilize market share liability:
Absent some authoritative signal from the legislature or the [state courts], we see no basis for even considering' the pros and cons of innovative theories.... We must apply the law of the forum as we infer it presently to be, not as it might come to be.
Tidler v. Eli Lilly & Co., 851 F.2d 418, 424 (D.C.Cir.1988) (ellipses in original) (quoting Dayton v. Peck, Stow & Wilcox Co., 739 F.2d 690, 694-95 (1st Cir.1984)).
An authoritative signal that a state’s highest court would modify existing state law may be gleaned from lower state court decisions, the decisions of other courts, and treatises or other scholarly works. See Pennsylvania Glass Sand, 652 F.2d at 1167. Although not dispositive, decisions of state intermediate appellate courts should be accorded significant weight in the absence of an indication that the highest state court would rule otherwise. See Rolick v. Collins Pine Co., 925 F.2d 661, 664 (3d Cir.1991); Wisniewski, 759 F.2d at 273-74. Cognizant of our limited role in diversity cases, we now survey Pennsylvania law with respect to each liability theory advocated by plaintiffs.
1. Market Share Liability
Under Pennsylvania law, a plaintiff usually must prove that the defendant was the proximate cause of plaintiffs injury. See Cuthbert v. City of Philadelphia, 417 Pa. 610, 209 A.2d 261, 263 (1965). In Cuthbert, the Pennsylvania Supreme Court reiterated its longstanding causation requirement:
[I]t remains a principle so fundamental as to require no authority that the mere existence of negligence and the occurrence of injury are insufficient to impose liability upon anyone. There remains to be proved the vitally important link of causation. And plaintiff has the burden of proving this link, that the defendant’s negligence was the proximate cause of her injury.
Id. (citations omitted). Proof of causation is equally necessary in products liability actions. A plaintiff must establish that a particular product of a defendant manufacturer caused her injuries. See Lilley v. Johns-Manville Corp., 408 Pa.Super. 83, 596 A.2d 203, 207 (1991), allocatur denied, 607 A.2d 254 (Pa.1992); Eckenrod v. GAF Corp., 375 Pa.Super. 187, 544 A.2d 50, 52, allocatur denied, 520 Pa. 605, 553 A.2d 968 (1988).
The highest courts of several states have qualified this generally applicable rule with an exception, entitled market share liability.10 Under the theory of market share *124liability, tortious manufacturers who produce a fungible and unidentifiable product that injures the plaintiff are held liable in proportion to their respective market shares. The pioneering California Supreme Court originally developed this theory. See Sindell v. Abbott Lab., 26 Cal.3d 588, 163 Cal.Rptr. 132, 607 P.2d 924, cert. denied, 449 U.S. 912, 101 S.Ct. 285, 66 L.Ed.2d 140 (1980). In Sindell, the plaintiff alleged that she contracted cancer as a result of her mother’s ingestion of diethylstilbestrol (“DES”) during her pregnancy. DES is a drug that many pharmaceutical companies manufactured from an identical formula. Due to the fungibility of the various brands of the drug and the long interval between its sale and the manifestation of cancer, the plaintiff was unable to identify, through no fault of her own, the company that produced the DES her mother used.
Acknowledging its departure from traditional tort law, the California Supreme Court fashioned a new theory of liability. The court reasoned that “as between an innocent plaintiff and negligent defendants, the latter should bear the cost of the injury.” Id., 607 P.2d at 936. Imposing liability on the manufacturers had the further benefit of providing-incentives for companies to produce safer products and affix clear warnings of potential harmful effects. Id. The court held that when a substantial percentage of DES manufacturers are named as defendants, each may be held liable for the percentage of damages represented by its market share, unless it demonstrates that it could not have produced the DES that caused plaintiffs injuries. Id. at 936-37. A defendant could exculpate itself, for example, by showing that they did not manufacturer DES during the relevant time or in the relevant geographic area.
