Agere Systems, Inc. v. Advanced Environmental Technology Corp.

602 F.3d 204, 602 F. Supp. 3d 204, 40 Envtl. L. Rep. (Envtl. Law Inst.) 20119, 70 ERC (BNA) 1385, 2010 U.S. App. LEXIS 7462, 2010 WL 1427582

• Court: Court of Appeals for the Third Circuit
• Decided: April 12, 2010
• Docket: 09-1814
• Status: Published

Opinion

OPINION OF THE COURT

JORDAN, Circuit Judge.

Table of Contents

I. Background...............................................................211

A. EPA Actions at the Boarhead Site.......................................211

i. The OU-1 Consent Decree..........................................212

ii. The OU-2 Consent Decree..........................................212

iii. Carpenter........................................................213

B. Present Suit..........................................................213

i. Stipulations......................................................214

ii. Bench Trial......................................................214

II. Statement of Jurisdiction and Standard of Review 216

III. Discussion ................................................................216

A. Statutory Background Law.............................................216

B. Issues on Appeal......................................................218

C. Cytec, Ford, SPS, and TPs § 113(f) Claim for Reimbursement of

Payments Made to the EPA for Past Costs .............................219

i. Background......................................................219

ii. The Statute of Limitations to Recover Past Costs......................219

iii. The Exceptions to the Three-Year Statute of Limitations Period........221

D. TI and Agere’s § 107(a) Claims to Recover Costs Paid to Other

Plaintiffs Pursuant to Settlement Agreements...........................224

i. Background......................................................224

ii. Section 107(a) Cost Recovery Claims ................................225

E. The District Court’s Equitable Allocation for the Plaintiffs’ Costs of

Performing Work Under the OU-1 and OU-2Consent Decrees ............227

i. Background......................................................227

ii. Sections 107(a) and 113(f)..........................................227

iii. The District Court’s Equitable Allocation Under § 113(f)...............229

iv. The June 23rd Stipulation is Not an Admission that is Admissible

Against Carpenter...............................................231

v. Other Evidence Regarding Waste Volumes ...........................233

F. Other Contentions Regarding the District Court’s Equitable Allocation.....234

i. NRM’s Waste.....................................................234

ii. The Culpability and Lack of Cooperation of Settling Defendants........235

iii. Settlement Amounts...............................................235

G. The Pennsylvania Hazardous Sites Cleanup Act..........................236

IV. Conclusion................................................................236

This appeal arises from nearly three decades of involvement by the Environmental Protection Agency (“EPA”) at the Boarhead Farms Superfund Site in Bucks County, Pennsylvania (the “Boarhead Site” or the “Site”). At issue in the underlying case was the disposal of millions of gallons of toxic waste, over a six-year time period, by more than twenty parties, with millions of dollars of cleanup costs at stake. Along with the factual issues born of that history, the case implicates the still developing distinctions between liability under § 107(a) and § 113(f) of the Comprehensive Environmental Response, Compensation, and Liability Act of 1980 and the Superfund Amendments and Reauthorization Act of 1986 (“SARA”), codified together at 42 U.S.C. §§ 9601-9675 (collectively “CERCLA”).

On June 18, 2002, five plaintiffs — Agere Systems, Inc. (“Agere”), Cytec Industries, Inc. (“Cytec”), Ford Motor Company (“Ford”), SPS Technology, LLC (“SPS”), and TI Automotive Systems LLC (“TI”) (collectively “plaintiffs” or “appellees”)— filed the present suit against twenty-three defendants for cost recovery and contribution under CERCLA and the Pennsylvania Hazardous Sites Cleanup Act (“HSCA”), 35 Pa. Stat. Ann. § 6020.101-104 et seq., to recover costs that the plaintiffs had paid to the EPA pursuant to certain consent decrees or that they had provided as a consequence of the cleanup of hazardous substances at the Boarhead Site. All of the defendants except one, Carpenter Technology Corporation (“Carpenter”), settled their liabilities with the plaintiffs or were otherwise dismissed from the suit after a bench trial. On August 22, 2008, the United States District Court for the Eastern District of Pennsylvania entered judgment against Carpenter, finding it liable for 80% of the costs paid by the plaintiffs as of December 31, 2007, plus prejudgment interest. The Court also entered a declara tory judgment that Carpenter is liable for 80% of all cleanup costs that the plaintiffs may incur after January 1, 2008. The District Court denied Carpenter’s motion to alter or amend the judgment, and Carpenter filed this appeal. For the following reasons, we will vacate the District Court’s judgment and remand for proceedings consistent with this opinion.

I. Background

Beginning in 1972, DeRewal Chemical Corporation (“DCC”), a business that removed, transported, and disposed of chemical waste generated by other companies, began illegally dumping its customers’ waste at the Boarhead Site. The dumping continued until 1976, when DCC was enjoined from bringing any chemicals to the Site. 1

A. EPA Actions at the Boarhead Site

Between 1984 and 1986, the EPA completed an initial investigation of the contamination at the Boarhead Site. Based on the results of that investigation, the Site was added to the EPA’s National Priorities List on March 13, 1989, 2 *and thus became a Superfund site. 3 Later that year, the EPA performed a remedial investigation of the Site to identify whether there were contaminants that posed a risk to human health and the environment. That investigation revealed a variety of hazardous substances in the soil, sediments, and groundwater. In response, the EPA conducted several small-scale cleanup actions 4 over the next three years to address immediate risks.

A report of the remedial investigation was published in January 1997, and, by the following July, the EPA issued a feasibility study that defined objectives for a larger-scale response with additional remedial actions. 5 In January 1998, the EPA produced its proposed remedial action plan, based on both the remedial investigation and the feasibility study. Then, on November 18, 1998, it issued its Record of Decision (“ROD”) respecting the Site. 6 *The ROD was to be implemented in two stages, which, in the argot of the EPA, are referred to as “operable units.” An operable unit is a “discrete action that comprises an incremental step toward comprehensively-addressing site problems.” 40 C.F.R. § 307.14. The two planned for the Boar-head Site were designated as Operable Unit One (“OU-1”) and Operable Unit Two (“OU-2”). 7

i. The OU-1 Consent Decree

On June 2, 2000, the EPA commenced a suit in the District Court against Cytec, Ford, and SPS, under § 107 of CERCLA, 42 U.S.C. § 9607, 8 by filing a complaint along with a consent decree that those parties had executed (the “OU-1 Consent Decree”). The District Court approved the OU-1 Consent Decree on September 28, 2000. Pursuant to the mandates of the decree, as well as an administrative order, 9 Cytec, Ford, and SPS were required to do the work contemplated for OU-1 and to reimburse the EPA for its administrative and oversight costs in connection with the OU-1 cleanup.

Cytec, Ford, and SPS subsequently entered into a separate settlement agreement with Agere, TI, and two other companies, whereby they all agreed to collectively fund and perform OU-1 work and to otherwise comply with the OU-1 Consent Decree. All seven of those companies, which, for convenience, we will call the “OU-1 group,” have contributed to trust accounts from which various contractors have been paid and will continue to be paid to perform the work required by the OU-1 Consent Decree.

ii. The OU-2 Consent Decree

On December 6, 2001, the EPA commenced another suit in federal court under § 107 against Cytec, Ford, SPS, and TI, by filing a complaint and a second consent decree that those parties had executed (the “OU-2 Consent Decree”). The District Court approved the OU-2 Consent Decree on March 14, 2002. Pursuant to that decree and, again, an administrative order, 10 the Court ordered Cytec, Ford, SPS, and TI to do four things: (1) do the work contemplated for OU-2; (2) reimburse the EPA for approximately $7 million in costs related to removal actions at the Boarhead Site that the EPA had incurred prior to July 2000; 11 (3) reimburse the EPA for a yet-to-be-determined amount of response costs incurred after July 2000; and (4) reimburse the EPA for its other future response costs in connection with OU-2 work.

Ford, Cytec, SPS, and TI subsequently entered into a separate, private settlement with Agere whereby the parties agreed to collectively fund and perform OU-2 work and to otherwise comply with the OU-2 Consent Decree. Again for convenience, we will refer to those five companies as “the OU2 group.” In a fashion similar to the OU-1 group, the five members of the OU-2 group contributed to group trust accounts from which they paid and will continue to pay various contractors to perform the work required by the OU-2 Consent Decree.

On March 30, 2007, in exchange for $400,000, Agere (which had never itself been sued by the EPA), assigned to Cytec, Ford, SPS, and TI its claims to recover the approximately one million dollars it had paid into the OU-1 and OU-2 group trust accounts. 12

iii. Carpenter

DCC collected and disposed of waste at the Boarhead Site from more than twenty customers between 1972 and 1977. Carpenter was one of those customers and is a Delaware corporation with its principal place of business in Reading, Pennsylvania.

Carpenter received a notice from the EPA concerning the Boarhead Site, dated September 28, 2000, asking it to resolve its potential liability to the EPA for past cleanup costs and further asking that it agree to perform work required by the EPA’s 1998 ROD, excluding the work outlined in the OU-1 Consent Decree. 13 The letter included a list of ten previous recipients of similar letters from the EPA concerning the Boarhead Site. Carpenter did not comply with the requests in the letter.

B. Present Suit

On June 18, 2002, the five plaintiffs in this suit — Agere, Cytec, Ford, SPS, and TI — filed their original complaint in the District Court against twenty-three defendants, 14 seeking cost recovery and contri bution under CERCLA and under the HSCA to recoup costs they had paid to the EPA pursuant to one or more consent decrees or had reimbursed to one another, all relating to the cleanup of hazardous substances at the Boarhead Site. By January 21, 2008, the plaintiffs had filed a fifth amended complaint and seven defendants remained. 15 When the bench trial commenced on June 23, 2008, only two defendants were left, Handy & Harman Tube Company, Inc. (“H & H”) and Carpenter. 16 After trial, H & H was granted an unopposed motion to dismiss, leaving Carpenter as the sole remaining defendant.

i. Stipulations

The parties 17 entered into several stipulations before and during trial, to narrow the factual and legal issues before the District Court. First, they signed a stipulation of background facts, many of which have been noted here already. 18 Second, on June 19, 2008, they stipulated to the volume of waste that DCC collected from fifteen other companies, fourteen of which were defendants. Carpenter’s waste was not included in that stipulation. Third, on June 23, 2008, the plaintiffs and defendant H & H stipulated to the volume of waste that DCC collected from three of the plaintiffs' — Cytec, Ford, and SPS — -and from H & H. Importantly, Carpenter did not join that stipulation. Finally, on July 1, 2008, the plaintiffs, Carpenter, and H & H stipulated to the fact that the waste described in the June 23rd stipulation was hazardous waste as defined by CERCLA. They also stipulated that DCC had not hauled any waste from TI or Agere to the Boarhead Site. Thus, while the effect of the July 1st stipulation is contested, it is undisputed that Carpenter never directly stipulated to the volume of waste that Cytec, Ford, SPS, and H & H had contributed to the Boarhead Site.

ii. Bench Trial

Pursuant to CERCLA, the District Court’s responsibility included “allocating] response costs among liable parties using such equitable factors as the court determine[d] [to be] appropriate.” 42 U.S.C. § 9613(f)(1). The Court decided that the parties’ various volumes of waste at the Boarhead Site constituted the most equitable basis upon which to allocate costs. Hence, the purpose of the bench trial, held in July of 2008, was to figure out as a matter of fact what those volumes were, including the volumes attributable to Carpenter. The Court also held that culpability and the degree of cooperation the parties had demonstrated in dealing with the cleanup should be considered in allocating response costs.

