Resolving “Resolved”: Covenants Not to Sue and the Availability of CERCLA Contribution Actions

The Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA)—as part of its dual goals of cleaning up hazardous-waste sites and ensuring that the polluter pays for that cleanup—gives private parties two mutually exclusive causes of action: cost recovery and contribution.

Contribution is available in limited circumstances, including if the party has “resolved” its liability with the government. But CERCLA does not define this operative term. Federal courts are split over how the structure of a settlement resolves liability. Several courts follow Bernstein v. Bankert, which held that any conditions precedent and nonadmissions of liability strongly suggest that a party has not yet resolved its liability. The Ninth Circuit’s recent case, ASARCO LLC v. Atlantic Richfield Co., said liability is resolved if the settlement determines the party’s obligations with “certainty and finality.” Bernstein deviates from CERCLA’s text and policy, leading to serious inconsistencies in the interpretation and application of the statute. ASARCO injects uncertainty into the statute, which disincentivizes settlements. When the stakes are the reallocation of billions of dollars and the amelioration of the most notorious environmental disasters, getting it right is paramount. This Note proposes a bright-line rule—liability is resolved when the settlement contains any covenant not to sue, conditional or unconditional—and argues that this reading cleans up many of the issues the current circuit split imparts on the statute.


Four years after Congress thought it had closed environmental law’s “last remaining loophole,” a chemical soup full of carcinogens began bubbling up in people’s basements in the Love Canal neighborhood of Niagara Falls, New York.1Robert V. Percival, Christopher H. Schroeder, Alan S. Miller & James P. Leape, Environmental Regulation: Law, Science, and Policy 409 (8th ed. 2018). In 1953, the local school board had bought what it knew to be essentially a loosely covered dump of industrial waste from the Hooker Chemical Company for one dollar.2Alexander Nazaryan, Love Hurts, Newsweek (Oct. 17, 2013), https://www​ []. After years of complaints of health problems, the problem reached crisis levels when heavy rains caused the toxins to seep into people’s homes.3See Colin Dabkowski, A History of the Love Canal Disaster, 1893 to 1998, Buff. News (Aug. 4, 2018), []. In August 1978, the State of New York declared a public health emergency, with the federal government stepping in five days later.4Nazaryan, supra note 2. Over the next two years, around 1,000 people left their homes, and most would never return.5Id. Love Canal was by no means unique. For example, in the ominously named “Valley of the Drums,” a massive collection of metal drums left in a field caught fire and burned for more than a week.6U.S. Env’t Prot. Agency, Sixth Five-Year Review Report for A. L. Taylor (Valley of the Drums) Superfund Site, Brooks, Bullitt County, Kentucky, at D-3 (2018), []; Valley of the Drums, Atlas Obscura, []. Or take Times Beach, where road crews used the notorious carcinogen dioxin as a dust suppressant.7William Powell, Remember Times Beach: The Dioxin Disaster, 30 Years Later, St. Louis Mag. (Dec. 3, 2012), [].

Despite the 1970s being the “environmental decade” when Congress legislated the bulk of the modern-day environmental regulatory regime, it left one gap.8Jeffrey M. Gaba, The Private Causes of Action Under CERCLA: Navigating the Intersection of Sections 107(a) and 113(f), 5 Mich. J. Env’t & Admin. L. 117, 121–22 (2015). It failed to address the pollution that predated these statutes.9Id. Love Canal, Valley of the Drums, Times Beach, and a slew of other incidents generated the political pressure necessary to pass federal legislation closing this gap.10See id. at 122. However, the election of Ronald Reagan and the Republican takeover of the Senate in 1980 left the Ninety-Sixth Congress mere weeks to pass any legislation on the topic.11Id. Thus, the legislative history is, at best, messy.12See, e.g., Carson Harbor Vill., Ltd. v. Unocal Corp., 270 F.3d 863, 885 (9th Cir. 2001) (en banc) (“Any inquiry into CERCLA’s legislative history is somewhat of a snark hunt.”). See generally Frank P. Grad, A Legislative History of the Comprehensive Environmental Response, Compensation and Liability (“Superfund”) Act of 1980, 8 Colum. J. Env’t L. 1 (1982) (providing a comprehensive overview of Congress’s abbreviated consideration of CERCLA). This rushed process resulted in the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA)1342 U.S.C. §§ 9601–9675.—a statute frequently and colorfully ridiculed for its complexity and poor drafting.14See, e.g., CP Holdings, Inc. v. Goldberg-Zoino & Assocs., Inc., 769 F. Supp. 432, 435 (D.N.H. 1991) (“[C]ourts seem to resort to . . . hoping that if they stare at CERCLA long enough, it will burn a coherent afterimage on the brain.”).

With a set of amendments in 1986 known as the Superfund Amendments and Reauthorization Act (SARA),15Superfund Amendments and Reauthorization Act of 1986, Pub. L. No. 99-499, 100 Stat. 1613 (codified as amended at 42 U.S.C. §§ 9601–9675). SARA, among other things, clarified the scope of liability and cleanup standards, created new government settlement authority, and created new private causes of action. Zygmunt J.B. Plater, Environmental Law and Policy: Nature, Law, and Society 713 (5th ed. 2016) (“[T]he generally pro-liability posture of SARA led some experts in the field to suggest that its acronym ought to be changed to RACHEL, because the Reauthorization Act Confirms How Everyone’s Liable.”); Gaba, supra note 8, at 122–23. CERCLA created a complicated structure designed to effect two goals: (1) make sure hazardous-waste sites are cleaned up in a timely manner and (2) make those who caused the contamination pay for the cleanup.16Burlington N. & Santa Fe Ry. Co. v. United States, 556 U.S. 599, 602 (2009) (quoting Consol. Edison Co. of N.Y. v. UGI Utils., Inc., 423 F.3d 90, 94 (2d Cir. 2005)). It addresses the first goal by authorizing the federal government to clean up a site and then sue those responsible17See 42 U.S.C. § 9607(a). or compel those parties to clean up the site themselves.18See id. § 9606(a). In addition, it created a tax on the chemical industry that would create a large fund to finance cleanups.19See I.R.C. § 9507; Martina E. Cartwright, Superfund: It’s No Longer Super and It Isn’t Much of a Fund, 18 Tul. Env’t L.J. 299, 308 (2005). “Superfund,” as this fund is commonly called, has become the nickname for the whole statute.20See Superfund Liability, U.S. Env’t Prot. Agency (updated July 18, 2019), []. The statute addresses the second goal by allowing parties who pay for cleanup efforts beyond the harm they caused to sue other polluters, which CERCLA calls “Potentially Responsible Parties” (PRPs).21See, e.g., ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1114 (9th Cir. 2017). Parties that incur cleanup costs directly can bring “cost-recovery” actions, while parties whose costs were imposed by a legal action can bring a “contribution” claim.22See 42 U.S.C. §§ 9607(a), 9613(f).

One circumstance in which a party may bring a contribution suit is when that party has already “resolved” its liability with the United States or a state.23Id. § 9613(f)(3)(B). And now CERCLA’s poor drafting rears its head: What does “resolved” mean? For example, if the government agreed to not sue only on the condition that the party completely finishes a cleanup (termed a “condition precedent”),24See infra text accompanying notes 60–67. is liability resolved if the cleanup is not yet finished? Essentially, the debate concerns how the structure of a settlement determines resolution of liability. Given the frequency of settlements25Gaba, supra note 8, at 124–25. and the gargantuan costs of cleanups,26See, e.g., Steven Ferrey, Superfund Chaos Theory: What Happens When the Lower Federal Courts Don’t Follow the Supreme Court, 6 Mich. J. Env’t & Admin. L. 151, 154 (2016) (“Addressing remaining contaminated sites in the U.S. would cost up to one-quarter trillion dollars, or an expenditure of -8 billion annually for forty years.” (footnote omitted)). Despite these costs, there can be massive public health benefits. E.g., Justin R. Pidot & Dale Ratliff, The Common Law of Liable Party CERCLA Claims, 70 Stan. L. Rev. 191, 196 (2018) (“[T]he EPA estimates that the additional risk of cancer will decrease significantly from four in 1000 to between one in 10,000 and one in 1 million people.”). resolving questions of liability in hazardous-waste cases is of critical importance.

This Note proposes a bright-line rule to determine if a settlement resolves liability. Liability is “resolved”—and thus a contribution action is available—if a settlement contains any covenant not to sue, conditional or otherwise; a party’s failure to admit or deny liability is irrelevant to this analysis. Part I provides an overview of CERCLA’s structure—namely who is liable for what and how the causes of action to assign this liability interact. Part II analyzes the circuit split created by this gap, concluding that both of the leading approaches produce unsatisfactory results. Part III proposes the bright-line rule described above and shows how that solution is consistent with CERCLA’s text, overall statutory structure, and policy objectives.