The Sindell Court refused to hold manufacturers jointly and severally liable when not all potential tortfeasors were joined, because of the risk that the offending company would escape liability while scapegoats would owe damages in excess of any harm their products caused. Id.' at 931. The court found it reasonable, however, to use a defendant’s market share to estimate the likelihood that it supplied the defective product. If a defendant is liable whenever the generic product it manufactures causes harm, but only for a percentage of the damages based on the likelihood that it produced the injury-causing product, the total damages it pays in all lawsuits should equal the amount of harm its product caused. Id. at 937 & n. 28.
The Supreme Court of Pennsylvania has never embraced, rejected, or even discussed market share liability. The only Pennsylvania state court that has allowed a plaintiff to utilize market share liability instead of identifying the manufacturer who produced the defective, injury-causing product was a trial court. See Erlich v. Abbott Lab., 5 Phila. 249 (C.P.1981).11 Erlich also was a DES case. There, the Court of Common Pleas found that market share liability is appropriate when: (1) all defendants are tortfeasors; (2) the allegedly harmful products are identical and share the same defective qualities; (3) plaintiff is unable to identify which defendant caused her injury through no fault of her own; and (4) the manufacturers of substantially all of the defective products in the relevant area and during the relevant time are named as defendants. Id. at 264-67. These are the same factors applied by the Sindell court. Finding that plaintiff met all four criteria, the Erlich court allowed her to use market share liability. Id.
*125Neither the Superior Court nor the Commonwealth Court, Pennsylvania’s intermediate appellate courts, has permitted a plaintiff to proceed using the theory of market share liability. The Superior Court, however, has discussed market share liability in several decisions. Plaintiffs contend that dicta in these opinions provides a clear signal that the Pennsylvania Supreme Court would accept market share liability. We disagree.
In Burnside v. Abbott Laboratories, 351 Pa.Super. 264, 505 A.2d 973, 986 (1985), another DES case, the Pennsylvania Superior Court decided that it was “unnecessary to consider the merits of the market share alternative liability concept,” because every defendant had proved that it did not supply the plaintiffs mother with DES. Although it described the version of market share liability adopted by the Supreme Court of Washington, the court did not endorse this theory. Indeed, it noted that “[tjhis theory has not been adopted by the courts of Pennsylvania.” Id., 505 A.2d at 985.
The Pennsylvania Superior Court again encountered market share liability in Cummins v. Firestone Tire & Rubber Co., 344 Pa.Super. 9, 495 A.2d 963 (1985). In Cum-mins, the plaintiff was injured when a multi-piece rim and tire assembly exploded. Because the multi-piece assembly involved in the accident was irretrievably lost, plaintiff was unable to identify its manufacturer and sued several companies that produced multi-piece rims, tires, or both. After outlining the Sindell model of market share liability, the court found the theory clearly inapposite because the products were not fungible and a substantial percentage of the manufacturers were not joined as defendants. Id., 495 A.2d at 971-72. Characterizing the concept as “novel to the courts of this Commonwealth,” the Cummins court did not state whether market share liability should ever be embraced. Id. at 972.
The most recent decision in which the Pennsylvania Superior Court discussed market share liability is Pennfield Corporation v. Meadow Valley Electric, Inc., 413 Pa.Super. 187, 604 A.2d 1082 (1992). In Pennfield, the plaintiff corporation owned swine that suffocated when a defective electrical cable failed, causing the ventilation system to shut down. The plaintiff sued the only two distributors from which it had purchased cables and argued that Pennsylvania had recognized market share liability, as articulated in Sindell and Erlich. Unlike DES, not all cables are defective. Therefore, one defendant most likely was an innocent party who had committed no negligence. Because tortious conduct by all defendants is the sine qua non of market share liability, the court easily refused to shift the burden of disproving causation to a potentially innocent party. Id., 604 A.2d at 1087. The court again did not reach whether market share liability was a viable theory under Pennsylvania law.