The District Court found that the plaintiffs had incurred costs totaling $13,678,378.55 as of December 31, 2007. 19 The Court then concluded that those costs had been incurred of necessity in complying with the requirements of the ROD. The Court also held that cleanup at the Boarhead Site must continue until the organic and inorganic contaminants in the groundwater reach an acceptable level as set forth by the EPA in the ROD. Thus, the Court noted, the plaintiffs will continue to incur costs for the foreseeable future.

To prove the waste volumes at the Site, the plaintiffs presented testimony from truck drivers who had been employed by DCC. Based on that testimony and the above-described stipulations, the Court calculated the volume of waste that DCC collected from each of the parties, and then calculated what percentage of that waste was disposed of at the Boarhead Site. Regarding Carpenter’s waste volumes, the Court also had the benefit of Carpenter’s business records, which confirmed parts of the testimony of the DCC drivers. The Court found that Carpenter’s waste constituted 62.6% of the hazardous waste disposed of at the Boarhead Site. 20

The plaintiffs also presented evidence that in 1969 Carpenter hired two waste collection companies, Revere Chemical Company (“Revere”) and Echo, Inc. (“Echo”), to haul its waste acids. Carpenter knew that Manfred DeRewal, who later founded DCC, was the president of Revere and Echo. The evidence showed that in 1970 Carpenter learned that Revere and Echo were shut down for unlawful polluting practices. 21 Nevertheless, in June 1973, Carpenter hired DCC, which Carpenter knew was owned by DeRewal, simply because DCC’s bid was significantly below the competing bids. Because “Carpenter relinquished its potent waste acids to a known polluter,” and because Carpenter did not cooperate with the EPA, the Court allocated to Carpenter an additional 17.4% of the total clean-up cost — 8.7% for repeatedly working with a known polluter and 8.7% for not cooperating with the EPA — holding that “culpability is an appropriate equitable factor in resolving contribution claims.” (App. at 1:A72.) 22 In other words, the Court adjusted Carpenter’s share of liability above what it had determined to be Carpenter’s volumetric share of waste, based on Carpenter’s significant culpability in creating a Superfund site and its lack of cooperation in cleaning it up.

On August 22, 2008, the District Court entered judgment against Carpenter, finding it liable for 80% (62.6% plus 8.7% plus 8.7%) of the $13,678,378.55 paid by the plaintiffs as of December 31, 2007, plus prejudgment interest. That resulted in a judgment of $10,942,702.84 against Carpenter. The remaining 20% of liability was allocated to the plaintiffs. The Court also entered a declaratory judgment that Carpenter is liable for 80% of all future costs, as measured from January 1, 2008, for response actions pursuant to the OU-1 and the OU-2 Consent Decrees. Finally, the Court found that Carpenter’s liability under the HSCA parallels its liability under CERCLA.

Carpenter filed a motion to alter or amend the judgment, which the District Court denied on February 20, 2009. Carpenter then filed a timely notice of appeal to our Court.

II. Statement of Jurisdiction and Standard of Review

The District Court had jurisdiction over the plaintiffs’ CERCLA claims pursuant to 42 U.S.C. § 9613(b) and 28 U.S.C. § 1331. It had supplemental jurisdiction over the HSCA claims under 28 U.S.C. § 1367. We have jurisdiction pursuant to 28 U.S.C. § 1291.

“We may set aside [a] district court’s findings of fact only if they are clearly erroneous.” FMC Corp. v. U.S. Dep’t. of Commerce, 29 F.3d 833, 838 (3d Cir.1994); cf. In re Fruehauf Trailer Corp., 444 F.3d 203, 210 (3d Cir.2006) (explaining that “[fjactual findings may only be overturned if they are completely devoid of a credible evidentiary basis or bear no rational relationship to the supporting data” (citation omitted) (internal quotation omitted)).

However, we review a district court’s allocation of CERCLA response costs for abuse of discretion. Beazer E., Inc. v. Mead Corp., 412 F.3d 429, 446 n. 18 (3d Cir.2005). “An abuse of discretion occurs when ‘the district court’s decision rests upon a clearly erroneous finding of fact, an errant conclusion of law or an improper application of law to fact.’ ” Id. (quoting Int’l Union v. Mack Trucks, Inc., 820 F.2d 91, 95 (3d Cir.1987)).

Finally, we exercise plenary review over a district court’s interpretation of CERCLA, including statute of limitations issues. United States v. Se. Pa. Transp. Auth, 235 F.3d 817, 822 (3d Cir.2000); see also New Castle County v. Halliburton NUS Corp., 111 F.3d 1116, 1120 (3d Cir.1997) (explaining that, in approaching a statute of limitations issue, we must “exercise plenary review over the district court’s interpretation of the relevant CERCLA ... provisions”).

III. Discussion

A. Statutory Background Law

All of Carpenter’s contentions on appeal involve the plaintiffs’ claims under CERCLA § 107(a) for cost recovery and § 113(f) for contribution. Thus, we present a brief overview of those two statutory remedies before addressing Carpenter’s specific contentions.

CERCLA provides two mechanisms that allow potentially responsible parties (“PRPs”) to recover costs they have expended to decontaminate a polluted site: § 107(a) cost recovery claims and § 113(f) contribution claims. The first option, § 107(a), provides that PRPs are liable for “any ... necessary costs of response incurred by any other person” consistent with CERCLA. 23 42 U.S.C. § 9607(a)(4)(B). Section 107(a) thus al lows private parties to bring cost recovery suits against other PRPs, 24 and courts have developed several principles for application in such suits. 4 The Law of Hazardous Waste § 14.01[2][c] (Susan M. Cooke, ed., 2009). Significantly, § 107(a) allows for complete cost recovery under a joint and several liability scheme. See N.J. Tpk. Auth. v. PPG Indus., Inc., 197 F.3d 96, 104 (3d Cir.1999). Initially, “[w]hen CERCLA was first enacted, [§ 107 cost recovery] was the only remedy available, and [c]ourts struggled with whether PRPs (themselves liable for some of the cleanup) could invoke § 107 for contribution from other PRPs for their proportionate share of the costs as opposed to full cost recovery.” Niagara Mohawk Pwr. Corp., 596 F.3d at 121.

Congress ultimately provided the language necessary to authorize contribution under CERCLA when it added § 113 to the statutory scheme with the passage of SARA. Id. at 121. Section 113(f) specifically is a second means of recouping cleanup costs, and it, in turn, provides two avenues of relief. Under § 113(f)(1), a PRP can seek contribution from another PRP during or following a CERCLA suit brought against the first PRP. 42 U.S.C. § 9613(f)(1); see also United States v. Atl. Research Corp., 551 U.S. 128, 139, 127 S.Ct. 2331, 168 L.Ed.2d 28 (2007) (explaining that “[s]ection 113(f)(1) authorizes a contribution action to PRPs with common liability stemming from an action instituted under [CERCLA]”). Likewise, under § 113(f)(3)(B), PRPs who resolve their liability to the United States or an individual State through an administratively or judicially approved settlement can seek contribution from another PRP. 42 U.S.C. § 9613(f)(3)(B); see also Atl. Research Corp., 551 U.S. at 139 n. 5, 127 S.Ct. 2331 (“Similarly, § 113(f)(3)(B) permits a PRP to seek contribution after it ‘has resolved its liability to the United States or a State ... in an administrative or judicially approved settlement ....’” (quoting 42 U.S.C. § 9613(f)(3)(B))).

In Atlantic Research, the Supreme Court endeavored to clarify the relationship between § 107(a) cost recovery claims and § 113(f) contribution claims, noting that those remedies are distinct. See Atl. Research Corp., 551 U.S. at 138, 127 S.Ct. 2331 (“[Sections] 107(a) and 113(f) provide two ‘clearly distinct’ remedies. ‘CERCLA provide[s] for a right to cost recovery in certain circumstances, § 107(a), and separate rights to contribution in other circumstances, §§ 113(f)(1), 113(f)(3)(B).’ ” (quoting Cooper Indus., Inc. v. Aviall Serv., Inc., 543 U.S. 157, 163 n. 3, 125 S.Ct. 577, 160 L.Ed.2d 548 (2004))). With regard to § 113(f) contribution claims, the Court held that a private party may not bring a contribution claim unless it has been sued under § 106 or § 107 of CERCLA. 25 See Cooper Indus., Inc. 543 U.S. at 168, 125 S.Ct. 577 (explaining that § 113(f) “authorizes contribution claims only ‘during or following’ a civil action under § 106 or § 107(a)”); see also Atl. Research Corp., 551 U.S. at 139, 127 S.Ct. 2331 (“Section 113(f)(1) authorizes a contribution action to PRPs with common liability stemming from an action instituted under [CERCLA].”). With regard to § 107(a) cost recovery claims, the Court held that a private party who voluntarily undertakes a cleanup action — and is therefore unable to sue under § 113(f) because it “remediated the hazardous material [ ] without the judicial spur of § 106 or § 107” — can seek recovery of response costs under § 107(a)(4)(B). Niagara Mohawk Pwr. Corp., 596 F.3d at 121 (describing Atl. Research Corp., 551 U.S. at 139, 127 S.Ct. 2331). Thus, a private party may recover under § 107(a) without any establishment of liability on its part to a third party, including the government. Atl. Research Corp., 551 U.S. at 139, 127 S.Ct. 2331.

In short, the Court summarized the distinction as follows:

[T]he remedies available in §§ 107(a) and 113(f) complement each other by providing causes of action to persons in different procedural circumstances.... Section 113(f)(1) authorizes a contribution action to PRPs with common liability stemming from an action instituted under ... § 107(a). And § 107(a) permits cost recovery (as distinct from contribution) by a private party that has itself incurred cleanup costs.

Id. (citations omitted) (internal quotations omitted). Despite this clarification, navigating the interplay between § 107(a) and § 113(f) remains a deeply difficult task. See New York v. Solvent Chem. Co., Inc., No. 83-CV-1401-JTC, 685 F.Supp.2d 357, 425, 2010 WL 376328, at *67 (W.D.N.Y. Jan.26, 2010) (“[R]eeent rulings have done little to provide the lower courts with useful guidance in determining which subsection of CERCLA provides a cause of action for parties seeking reimbursement of response costs in differing situations.”); see also id. at 422, 2010 WL 376328 at *64 (“Perhaps most perplexing is the interplay between the two cost recovery provisions which this court must apply in resolving the difficult factual and legal issues.”). The sometimes blurry relationship between § 107(a) cost recovery claims and § 113(f) contribution claims is a theme to which we will return repeatedly in this opinion.

B. Issues on Appeal

The issues raised by Carpenter on appeal can be grouped into six overarching contentions.

The first is that plaintiffs Cytec, Ford, SPS, and TI cannot bring a § 113(f) contribution claim for the approximately $7 million they paid to the EPA under the OU-2 Consent Decree to reimburse the EPA for its response costs because the EPA’s suit to enforce the OU-2 Consent Decree was time-barred. According to Carpenter, it does not share a common liability with those plaintiffs because none of the plaintiffs could have been sued by the EPA at the time they chose to settle with the agency.