I.      An Overview of CERCLA Liability

CERCLA liability is complex and broad. The statute uses intricate pathways to make a wide range of parties liable for an even wider range of damages.27See Jerry L. Anderson, The Hazardous Waste Land, 13 Va. Env’t L.J. 1, 6–8 (1993) (“The CERCLA liability system has become a black hole that indiscriminately devours all who come near it.” (footnote omitted)). This Part provides an overview of how CERCLA’s liability provision works. Section I.A outlines the statute’s basic liability structure and how the government leverages that liability to effect cleanups, while Section I.B describes how CERCLA’s two causes of action function and interact with each other.

A.    The Parties Under CERCLA and Their Liabilities

Under CERCLA, there are multiple categories of parties that can be liable (the PRPs), several types of recoverable damages, and a plethora of ways the government uses this liability to clean up hazardous pollution. This Section will walk through each one in turn.

First, who is liable under CERCLA? For a party to be liable, it must meet the definition of “person” under the statute, which is easy given its broad definition.28See 42 U.S.C. § 9601(21) (defining “person” to include: individuals, business entities, municipalities, states, and even the United States). There are four types of “persons” who can be held liable as PRPs under the statute. First, there are the current owners or operators of a hazardous-waste site.29Id. § 9607(a)(1). The term “owner or operator” is, unhelpfully, defined as “any person owning [or] operating” a facility or vessel.30Id. § 9601(20)(A)(i). This bit of “tautology” serves as just another example of CERCLA’s poor drafting. United States v. Bestfoods, 524 U.S. 51, 56 (1998). Courts have fleshed out the meanings of those words elsewhere. See id. at 66–67. Members of this group are liable regardless of whether they actually caused the pollution. For instance, the current owner of a landfill that spews toxic materials would be designated as a PRP under the statute, even if it recently purchased the site and played no role in creating the hazardous conditions.31See, e.g., New York v. Shore Realty Corp., 759 F.2d 1032, 1044 (2d Cir. 1985) (quoting 42 U.S.C. § 9607(a)(2)). To blunt this otherwise harsh result, Congress created a kind of “due diligence” defense: a prospective property purchaser can avoid liability if they took certain steps to identify contaminants and cooperate with cleanup activities if hazardous waste is found. See Pidot & Ratliff, supra note 26, at 202. The second category is past owners or operators. To be liable, the past owner or operator must have owned or operated the site at the time the hazardous chemicals were disposed of.3242 U.S.C. § 9607(a)(2). Members of the third group of PRPs, known as “arrangers,”33See Burlington N. & Santa Fe Ry. Co. v. United States, 556 U.S. 599, 606–07 (2009). are liable if they paid or coordinated with someone else to dispose of the waste.3442 U.S.C. § 9607(a)(3). “Arranging” requires an intent to enter the transaction for the purpose of disposing of the hazardous substance.35Burlington N., 556 U.S. at 612 (“[K]nowledge alone is insufficient to prove that an entity ‘planned for’ the disposal, particularly when the disposal occurs as a peripheral result of the legitimate sale of an unused, useful product. In order to qualify as an arranger, [a PRP] must have entered into the sale of [the hazardous substance] with the intention that at least a portion of the product be disposed of during the transfer process . . . .”). The fourth and final category is transporters of hazardous wastes, but only if they had a role in selecting the site of disposal.3642 U.S.C. § 9607(a)(4); see also Pidot & Ratliff, supra note 26, at 202 n.52.

And what are they liable for? There are three categories of recoverable costs from liable parties under CERCLA. First, PRPs are liable for the costs of the cleanup. That liability is owed to any party that incurred costs as part of the cleanup, including but not limited to the federal government, states, and Native American tribes.3742 U.S.C. § 9607(a)(4)(A)–(B). CERCLA uses the term of art “response” to describe these cleanups,38ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1115–16 (9th Cir. 2017). which are divided into two types: removal actions and remedial actions.3942 U.S.C. § 9601(23)–(24). Both actions need to address a “release” of a “hazardous substance,” or the threat of one, into the “environment.”40Id. CERCLA defines all three of these operative terms quite broadly. For example, the “environment” includes, basically, any water (navigable or otherwise), land (above or below ground), or air under the jurisdiction of the United States. Id. § 9601(8); see also id. § 9601(22) (defining “release” broadly to include spilling, leaking, or otherwise discharging any hazardous substance into the environment). A notable exception to this broadness is that the definition of “hazardous substance” excludes petroleum and natural gas. Id. § 9601(14)(F). As to the difference between the two, removal actions are “‘those taken to counter imminent and substantial threats to public health and welfare,’ while remedial actions ‘are longer term, more permanent responses.’”41Morrison Enters., LLC v. Dravo Corp., 638 F.3d 594, 608 (8th Cir. 2011) (quoting Minnesota v. Kalman W. Abrams Metals, Inc., 155 F.3d 1019, 1024 (8th Cir. 1998)); see 42 U.S.C. § 9601(23)–(24) (defining “removal” and “remedial action”). Second, PRPs must cover the cost of health assessments for, or public health studies of, the people who might later suffer adverse health consequences from the toxic substances.4242 U.S.C. § 9607(a)(4)(D). Third, PRPs are liable for natural resource damages.43Id. § 9607(a)(4)(C). See generally 2 Karen A. Gottlieb, Toxic Torts Practice Guide § 25:8 (Fall 2019 ed.). This provision goes beyond response costs to cover damages to the environment that remain once pollution is removed.44See Patrick E. Tolan, Jr., Natural Resource Damages Under CERCLA: Failures, Lessons Learned, and Alternatives, 38 N.M. L. Rev. 409, 413 (2008). However, the scope of environmental damage can be quite difficult to quantify.45See id. at 414 n.40 (collecting sources). Notably, CERCLA does not cover damages related to personal injury or property value, nor does it allow successful plaintiffs to recover attorney or expert fees.46Alexandra B. Klass, CERCLA, State Law, and Federalism in the 21st Century, 41 Sw. L. Rev. 679, 685 (2012).

There are five ways that the government can get a site cleaned up at the polluter’s expense under CERCLA.47Pidot & Ratliff, supra note 26, at 204–08. The first two options involve going to court first in order to secure a cleanup, while the last three involve cleaning up or agreeing to do so before any litigation. As its first two options, EPA can file an abatement action seeking to judicially enjoin a party to clean up, or it may issue a Unilateral Administrative Order (UAO) ordering the same, if there is “an imminent and substantial” threat to public health or the environment posed by a current or potential release.4842 U.S.C. § 9606(a); accord Gen. Elec. Co. v. Jackson, 610 F.3d 110, 114 (D.C. Cir. 2010); Agere Sys., Inc. v. Advanced Env’t Tech. Corp., 602 F.3d 204, 217 n.25 (3d Cir. 2010). EPA tends to use UAOs more than filing an abatement action in court because they are powerful tools. See Pidot & Ratiff, supra note 26, at 206 n.77, 207 (discussing lack of judicial review for, and fines of, UAOs). Third, CERCLA authorizes EPA to clean up the site itself, financed by the Superfund,49I.R.C. § 9507; 42 U.S.C. § 9604(a). Using the Superfund to finance cleanups used to be EPA’s preferred method. Ian G. John, Note, Too Much Waste: A Proposal for Change in the Government’s Effort to Clean Up the Nation, 70 Ind. L.J. 951, 965 (1995). However, the tax expired in 1995 and the fund ran out of money eight years later, leaving EPA with overall fewer funds that Congress appropriates. See Cartwright, supra note 19, at 300–01. and then to sue the PRPs to replenish the Superfund.50See 42 U.S.C. § 9607(a)(4)(A). Besides the tax, the Superfund is funded by recovery from CERCLA suits brought by the United States. See I.R.C. § 9507(b)(2). Fourth, a PRP may decide to “voluntarily” clean up the site in anticipation of the government compelling the PRP via litigation or settlement.51Pidot & Ratliff, supra note 26, at 208–09.