After canvassing all relevant Pennsylvania law, we have found only a decade-old, trial court opinion that has permitted a plaintiff to proceed under a theory of market share liability. This lone decision is insufficient to establish market share liability as a recognized exception to the proximate cause requirement under Pennsylvania law. Nor has there been a clear authoritative signal that the Pennsylvania Supreme Court would adopt market share liability.12 No nationwide consensus that-market share liability should be embraced exists, as evidenced by the split of authority among the highest state courts that have addressed this issue.13 Fur*126ther, the few courts to evaluate the applicability of market share liability to the lead pigment industry have found it inappropriate. See Swartzbauer v. Lead Indus. Ass’n, Inc., 794 F.Supp. 142, 146 (E.D.Pa.1992); Santiago v. Sherwin-Williams Co., 782 F.Supp. 186, 191-95 (D.Mass.1992).
On a broader scale, we discern no current trend in Pennsylvania to expand the parameters of tort law. The Pennsylvania Supreme Court recently declined to allow recovery for emotional distress to persons outside the zone of danger of an accident. See Mazzagatti v. Everingham, 512 Pa. 266, 516 A.2d 672, 679 (1986). In a concurring opinion, Justice Flaherty warned of the dangers of expanding tort recovery:
As it is with everything, a balance must be struck — certain limits drawn. We are, in the end, dealing with money, and that money must come from somewhere — from someone: the public pays for the very most part by increased insurance premiums, taxation, prices paid for consumer goods, medical services, and in loss of jobs when the manufacturing industry is too adversely affected. A sound and viable tort system — generally what we now have — is a valuable incident of our free society, but we must protect it from excess lest it becomes unworkable and alas, we find it replaced with something far less desirable.
Id., 516 A.2d at 680 (Flaherty, J., concurring) (emphasis in original). This passage is often quoted by Pennsylvania appellate courts refusing to expand tort law. See, e.g., White v. Weiner, 386 Pa.Super. 111, 562 A.2d 378, 385 (1989) (no duty to warn for entities in distribution chain other than final drug manufacturer), aff'd, 525 Pa. 572, 583 A.2d 789 (1991); Steiner v. Bell Tel. Co. of Pa., 358 Pa.Super. 505, 517 A.2d 1348, 1357 (1986) (no cause of action for loss of parental consortium), aff'd, 518 Pa. 57, 540 A.2d 266 (1988).
The Pennsylvania Supreme Court one day may adopt market share liability. There are compelling policy arguments pro and con. Market share liability furthers a primary objective of tort law, which is to compensate innocent, injured victims. The cost of injury from a defective product, while overwhelming to an individual, may be insured by manufacturers and distributed among the consuming public as a cost of doing business. Market share liability also provides incentives for manufacturers to produce safer goods and to warn of potential dangers.
Compensation, however, is not the only goal of a tort system. Market share liability compromises fairness . to defendants who must incur oftentimes staggering litigation costs as they are forced to defend all claims involving their product irrespective of their market share. Some courts and commentators have predicted market share liability will inhibit cutting-edge research, especially by pharmaceutical companies. See Zafft v. Eli Lilly & Co., 676 S.W.2d 241, 247 (Mo.1984); K. Miller & J. Hancock, Perspectives On Market Share Liability: Time For a Reassessment?, 88 W.Va.L.Rev. 81, 102-03 (1985). Manageability concerns for the judicial system also are implicated.
Pennsylvania law has not yet endorsed market share liability or sent a clear authoritative signal that it would do so in an appropriate case. We are unsure how the Pennsylvania Supreme Court would balance the policy arguments favoring and militating against the adoption of market share liability. We are certain, however, that a federal court in a diversity case should not perform the delicate balancing act for it.