Second, Carpenter argues that Agere and TI cannot bring a § 107(a) cost recovery claim for sums they paid pursuant to private settlements with other plaintiffs, as opposed to sums paid to the EPA directly, because § 107(a) does not allow for the recovery of such payments.

Third, Carpenter argues that those plaintiffs who signed the OU-1 and OU-2 Consent Decrees — Cytec, Ford, and SPS — as well as TI, who signed the OU-2 Consent Decree only, do not have § 107(a) cost recovery claims for costs expended pursuant to the OU-1 and OU-2 Consent Decrees because, once a party has been sued or has settled under CERCLA, that party may pursue only a § 113(f) contribution claim. Thus, Carpenter contends, those plaintiffs have nothing but § 113(f) available as a vehicle for trying to recover the amounts they expended to comply with the OU-1 and OU-2 Consent Decrees.

Fourth, recognizing that the District Court proceeded under § 113(f) with regard to those claims, and not under § 107(a), Carpenter argues that Cytec, Ford, SPS, and TI failed to present evidence of the total volume of waste disposed of at the Boarhead Site and, therefore, the District Court’s allocation of liability among the parties under § 113(f) cannot stand. Specifically, Carpenter argues that, other than the stipulation that it refused to sign, the plaintiffs have failed to introduce any evidence of the volume of waste for which Cytec, Ford, SPS, and H & H are responsible. Without evidence as to those parties, says Carpenter, it is impossible to know the total volume of waste disposed of at the Boarhead Site.

Fifth, Carpenter argues that the District Court erred in three ways in its cost allocation. First, Carpenter contends that the Court miscalculated the volumetric share of waste that former defendant NRM sent to the Boarhead Site. Next, Carpenter says that the Court did not consider other defendants’ culpability or lack of cooperation with the EPA to the same extent that it considered Carpenter’s lack of cooperation and culpability. As its last cost allocation point, Carpenter contends that the District Court wrongly considered the settling defendants’ volumetric share of waste, rather than the dollar amounts that the plaintiffs obtained from those settling defendants, when the Court performed its equitable allocation.

Sixth and finally, Carpenter argues that, since § 702(a)(3) of the HSCA mirrors § 107(a)(4)(B) of CERCLA, the plaintiffs do not have a viable claim under that section of the HSCA, for the same reasons that Carpenter believes the plaintiffs do not have a claim under § 107 of CERCLA.

C. Cytec, Ford, SPS, and TPs § 113(f) Claim for Reimbursement of Payments Made to the EPA for Past Costs

i. Background

The OU-2 Consent Decree obligated the plaintiffs who signed it — Cytec, Ford, SPS, and TI — to reimburse the EPA for a percentage of past response costs that the EPA had incurred in various removal efforts at the Boarhead Site, including work related to the OU-1 Consent Decree. Specifically, pursuant to the OU-2 Consent Decree, Cytec, Ford, SPS, and TI agreed, among other things, to reimburse the EPA for approximately $7 million in response costs incurred and accounted for prior to July 2000. In their fifth amended complaint, those four plaintiffs asserted a contribution claim under § 113(f), seeking reimbursement from Carpenter for that $7 million.

On appeal, Carpenter argues that the plaintiffs cannot recover the $7 million because there was no “common liability” among Carpenter and the plaintiffs at the time the plaintiffs settled with the EPA, as required by Atlantic Research. More particularly, Carpenter contends that no common liability exists because, when the EPA brought suit against the plaintiffs on December 6, 2001 to enforce the OU-2 Consent Decree, that suit, and therefore any hypothetical suit against Carpenter, was time-barred. The plaintiffs, of course, say that Carpenter is mistaken about the statute of limitations.

ii. The Statute of Limitations to Recover Past Costs

Contribution claims under § 113(f) require a “common liability” among PRPs at the time the underlying claim is re solved. See Atl. Research Corp., 551 U.S. at 139, 127 S.Ct. 2331 (“[A] PRP’s right to contribution under § 113(f)(1) is contingent upon ... common liability among liable parties.”); In re Reading Co., 115 F.3d 1111, 1124 (3d Cir.1997) (“Contribution, by its own definition, requires a common liability for the same injury.”), abrogation on other grounds recognized by E.I. DuPont De Nemours & Co. v. United States, 460 F.3d 515, 518 (3d Cir.2006). Thus, Carpenter correctly asserts that, in order for it to be liable under § 113(f), it must have had a “common liability” with the plaintiffs at the time the EPA filed suit to enforce the OU-2 Consent Decree. In other words, the EPA claim must have been viable against both the plaintiffs and Carpenter at the time the EPA sought to enforce the OU-2 Consent Decree; otherwise, there is no “ ‘common liability’ to serve as the basis of a contribution action.” (Appellant’s Op. Br. at 14.) Hence, this contention boils down to a question about the applicable statute of limitations.

The timeliness of an EPA suit to recover costs associated with a removal action is governed by § 113(g)(2), which requires that any such suit be brought “within 3 years after completion of the removal action.” 42 U.S.C. § 9613(g)(2)(A). At the same time, the statute provides for three exceptions that extend the limitations period. Before addressing those, however, we must describe more fully Carpenter’s statute of limitations argument and an essential flaw in the District Court’s decision.

Carpenter argues that the EPA’s claim is time-barred because the EPA completed its removal action on November 18, 1998, but did not initiate its lawsuit against the plaintiffs until December 6, 2001, more than three years after the completion of the removal action. As Carpenter sees it, since none of the exceptions to the three-year statute of limitations apply, the EPA’s lawsuit against the plaintiffs was untimely, and, if Carpenter had been named as a defendant, the EPA’s claims against it would have been untimely too. Carpenter has rightly identified the completion of the removal action as the point from which the statute of limitations begins to run. The District Court held that the EPA “conducted a single ‘removal action’ which ended, at the earliest, on November 18, 1998, when EPA issued the ROD.” (App. at 1:A64.) In an accompanying footnote in its opinion, however, the Court said, “[i]t is more likely that EPA’s removal action was ongoing as of September 28, 2000.” (App. at 1:A64 n. 38.) Despite its “at the earliest” caveat when citing the November 18, 1998 date, and also despite its comment on the “more likely” course of events, the District Court went on to say that, in order to give Carpenter every reasonable consideration, it would “select! ]” the November 18, 1998 date as the date that the EPA completed the removal action. (Id.)

Given the magnitude and complexity of the record, as well as the decades of EPA involvement at the Boarhead Site, we can appreciate the difficulty in determining the time frame within which the EPA completed its removal action. Nevertheless, by stating that it was “more likely” that removal work was ongoing on September 28, 2000, but then “selectfing]” the earlier date of November 18, 1998 as the date that work was completed, the Court has left a legally significant ambiguity in the record. There is no clear finding of fact as to a date or even a range of dates within which the EPA’s removal work was completed.

Of course, a court may, for analytical purposes, assume facts in favor of a party to demonstrate that, even under that party’s most advantageous view of the record, legal relief is unavailable. That appears to have been the District Court’s intention here. Nevertheless, the opinion is not entirely clear and there is a strong implication that the completion date of the EPA’s removal work was later than the selected date, which, if true, is of real importance in this case. If the removal work was ongoing as of September 28, 2000, as indicated by the Court, then the applicability of complicated exceptions to the three-year statute of limitations becomes irrelevant, and the EPA’s suit was timely regardless of those exceptions.

Because the District Court has not made a factual finding sufficient to determine whether or not the EPA’s December 6, 2001 enforcement suit for the OU-2 Consent Decree was timely under the three-year statute of limitations, we will vacate the judgment and remand for the Court to make an unequivocal finding as to the date or time period within which the EPA completed its removal action, if such a finding is possible. 26 To the extent the District Court needs to reopen the record in order to resolve this ambiguity, it may do so, since Carpenter’s statute of limitations defense makes the date on which the EPA completed its removal action a potentially claim-dispositive fact. 27

iii. The Exceptions to the Three-Year StaUite of Limitations Period

As the District Court rightly recognized, the EPA’s suit to recover past costs could still be timely even if the earlier November 18, 1998 date is correct, assuming that an exception to the statute of limitations applies. Though there are three exceptions generally available, the only one possibly of consequence is found in 42 U.S.C. § 9613(g)(2)(B), which is the one the District Court applied. 28

Section 9613(g)(2)(B) states that removal costs may be recovered as part of a remedial action suit, which effectively extends the statute of limitations to six years from the initiation of physical on-site construction of the remedial action. However, to recover such costs, the remedial action suit must be initiated within three years after the completion of the removal action. 29 Id. In other words, the limitations period may be extended if the EPA attempts to recover removal costs during a remedial action suit, rather than in a removal action suit. See id. But, if the EPA seeks to do so, it must bring that remedial action suit “within 6 years after [the] initiation of physical on-site construction of the remedial action,” and the remedial action must be initiated within three years after the completion of the removal action. Id. Given the language of § 9613(g)(2)(B), we read the requirement that remedial action (i.e., remedial work at the site) be “initiated” within three years of completion of the removal action (i.e., removal work at the site) to mean that the physical on-site construction of the remedial action has to be initiated within that three-year time period. As Carpenter notes, any other interpretation of that language would create two different meanings of the word “initiate” in § 9613(g)(2)(B), and, more importantly would eviscerate the three-year statute of limitations period for removal actions contained in § 9613(g)(2)(A). 30 Thus, whether the § 9613(g)(2)(B) exception applies in the present case is again dependent upon the time that the EPA completed its removal work.

Carpenter argues that the EPA completed its removal action on November 18, 1998, when it issued the ROD, because case law supports the idea that, for statute of limitations purposes, a removal action is complete upon the issuance of a ROD. (See Appellant’s Reply Br. at 9 (citing, inter alia, Kelley v. E.I. DuPont de Nemours & Co., 17 F.3d 836, 843-44 (6th Cir.1994)).) Plaintiffs, however, note that in the OU-1 Consent Decree, filed September 28, 2000, the EPA stated that it “conducted [a] removal action at the Site beginning in the year 1992 and continuing to the present. ...” (App. at 5:A2132, ¶ G.) Thus, say the plaintiffs, the EPA has essentially affirmed that its removal action was ongoing as of September 28, 2000.

Again, as with the effect of the three-year statute of limitations, we cannot tell whether this exception applies because the District Court has not made a definitive factual finding as to when the EPA completed its removal action. Thus, we cannot determine whether the EPA initiated on-site physical construction of the remedial action within three years of the completion of its removal action.

In sum, the EPA’s lawsuit to recover past costs was timely if one of two circumstances is found to exist: either the date or time period that the EPA completed its removal action fell within three years of the EPA’s OU-2 enforcement suit, or the EPA initiated on-site physical construction of the remedial action within three years of the completion of its removal action. If the first of these requirements is met, the suit was timely under the three-year statute of limitations. If the second is met, the six-year statute of limitations exception contained in § 113(g)(2)(b) would apply, allowing the filing to be timely.