Fifth, and most relevant to this Note, a PRP may enter a settlement or a consent decree with the government, agreeing to do all or part of a cleanup.52See 42 U.S.C. § 9622. While settlements and consent decrees are of different legal character, Courtney R. McVean & Justin R. Pidot, Environmental Settlements and Administrative Law, 39 Harv. Env’t L. Rev. 191, 199–201 (2015), they do not differ in ways significant to this Note. As such, it refers to both of them interchangeably. A common approach is for EPA to conduct an initial investigation and then threaten PRPs with lawsuits, UAOs, or abatement actions.53Gaba, supra note 8, at 124. This carrot-and-stick approach works because CERCLA lawsuits and UAOs are incredibly hefty sticks,54See supra note 48 (discussing UAOs and abatement actions); infra note 73 and accompanying text (discussing lawsuits). while settlements are chock full of carrots, such as contribution protection (also known as the “settlement bar”),55Joanna M. Fuller, Note, The Sanctity of Settlement: Stopping CERCLA’s Volunteer Remediators from Sidestepping the Settlement Bar, 34 Colum. J. Env’t L. 219, 227, 245, 248 (2009); see 42 U.S.C. § 9613(f)(2). covenants not to sue (i.e., releases of liability),56Fuller, supra note 55, at 248; see 42 U.S.C. § 9622(f). and causes of action for suits against other PRPs.57See 42 U.S.C. § 9613(f)(3)(B). In fact, EPA has restructured its settlements to make sure settling PRPs have this cause of action available based on recent case law. Gaba, supra note 8, at 124 n.33. CERCLA does limit how generous the government can be. For example, the government cannot release a PRP from future liability until the president certifies that PRP has cleaned up satisfactorily.5842 U.S.C. § 9622(f)(3). Settlements are now EPA’s preferred approach “[b]ecause [they] cost far fewer taxpayer dollars than EPA-led cleanups, reduce litigation risks, and promote cooperation between the EPA and PRPs.”59Pidot & Ratliff, supra note 26, at 205.

There are different ways to structure these settlements, and the distinctions are subtle but key. Namely, the different configurations focus on the sequencing of a PRP’s obligations and benefits. On one hand, the PRP could first have to finish cleaning up before EPA is bound by its covenant not to sue the PRP anymore.60See, e.g., Bernstein v. Bankert, 733 F.3d 190, 204 (7th Cir. 2013) (“[W]hen the [PRPs] completed performance of their obligations under the [settlement] . . . the conditional covenants not to sue contained therein went into effect.” (emphasis added)). This structure makes completely cleaning up the condition precedent for the covenant because the cleanup is “[a]n act or event . . . that must exist or occur before a duty to perform something promised arises.”61Condition Precedent, Black’s Law Dictionary (11th ed. 2019). On the other hand, EPA could bind itself not to sue immediately, with that obligation disappearing later if the party fails to clean up.62See, e.g., ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1114 (9th Cir. 2017) (“[S]o long as [the PRP] funds the Custodial Trust Accounts, it is released from liability . . . .” (footnote omitted)). In this case, cleaning up is the condition subsequent to the covenant because the failure of a cleanup is “[a] condition that, if it occurs, will bring something else to an end.”63Condition Subsequent, Black’s Law Dictionary, supra note 61. Finally, EPA could make the covenant unconditional.64See, e.g., Hobart Corp. v. Waste Mgmt. of Ohio, Inc., 758 F.3d 757, 769 (6th Cir. 2014). In theory, EPA is bound immediately to not sue the PRP, and that obligation never ends. Of course, if the PRP fails to clean up, EPA could sue for a breach of contract.65See id. at 768–69. Very few courts use this contract terminology.66But see Fla. Power Corp. v. FirstEnergy Corp., 810 F.3d 996, 1017–18 (6th Cir. 2015) (Suhrheinrich, J., dissenting) (classifying covenants not to sue in various CERCLA settlements as conditions precedent or subsequent). However, classifying various CERCLA settlements among these categories is crucial to understanding the problems with current interpretations of CERCLA and the validity of this Note’s bright-line rule.67See infra Parts II–III.

The upshot of this Section is that CERCLA liability is broad. A large number of parties, even those tangentially associated with a hazardous-waste site, can be liable for massive amounts of damages, and the government has numerous ways to enforce this liability regime.

B.    The CERCLA Causes of Action

This Section discusses the two causes of action parties can use to seek the monetary damages recoverable under CERLCA: cost recovery and contribution.68Some courts call cost-recovery actions “ ‘section 107(a)’ action[s],” and the two types of contribution actions either section 113(f)(1) actions or “ ‘section 113(f)(3)(B)’ action[s],” respectively. See, e.g., Gov’t of Guam v. United States, 341 F. Supp. 3d 74, 77 n.1 (D.D.C. 2018), rev’d on other grounds, 950 F.3d 104 (D.C. Cir. 2020); Gaba, supra note 8, at 133. This Note uses the “cost-recovery” and “contribution” labels. However, contribution actions have two separate sufficient triggers. See 42 U.S.C. § 9613(f)(1), (3)(B). This Note focuses mostly on the trigger in section 113(f)(3)(B). Each remedy is distinct and operates in different procedural circumstances.69United States v. Atl. Rsch. Corp., 551 U.S. 128, 138–39 (2007) (quoting Cooper Indus., Inc. v. Aviall Servs., Inc., 543 U.S. 157, 163 n.3 (2004), and Consol. Edison Co. of N.Y. v. UGI Utils., Inc., 423 F.3d 90, 99 (2d Cir. 2005)). In that vein, every circuit to address this issue has found that cost recovery and contribution are mutually exclusive causes of action.70ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1117 (9th Cir. 2017) (collecting cases).

A cost-recovery action is available to anyone who has incurred costs in cleaning up a contaminated site.71Atl. Rsch., 551 U.S. at 136. Recall that a “person” includes private and governmental entities. See supra note 28 and accompanying text. Furthermore, the plaintiff does not have to be a PRP; an innocent party can clean up and then sue any PRP. Bernstein v. Bankert, 733 F.3d 190, 201 (7th Cir. 2013).Recovery is limited only to costs a party spent actually cleaning up a site; money paid for settlements or judgments is excluded. 72Atl. Rsch., 551 U.S. at 139 (“Rather, [that money] reimburses other parties for costs that those parties incurred.”).

Cost recovery is a powerful tool. CERCLA is often described as imposing liability that is strict, joint, several, and retroactive, despite all of those words being absent from the statute.73See, e.g., ASARCO LLC v. Atl. Richfield Co., 73 F. Supp. 3d 1285, 1288 (D. Mont. 2014) (“CERCLA is a unique and powerful statute, imposing strict and joint and several liability on countless parties for contamination reaching back to the Nineteenth Century.”), vacated on other grounds, 866 F.3d 1108, 1114 (9th Cir. 2017); Superfund Liability, supra note 20. Ironically, the statute’s drafters actually struck those words from an early version of CERCLA because joint and several liability proved too controversial. Pidot & Ratliff, supra note 26, at 216. But a key drafter suggested that instead “issues of liability not resolved by [CERCLA], if any, shall be governed by traditional and evolving principles of common law.” 126 Cong. Rec. 30,932 (1980) (statement of Sen. Randolph). In the “seminal opinion” on the issue,74Burlington N. & Santa Fe Ry. Co. v. United States, 556 U.S. 599, 613 (2009). a district court adopted a common law approach when interpreting CERCLA liability as such,75See United States v. Chem-Dyne Corp., 572 F. Supp. 802, 808 (S.D. Ohio 1983). which now all federal courts follow.76Burlington N., 556 U.S. at 613–14 (collecting cases). Using those traditional notions of common law, courts apply joint and several liability when the harm is indivisible.77E.g., Chem-Dyne, 572 F. Supp. at 810. And when there is a reasonable method to divide the harm, courts apportion it among the PRPs.78See, e.g., Burlington N., 556 U.S. at 614 (quoting Restatement (Second) of Torts § 433A(1)(b) (Am. Law Inst. 1965)). Regarding strict liability, CERCLA adopts the standard of liability used in the Federal Water Pollution Control Act.7942 U.S.C. § 9601(32). Courts have also interpreted that Act as imposing strict liability, so that standard is imputed to CERCLA.80Pidot & Ratliff, supra note 26, at 210 n.102; see also Chem-Dyne, 572 F. Supp. at 805.