Our conviction is strengthened by the numerous disparate versions of market share liability in existence. Some states apportion liability based on market share, while other states hold defendants liable for the amount of risk they created that the plaintiff would be injured, with market share being a relevant factor in this inquiry. Compare Collins v. Eli Lilly Co., 116 Wis.2d 166, 342 N.W.2d 37, 52-53 (risk approach), cert. denied, 469 U.S. 826, 105 S.Ct. 107, 83 L.Ed.2d 51 (1984) with Hymowitz v. Eli Lilly & Co., 73 N.Y.2d *127487, 541 N.Y.S.2d 941, 539 N.E.2d 1069, 1077-78 (market share approach), cert. denied, 493 U.S. 944, 110 S.Ct. 350, 107 L.Ed.2d 338 (1989). To calculate the market share of defendants, some states utilize sales in the national market, while others examine local markets. Compare Smith v. Cutter Biological, Inc., 72 Haw. 416, 823 P.2d 717, 728 (Haw.1991) (national market) and Hymowitz, 541 N.Y.S.2d at 949, 539 N.E.2d at 1077 (same) with Conley v. Boyle Drug Co., 570 So.2d 275, 283 (Fla.1990) (local market) and George v. Parke-Davis, 107 Wash.2d 584, 733 P.2d 507, 512 (1987) (in banc) (same). Although all states use only several and not joint and several liability, their approaches vary greatly. In some states, a plaintiff must prove the actual market share of each defendant. See, e.g., Sindell, 607 P.2d at 935-37; Hymowitz, 541 N.Y.S.2d at 949-50, 539 N.E.2d at 1077-78. Other states impose a rebuttable presumption that all defendants have an equal market share, total-ling one hundred percent. See, e.g., Martin v. Abbott Lab., 102 Wash.2d 581, 689 P.2d 368, 383 (1984) (in banc); Conley, 570 So.2d at 286. Each defendant may rebut this presumption by showing that its actual market share was less. Martin, 689 P.2d at 383; Conley, 570 So.2d at 286. The liability of defendants that cannot prove their actual market share then inflates so plaintiff receives a total recovery. Martin, 689 P.2d at 383; Conley, 570 So.2d at 286.
The differences go on and on. And each highest state court that has adopted market share liability has harshly criticized its predecessors for either distorting defendants’ liability or creating administratively unworkable schemes. If we predicted that the Pennsylvania Supreme Court would adopt market share liability in this case, we or the district court on remand then would be obligated to articulate a comprehensive theory of’ market share liability from among the many divergent approaches. Plaintiffs’ request that we formulate a new area of Pennsylvania tort law without any guidance from state authorities simply reinforces that we are not the appropriate tribunal to perform this formidable task.
Plaintiffs attempt to minimize the extent of the departure they seek by characterizing the application of market share liability in this case as not “far afield from traditional causation requirements.” PHA Brief at 24. Market share liability, however, clearly represents a significant extension of Pennsylvania tort law. -Even courts in other states that have embraced market share liability have forthrightly labeled their actions a modification or deviation from prior law.14 Federal courts also consistently have characterized market share liability as novel and even radical.15 Plaintiffs cannot escape the fact that market share liability is a significant departure from Pennsylvania’s traditional requirement that a plaintiff prove proximate cause. We will apply current Pennsylvania law and reject the applicability of market share liability.
2. Alternative Liability
Having rejected the use of market share liability, we easily can dismiss plaintiffs’ second theory of collective liability, alternative liability. Alternative liability holds all tortfeasors who are unable to exculpate themselves jointly and severally liable for the plaintiffs injury. Snoparsky v. Baer, 439 Pa. 140, 266 A.2d 707, 709 (1970). The Pennsylvania Supreme Court adopted the definition of alternative liability espoused by the Restatement (Second) of Torts:
Where the conduct of two or more actors is tortious, and it is proved that harm has been caused to the plaintiff by only one of them, but there is uncertainty as to which *128one has caused it, the burden is upon each such actor to prove that he has not caused the harm.
Id. (quoting Restatement (Second) of Torts § 433B(3) (1965)).