D. TI and Agere’s § 107(a) Claims to Recover Costs Paid to Other Plaintiffs Pursuant to Settlement Agreements

Agere was not a party to either the OU-1 or the OU-2 Consent Decrees. However, Agere entered into a private settlement agreement with the plaintiffs who signed the OU-1 Consent Decree and a second private settlement agreement with the plaintiffs who signed the OU-2 Consent Decree. Pursuant to those agreements, Agere contributed to group trust accounts that funded the work associated with the consent decrees. Agere’s total costs from contributing to both trust accounts were found to be $902,152.49.

Although TI was a party to the OU-2 Consent Decree, it was not a party to the OU-1 Consent Decree. However, TI entered into a private settlement agreement with the plaintiffs who did sign the OU-1 Consent Decree, and, pursuant to that agreement, TI contributed to a group trust account that funded the work required by the OU-1 Consent Decree. TI’s total costs from contributing to the OU-1 group trust account were found to be $308,961.37.

In their fifth amended complaint, Agere and TI asserted § 107(a) cost recovery claims for the amounts that they had paid pursuant to their settlement agreements with the other plaintiffs, and the District Court held that Agere and TI did indeed have such claims. On appeal, Carpenter argues that the District Court erred in that conclusion and that Agere and TI should be barred from recovering those sums. 31

ii. Section 107(a) Cost Recovery Claims

The cost recovery available under § 107(a) is founded on the statutory language providing that PRPs shall be liable for “any other necessary costs of response incurred by any other person” consistent with CERCLA. 42 U.S.C. § 9607(a)(4)(B). The Supreme Court has held that “ § 107(a) permits a PRP to recover only the costs it has ‘incurred’ in cleaning up a site.” Atl. Research Corp., 551 U.S. at 139, 127 S.Ct. 2331 (quoting 42 U.S.C. § 9607(a)(4)(B)). The Court has further explained that, “[w]hen a party pays to satisfy a settlement agreement or a court judgment, it does not incur its own costs of response. Rather, it reimburses other parties for costs that those parties incurred.” Id.

Agere and TI are seeking to recoup costs that each paid pursuant to private settlement agreements with other plaintiffs — TI with regard to the OU-1 Consent Decree and Agere with regard to both OU-1 and OU-2 Consent Decrees. Their payments under the private settlement agreements were combined into group trust accounts with money paid by the signatories to the consent decrees, and those group trust accounts were then used to fund the OU-1 and OU-2 work. Carpenter argues that, in making those payments into group trust accounts, Agere and TI did not “incur” their own costs of response. Instead, says Carpenter, Agere and TI paid into the group trust accounts only to “satisfy a settlement agreement,” and, pursuant to Atlantic Research, those payments cannot be considered “costs incurred,” as required under § 107(a).

We disagree and will affirm the District Court in allowing Agere and TI to pursue § 107(a) claims for the amounts they have contributed to trust accounts funding the OU-1 and OU-2 work. We do not think the Supreme Court intended to deprive the word “incurred” of its ordinary meaning. Agere and TI put their money in the pot right along with the money from the signers of the consent decrees. The costs they paid for were incurred at the same time as the costs incurred by the signers of the consent decrees and for the same work. Those costs were incurred in the ordinary sense that a bill one obligates oneself to pay comes due as a job gets done. While the Supreme Court in Atlantic Research did hold that § 107(a) permits a PRP to recover only costs it has “incurred,” and did suggest that costs paid pursuant to a settlement agreement are not such costs, those statements were not made in the context of payments made for on-going work.

Moreover, the Court appears to have made its statement about the unavailability of § 107(a) relief on the assumption that “a PRP that pays money to satisfy a settlement agreement ... may pursue a § 113(f) contribution.” Id. In other words, while the Court indicated that parties seeking reimbursement for settlement payments do not have a § 107(a) claim, a basic premise of that holding was that those parties do have a § 113(f) contribution claim for their settlement amounts. We do not believe that the Court intended its holding to reach a circumstance like this, where Agere and TI do not have § 113(f) contribution claims for the settlement sums because those parties were never themselves sued for those amounts by the EPA or by other PRPs. See Cooper Indus., Inc., 543 U.S. at 168, 125 S.Ct. 577 (explaining that § 113(f) authorizes contribution claims only for PRPs who have been subject to a civil action under CERCLA).

If we were to hold that Agere and TI cannot pursue § 107(a) claims for their settlement payments, they would be com pletely barred from recovering those amounts under CERCLA. To accept that outcome, one must imagine that Congress intended to penalize cooperative cleanup efforts by excluding from CERCLA’s broad recovery provisions all PRPs who, like Agere and TI, agree to come forward and assist in a cleanup even though they have not been subjected to a cost recovery suit. Such an intent is extremely unlikely, since the goal of CERCLA is “to encourage private parties to assume the financial responsibility of cleanup by allowing them to seek recovery from others.” Key Tronic Corp. v. United States, 511 U.S. 809, 819 n. 13, 114 S.Ct. 1960, 128 L.Ed.2d 797 (1994); see also W.R. Grace & Co. v. Zotos Int’l, Inc., 559 F.3d 85, 88 (2d Cir.2009) (“[I]t has now been over twenty-five years since CERCLA’s enactment, and although many of the provisions remain perplexing, the statute’s primary purposes are axiomatic: (1) to encourage the timely cleanup of hazardous waste sites; and (2) to plac[e] the cost of that [cleanup] on those responsible for creating or maintaining the hazardous condition.” (alterations in original) (citations omitted) (internal quotations omitted)).

When a company in the position of Agere and TI has not yet been sued by the EPA but appreciates that it bears some responsibility for cleaning up hazardous waste, the language of CERCLA, which is intended to encourage cleanup, ought not be interpreted to discourage participation in cleanup if a more consistent construction of the statute is plausible. Private actors are not likely to settle and step forward unless they know that they can seek some of the amounts they will contribute, just like those who have been sued by the EPA or a PRP, or those who voluntarily clean up a site in the first instance. To encourage participation in environmental cleanup, the statute should be read in a way that assures PRPs like Agere and TI that they can later bring a § 107(a) cost recovery claim for the amounts they pay to help with the cleanup, even if those costs are related to a settlement obligation. Cf. Steven Patrick, Superfund: Second Circuit Says EPA Settlement Approval Is Unnecessary to Trigger Contribution Right, 41 Daily Envtl. Rep. (BNA) A-10 (Mar. 4, 2010) (citing a Department of Justice amicus brief filed in Niagara Mohawk Pwr. Corp., stating that “[i]t is important that PRPs ... that ... engage in response activities in settlements with states have appropriate CERCLA claims for contribution against other PRPs [because otherwise PRPs will decline to enter into administrative settlements and instead wait for the filing of civil actions to ensure they can sue for contribution under Section 113(f)(1).”). Our holding is buttressed by the Supreme Court’s description of § 107(a) and § 113(f) as “overlapping” remedies. Atl. Research Corp., 551 U.S. at 139 n. 6, 127 S.Ct. 2331. It would be a stretch to describe the remedies as “overlapping” if they are actually intended to exclude an entire group of PRPs from both remedies.

Accordingly, we affirm the District Court’s holding that TI (with regard to OU-1) and Agere (with regard to OU-1 and OU-2) 32 have § 107(a) cost recovery claims to recoup costs paid for cleanup work, even though the payments were made pursuant to settlement agreements. Those claims may thus go forward on remand.

E. The District Court’s Equitable Allocation for the Plaintiffs’ Costs of Performing Work Under the OU-1 and OU-2 Consent Decrees

The District Court concluded that Cytec, Ford, and SPS had both § 107(a) and § 113(f) claims for the costs of performing work pursuant to the OU-1 and OU-2 Consent Decrees, and that TI had such claims with respect to the OU-2 Consent Decree. Carpenter argues that those plaintiffs have § 113(f) claims only, for the costs they incurred pel-forming the work required by the consent decrees. They do not have § 107(a) claims, Carpenter says, because once a party has been sued or has settled under CERCLA its sole avenue of relief is a § 113(f) claim.

ii. Sections 107(a) and 118(f)

The Supreme Court, in Atlantic Research, left open the precise question raised by Carpenter’s argument, namely, whether plaintiffs in the position of Cytec, Ford, SPS, and TI can bring a § 107(a) claim in addition to a § 113(f) claim. The Court said,

We do not suggest that §§ 107(a)(4)(B) and 113(f) have no overlap at all. For instance, we recognize that a PRP may sustain expenses pursuant to a consent decree following a suit under ... § 107(a). In such a case, the PRP does not incur costs voluntarily but does not reimburse the costs of another party. We do not decide whether these compelled costs of response are recoverable under § 113(f), § 107(a), or both. For our purposes, it suffices to demonstrate that costs incurred voluntarily are recoverable only by way of § 107(a)(4)(B), and costs of reimbursement to another person pursuant to a legal judgment or settlement are recoverable only under § 113(f). Thus, at a minimum, neither remedy swallows the other....

Id. (citations omitted); see also Key Tronic Corp. 511 U.S. at 816, 114 S.Ct. 1960 (stating that the provisions provide “similar and somewhat overlapping remedies]”). Claims by Cytec, Ford, and SPS for costs of performing the OU-1 and OU-2 work, and TI’s costs associated with OU-2, fall into the region of doubt identified by the Supreme Court in Atlantic Research. The costs were neither “incurred voluntarily,” because the parties were in fact sued by the EPA, nor were they “reimbursed to another party,” because they were expended in performing the OU-1 and OU-2 work directly. Rather, they are costs incurred pursuant to a consent decree following a CERCLA suit. See W.R. Grace & Co., 559 F.3d at 93 n. 7 (“As the Supreme Court suggested, it may well be that a party who sustains expenses pursuant to a consent decree following a suit under [CERCLA] may have a cause of action under either section 113(f), section 107(a), or both.”).

The District Court correctly recognized that the claims of Cytec, Ford, SPS, and TI associated with the consent decrees fall within an “open question of law.” (App. at 1:A59). Without explanation, the Court concluded that those plaintiffs have both § 113(f) and § 107(a) claims. Thus, while it appears that the Court went on to allocate liability under § 113(f), 33 we are squarely confronted with the issue left open in Atlantic Research: whether, in addition to § 113(f) claims, plaintiffs such as these have § 107(a) claims for expenses sustained pursuant to a consent decree following a CERCLA suit. This is not simply a matter of academic interest. Were we to find that these parties have § 107(a) claims in addition to § 113(f) claims, the District Court could, on remand, decide to allocate liability under § 107(a) and potentially impose 100 percent of the liability on Carpenter pursuant to the joint and several liability available under that provision of the statute. For that reason, the parties have heavily briefed and vigorously argued the issue.