Where cost recovery is blunt and harsh, contribution claims are flexible and less severe. Contribution claims are designed to “promote quicker and fairer settlements, decrease litigation, and facilitate cleanups.”81Kenneth K. Kilbert, Neither Joint nor Several: Orphan Shares and Private CERCLA Actions, 41 Env’t L. 1045, 1070 (2011). They do so, in part, by providing courts broad discretion to use equitable factors to allocate costs between PRPs, unlike the much harsher standard of its cost-recovery cousin.82See 42 U.S.C. § 9613(f)(1). While the statute does not require any particular test, courts generally use one of two types of tests, known as the Gore factors or Torres factors. Pidot & Ratliff, supra note 26, at 259. Besides a different standard of liability, contribution claims have other procedural differences from cost-recovery claims. First, there is contribution protection: a PRP that has resolved liability with the federal or state government cannot be sued via contribution.8342 U.S.C. § 9613(f)(2). This provides a powerful incentive to settle and thus serves as a specific example of how contribution fits into CERCLA’s overall scheme of promoting cleanups. See Pidot & Ratliff, supra note 26, at 213–14. But a PRP that settled with the government is still vulnerable to a cost-recovery action.84United States v. Atl. Rsch. Corp., 551 U.S. 128, 138–39 (2007). Commentators have lambasted the policy implications resulting from this particular portion of this case. See infra notes 225–232 and accompanying text. Second, the statutes of limitations differ. A cost-recovery action expires three years after the completion of a removal action or six years after a remedial action,8542 U.S.C. § 9613(g)(2). The policy justification behind the shorter statute of limitations for contribution actions is “to ensure that the responsible parties get to the bargaining—and clean-up—table sooner rather than later.” ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1117 (9th Cir. 2017) (quoting RSR Corp. v. Com. Metals Co., 496 F.3d 552, 559 (6th Cir. 2007)). whereas contribution actions must be brought within three years of the date of judgment or settlement.8642 U.S.C. § 9613(g)(3). For additional differences between cost-recovery and contribution actions, see Gaba, supra note 8, at 127–29.

There are two triggers sufficient to bring a contribution action. First, a PRP may bring a contribution claim during or following a UAO, abatement action, or cost-recovery action.8742 U.S.C. § 9613(f)(1); Cooper Indus., Inc. v. Aviall Servs., Inc., 543 U.S. 157, 166 (2004) (“The natural meaning of [section 113(f)(1)] is that contribution may only be sought subject to specified conditions, namely, ‘during or following’ a specified civil action.”). This claim essentially functions as a way for a defendant facing one of the aforementioned actions to crossclaim or counterclaim against other PRPs.88As a hypothetical, imagine three PRPs—M, N, and O—who caused 60, 10, and 30 percent of the contamination at a site, respectively. M cleans up all the waste and brings a cost-recovery action against N. Technically, M might be able to recover all of the cleanup costs even though it caused the majority of the pollution given that cost-recovery liability is strict, joint, and several. Given that this is now “during” a cost-recovery action, one of the triggers for a contribution claim has been met, so N can file a contribution counterclaim so that it is not unfairly stuck footing the entire bill for a problem that it caused the least of. Likewise, it can crossclaim to rope in O, who likewise caused significantly more of the problem than N.Second, a PRP may seek contribution from other PRPs if it has “resolved its liability to the United States or a State for some or all of a response action or for some or all of the costs of such action in an administrative or judicially approved settlement.”8942 U.S.C. § 9613(f)(3)(B) (emphasis added). But CERCLA does not define what it means to “resolve” liability.90Gov’t of Guam v. United States, 341 F. Supp. 74, 85 (D.D.C. 2018), rev’d on other grounds, 950 F.3d 104 (D.C. Cir. 2020). This absence undergirds a circuit split. Specifically, the split is over what a settlement between the government and a PRP needs to contain to resolve liability and trigger the availability of a contribution action.

II.     An Unresolved Circuit Split

This Part analyzes the circuit split that emerged over what it means to “resolve” liability with the government in order to trigger the availability of a contribution action. This Part starts with the most popular position, adopted by the Sixth and Seventh Circuits. Those courts have held that settlements most likely do not resolve liability if they include nonadmission of liability or if the covenant not to sue goes into effect only after a party finishes cleaning up.91See infra Section II.A. In other words, if cleaning up is the condition precedent to the covenant not to sue, liability is not resolved.92See supra text accompanying notes 60–67. This Part then discusses the Ninth Circuit’s test that liability must be determined with “certainty and finality” in order to be resolved.93See infra Section II.B. After this Note entered the publication process with the Michigan Law Review, the D.C. Circuit entered the split with a new, third test. See Gov’t of Guam, 950 F.3d 104. After dealing with one novel and two common threshold issues, see id. at 110–14, Guam addressed the merits with a heavily textualist analysis. See id. at 114–15. It held that all a settlement needs to do to “resolve” liability is obligate a party to take some action that is covered under the expansively broad definition of a response action—that is, a cleanup. Id. at 115–16. That definition includes “‘confinement’ of substances.” Id. at 116 (quoting 42 U.S.C. § 9601(24)). Thus, liability had been resolved given the settlement’s requirement that the PRP design a system to contain contaminated water that had percolated through the hazardous wastes. Id. Similar to this Note’s proposal, Guam’s approach seems to be a bright-line rule, albeit with the line drawn elsewhere; thus, Guam is more consistent with CERCLA’s text, structure, and policy than its sister circuits’ approaches. Cf. infra Part III. For a more detailed factual and legal analysis of Guam, see Austin W. Manning, D.C. Circuit Follows Trend on Non-CERCLA Settlements that Leaves Guam Footing 0M CERCLA Bill, Manko Gold Katcher Fox: Litig. Blog (Mar. 3, 2020),​/guam_navy_superfund_Ordot_limitations_contribution_settlement_decree [​/LWV8-DU2A].

A.    Conditions Precedent and Nonadmissions of Liability: The Sixth and Seventh Circuits’ Approach

In Bernstein v. Bankert,94733 F.3d 190 (7th Cir. 2013). the Seventh Circuit established the most popular test: covenants not to sue that are conditions precedent and/or nonadmissions of liability are strong indicators that the PRP has not yet resolved liability with the government. In that case, the company known as “Enviro-Chem” conducted waste-handling and disposal operations in a suburb of Indianapolis.95Bernstein, 733 F.3d at 195. It shut down in 1982, leaving considerable pollution behind.96Id. at 195–96. The pollution site was draining into a waterway that partially supplied Indianapolis’s drinking water.97Id. at 197. EPA first issued a UAO to prevent drinking supply contamination, and, with the pollution contained, EPA turned to assigning liability for actually cleaning up the site.98Id. EPA and the PRPs entered the first Administrative Order by Consent (AOC)—a type of settlement—where the PRPs agreed to do two things: (1) conduct a study to analyze the various cleanup options available and (2) form a trust to fund that study and any subsequent cleanups.99Id. at 203–04. Three years later, in a second AOC, EPA and the PRPs agreed to further fund the trust.100Id. at 203, 207. Regarding the status of the PRPs’ liability, there were two important provisions, which were identical for both AOCs: (1) cleaning up was the condition precedent for the PRP’s release of liability because the release was not triggered until the cleanup was certified complete by the EPA101See supra text accompanying notes 60–67. and (2) the PRPs neither admitted nor denied liability.102Bernstein, 733 F.3d at 197, 203–04, 207. The cleanup began, and, at the time of suit, EPA had not issued any notice of approval.103Id. at 197.

The fund trustees filed a cost-recovery action against some PRPs and their insurers who had not paid into the trust or fulfilled other obligations under the AOCs.104Id. at 198. The defendant PRPs moved for summary judgment on statute of limitation grounds.105Id. First, the district court found that the trustees could not bring a cost-recovery action; rather, contribution was their only option.106Id. at 198, 206–07. Given that the statute of limitations for contribution had expired three years prior, the court granted the motion.107Id.

On appeal, the Seventh Circuit broke up the first and second AOCs, reaching different results for each. For the second AOC, the court looked to the plain meaning of the word “resolved.”108Id. at 210–11. It canvassed dictionaries and court opinions to find the word’s meaning.109Id. at 211–12, 211 n.12 (collecting sources). After its survey, the court defined “resolved” as when “[a]n issue . . . is decided, determined, or settled—finished, with no need to revisit.”110Id. at 211. Applying that standard, it was clear to the court the parties had not “resolved liability”; even though the parties had “settled” (one element of Bernstein’s definition of “resolve”), they had not satisfied the other element (“no need to revisit”). The covenants not to sue had not kicked in yet because the cleanup was ongoing.111Id. at 207. Because the government could still sue until the condition precedent (i.e., complete cleanup) had been met, liability had not been decided with finality.112Id. at 212. That potential to “revisit” the issue meant nothing had been “resolved.” However, a cost-recovery suit was available because, while the trustees had not resolved liability, they had met the paradigmatic trigger for cost recovery: incurring costs while cleaning up a site.113Id. at 207.