The California Supreme Court originally formulated the alternative liability theory in the landmark case of Summers v. Tice, 33 Cal.2d 80, 199 P.2d 1 (1948). In Summers, the plaintiff was injured when two hunters simultaneously fired gunshots in his direction. Although the plaintiff could not identify which defendant had fired the shot that hit him, the Summers Court held that each defendant would be jointly and severally liable unless he could prove that he had not caused plaintiffs injury. 199 P.2d at 3-4. The rationale for shifting the burden of proof in such cases is that defendants usually are in a better position to determine who caused the injury. Id. 199 P.2d at 4. An innocent plaintiff should not be left without a remedy because defendants refuse to divulge information in their possession.
The Pennsylvania Supreme Court twice has sanctioned the use of alternative liability. In Snoparsky, a minor child alleged that several boys threw stones at her, one of which hit and injured her. The Pennsylvania Supreme Court allowed alternative liability to establish causation with respect to each negligent defendant. Snoparsky, 266 A.2d at 709. Several years later, a child who lost his eyesight in a spitball fight on a school bus collected damages from all students who had thrown spitballs and were unable to prove that they had not caused plaintiffs injury. See Sommers v. Hessler, 227 Pa.Super. 41, 323 A.2d 17, 19-20 (1974).
The Pennsylvania Supreme Court has approved alternative liability only when each defendant’s tortious conduct was simultaneous and identical, and all potential tortfea-sors were joined as defendants. Defendants argue that these conditions are prerequisites to the imposition of alternative liability. Alternate liability is inappropriate in this products liability case, they reason, since defendants manufactured and marketed their allegedly defective product over a long period of time, and not all potential tortfeasors are named as defendants.16 Plaintiffs reply that the parameters of alternative liability must be flexible to respond to the everchanging world of tort litigation in a modern industrialized society. To support their assertion that alternative liability should be applied in products liability cases, and in this case in particular, plaintiffs point to the Restatement (Second) of Torts, which provides:
The cases thus far decided in which the rule stated in Subsection (3) has been applied all have been cases in which all of the actors involved have been joined as defendants. All of these cases have involved conduct simultaneous in time, or substantially so, and all of them have involved conduct of substantially the same character.... The rule stated in Subsection (3) is not intended to preclude possible modification if such situations call for it.
Restatement (Second) of Torts § 433B(3) cmt. h (1965) (emphasis added).
The modification that plaintiffs advocate would transform alternative liability into market share liability with one glaring difference: defendants who are unable to exculpate themselves would be jointly and severally liable instead of liable only for a percentage of the judgment based on their market share.. The plaintiffs concede as much. In Erlich, the court expanded alternative liability and applied it to a products liability case in which not all manufacturers of the allegedly defective product were joined as defendants. Plaintiffs acknowledge that “market share liability is almost identical to the theory of alternative liability adopted by the Er-lich court.” PHA Brief at 37; see also Vi-giolto, 643 F.Supp. at 1461 (Erlich “adoptfed] a modification of § 433B(3) which was quite similar to the Sindell theory”). Because alternative liability holds defendants jointly and severally liable, the modified alternative liability plaintiffs propose departs further from established Pennsylvania tort principles than does market share liability. Therefore, for the reasons discussed above, see discussion supra part II.B.l, plaintiffs may not proceed on a theory of alternative liability.
*1293. Enterprise Liability
Enterprise liability, plaintiffs’ final theory of collective liability, first was articulated in Hall v. E.I. Du Pont De Nemours & Co., 345 F.Supp. 353, 376-78 (E.D.N.Y.1972). In Hall, thirteen children injured in accidents with blasting caps sued six blasting cap manufacturers, who comprised virtually the entire industry in the United States, and their trade association. Plaintiffs could not specifically link any manufacturer to any particular injury. Applying enterprise liability, the court permitted plaintiffs’ claims to survive a motion to dismiss because all defendants who could not exonerate themselves could be held jointly and severally liable if they had adhered to an unreasonable, industry-wide safety standard set by their trade association. The Hall court found that enterprise liability exists when (1) a small number of manufacturers, virtually all of which are named defendants, produced the injury-causing product; (2) the defendants had joint knowledge of the risks inherent in the product and possessed a joint capacity to reduce the risks; and (3) each delegated the responsibility to set safety standards to a trade association, which failed to reduce the risk. Id. at 378.