In Atlantic Research, immediately after setting forth this unanswered question, and after suggesting that there is, in fact, some “overlap” between § 107(a) and § 113(f), the Supreme Court stated that “a defendant PRP in such a § 107(a) [joint and several liability] suit could blunt any inequitable distribution of costs by filing a § 113(f) counterclaim.” 551 U.S. at 140, 127 S.Ct. 2331. The Court explained that any fear that “PRPs will eschew equitable apportionment under § 113(f) in favor of joint and several liability under § 107(a)” is mitigated by the fact that “a § 113(f) counterclaim would necessitate the equitable apportionment of costs among the liable parties, including the PRP that filed the § 107(a) action.” Id. at 138, 40, 127 S.Ct. 2331. Unfortunately, the potential for an inequitable result remains in this ease because Carpenter would not be able to bring a § 113(f) counterclaim against Cytec, Ford, SPS, and TI. Section 113(f)(2) would prevent it. That subsection states that “[a] person who has resolved its liability to the United States or a State in an administrative or judicially approved settlement shall not be liable for claims for contribution regarding matters addressed in the settlement.” 42 U.S.C. § 9613(f)(2). Thus, if we allowed Cytec, Ford, SPS, or TI to bring a § 107(a) claim here and assert joint and several liability against Carpenter, Carpenter would be barred from then bringing a contribution counterclaim against them because those plaintiffs have entered into consent decrees, i.e., judicially approved settlements with the EPA. As a result, those plaintiffs would be able to recover 100 percent of their own costs against Carpenter, even though they themselves are actually responsible for, and have stipulated that they are responsible for, a significant portion of the contamination at the Boarhead Site.

As Carpenter summarized it, “[Plaintiffs who had entered into consent decrees with the EPA] would not be subject to equitable allocation. They would have no liability because they would be able to assert joint and several liability against the other parties.” (Oral Argument Tr. at 9-10.) This is a perverse result, since a primary goal of CERCLA is to make polluters pay. Indeed, the Supreme Court expressed concern about this very type of result when it recognized that, without defendants having the availability of a § 113(f) counterclaim, PRP plaintiffs might “eschew equitable apportionment ... in favor of joint and several liability under § 107(a).” Atl. Research Corp., 551 U.S. at 138, 127 S.Ct. 2331. The Court’s explanation that the harshness of joint and several liability is mitigated by the possibility of a § 113(f) counterclaim is of no comfort to defendants like Carpenter.

Further, even under joint and several liability, a district court is not supposed to fashion an award that allows a plaintiff to recover from a defendant costs associated with the cleanup of waste that the plaintiff itself has contributed to the site. See The Uniform Comparative Fault Act § 2 cmt. [hereinafter “U.C.F.A.”] (indicating that even when imposing joint and several liability, courts must nevertheless “set forth ... the equitable share of the total obligation to the claimant for each party, based on his established percentage of fault ... [so that] each party should eventually be responsible for [that amount] as a result of the rules of contribution.”). That is to say, while joint and several liability allows a plaintiff to collect from a single defendant the collective liability of all defendants, it does not permit a plaintiff to recover from a defendant the costs to undo what the plaintiff itself has done. Yet this is precisely what would occur if Carpenter were barred from bringing a contribution counterclaim against the plaintiffs after the plaintiffs recovered from Carpenter under joint and several liability.

Thus, we hold that plaintiffs in the position of Cytec, Ford, SPS, and TI, who if permitted to bring a § 107(a) claim would be shielded from contribution counterclaims under § 113(f)(2), do not have any § 107(a) claims for costs incurred pursuant to consent decrees in a CERCLA suit. Cf. Niagara Mohawk Pwr. Corp., 596 F.3d at 127 (holding that a PRP who had settled its CERCLA liability by consent order with a state environmental agency had a § 113(f)(3)(B) claim but not a § 107(a) claim, and saying, “[cjlearly, the two sections have differing restrictions and different purposes.”). We need not decide the contours of the overlap postulated in Atlantic Research because, regardless of whether § 107(a) and § 113(f) remedies overlap at all, they cannot properly be seen to overlap here.

iii. The District Court’s Equitable Allocation Under § 113(f)

As discussed above, Cytec, Ford, and SPS, have only § 113(f) claims for costs they incurred pursuant to the OU-1 and OU-2 Consent Decrees, and TI is in the same position with its OU-2 costs. Because the District Court resolved those claims under § 113(f) “by determining the equitable shares of [all of the parties]” (App. at 1:A61), we must now address the Court’s allocation of cleanup costs.

The plaintiffs concede that they had to “put on evidence sufficient for the District Court to determine which equitable factors were appropriate and to apply those factors to the facts to determine each party’s share.” (Appellees’ Ans. Br. at 38.) At trial, the “[plaintiffs’ theory of the case was that the volume of each party’s waste disposed of at the Site was the most appropriate equitable factor for allocation.” (Id.) The District Court adopted that theory, holding that the “[v]olume of waste disposed at the Site is the most equitable manner of cost allocation and the Court is not hesitant to use this as its primary equitable factor.” (App. at 1:A70-A71.) The Court further held that, “because culpability and cooperation are significant equitable considerations, they will be considered in the allocation equation.” (App. at 1A71.)

On appeal, Carpenter argues that the only evidence the plaintiffs introduced as to the volume of waste attributable to parties other than NRM and Carpenter consisted of the following three stipulations: (1) the June 19, 2008 stipulation among all of the parties as to the waste that DCC collected from fifteen of the defendants (not including Carpenter’s, NRM’s, or H & H’s waste volumes); (2) the June 23, 2008 stipulation between the plaintiffs and H & H (that Carpenter did not sign) as to the volume of waste that DCC collected from plaintiffs Ford, Cytec, and SPS, and from former defendant H & H; and (3) the July 1, 2008 stipulation between the plaintiffs, Carpenter, and H & H, stipulating to the fact that the waste described in the June 23rd stipulation was CERCLA hazardous waste. Carpenter asserts that the June 23rd “stipulation is not competent evidence of the volumes of waste that DCC acquired from [the] plaintiffs and [from] H & H, because Carpenter, the only party against whom it [is] being used, did not stipulate to those volumes.” 34 (Appellant’s Op. Br. at 43.) Thus, the argument runs, “if you don’t have the plaintiffs’ volume ... you don’t know the total volume at this site. If you don’t know the total volume at the site, you can’t allocate between all the parties.” (Oral Argument Tr. at 17.)

The plaintiffs, in response, argue that they offered sufficient evidence for the Court to determine the volume of waste that each party left at the site since they “offerfed] into evidence the stipulations of all parties to the volumes of waste hauled by DCC from each [p]laintiff and each [defendant (other than Carpenter and NRM).... ” (Appellee’s Ans. Br. at 39.) The plaintiffs contend that they “thus had only to offer additional evidence sufficient for the District Court to determine the volumes of Carpenter and NRM waste that were disposed of at the Site, thereby finding all of the facts necessary to apply its primary equitable factor [i.e., waste volume].” {Id. at 40.) Specifically, the plaintiffs argue that “[a]ll of these stipulations were properly admitted into the record, whether or not Carpenter was an original signatory to the June 23rd stipulation.” {Id. at 42-43 n. 14.)

Forced to confront the fact that Carpenter did not sign the June 23, 2008 stipulation, and thus never stipulated to the plaintiffs’ and H & H’s waste volumes, the plaintiffs advance the following theory on appeal:

Plaintiffs and H & H stipulated on June 23, 2008 as to the nature and volume of H & H waste disposed of at the Site and the volumes of waste[ ] of Plaintiffs that were hauled by DCC. Then Plaintiffs, H & H, and Carpenter entered into a July 1, 2008 stipulation clarifying that the June 23rd stipulation was intended (by referencing no volumes for TI and Agere) to include the conclusion that no waste from TI and Agere was hauled by DCC, and that the wastes of the other three Plaintiffs referenced in the June 23rd stipulation ... contained CERCLA hazardous substances. By executing the second [July 1st] stipulation, Carpenter also joined in the first one as well”

(Id. (internal citations omitted) (emphasis added).) The plaintiffs thus argue that, because the July 1st stipulation refers to the June 23rd stipulation, the later one was “intended” to incorporate the earlier. 35 (Id.)

The District Court accepted that theory, stating that “[although Carpenter did not enter into [the June 23rd stipulation], it is nonetheless evidence of plaintiffs volumetric shares.” (App. at 1:A71 n. 43.) From the Court’s opinion, it appears that the stipulations were the only evidence the Court considered with regard to the plaintiffs’ and H & H’s waste. All other volume evidence discussed by the District Court relates to Carpenter’s or NRM’s waste.

iv. The June 23rd Stipulation is Not an Admission that is Admissible Against Carpenter

The June 23rd stipulation is the only one that included the plaintiffs’ and H & H’s volumes of waste, and Carpenter did not sign it. The July 1, 2008 stipulation does not incorporate by reference the June 23, 2008 stipulation, nor does it even address the volumes of waste in the earlier stipulation. It reads, in relevant part, that “[t]he waste from Plaintiff companies referenced in paragraph 5 of the June 23, 2008[s]tipulation ... contained at least some CERCLA hazardous substances. There were no wastes hauled by [DCC] from [TI, Agere,] or their predecessors.” (App. at 14:A6398.) When plaintiffs moved to admit the June 23rd stipulation at trial, Carpenter pointedly objected, as it had before, to any substantive use of the stipulation, stating “[w]e’re just making a record that we do not agree or stipulate or we’re not waiving any kind of objection to paragraph 5 of [the June 23rd] stipulation.” (App. at 3A1199; see also supra note 34.)

It would be extraordinary, in light of Carpenter’s decision not to sign the June 23rd stipulation and its repeated and emphatic objections, to allow the use of the stipulation to bind it, and to permit the stipulation to become exactly what Car penter had a right to say it was not, namely evidence against it of the volumes of waste at issue. The District Court should not have decided the plaintiffs’ and H & H’s volume of waste from a stipulation that Carpenter did not join. Particularly in a case like this, where the question is who bears what share of the responsibility among those who are culpable, it is no mere technical violation of the rules of evidence to allow some parties to stipulate to their share of responsibility and then later have that stipulation be used as the basis of liability against another party. No matter how culpable a polluter may be — and in this instance, Carpenter gave the District Court ample reason to view it as highly culpable — liability can only be assigned on the basis of procedures consistent with the Due Process Clause of the Fifth Amendment of the United States Constitution and the Federal Rules of Evidence. Cf. Dawson v. Delaware, 503 U.S. 159, 178, 112 S.Ct. 1093, 117 L.Ed.2d 309 (1992) (“The Due Process Clause ... traditionally has regulated questions about the improper admission of evidence.”).

Before us, plaintiffs’ counsel argued that the stipulation was “properly admitted into evidence” because it constitutes an admission by the plaintiffs. 36 This, of course, misapprehends the issue entirely. The question isn’t whether the stipulation is an admission by the plaintiffs and therefore binding as to them. The question is whether the plaintiffs’ self-serving stipulation can bind Carpenter.

Taking a stipulation that a party has chosen not to sign and using it as evidence against the non-signatory violates at least the rule against hearsay. In general, hearsay “is a statement, other than one made by the declarant while testifying at the trial or hearing, offered in evidence to prove the truth of the matter asserted.” Fed.R.Evid. 801(c). Hearsay is not admissible unless it falls under one of the exceptions to the hearsay rule contained in the Federal Rules of Evidence. Fed.R.Evid. 802. One thing which the rules define as non-hearsay is an admission by a party-opponent, defined as a “party’s own statement, in either an individual or representative capacity” that is “offered against” that party. Fed.R.Evid. 801(d)(2)(A). Thus, whether a statement constitutes inadmissible hearsay, or is instead considered not to be hearsay at all because it is an admission of a party-opponent, depends on both who made the statement and whom it is being offered against. Here, while the stipulation might well be admissible non-hearsay if it were being offered against the parties who signed it, it should not have been admitted against Carpenter, since Carpenter did not sign it and no exception to the hearsay rule was advanced by the plaintiffs. 37 Cf. Brzozowski v. Corr. Physician Servs., Inc., 360 F.3d 173, 179 (3d Cir.2004) (cautioning that a plaintiff could not rely on a consent judgment against a party because that party did not sign the “stipulation ... and must be afforded the opportunity to defend itself against the claim de novo.”).