The court also addressed two counterarguments. First, the defendant PRPs claimed that in United States v. Atlantic Research Corp.,114551 U.S. 128 (2007). the Supreme Court held that cost recovery was only available for costs incurred “voluntarily.”115Bernstein, 733 F.3d at 208. The court dismissed this argument by saying that the defendant PRPs misread Atlantic Research’s interpretation of CERCLA, and, more importantly, there was no statutory basis for this distinction.116Id. at 208–10; see also Gaba, supra note 8, at 146. Second, as a policy argument, the PRPs who had not settled claimed that this reading of the word “resolved” discouraged settlements. They argued that the availability of a cause of action is the reason a PRP agrees to settle in the first place, so its unavailability decreases a PRP’s desire to settle.117Bernstein, 733 F.3d at 214. The court disagreed. Its interpretation did not leave the trustees with no legal recourse; in the absence of a contribution action, a cost-recovery action was available.118Id. Further, cost recovery has a longer statute of limitations, making it preferable to contribution.119Id. Moreover, if the availability of a contribution action were crucial to getting settlements, EPA could remove the condition precedent and structure the covenants not to sue to kick in immediately.120Id. EPA has since amended its model settlement to conform to Bernstein’s requirements. Gaba, supra note 8, at 158 & n.216.

Regarding the first AOC, the covenants not to sue had gone into effect, so liability had been resolved.121Bernstein, 733 F.3d at 204. Thus, the trustees could only bring a contribution action for the first AOC. But because the PRPs entered the AOC in 1999, the three-year statute of limitations ran in 2002, which the PRPs missed by six years, and so that claim was time-barred.122Id. at 207. A cost-recovery action was also unavailable because Bernstein, on a matter of first instance for the Seventh Circuit, determined that if a contribution action is available, it is the exclusive remedy. See id. at 206; see also supra text accompanying note 70.

Meanwhile, the Sixth Circuit seems to have an inconsistent approach regarding this question. On one hand, in RSR Corp. v. Commercial Metals Co., it held that an agreement resolved liability where the covenant not to sue was a condition precedent “even [though] the covenant . . . did not take effect until the remedial action was complete.”123496 F.3d 552, 558 (6th Cir. 2007). On the other hand, in cases following RSR Corp., Sixth Circuit panels adopted Bernstein’s logic and found that conditions precedent do not resolve liability. In Florida Power Corp. v. FirstEnergy Corp., the plaintiff utility company entered into two AOCs with EPA in 1998 and 2003.124810 F.3d 996, 998–99 (6th Cir. 2015). As in Bernstein and RSR Corp., cleaning up was the condition precedent to the covenant not to sue.125Fla. Power, 810 F.3d at 1003–04; supra text accompanying notes 60–67. In 2011, the utility sued another company under both cost recovery and contribution.126Fla. Power, 810 F.3d at 999. The defendant moved for judgment on the pleadings, arguing that the claim was time-barred because more than three years had passed since the most recent settlement.127Id. The district court agreed and dismissed the case.128See id.

On appeal, the Sixth Circuit found Bernstein “especially illuminating” and noted that the structure of the settlement there resembled the one before the court.129Id. at 1002. Next, the court distinguished RSR Corp. because it had not actually interpreted the various provisions of the consent decree at issue there.130Id. at 1007. In other words, the issue of whether the condition-precedent provisions resolved liability was not even before the court in RSR Corp.131Id. The Sixth Circuit did have the chance to consider this issue in Hobart Corp. v. Waste Management of Ohio, Inc.132758 F.3d 757 (6th Cir. 2014). There, the covenant went into effect immediately.133Hobart Corp., 758 F.3d at 769. There was no condition that would end the government’s obligation not to sue. The PRP was released from liability right away. As such, the court concluded that liability had been resolved.134Id. Florida Power combined Bernstein, Hobart Corp., and RSR Corp. to distinguish between covenants that go into effect immediately to potentially be undone by a condition subsequent and those that are later triggered by a condition precedent.135Fla. Power, 810 F.3d at 1007–09; supra text accompanying notes 60–67.

The Florida Power dissent’s main critique was that the distinctions the majority drew did not matter. The dissenting judge attacked the claim that there was a reasonable distinction between covenants that were immediately effective versus ones that would be triggered later, concluding that it “does not represent a rational basis for deciding which settlement agreements give rise to a contribution action.”136Fla. Power, 810 F.3d at 1019 (Suhrheinrich, J., dissenting). Additionally, the dissent also critiqued the court for giving weight to nonadmissions of liability.137Id. at 1016.

The flaws in Bernstein’s logic operate at the textual, statutory, and policy levels. Starting at the smallest unit of analysis, the plain meaning of the word “resolve” that the court articulated is incomplete. The court heavily cited past cases where “resolved” meant “someone was found liable.”138Bernstein v. Bankert, 733 F.3d 190, 211–12 (7th Cir. 2013) (collecting sources). For example, a cited case found the “question of liability was ‘resolved’ by the district court’s determination . . . that the defendant was liable.” Id. (citing Guzman v. City of Chicago, 689 F.3d 740, 745 (7th Cir. 2012)). Courts’ main job is to resolve cases by adjudicating them and assigning liability. But by focusing solely on courts, Bernstein ignored half of the definition it pulled from the dictionary. Though resolving something can mean “‘[t]o answer (a question),’”139Id. at 211 n.12 (alteration in original) (citing Resolve, Oxford English Dictionary (June 2020), (on file with the Michigan Law Review)). it can also mean to “‘settle or find a solution to (a problem, dispute, or contentious matter).’”140Id. (quoting Resolve, New Oxford American Dictionary (Angus Stevenson & Christine A. Lindberg eds., 3d ed. 2010)). At a general level, the problem in any CERCLA dispute is that EPA says a PRP is liable and the PRP disagrees, either on the existence or the amount of liability. This problem can be solved without PRPs admitting their liability or having a court conclusively decide they are liable.141By way of analogy, imagine two roommates have a disagreement. X thinks the apartment is too hot and thus leaves the window open, while Y thinks the apartment is too cold. Y gets a thick blanket and is now comfortable in the apartment with the window open. They have “resolved” their differences without actually having either: (1) a neutral third party (an RA, for example) weigh in on what the right temperature is or (2) X or Y admit that their individual temperature preference was incorrect. The court’s selective use of dictionary definitions is also wrong in another sense. “Admission” and “resolve” have different definitions.142See Fla. Power, 810 F.3d at 1016–17 (Suhrheinrich, J., dissenting) (“Admitting liability is not the same as resolving liability.”). The former focuses on someone stating that they have done something wrong, while the latter is merely finding an acceptable solution to a problem, without any mention of self-assigning misdeeds.143Compare Admission, Black’s Law Dictionary, supra note 61, with Resolve, Black’s Law Dictionary, supra note 61. Bernstein conflates the two by requiring an admission to achieve a resolution.

Even if one grants this myopic definition, Bernstein does not apply it correctly. The court said liability is resolved when there is no need to revisit that issue.144Bernstein, 733 F.3d at 211. Its logic was that making the completed cleanup a condition precedent to the covenant not to sue meant a “revisiting” could still happen because the government could sue at any time until the cleanup was complete. But Bernstein’s solution has the same problem. The court said that if the covenant not to sue became a condition subsequent—meaning that the government’s commitment to not sue took effect before cleanup was complete—that would resolve liability.145Id. at 213–14. But in fact, there could still be a need to revisit a condition-subsequent settlement. If the party fails to clean, then the covenant is no longer in effect and the government could file a suit, thereby revisiting the issue.146See Fla. Power, 810 F.3d at 1018 (Suhrheinrich, J., dissenting). For a case where litigation had to “revisit” an issue from a settlement, even though the covenant not to sue was a condition subsequent, see, for example, NCR Corp. v. George A. Whiting Paper Co., 768 F.3d 682, 689, 692 (7th Cir. 2014). Further, even an unconditional covenant not to sue does not end things; if a party violates that settlement, then the government will sue for breach of contract.147Fla. Power, 810 F.3d at 1018 n.10 (Suhrheinrich, J., dissenting). For cases where the government sued for breach of contract, see, for example, Hobart Corp. v. Waste Management of Ohio, Inc., 758 F.3d 757, 769 (6th Cir. 2014), and Refined Metals Corp. v. NL Industries, Inc., No. 1:17-cv-02565-SEB-TAB, 2018 WL 4592110, at *5–6 (S.D. Ind. Sept. 25, 2018). Overall, there is no satisfying justification for making the availability of a contribution claim hinge on a condition subsequent versus precedent; no matter how you structure a settlement, one party could always fail to live up to its obligations and require litigation to revisit the issue.148See Fla. Power, 810 F.3d at 1019 (Suhrheinrich, J., dissenting). While EPA could solve this problem by changing the drafting of agreements, see Gaba, supra note 8, at 158, there is no basis in law to require this solution in the first place.