Pennsylvania courts consistently have rejected this theory of joint liability. In Cum-mins, the Pennsylvania Superior Court noted that enterprise liability has been rejected by virtually every jurisdiction. 495 A.2d at 971 n. 6. The court then stated “this cause of action has heretofore not been recognized by the courts of this Commonwealth ... and will not now be so adopted.” Id. at 971 (quoted in Burnside, 505 A.2d at 985). The Pennsylvania Superior Court most recently rejected a plaintiffs attempt to invoke enterprise liability because “no Pennsylvania appellate court has endorsed the theory.” Burman v. Golay & Co., 420 Pa.Super. 209, 616 A.2d 657, 660 (1992).
The Cxmimins court explained its reason for refusing to analyze the policy arguments favoring and militating against enterprise liability: “Extensive policy shifts of this magnitude should not be initiated by an intermediate appellate (sic) court. The appropriate tribunal to accomplish such drastic changes is either the Supreme Court or the Legislature.” 495 A.2d at 971 n. 6 (quoting Namm v. Charles E. Frosst & Co., 178 N.J.Super. 19, 427 A.2d 1121, 1129 (App.Div.1981)). We agree with the district court that “[i]f a state appellate court is not the correct tribunal for recognizing a drastic change in state law, then neither is a federal district court sitting in diversity.” City of Philadelphia, slip op. at 35 (April 23,1992). We once again refuse to articulate a new tort principle under Pennsylvania law.
We hold that none of the theories advocated by plaintiffs — market share liability, alternative liability or enterprise liability — may be invoked to impose liability on the lead pigment industry for the costs of lead-based paint abatement. We therefore will affirm the district court’s dismissal of the amended complaint in its entirety.17 The plaintiffs in this case had a choice to litigate their claims either in state or federal court. They selected the federal court system to urge modifications of state tort law based on the equities of the case. Plaintiffs ask us to construct a collective liability theory “out of whole cloth, using for a pattern only bits and pieces of the decisions of courts in remote states.” Tidler v. Eli Lilly & Co., 851 F.2d 418, 424 (D.C.Cir.1988). If plaintiffs seek a modification of state law, they must argue their case to the appropriate state authority, either judicial or legislative. “It is decidedly not the business of the federal courts to alter or augment state law to meet the felt necessities of the case; to suggest otherwise is to ignore fundamental principles of comity inherent in our federal system of government.” Id. at 425.
III. CONCLUSION
For the foregoing reasons, we will affirm the dismissal of plaintiffs’ complaint in its entirety.
. Plaintiffs also bring this class action on behalf of:
The City of Philadelphia and all other cities in the United States with a population over 100,-000 persons, the Philadelphia Housing Authority, and all housing authorities and/or public health authorities affiliated with such cities (collectively the "Cities” or “Class”) engaged in and/or contemplating a program:
a. for the inspection, testing, monitoring or abatement of lead paint in properties within the respective jurisdictions of the Cities, which have been built or whose interiors have been painted within the period from the early 1900s through 1977, which are owned and/or managed by the Cities, and/or privately owned or managed where a known lead paint hazard exists and private owners refuse to abate such hazard; and/or
b. to screen, test, diagnose and treat the residents of the Cities for exposure to lead paint and to educate them about its hazards.
Amended Complaint ¶ 23, App. at 271-72.
. Eagle-Pitcher Industries, originally a defendant, was not named as a defendant in the amended complaint because it had sought protection pursuant to Chapter 11 of the Bankruptcy Code.