Because the June 23rd stipulation should not have been admitted against Carpenter, the District Court did not rely on competent evidence as to four of the parties’ waste volumes. As a result, Carpenter is correct in its assertion that the Court did not articulate an adequate basis for its decision as to the total volume of waste at the Boarhead Site. The plaintiffs concede, as they must, that it was their burden to prove each party’s “fair share” or “size of the pie.” (Oral Argument Tr. at 24-25.) They did not meet that burden, and, without a mention of any other evidence of the matters covered by the June 23rd stipulation, it was impossible for the District Court to perform an equitable allocation of waste volumes among Carpenter and the other PRPs. Thus, the District Court’s volume allocation constitutes an abuse of discretion and cannot stand.

v. Other Evidence Regarding Waste Volumes

That conclusion, however, does not end the matter. While the stipulations are inadmissible unless a hearsay exception applies, there may well be other evidence of record that could be dispositive as to plaintiffs’ and H & H’s waste volumes. For example, in a footnote, the District Court indicated that, because the plaintiffs stated their volumes of waste and other parties’ volumes of waste in answers to contention interrogatories, and because Carpenter moved those answers into evidence, “these answers corroborate the volumes of plaintiffs’ waste disposed of at the Site.” (App. at 1:A71 n. 43; see also App. at 14:A6416-6498; 6485.) Carpenter concedes that it did, indeed, introduce into evidence plaintiffs’ responses to contention interrogatories. (See Appellant’s Reply Br. at 25.) Moreover, in another part of its argument — concerning alleged error by the District Court in determining defendant NRM’s waste volumes — Carpenter relied, and continues to rely, on those very same responses to argue its own point. 38 (See App. at 1:A39 n. 21; Appellant’s Reply Br. at 25.) It thus may be that the answers to contention interrogatories constitute admissions that Carpenter has adopted. 39

The District Court, which is far more familiar with the context in which Carpen ter moved to admit those responses, and the ways in which Carpenter argued their relevance, is in a better position to consider that issue. The Court may also decide that, in the unusual circumstances of this case, the record will need to be reopened to consider the equitable allocation again, and, in that context, it may review any admissible evidence as to the plaintiffs’ and H & H’s waste volumes. Since it may be necessary to reopen the record to address other questions, like the time period during which the EPA completed its removal action, issues of waste volumes could be addressed at the same time.

F. Other Contentions Regarding the District Court’s Equitable Allocation

Carpenter contends that, in addition to the improper admission of the June 23rd stipulation as volume evidence for the plaintiffs and H & H, the District Court erred in its equitable allocation analysis in three other ways. First, Carpenter argues that the Court miscalculated the volumetric share of waste that former defendant NRM sent to the Boarhead Site. Second, it argues that the District Court did not consider other defendants’ culpability and lack of cooperation with the EPA to the same extent it considered Carpenter’s behavior. Finally, Carpenter says that, in allocating liability, the Court should have considered the dollar amounts that the plaintiffs obtained from settling defendants, rather than those defendants’ volumetric share of waste. Each of those contentions is unpersuasive.

i. NRM’s Waste

Because the parties did not stipulate to NRM’s waste, the Court heard testimony from several DCC drivers, all of whom were questioned about transporting NRM’s waste to the Boarhead Site. The testimony produced the following evidence. One driver picked up waste from NRM but transported all of it to another location, not the Boarhead Site. A second driver never went to NRM to pick up waste. Two other drivers transported all NRM waste to a third location, not the Boarhead Site. June Stephens, another driver, was unclear as to whether she had driven a truck containing NRM waste, and, if she had, whether or not she had disposed of that waste at the Boarhead Site. Finally, a driver named Manfred (“Freddie”) DeRewal, Jr. testified that he took six to ten tankers of NRM waste to the Boarhead site, but took the remainder of NRM’s waste to other locations.

Based on that testimony, the District Court found that 32,000 gallons of NRM’s waste was disposed of at the Boarhead Site. According to the Court,

Freddie DeRewal took six to ten tankers of NRM waste to the [Site].... As a tanker truck held approximately 4,000 gallons of waste, DCC disposed of between 24,000 and 40,000 gallons of NRM waste at the Site.... The Court finds that DCC disposed of 32,000 gallons of NRM waste at the Site during the Gap Period.

(App. at 1:A36.) The District Court thus chose the halfway point in the 24,000 and 40,000 gallon range.

Carpenter offers two specific contentions with regard to the NRM waste figure. First, it argues that the plaintiffs gave more definite figures in response to contention interrogatories and that the Court should have used those figures. Second, it argues that June Stephens’s testimony demonstrates that she did, in fact, collect waste from NRM. As to the first point, there was nothing improper in the District Court’s decision to give greater weight to witness testimony than to contention interrogatory answers. As to the second, Stephens stated explicitly that, instead of the waste she collected being NRM’s, it “could have been [waste belonging to] another one of these outfits.” (App. at 15A878.) Given the indecisiveness of Stephens’s testimony, the District Court did not clearly err in interpreting her testimony the way that it did. Considering the testimony before it regarding the handling of NRM’s waste, the District Court could properly determine that 32,-000 gallons of waste were fairly attributable to NRM.

ii. The Culpability and Lack of Cooperation of Settling Defendants

The District Court concluded that, in addition to volume, culpability and cooperation with the government should also be considered in allocating response costs. Because “Carpenter relinquished its potent waste acids to a known polluter,” (App. at 1:A72), and because Carpenter did not cooperate with the EPA, the Court decided to allocate to Carpenter an additional 17.4% of the total clean-up cost: 8.7% for working with a known polluter and 8.7% for not cooperating with the EPA. (App. at LA73-A74.)

Carpenter does not dispute the District Court’s findings of fact concerning its repeated business dealings with DCC and related companies or the refusal to participate in cleaning up the Boarhead Site when asked to do so by the EPA. Rather, Carpenter contends on appeal that the District Court erred in ignoring the same kind of evidence with regard to other parties.

We disagree. The District Court found that Carpenter’s culpability was of a different character than any other defendant’s because Carpenter had actual knowledge of DeRewal’s polluting activities and yet twice chose to hire his companies to dispose of toxic waste. Carpenter introduced no evidence demonstrating that any other party acted with the same callous disregard as it did in its dealings with DeRewal’s companies or with the EPA. The record reveals that Carpenter was particularly contemptuous of the law and public safety in the way it stonewalled participation in the government’s cleanup efforts at the Boarhead Site. These are valid bases for viewing Carpenter as occupying a category of culpability all its own in this case. Thus, the District Court’s decision to allocate an additional 17.4% of liability to Carpenter was certainly not an abuse of discretion, and the increased liability need not be revisited on remand.

iii. Settlement Amounts

The District Court chose to use the waste volumes of the settling defendants, rather than the dollar amounts that those settling defendants paid, as the primary equitable factor for its allocation. The District Court explained that it was adopting the U.C.F.A. to determine the settling parties’ shares of liability. That Act states that “[t]he claim of the releasing person against other persons is reduced by the released person’s equitable share of the obligation----” U.C.F.A. § 6. Thus, the Court concluded that, since it had used waste volumes to determine the equitable shares of the remaining PRPs, the settling defendants’ liability should be “reduced by the settling party’s equitable share of the liability, not by the dollar value of the settlement.” (App. at 1:A71 n. 43.) The District Court had in fact held four years before trial that “the liability of the Non-Settlers [will be] determined without regard to the dollar amounts of previous settlements.” (App. at 1:A85.)

Carpenter argues that the Court erred in concluding that the dollar value of the settlements were not relevant to the equitable allocation. But the District Court did not err. First, practically speaking, the settlement amounts are not of record and so the District Court could not have used them as an equitable factor. Second, as the District Court noted, the plaintiffs have every incentive to settle for as close to the settling parties’ shares as possible because the plaintiffs bear the risk that those parties’ shares of the cleanup costs may be greater than the settlement amount. Thus, the volume allocation likely reflects the dollar amounts. Finally, to the extent waste volumes and settlement dollars diverge, it was well within the Court’s discretion to determine that the former, which are historical facts developed before litigation, are a better measure of fault than the latter, which are necessarily colored by litigation concerns. In short, while the District Court must revisit certain PRPs’ volumes for the reasons described above, there was nothing wrong with its choosing to use volume, rather than the dollar value of the settlements, as the basis for allocating liability among the parties.

G. The Pennsylvania Hazardous Sites Cleanup Act

Lastly, Carpenter argues that plaintiffs Cytec, Ford, SPS, and TI do not have viable cost recovery claims under § 702(a) of the HSCA for the same reasons that they do not have a § 107(a) claim under CERCLA. Neither party disputes that liability under the HSCA mirrors liability under CERCLA and that § 702(a) of the HSCA mirrors § 107(a) of CERCLA. Rather, Carpenter simply repeats its contentions as to why it believes the District Court erred in holding that Cytec, Ford, SPS, and TI have § 107(a) claims for the costs they incurred under the OU-1 and/or OU-2 Consent Decrees.

The District Court correctly held that Carpenter’s liability “is neither greater nor lesser under the HSCA.” (App. at 1:A76.) Indeed, the cost recovery and contribution provisions in HSCA are virtually identical to those in CERCLA. Compare 35 Pa. Stat. Ann. §§ 6020.702(a)(3), 705(a), with 42 U.S.C. §§ 9607(a)(4)(B), 9613(f). Thus, on remand, the District Court should continue to address the CERCLA and HSCA issues in this case identically. 40

IV. Conclusion

We recognize the many difficulties inherent in adjudicating this highly complex and contentious case and express appreciation for the clarity with which the District Court set forth its reasoning in support of the equitable allocation it determined. However, for the reasons discussed above, we must vacate the judgment and remand the case.

In summary, first, we ask the District Court to make a clear and unequivocal finding, if possible, as to when the EPA completed its removal action. To the extent the District Court decides it must reopen the record in order to make that finding, it may do so. Such a finding will allow the District Court to determine whether the EPA’s December 6, 2001 filing of the suit to enforce the OU-2 Consent Decree was timely, and thus whether there is a time-bar to plaintiffs’ recovering the approximately $7 million they paid to reimburse the EPA for past costs. Further, if the District Court finds that the EPA initiated “on-site physical construction” of the remedial action within three years of the completion of its removal action, it may apply the six-year statute of limitations exception contained in § 113(g)(2)(B). If the District Court finds that the EPA’s December 6, 2001 suit was not time-barred, and that the plaintiffs are able to recover for past costs, the Court should make a more exact finding as the amount that the plaintiffs paid to reimburse the EPA for past costs.

Second, the District Court should permit TI (with regard to OU-1) and Agere 41 (with regard to OU-1 and OU-2) to go forward with their § 107(a) cost recovery claims to recoup costs paid as part of the shared expense of cleaning up the Boar-head Site.