Bernstein’s definition of “resolved” also leads to four serious inconsistencies across the whole of CERCLA, violating the maxim to read statutes as a whole.149See United States v. Atl. Rsch. Corp., 551 U.S. 128, 135 (2007); see also Nina A. Mendelson, Change, Creation, and Unpredictability in Statutory Interpretation: Interpretive Canon Use in the Roberts Court’s First Decade, 117 Mich. L. Rev. 71, 101 (2018) (noting the importance of the whole act canon). First, CERCLA provides that participation in a settlement cannot be considered an admission of liability.15042 U.S.C. § 9622(d)(1)(B). It is odd to say that a PRP need not worry about being found liable for attempting to resolve liability with the government, but then to force them to admit liability to find that resolution.151See Gaba, supra note 8, at 157–58. Second, EPA cannot forfeit its ability to sue unless the president certifies that the cleanup was completed satisfactorily.15242 U.S.C. § 9622(f)(3). Yet Bernstein would take that requirement from section 122 and essentially graft it onto section 113(f)(3)(B), where that language is not found at all.153Gaba, supra note 8, at 158. Third, besides simply being strange, such a requirement would make a contribution claim impossible. The statute of limitations for a section 113(f)(3)(b) claim begins to run when the PRPs enter settlement.15442 U.S.C. § 9613(g)(3)(B). Given that cleanups often take years,155E.g., Pidot & Ratliff, supra note 26, at 195. Bernstein means that one’s contribution action is not available until several years after the statute of limitations has expired.156ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1124 n.8 (9th Cir. 2017). That result is incongruous.157ASARCO LLC v. Atl. Richfield Co., 73 F. Supp. 3d 1285, 1288 (D. Mont. 2014) (“Neither the law nor common logic supports the concept that a statute of limitations could run on a claim that has not yet accrued.”), vacated on other grounds, 866 F.3d 1108.

Fourth, and finally, Bernstein might require a PRP to change its claim midsuit. Recall that every circuit, including the Seventh in Bernstein, agrees that contribution and cost recovery are mutually exclusive.158ASARCO, 866 F.3d at 1117 (collecting cases). But consider this scenario: The plaintiffs in Bernstein commence with their cost-recovery action. During the litigation, the plaintiffs finish cleaning up and get certification, and then the covenants kick in.159This scenario is not out of the realm of possibility given that often one of the only things that takes longer than cleaning up a Superfund site is litigating over it. Compare ASARCO, 73 F. Supp. 3d at 1291 (remarking that CERCLA is known as “the lawyer employment act”), with Pidot & Ratliff, supra note 26, at 195 (describing a cleanup in New Jersey anticipated to take six years). What happens next is unclear. Potentially, the cost-recovery action is left unscathed—despite now being technically invalid. Maybe it gets converted into a contribution action. If neither happens, plaintiffs could become vulnerable to a dispositive motion for their cost-recovery claim because they now have a contribution action, which is mutually exclusive with cost recovery. If they tried to file a brand-new contribution action, the three-year statute of limitations has likely passed. This irregularity—which this Note terms “claim slippage”—creates a glaring hole in how the two causes of action interact.

Bernstein’s reading is also counter to the policy undergirding section 113(f)(3)(B): encouraging settlements.160See Kilbert, supra note 81, at 1070. The court said its reading achieved CERCLA’s goals by still giving the trustees a cause of action—the better one.161See supra text accompanying notes 117–119. But Bernstein confused contribution’s means—settlement—with its ends—ensuring sites actually get cleaned.162See Kilbert, supra note 81, at 1070. Though it is true that Bernstein still gives parties a cause of action, which is an incentive to settle, it undermines the end of cleanup. Under Bernstein, PRPs are disincentivized from actually fulfilling their obligations under a settlement with EPA because if they clean up a site, they lose the very cause of action that Bernstein grants.163As a hypothetical, imagine two settling PRPs. E just received certification that it cleaned up well enough. E now loses its cost-recovery claim, because it is mutually exclusive with contribution. See supra text accompanying note 70. Meanwhile, F has not yet received certification that it cleaned up well, but it has spent money cleaning up. So, F has a cost-recovery claim. See supra text accompanying note 71. Bernstein says that E’s reward for its certifiably good behavior is the weaker cause of action. F’s “reward” is the stronger cause of action, even though F has not yet proven itself. Further, contribution protection only bars contribution claims. So, (the questionable PRP) F can sue (the good actor) E, while the good actor has no cause of action against the questionable one (initially at least—there is the contribution counterclaim). Note that this particular scenario is not rare; in fact, “the private plaintiff in a [cost-recovery] case may be one of the most significant, if not the most significant, contributor to contamination at the site.” Kilbert, supra note 81, at 1075. Further, Bernstein shifts more private-party CERCLA cases to cost recovery rather than contribution, which commentators view as the incorrect direction.164See infra notes 225–232 and accompanying text.

B.    “Certainty and Finality”: The Ninth Circuit’s Approach

To “resolve” liability in the Ninth Circuit, the PRP and the government must have determined liability with “certainty and finality.”165ASARCO LLC v. Atl. Richfield Co., 866 F.3d 1108, 1124 (9th Cir. 2017). In ASARCO LLC v. Atlantic Richfield Co., the plaintiff ASARCO operated a lead smelter in Montana from 1888 to 2001.166Id. at 1114. The Anaconda Mining Company, the predecessor of the defendant Atlantic Richfield, operated a zinc fuming plant in the same area.167Id. After over a century of production, the area was heavily polluted with hazardous waste.168Id. In 1998, the United States brought claims against ASARCO under the Resource Conservation and Recovery Act (RCRA)169Resource Conservation and Recovery Act of 1976, Pub. L. No. 94-580, 90 Stat. 2795 (codified in scattered sections of 42 U.S.C.). and the Clean Water Act17033 U.S.C. §§ 1251–1387. (instead of CERCLA), and ASARCO and the United States entered their first consent decree.171See ASARCO, 866 F.3d at 1114. The settlement only dealt with claims for civil penalties. Id. at 1126. The United States reserved its rights, multiple times, to sue for claims related to responses actions or their costs. Id. This distinction turned out to be crucial for the disposition of this case. See infra notes 196–197 and accompanying text. ASARCO failed to meet its obligations under that 1998 decree and filed for bankruptcy in 2005.172ASARCO, 866 F.3d at 1114–15. In 2009, ASARCO and the United States entered a new consent decree that established a trust fund to oversee the cleanup, into which ASARCO paid nearly $100 million.173Id. at 1115. The new 2009 decree had a covenant not to sue that was effective immediately, conditioned on ASARCO fulfilling its obligations.174Id. at 1128. This was a condition subsequent, even if the court did not use those words. See supra text accompanying notes 60–67.

On June 5, 2012, exactly three years later,175Remember that contribution actions have a three-year statute of limitations. 42 U.S.C. § 9613(g)(3). ASARCO brought a contribution action against Atlantic Richfield.176ASARCO, 866 F.3d at 1115. Atlantic Richfield moved for summary judgment, arguing that the statute of limitations had begun to run with the 1998 decree and thus ASARCO’s claim had expired over a decade earlier.177Id. The district court held that the 1998 RCRA decree did trigger the availability of ASARCO’s CERCLA contribution claim and dismissed the claim for being untimely.178Id. On appeal, the case presented three issues of first impression for the Ninth Circuit, two of which waded into a circuit split.179Id. at 1113; see also Eric A. Rey, CERCLA Contribution: Ninth Circuit Addresses Two Circuit Splits, 18 Pratt’s Energy L. Rep. 3, 3 (2018). As a preliminary matter, the court held “that a non-CERCLA settlement agreement may form the necessary predicate for a § 113(f)(3)(B) contribution action.”180ASARCO, 866 F.3d at 1120–21.

Having determined that a non-CERCLA agreement could trigger a contribution action, the court then considered whether either the 1998 Decree or the 2009 one in fact did. The court zeroed in on the same issue that the Seventh Circuit did in Bernstein: “[I]s [a PRP’s] liability ‘resolved’ where the government reserves certain rights, or where the party refuses to concede liability?”181Id. at 1122–23. The court answered this question by saying liability is “resolved” when “a settlement agreement . . . determine[s] a PRP’s compliance obligations with certainty and finality.”182Id. at 1124 (emphasis added). It reached this conclusion by citing both dictionaries, including some of the same ones that Bernstein cited,183Compare id. at 1122, with Bernstein v. Bankert, 733 F.3d 190, 210–11 n.12 (7th Cir. 2013). and Bernstein’s own summation of the varying judicial definitions of “resolved.”184ASARCO, 866 F.3d at 1122 (quoting Bernstein, 733 F.3d at 211). Despite using common sources, the ASARCO court disagreed with the Sixth and Seventh Circuits’ conclusion.185Id. at 1124.