. The City then failed to comply with its own ordinance. In 1977, it signed a consent decree in which it reaffirmed its duty to abate lead paint in city-owned housing when it posed a hazard. See City-Wide Coalition Against Childhood Lead Paint Poisoning v. Philadelphia Housing Auth., 356 F.Supp. 123 (E.D.Pa.1977).
. This decision, in response to defendants’ motion for reconsideration, amended the district court's earlier opinion dated August 23, 1991. See City of Philadelphia v. Lead Indus. Ass’n, No. 90-7064, 1991 WL 170810 (E.D.Pa. Aug. 23, 1991).
. Plaintiffs do not appeal the dismissal of their concert of action and conspiracy theories.
. HUD regulations requiring abatement of lead were enacted on August 1, 1986, see 51 Fed.Reg. 27,787-88 (1986) (codified at 24 C.F.R. §§ 35.1-35.24), while HUD’s Interim Guidelines for Hazard Identification and Abatement in Public and Indian Housing became effective on April 1, 1990, see 55 Fed.Reg. 14,556 (1990). We will assume arguendo that the City became obligated to expend public funds to abate lead on April 1, 1990.
. The breach of warranty claim accrued when delivery of the paint was tendered. 13 Pa.Cons. Stat.Ann. § 2725(b) (1984). The discovery rule does not apply. Northampton County Area Community College, 566 A.2d at 599.
. In 1966, the City enacted regulations prohibiting the use in dwellings of paint with a lead content of over 1 percent. See Philadelphia, Pa., Code § 6-403(1 )(a), (2)(a) (1966); Philadelphia Dep’t of Public Health, Regulations Relating to Labeling, Application and Removal of Lead Paint § 6-403(l)(d), (2) (June 27, 1966). In 1971, the original version of 42 U.S.C. § 4841 defined lead-based paint as paint with a lead content of over 1 percent. See Pub.L. No. 91-695, § 501(3), 84 Stat. 2080 (1971). In 1974, the maximum safe level was changed to .5 percent, see Pub.L. No. 93-151, § 6, 87 Stat. 567 (1973), and in 1976 to .06 percent, see Pub.L. No. 94-317, § 204(c)(1), 90 Stat. 706 (1976).
The City therefore was aware of the hazards of 1 percent lead paint by 1966, of .5 percent lead paint by 1974, and of .06 percent lead paint by 1976. Though the City's claims for damages caused by paint with lead contents between .5 and 1 percent accrued before 1976, we will assume arguendo that all of the City’s claims accrued in 1976.
. Defendants do not challenge the district court's determination that the City's restitution and indemnification claims are not lime-barred.
. The highest state courts in six states, Hawaii, New York, Wisconsin, Washington, Florida and California, have adopted varying versions of market share liability. See Smith v. Cutter Biological, Inc., 72 Haw. 416, 823 P.2d 717 (1991); Hymowitz v. Eli Lilly & Co., 73 N.Y.2d 487, 541 *124N.Y.S.2d 941, 539 N.E.2d 1069, cert. denied, 493 U.S. 944, 110 S.Ct 350, 107 L.Ed.2d 338 (1989); Collins v. Eli Lilly Co., 116 Wis.2d 166, 342 N.W.2d 37, cert. denied, 469 U.S. 826, 105 S.Ct. 107, 83 L.Ed.2d 51 (1984); Martin v. Abbott Lab., 102 Wash.2d 581, 689 P.2d 368 (1984) (en banc); Conley v. Boyle Drug Co., 570 So.2d 275 (Fla.1990); Sindell v. Abbott Lab,, 26 Cal.3d 588, 163 Cal.Rptr. 132, 607 P.2d 924, cert. denied, 449 U.S. 912, 101 S.Ct. 285, 66 L.Ed.2d 140 (1980).