Third, because we hold that Cytec, Ford, and SPS, as well as TI (with regard to OU-2), are shielded from contribution counterclaims under § 113(f)(2) and therefore do not have § 107(a) claims for costs incurred pursuant to the consent decrees, the District Court should again proceed solely under § 113(f) as to those claims.

Fourth, while proceeding under § 113(f) to allocate liability among the parties, the District Court may not consider the June 23rd stipulation as evidence against Carpenter, at least not without addressing the evidentiary problems noted herein. Beyond the stipulation, there may be evidence in the record that the District Court can rely on to perform its equitable allocation, but the District Court may also reopen the record if it deems that necessary.

For the foregoing reasons, we vacate the District Court’s judgment and remand for further proceedings consistent with this opinion.

1 . Between 1973 and 1976, the Bucks County Department of Health filed over fifteen Waste Discharge Inspection Reports regarding the Boarhead Site. On October 15, 1976, an injunction was issued preventing DCC from bringing any chemicals to the Site. In addition to the Boarhead Site, DCC disposed of its customers’ waste at two other sites, known to the parties as the Ontario Street property and the Wissinoming Industrial Park. The former was a rental property that DCC took possession of on November 15, 1973, on Ontario Street in Philadelphia, Pennsylvania. On June 13, 1975, the Philadelphia Water Department sealed the lateral connection between the Ontario Street property and the city sewer system after the Department discovered that DCC was improperly disposing of chemical waste into the city sewer system from the property. On May 6, 1976, the owner of DCC leased another site, through its subsidiary, located in the Wissinoming Industrial Park in Philadelphia. DCC eventually ceased use of that site as a result of concerns raised by the Philadelphia Water Department, the Philadelphia Police Department, and the EPA. Those two disposal sites are not the focus of the present appeal.

2 . "The NPL [National Priorities List] is the list of priority releases for long-term remedial evaluation and response.” 40 C.F.R. § 300.425(b).

3 . Superfund sites are those hazardous waste sites listed on the EPA's National Priorities List. See U.S. Environmental Protection Agency, Superfund, http://www.epa.gov/superfund/ index.htm (last visited March 5, 2010).

4 . In general, when we use the term "action” throughout this opinion, we are using it to refer to environmental cleanup work, and not as a synonym for a lawsuit.

5 . See infra note 29 regarding the definitions of remedial actions and removal actions.

6 . "A ROD provides the justification for the remedial action (treatment) chosen at a Su perfund site. It also contains site history, site description, site characteristics, community participation, enforcement activities, past and present activities, contaminated media, the contaminants present, scope and role of response action and the remedy selected for cleanup.” U.S. Environmental Protection Agency, Superfund Information Systems, Record of Decision System, http:Avww.epa.gov/ superfund/sites/rods/ (last visited Dec. 30, 2009).

7 . OU-1 addressed groundwater extraction, metal precipitation, air stripping, the installation of wells, residential water treatment, phytoremediation, and the installation of institutional controls and monitoring for OU-1. OU-2 addressed soil aeration and treatment of volatile organic compound hotspots, the excavation and off-site disposal of buried drums, and the implementation of institutional controls and monitoring for OU-2.

8 . As more fully discussed herein, § 107 authorizes the United States or a state or "any other person” to seek reimbursement for removal or remedial costs incurred in responding to releases of hazardous substances as defined by CERCLA, provided that those actions are consistent with the national contingency plan. See 42 U.S.C. § 9607(a). The national contingency plan has been called "the federal government’s roadmap for responding to the release of hazardous substances.” Niagara Mohawk Pwr. Corp. v. Chevron U.S.A., Inc., 596 F.3d 112, 121 (2d Cir.2010).

9 . Plaintiffs Cytec, Ford and SPS are also signatories to an EPA Administrative Order for Consent for Remedial Design executed in February 2000.

10 . Plaintiffs Cytec, Ford, SPS, and TI are also signatories to an EPA Administrative Order for Consent for Remedial Design, executed on October 17, 2001.

11 . The significance of July 2000 as a pivotal point in the timing of EPA's costs and calculations is not clear from the record.

12 . While none of the parties has raised the issue, we are left to wonder whether Agere has standing to bring the present suit against Carpenter after having assigned to others its right to recover the amounts it paid into the OU-1 and OU-2 group trust accounts. Whether or not Agere has standing, however, we refer to the claims which originated with it as being "Agere’s claims,” since, whether held by Agere or assignees, the claims themselves remain. On remand, the District Court should determine who holds these claims and, specifically, to what extent, if any, Agere retains standing. This opinion and accompanying judgment order are, obviously, to be underslood as permitting recovery by the parties in interest only.

13 . The basis for that exclusion has not been explained on appeal.

14 . The original defendants included the following parties: Advanced Environmental Technology Corporation; Ashland Chemical Company; Boarhead Corporation; Carpenter; Crown Metro, Inc.; Diaz Chemical Corporation; Etched Circuits, Inc.; fcg, Inc.; Globe Disposal Co., Inc.; Handy & Harman Tube Company, Inc.; Knoll, Inc.; Merit Metals Products Corporation; Novartis Corporation; NRM Investment Company; Plymouth Tube Company; Quickline Design and Manufacturing Company; Rahns Specialty Metals, Inc; Rohm and Haas Company; Simon Wrecking Co., Inc.; Techalloy Co., Inc.; Thomas & Betts Corporation; Unisys Corporation; and the United States of America Department of Navy.

15 . At the time of the fifth and final amended complaint, the defendants included Advanced Environmental Technology Corporation; Ash-land Chemical Company; Carpenter; Diaz Chemical Corporation; fcg, Inc; Handy & Harman Tube Company, Inc.; and NRM Investment Company.

16 . By the time of the bench trial, all of the other defendants had either settled or had been granted unopposed motions to dismiss or unopposed summary judgment motions.

17 . From this point forward, references to "the parties” are to the named plaintiffs in this suit and Carpenter, unless the context dictates otherwise.

18 . The exact date on which the parties entered into this stipulation is not entirely clear from the record. The signature page, however, suggests that the date might have been June 18, 2008. (See App. at 14:A6390.)

19 . The costs broke down as follows:

• Agere: $902,152.49

• Cytec: $3,368,551.07

• Ford: $3,354,122.22

• SPS: $3,354,122.22

• TI: $2,699,430.55

20 . The 62.6% was calculated as follows: the Court found that the volume of waste disposed of at the Boarhead Site from all of the plaintiffs, and the settled and dismissed parties, as well as Carpenter, was 1,594,668 gallons. The Court found that the total volume of waste disposed of at the Boarhead Site by Carpenter was 998,284 gallons. Thus, the Court determined that Carpenter's percentage share of the total volume was 62.6 %.

21 . The record is not clear as to what regulatory authority or authorities decided that Revere and Echo could no longer operate.

22 . We refer to the appendix on appeal by the volume number followed by a colon and the page number on which the cited material appears, followed by specific paragraphs, if applicable.

23 . Section 107(a) actually says “consistent with the national contingency plan,” see 42 U.S.C. § 9607(a), which, as noted above (supra note 8), is the federal government’s plan for responding to releases of hazardous substances. Niagara Mohawk Pwr. Corp., 596 F.3d at 121.

24 . Section 107 also provides the EPA with a mechanism to recover costs from PRPs, stating that PRPs are liable for “all costs of removal or remedial action incurred by the United States Government or a State or an Indian tribe not inconsistent with the national contingency plan.” 42 U.S.C. § 9607(a)(4)(A). This aspect of § 107 is not at issue in the present matter.

25 . Section 106 authorizes the United States to bring a suit seeking injunctive relief to abate “imminent and substantial endangermenl” resulting from the release of a hazardous substance. The section also authorizes the EPA to issue an administrative order to abate such a condition. See 42 U.S.C. § 9606; see also 4 The Law of Hazardous Waste § 14.02[l][c] (Susan M. Cooke, ed., 2009).

26 . We do not discount the possibility that the District Court did not make a definitive finding because the record would not support one. Also, if the District Court’s finding of fact turns out to be that the removal action ended sometime after September 28, 2000, that will be sufficient, since, with that finding, Cytec, Ford, SPS, and TI's claims to recover for their approximately $7 million reimbursement to the EPA for past response costs would be timely, as is more fully discussed herein. See infra Section Ill.C.iii.

27 . We also ask the District Court to make a definitive finding as to the amount that the OU-2 plaintiffs paid to reimburse the EPA for past costs. While the stipulated facts cite the amount as $7 million in one place, in another place it appears to be approximately $7.4 million, and Carpenter cites the amount as $7.4 million. (Compare App. at 14:A6379 ¶ 40, and App. at 14:A6383 ¶ 59, with Appellant’s Op. Br. at 13.)

28 . The two other exceptions to the three-year statute of limitations are of no practical effect. The first exception, the “subsequent action” exception, is simply not applicable. That exception pertains to situations in which the EPA has brought more than one legal action (and here the term "action” is being used as a synonym for "lawsuit”) against the plaintiffs to recover costs. The statute provides that if the EPA’s suit is a "subsequent action” to recover further response costs (i.e., there has already been an "initial action” to recover response costs), the suit must be commenced no later than 3 years after the date of completion of the initial action. See 42 U.S.C. § 9613(g)(2). A "subsequent” suit for further response costs can "be maintained at any time during the response action, but [it] must be commenced no later than 3 years after the date of completion of the response action.” Id. § 9613(g)(2)(B). The plaintiffs argue that the EPA’s 2001 suit was a timely subsequent action, because the three-year limitations period applies only to the EPA’s initial action, which they identify as the EPA's June 2, 2000 suit against Cytec, Ford, and SPS to enforce the OU-1 Consent Decree. Appellees contend that "[i]t does not matter that Carpenter was not a defendant in the 2000 EPA action” because the statute does not indicate that the defendant must have been a party to the initial action. (Appellees' Ans. Br. at 23.) That argument fails. The EPA sued Cytec, Ford, and SPS in 2000, but not Carpenter. Thus, the only parties against which the EPA brought a subsequent action in 2001 are Cytec, Ford, and SPS. As a result, the "subse quent action” exception is inapplicable here. See United States v. Navistar Int’l Transp., 152 F.3d 702, 710 (7th Cir.1998) ("We believe that, from the language and structure of the statute, a 'subsequent action ... for further response costs' must be one brought against the same party or parties against which an 'initial action' to recover such costs has been maintained.” (quoting 42 U.S.C. § 9613(g)(2)).).