The ASARCO court first argued that because the president has to sign off on every cleanup before a covenant not to sue can go in effect, “it is unlikely that a settlement agreement could ever resolve a party’s liability.”186Id. The court was unclear why exactly this would never happen; potentially there is an unstated assumption that presidential certification is rare. Second, the ASARCO court pointed out the problem of claim impossibility187See supra text accompanying notes 154–157.: it would defeat CERCLA’s purpose to have the statute of limitations begin to run before the claim is even available.188ASARCO, 866 F.3d at 1124 n.8. Third, the ASARCO court discussed legislative history. A House committee had identified a goal of using contribution as a settlement incentive, while at the same time expressing distaste for EPA’s use of releases of liability as settlement incentives because the releases undermined public health and the environment.189Id. at 1125 (quoting H.R. Rep. No. 99-253, pt. 1, at 80 (1985)). Under the Bernstein court’s reading, the House supposedly wanted contribution to be available without any releases of liability but wrote a statute requiring the two to be together. According to the ASARCO court, this result would be illogical.190Id. Fourth, the ASARCO court noticed the mismatch between requiring a party to admit liability and statutory purpose. SARA’s purpose was to encourage settlement and quick cleanups.191Id. Whether or not a PRP admits liability has nothing to do with either of those goals.192Id. In fact, requiring an admission could potentially discourage settlements because admitting liability opens a PRP up to a cost-recovery action,193Id. (citing 42 U.S.C. § 9607). for which there is no contribution protection.194United States v. Atl. Rsch. Corp., 551 U.S. 128, 140 (2007).

Having determined that liability is resolved when the settlement determines a PRP’s obligations with “certainty and finality,”195ASARCO, 866 F.3d at 1124. the court went on to apply this standard to the two settlements between ASARCO and the government. The 1998 RCRA settlement did not resolve liability for a response action.196Id. at 1125. That release was only for civil penalties. Id. at 1126. In fact, the court sounds almost exasperated recounting the number of times that the liability release was cabined not to include cleanup costs or actions. See id. (“Lest there be any doubt, the Decree makes the point at least three more times.”). Because liability had not been resolved, no contribution claim attached, nor could the statute of limitations have expired on a nonexistent claim.197Id. at 1126.

But the 2009 CERCLA settlement did resolve liability.198Id. at 1128. The ASARCO court cited several elements of the settlement that satisfied this new “certainty and finality” test. First, the covenant not to sue went into effect right away.199Id. Again, the court did not call this a condition subsequent, but it was one. See supra text accompanying notes 60–67. Second, the reservation of rights only allowed suits for future acts that create liability.200ASARCO, 866 F.3d at 1128–29. Thus, using an implicit expressio unius argument, the court determined that because the 2009 Decree said the government could sue only for future acts, all current or past acts had been resolved with certainty and finality.201See id.; Expressio unius est exclusio alterius, Black’s Law Dictionary, supra note 61. Third, the Decree capped ASARCO’s total financial obligations at a specific amount.202ASARCO, 866 F.3d at 1129. Fourth, ASARCO received contribution protection, which is further evidence that a PRP intended to resolve liability.203Id. (citing Hobart Corp. v. Waste Mgmt. of Ohio, Inc., 758 F.3d 757, 768–69 (6th Cir. 2014)). As a result, ASARCO had a contribution claim available, which it timely filed on the last possible day before the statute of limitations expired.204Id. at 1115, 1127. Accordingly, the Ninth Circuit remanded the case to the district court for disposition on the merits.205Id. at 1129. On remand, the district court awarded ASARCO over million in addition to prejudgment interest and attorney fees. ASARCO LLC v. Atl. Richfield Co., 353 F. Supp. 3d 916, 958 (D. Mont. 2018).

The problem with ASARCO is that it creates an unclear standard that does not seem to practically differ from the Bernstein standard. ASARCO did not define the phrase “certainty and finality” in more detail than discussed in this Section. Some observers have tried to provide guidance for litigants,206See, e.g., Rey, supra note 179, at 8; Whitney Jones Roy, Whitney Hodges & Alison N. Kleaver, Ninth Circuit Weighs in on Circuit Split Regarding CERCLA Contribution Claims After Settlement and the Statute of Limitation, Env’t Litig. & Toxic Torts Comm. Newsl., Dec. 2017, at 3, 4–5. but no court has provided a detailed analysis.207In fact, the only real post-ASARCO interpretation of this standard happened when a D.C. district court rejected it as a valid test to apply, and even then, the appellate court reversed that ruling. See Gov’t of Guam v. United States, 341 F. Supp. 3d 74 (D.D.C. 2018), rev’d on other grounds, 950 F.3d 104 (D.C. Cir. 2020); see also supra note 93. Even then, when one dives into the particular facts of this split, it is not clear that there is that much disagreement.208But see, e.g., Cooper Crouse-Hinds, LLC v. City of Syracuse, 16-CV-1201, 2018 WL 840056, at *6 (N.D.N.Y. Feb. 12, 2018) (“There is a great deal of inconsistency among the cases addressing consent orders that conditionally resolve a PRP’s CERCLA liability and whether those consent orders trigger the requirement to proceed under § 113(f)(3)(B).”); Rey, supra note 179, at 3, 6. In ASARCO, the covenant not to sue was effective immediately. As long as ASARCO fulfilled its obligations under the settlement, it would be released from any response obligations.209ASARCO, 866 F.3d at 1128. That is squarely a condition subsequent.210See supra text accompanying notes 60–67. When the Seventh Circuit had a chance to visit the same issue, it interpreted Bernstein to mean the same thing: conditions subsequent resolve liability.211See NCR Corp. v. George A. Whiting Paper Co., 768 F.3d 682, 692 (7th Cir. 2014). Because both the Ninth and the Seventh Circuits’ tests would lead to the same result in ASARCO, it is uncertain what “certainty and finality” adds to the result. This leaves parties further in the dark, and confusion only further disincentivizes settlements.212Kilbert, supra note 81, at 1069.

III.   The Bright-Line Rule “Resolves” the Circuit Split

This Note rejects both the hard-line approach of Bernstein and the open-ended standard in ASARCO. Instead, it proposes a liberal bright-line approach: any conditional or unconditional covenant not to sue resolves liability and triggers a section 113(f)(3)(B) contribution action. This Part will discuss how this bright-line rule is consistent with CERCLA’s text, structure, and policy.

This bright-line reading of the word “resolved” reflects its plain meaning. Bernstein put too much weight on the “no need to revisit” portion of “resolve” and not enough on the “to solve a problem” part.213See supra text accompanying notes 138–148. Before a PRP settles, its potential liability is nearly boundless. Given CERCLA’s joint and several liability, the PRP could be found liable for all damages allowed under CERCLA. In contrast, a settlement, whether conditional or not, “converts a party’s ‘joint and several’ liability to a discrete set of requirements that define the totality of the obligation of the settling party with respect to the matters addressed in the settlement.”214Gaba, supra note 8, at 157 n.212. Even if a party violates the agreement, the subsequent suit will be about whether the party followed that agreement, rather than the complete gamut of CERCLA liabilities. In other words, the initial problem was determining the extent of the PRP’s liability, and the settlement resolved that problem. To use Bernstein’s phrasing, a settlement, even one with a condition precedent, largely sets the terms of the debate with “no need to revisit.”215Bernstein v. Bankert, 733 F.3d 190, 211 (7th Cir. 2013). Relatedly, whether the PRP admitted liability is irrelevant in this analysis. Admitting or denying liability does not change the substance of a PRP’s obligations under the settlement.216See Gaba, supra note 8, at 158.

The bright-line rule also solves the macro statutory problems that Bernstein’s reading poses.217For an analysis of these problems, see supra text accompanying notes 149–159. First, if admission and resolution would no longer be tied together, it would avoid conflict with section 122’s evidentiary ban on using the existence of settlements to infer an admission of liability. Second, the requirement that the president sign off on a cleanup is then no longer improperly read as a requirement for a contribution action. Thus, section 122 would no longer be grafted onto section 113. Third, this reading solves Bernstein’s claim-impossibility problem. If a covenant not to sue is itself enough to trigger resolution, then a contribution claim is available right upon entering settlement. That time matches when the statute of limitations begins to run and solves the disharmony of having the statute of limitations run before the claim is even available. Fourth, there is no longer any claim slippage. Currently, there is no statutory guidance as to what happens to a PRP’s cause of action if liability is suddenly resolved midsuit. But here, because the rule ties resolution to the covenant’s existence and not its condition precedent or subsequent, liability will always be resolved presuit.