. The Erlich court professed to endorse an extension of alternative liability, but really relied on market, share liability. Plaintiffs concede this point in their brief — “market share liability is almost identical to the theory of alternative liability adopted by the Erlich court.” PHA Brief at 37. Subsequent cases discuss Erlich as if it applied market share liability. See Pennfield Corp. v. Meadow Valley Elec., Inc., 413 Pa.Super. 187, 604 A.2d 1082, 1087 (1992); Vigiolto v. Johns-Manville Corp., 643 F.Supp. 1454, 1461-62 (W.D.Pa.1986), aff'd, 826 F.2d 1058 (3d Cir.1987).
. Occasionally, a federal district court in this circuit has endorsed the use of market share liability under Pennsylvania law. In Karibjanian v. Thomas Jefferson University Hospital, a district court allowed a claim that the drug Thorostat was defective to survive a motion to dismiss, stating that "the manufacturers might also be liable under a market-share theory....” No. 89-1891, 1989 WL 80317 at *1, 1989 U.S.Dist. LEXIS 8068 at *4 (E.D.Pa. July 11, 1989). The Karibjanian court, however, jumped ahead of the current state of Pennsylvania law. Another district court predicted that "when faced with an appropriate factual situation," the Pennsylvania Supreme Court "will recognize the market share liability theory,” but acknowledged that this move would "expand existing law." Vigiolto v. Johns-Manville Corp., 643 F.Supp. 1454, 1462 (W.D.Pa.1986), aff’d, 826 F.2d 1058 (3d Cir.1987). Although the Vigiolto court's prediction one day may come true, we will not perform the expansion of Pennsylvania law.
. The Rhode Island, Illinois, Iowa and Missouri Supreme Courts have rejected market share liability. See Gorman v. Abbott Lab., 599 A.2d 1364 *126(R.I.1991); Smith v. Eli Lilly & Co., 137 Ill.2d 222, 148 Ill.Dec. 22, 560 N.E.2d 324 (1990); Mulcahy v. Eli Lilly & Co., 386 N.W.2d 67 (Iowa 1986); Zafft v. Eli Lilly & Co., 676 S.W.2d 241 (Mo.1984) (en banc). For states that have *127adopted a version of market share liability, see supra note 10.
. See, e.g., Martin, 689 P.2d at 381 ("deviat[ion] from traditional notions of tort law”); Collins, 342 N.W.2d at 45 (same); Hymowitz, 541 N.Y.S.2d at 947, 539 N.E.2d at 1075 ("modification of existing doctrine”); Sindell, 607 P.2d at 936 (same); Ferrigno v. Eli Lilly & Co., 175 N.J.Super. 551, 420 A.2d 1305, 1322 (App.Div.1980) ("the views expressed in this opinion arc novel”).
. See, e.g., Tidler, 851 F.2d at 425 ("novel”); Blackston v. Shook & Fletcher Insulation Co., 764 F.2d 1480, 1483 (11th Cir.1985) ("novel”); Thompson v. Johns-Manville Sales Corp., 714 F.2d 581, 583 (5th Cir.1983) ("radical departures from traditional theories”), cert. denied, 465 U.S. 1102, 104 S.Ct. 1598, 80 L.Ed.2d 129 (1984); Morton v. Abbott Lab., 538 F.Supp. 593, 595 (M.D.Fla.1982) ("novel theories”); Mizell v. Eli Lilly & Co., 526 F.Supp. 589, 596 (D.S.C.1981) ("radical departure”).
. Plaintiffs allege only that substantially all lead pigment manufacturers are named as defendants. At least one company that produced lead pigment during the relevant time, Eagle-Pitcher, is not a party to this action because it is bankrupt. Defendants' conduct was hardly simultaneous since plaintiffs define the relevant period as the early 1900s through 1977.
. Because the plaintiffs failed to establish the necessary causation for all of their causes of action, we need not reach whether any individual claim should be dismissed for other reasons pursuant to Federal Rule of Civil Procedure 12(b)(6).