The second exception is an alternative limitations period that may well be applicable here. That exception states that when the EPA grants a consistency waiver under § 104(c)(1)(C) to extend the removal action, the limitations period is six years, rather than three years, from the date of a determination to grant that waiver. See 42 U.S.C. § 9613(g)(2)(A). In other words, the EPA must bring its suit within six years of the grant of a consistency waiver, regardless of when its removal action is completed. The parties have not defined the term “consistency waiver,” but publicly available information from the EPA indicates that a consistency waiver "is an exemption to the statutory limits of two million dollars in expenditures ... for removal actions.” Memorandum from Lon Biasco, On-Scene Coordinator with the U.S. Envtl. Prot. Agency, to Myron O. Knudson, Region 6 Superfund division Dir. of the U.S. Envtl. Prot. Agency, 9, http://www.epa. gov/region6/6si/louisiana/ag_street/importan1_ documents/action_memo_agstreet.pdf. The EPA performed a 1992 removal action at the Site pursuant to a consistency waiver it had granted on September 4, 1992. Under § 113(g)(2)(A), then, the EPA had to file its suit to recover those costs no later than September 4, 1998, six years from the grant of the waiver. Carpenter argues that since the EPA did not file the OU-2 suit until after 1998, the suit is barred under the “consistency waiver” exception. However, Carpenter concedes that the EPA's 1992 consistency waiver only applied to one specific remedial action, which cost $4.3 million. All of the EPA's other costs are not subject to the "consistency waiver” exception. Plaintiffs and Carpenter acknowledge that the EPA incurred costs on the order of $14 million and only sought reimbursement for $7.4 million. Thus, even accepting Carpenter's argument that $4.3 million of the EPA’s past costs are time-barred, the EPA still has claims for approximately $10 million which are not time-barred under the consistency waiver exception. Therefore, the EPA's $7.4 million claim remains recoverable.

29 . Removal actions are defined as

the cleanup or removal of released hazardous substances from the environment, such actions as may be necessary taken in the event of the threat of release of hazardous substances into the environment, such actions as may be necessary to monitor, assess, and evaluate the release or threat of release of hazardous substances, the disposal of removed material, or the taking of such other actions as may be necessary to prevent, minimize, or mitigate damage to the public health or welfare or to the environment, which may otherwise result from a release or threat of release. The term includes, in addition, without being limited to, security fencing or other measures to limit access, provision of alternative water supplies, temporary evacuation and housing of threatened individuals not otherwise provided for, action taken under section 9604(b) of this title, and any emergency assistance which may be provided under the Disaster Relief and Emergency Assistance Act [42 U.S.C. § 5121 et seq.].

42 U.S.C. § 9601(23).

Remedial actions are defined as those actions consistent with permanent remedy taken instead of or in addition to removal actions in the event of a release or threatened release of a hazardous substance into the environment, to prevent or minimize the release of hazardous substances so that they do not migrate to cause substantial danger to present or future public health or welfare or the environment. The term includes, but is not limited to, such actions at the location of the release as storage, confinement, perimeter protection using dikes, trenches, or ditches, clay cover, neutralization, cleanup of released hazardous substances and associated contaminat ed materials, recycling or reuse, diversion, destruction, segregation of reactive wastes, dredging or excavations, repair or replacement of leaking containers, collection of leachate and runoff, onsite treatment or incineration, provision of alternative water supplies, and any monitoring reasonably required to assure that such actions protect the public health and welfare and the environment. The term includes the costs of permanent relocation of residents and businesses and community facilities where the President determines that, alone or in combination with other measures, such relocation is more cost-effective than and environmentally preferable to the transportation, storage, treatment, destruction, or secure disposition offsite of hazardous substances, or may otherwise be necessary to protect the public health or welfare; the term includes offsite transport and offsite storage, treatment, destruction, or secure disposition of hazardous substances and associated contaminated materials.

Id. § 9601(24).

30 . The text of the relevant provision is as follows:

An initial action for recovery of costs ... must be commenced ... (B) for a remedial action, within 6 years after initiation of physical on-site construction of the remedial action, except that, if the remedial action is initiated within 3 years after the completion of the removal action, costs incurred in the removal action may be recovered in the cost recovery action brought under this subparagraph.

42 U.S.C. § 9613(g)(2). When the drafters first used a form of the word '‘initiate” in this section, it is specified as being the "initiation of physical on-site construction of the remedial action.” Thus, when the word "initiate” is used a second time, in the very next sentence, we read the statute to again mean the "initiation of physical on-site construction,” rather than the initiation of some other environmental cleanup endeavor. As Carpenter persuasively argues, to read the statute otherwise undermines the statute of limitations:

If EPA could use the six year limitation period in § 113(g)(2)(B) simply by incurring any type of response costs ... after issuance of the ROD, but before initiation of on-site construction of the remedial action, the three year limitations period for removal actions would never apply at any site at which remedial action occurs.

(Appellant’s Reply Br. at 12.) This is because, by using the definition of "initiate” that the plaintiffs suggest (i.e., incurring any type of further response costs), the EPA could simply take any one of a variety of later steps that it could claim constitutes the commencement of a remedial action.

31 . Plaintiffs argue that Carpenter waived this argument by failing to raise it before the District Court. Even if Carpenter had not raised the issue below, the District Court ruled on it when it held that Agere and TI had § 107(a) claims for costs they had contributed via the private settlement agreements. More to the point, though, Carpenter adequately preserved the issue when it argued, in its reply to the plaintiffs' proposed findings of fact and conclusions of law, that Agere and TI could not bring those § 107(a) claims.

32 . We recognize that the OU-2 Consent Decree included a lump payment of $7.4 million in “past costs” to the EPA, and that Agere accepted responsibility for this amount as well. Specifically, Agere’s share for these past costs was $83,000. While this specific amount could arguably be considered reimbursement rather then co-funding, the parties do not address this issue explicitly. Particularly, in light of the policy reasons outlined above, however, we are content to give the benefit of the doubt to Agere and believe that it is entitled to recover that amount.

33 . We read the District Court’s opinion as proceeding under § 113(f) for the following reasons. First, the Court said as much when it held that "cooperation with the government is also an appropriate equitable factor to consider in allocating response costs under § 113(f).” (App. at 1:A73.) Second, the Court spent two pages of its opinion outlining the elements a plaintiff must prove under § 113(f), and then organized its analysis based on those elements. And, third, the Court held that "[c]ulpability is an appropriate equitable factor in resolving contribution claims” (App. at 1:A72), which, by use of the word "contribution,” suggests that the District Court allocated liability under § 113(f). In addition, despite arguing that they have § 107(a) claims in addition to their § 113(f) claims, the plaintiffs concede that "the District Court determined the shares of each party without regard to those § 107(a) claims.” (Appellees' Ans. Br. at 47.)

34 . Despite the plaintiffs' argument to the contrary, Carpenter preserved this argument at trial. During trial, when the plaintiffs moved to admit the June 23rd stipulation into evidence, Carpenter objected by reminding the Court that it had not joined in the stipulation. In response, the plaintiffs' counsel stated to the Court, “Your honor, we're not asking Defendant Carpenter to stipulate to any of these facts." (App. at 3:A1198.) Carpenter then responded, “That's fine so long as it is not a stipulation among all of the parties relating to the drawing of waste that was hauled by [DCC] from the same company.” {Id.) Moreover, when the Court confirmed that it was "a stipulation of limited scope which doesn't involve [Carpenter],” Carpenter responded, "[W]e're just making a record that we do not agree or stipulate or we’re not waiving any kind of objection....” (App. at 3:A 1199.) The Court then asked the plaintiffs’ counsel whether he thought Carpenter had "waived any objection,” to which the plaintiffs' counsel responded, “No your Honor.” {Id.) Finally, at oral argument before us, the plaintiffs’ counsel conceded that "Carpenter chose not to agree” with the stipulation, and that Carpenter had voiced its objection at trial. (Oral Argument Tr. at 28.) The issue was thus amply preserved in the District Court.

35 . After oral argument, the plaintiffs’ counsel sent a letter to our Court pursuant to Fed. R.App. P. 28(j) to address what he described as a “misstatement during oral argument ... that Appellees did not introduce evidence at trial concerning the volumes of Cytec's, SPS's, and Ford’s wastes that were disposed of at the Boarhead Farms Superfund Site.” Letter from Glenn A. Harris, Counsel for Plaintiffs, to Marcia M. Waldron, Clerk of the United States Court of Appeals for the Third Circuit' (Dec. 18, 2009) (on file with the Court). Specifically, the plaintiffs’ counsel points to several exhibits moved into evidence at trial, including purchase orders from Ford and SPS to DCC, DCC invoices to Ford and SPS, and filings made with the New Jersey Department of Environmental Protection that evidence certain volumes of Cytec's waste disposed of at the Site. (Id.) The plaintiffs' position is that “[t]he exhibits are in the trial record and thus can be relied upon” by our Court. (Id.)

However, even if we were to consider those exhibits, they are not evidence of the plaintiffs’ waste volumes transported by DCC to the Boarhead Site during the relevant time period. They appear to be only a collection of invoices and purchase orders with no testimony to explain what they mean. Moreover, plaintiffs have not directed us to anything indicating that they argued at trial that those documents were evidence of their waste volumes. In fact, in the plaintiffs' reply to Carpenter’s proposed findings of fact, they stated that the volume evidence for plaintiffs and H & H came from the stipulations. The District Court’s opinion reveals that the Court relied solely on those stipulations to make findings of fact as to the plaintiffs’ waste volumes and did not rely on the evidence discussed in the recent 28(j) letter to our Court.

36 . The argument went as follows:

Plaintiffs’ Counsel: [T]he stipulation is admitted into evidence as proof against my client of the three entities’ volumes. They could have disputed it. But its not that there's no evidence.

The Court: Your argument is this was, in effect, an admission.

Plaintiffs’ Counsel: Exactly.

Plaintiffs’ Counsel: That's exactly correct. We admitted and we stipulated with [H & H] that those are the volumes for those three companies.... [I]t was in evidence. The judge used it. I mean he counted on it. Look on, look at the opinion. He counts those volumes against me.

(Oral Argument Tr. at 23-24.)

37 . Although no exception to the hearsay rule is immediately apparent as being applicable here, we do not decide that none applies to the June 23rd stipulation, since that issue is not before us. We decide only that the June 23rd stipulation was not incorporated into any stipulation Carpenter did sign and that it is not admissible against Carpenter as an admission because Carpenter admitted nothing by it.

38 . See infra Section III.F.i.

39 . An adoptive admission refers to “evidence of other conduct of a party manifesting circumstantially the party's assent to the truth of a statement made by another.” 2 George E. Dix, et al, McCormick on Evidence § 261 (Kenneth S. Broun, ed., 6th ed., 2009) [hereinafter “McCormick”]. Adoptive admissions in federal courts are governed by Federal Rule of Evidence 801(d)(2)(B), which provides that a statement is non-hearsay if “[t]he statement is offered against a party and is ... (B) a statement of which the party has manifested an adoption or belief in its truth.” Fed.R.Evid. 801(d)(2)(B). Here, Carpenter not only cited to plaintiffs’ responses to contention interrogatories, but it moved those responses into evidence, relied on them, and continues to rely on them on appeal. In other words, Carpenter urged the District Court — and now urges us — to rely on those responses as competent evidence. Carpenter thus may be viewed as showing its "assent to the truth of a statement made by another.” McCormick, supra at § 261 (explaining that "the introduction of evidence by a party” may constitute an adoptive admission “depend[ing] upon whether the particular circumstances warrant the conclusion that adoption in fact occurred”). However, we leave that determination to the District Court in the first instance.

40 . Carpenter also argues that even if the plaintiffs could state a § 702(a) cost recovery claim under the HSCA, that claim would be barred by HSCA's statute of limitations provision, contained in 35 Pa. Stat. Ann. § 6020.1114. We need not address that contention because no statute of limitations argument has been made pertaining to § 107(a). All the statute of limitations arguments that the parties have made to us pertain to § 113(f).

41 . Again (see supra n. 12), we note that the District Court must determine on remand whether Agere’s claims are held by it or by assignees.