Although recent CERCLA case law has ignored policy concerns,218Gaba, supra note 8, at 138. the bright-line rule still supports CERCLA’s policy goals—in this case, to promote settlements.219Kilbert, supra note 81, at 1070. Increasing settlements, which will lead to prompter cleanups, is a worthy goal of both CERCLA and the judicial system writ large.220McVean & Pidot, supra note 52, at 206. A contribution claim is an incentive to settle,221See supra text accompanying note 55. and anything that strengthens this incentive will further CERCLA’s goals.222Cf. Fuller, supra note 55, at 248 (describing how weakening a different settlement incentive would decrease settlements). Making this benefit more certain to PRPs ex ante is a way to strengthen it.223Cf. id. (“Private parties benefit from . . . the certainty of entering into a negotiated settlement . . . .”). This bright-line rule is administrable—the full inquiry is whether or not there is a covenant not to sue—making contribution’s availability all but certain.224The same logic can apply also to contribution protection. Given that it is another settlement incentive, see, e.g., id. at 248–49, strengthening it would promote CERCLA’s policy, and making its availability more certain would strengthen it. Like a contribution action, the settlement bar only covers PRPs that have “resolved” their liability. 42 U.S.C. §§ 9613(f)(3)(B), (f)(2). Identical terms within different parts of the same statute are generally given similar meaning. E.g., 2A Norman J. Singer & Shambie Singer, Sutherland Statutes and Statutory Construction § 46:6, Westlaw (database updated Apr. 2020). Thus, covenants not to sue bring a PRP under the settlement bar’s ambit—the same clarity-promoting rule as for contribution actions.

Additionally, this bright-line proposal would encourage private CERCLA claimants to file contribution rather than cost-recovery claims—a goal supported by the vast majority of commentators and legal scholars.225See, e.g., Kilbert, supra note 81, at 1069–77; Pidot & Ratliff, supra note 26, at 229–61. Mainly, they disfavor cost-recovery claims because giving private parties an end-run around contribution protection (also known as the “settlement bar”) vitiates much of CERCLA. Recall that United States v. Atlantic Research Corp. held that contribution protection gives settling PRPs immunity from only contribution, not cost-recovery, actions.226551 U.S. 128, 140 (2007). Ex post, this holding upset cases settled pre-Atlantic Research where PRPs thought they had already “bought their peace”227Fuller, supra note 55, at 244. with CERCLA; ex ante, the now-incomplete protection of the settlement bar disincentivizes settlement, both in the total number and the average value of each.228Pidot & Ratliff, supra note 26, at 223. Settling PRPs are not just victims in this scheme; there is a procedural machination for them to game the system as well. See Kilbert, supra note 81, at 1079–80; Pidot & Ratliff, supra note 26, at 226.

The severity of this problem is contested. Atlantic Research said that its holding would not “eviscerate” contribution protection because a settling PRP sued for cost recovery could file a contribution counterclaim to trigger equitable appointment.229Atl. Rsch., 551 U.S. at 140–42. Applying equitable principles, “a district court . . . would undoubtedly consider any prior settlement [as part of] the liability calculus.”230Id. at 142. But a benefit of the settlement bar is immunity from more CERCLA litigation, not just an advantage on the merits.231See Pidot & Ratliff, supra note 26, at 250 n.313. Further, subjecting courts to more contribution counterclaims wastes judicial resources by forcing parties to undertake additional procedural moves and merits arguments.232Kilbert, supra note 81, at 1079–80. Granted, the proposals cited here solve this problem by making a cost-recovery claim unavailable to a private party in the first instance, while the bright-line proposal would make contribution claims easier to obtain—leading to fewer cost-recovery actions because the two are mutually exclusive.233See supra text accompanying note 70. Still, this bright-line reading of “resolved” would make contribution actions easily available, and the fewer private cost-recovery actions, the better.

The bright-line approach makes it easier for settlements to resolve liability, but it does not impermissibly equate “resolve” with “settle.”234But see Atlanta Gas Light Co. v. UGI Utils., Inc., 463 F.3d 1201 (11th Cir. 2006). The court’s entire forty-four-word analysis essentially equated “resolving” with “settling.” See id. at 1204. This reading is wrong on its face. If Congress meant that simply “settling” is a sufficient trigger for a contribution action, it would have used the word “settled” instead of “resolved.” Cf. Singer & Singer, supra note 224, § 46:6 (“Different words used in the same . . . statute are assigned different meanings . . . .”). This served more as a cursory way to find subject-matter jurisdiction than an actual entry into this circuit split. See Atlanta Gas Light, 463 F.3d at 1203–04. In fact, the Eleventh Circuit has only cited Atlanta Gas Light for this proposition one other time. See Solutia, Inc. v. McWane, Inc., 672 F.3d 1230, 1236 (11th Cir. 2012). Moreover, none of the cases in the actual split cite it, either. See supra Part II. The vast majority of CERCLA settlements with the government will have covenants not to sue;235See Gaba, supra note 8, at 157 (“Resolution of liability with the United States is generally expressed in CERCLA settlements by inclusion of a ‘covenant not to sue’ . . . .”). so, most CERCLA settlements would, under the bright-line rule, resolve liability. However, the covenants serve an additional function that a settlement without them would not address. A covenant attempts to end the whole dispute while also cabining the scope of any future dispute that might arise from the PRP failing to meet an obligation under the settlement. That limited scope decreases the complexity of a potential CERCLA suit should one reemerge between a PRP and the government. Given the cost and complexity of contribution suits,236Sean Murphy, Duck, Duck, Sued!—CERCLA’s Game of Contribution Tag, Geo. Env’t L. Rev. Online (Apr. 27, 2019), []. setting the terms of debate ahead of time is crucial. So, tying covenants to the meaning of “resolved” will make a resolution limit future complexity, which merely settling might not do. Further, Congress designed contribution suits to be easy to obtain and covenants not to sue harder.237See supra text accompanying notes 189–190. So, an approach that brings the meaning of “resolve” closer to “settle” is appropriate.

Likewise, though a bright-line rule makes it easier to resolve liability, and so obtain a contribution action or the settlement bar’s protection, it does not allow the government to provide undue sweetheart deals to polluters. CERCLA has detailed procedures for how the United States can settle with PRPs.238McVean & Pidot, supra note 52, at 206. For example, a longer-term remedial action must be entered as a consent decree and judicially approved.23942 U.S.C. § 9622(d)(1)(A). The agency must also run a proposed settlement through a notice-and-comment procedure where it must “consider any comments filed . . . in determining whether or not to consent to the proposed settlement.”240Id. § 9622(i)(3). Other environmental statutes have similar notice-and-comment provisions. McVean & Pidot, supra note 52, at 206–07. These procedures allow courts to step in when an agency oversteps its bounds241See generally McVean & Pidot, supra note 52 (cataloging various types of settlement agreements in the context of environmental protection and identifying how each category allows for effective judicial review).—in this case failing to use CERCLA to protect the environment.


In advancing its goal of cleaning up toxic waste sites on the polluter’s dime, CERCLA plays with some high stakes: it reallocates billions of dollars while ameliorating the most acute public health crises and environmental disasters. It is a complicated—often poorly drafted—statute that uses many tools to reach its goals. One of these tools includes a cause of action for private parties who paid more than their fair share to recoup that overpayment. But first that party must seek an apparently elusive resolution of its liability with the government. Courts’ attempts to define the nature of this resolution have polluted the statute, either by creating uncertainty for litigants or undermining its statutory scheme.

This Note responds by proposing a new bright-line reading: any covenant not to sue, conditional or otherwise, resolves liability and thus leads to a contribution action. Overall, this rule would create administrable certainty and bring statutory cohesiveness to CERCLA. For a complex statute that is about the equally byzantine process of hazardous-waste remediation, it would be fitting for CERCLA to be cleaned up with a simple solution.

*J.D. Candidate, May 2021, University of Michigan Law School. Writing is not a solitary process, and this piece would not have come to fruition without the edits and critiques of Professor Oday Salim, Sophia Gonzalez and the Volume 118 Michigan Law Review Notes Office, editors at all levels of Volume 119, and my friend in copy editing, Zachary Beck. Of course, I also thank my family for their unorthodox, but always effective, love and support.