United States of America v. Cannons Engineering Corp., Appeal of Olin Hunt Specialty Products, Inc., United States of America v. Cannons Engineering Corp., Appeal of Cyn Oil Corp., United States of America v. Cannons Engineering Corp., Appeal of Beggs & Cobb Corporation, Etc., United States of America v. Cannons Engineering Corp., Appeal of Scott Brass, Inc., United States of America v. Cannons Engineering Corp., Appeal of Kingston-Warren Corporation, United States of America v. Cannons Engineering Corp., Appeal of Crown Roll Leaf, Inc.

"Superfund" sites are those which require priority remedial attention because of the presence, or suspected presence, of a dangerous accumulation of hazardous wastes. Expenditures to clean up such sites are specially authorized pursuant to 42 U.S.C. Sec. 9611 (1987). After the federal government, through the United States Environmental Protection Agency (EPA),1 identified four such sites in Bridgewater, Massachusetts, Plymouth, Massachusetts, Londonderry, New Hampshire, and Nashua, New Hampshire (collectively, the Sites), the EPA undertook an intensive investigation to locate potentially responsible parties (PRPs). In the course of this investigation, the agency created a de minimis classification (DMC), putting in this category persons or firms whose discerned contribution to pollution of the Sites was minimal both in the amount and toxicity of the hazardous wastes involved. See 42 U.S.C. Sec. 9622(g) (1987). The agency staked out the DMC on the basis of volumetric shares, grouping within it entities identifiable as generators of less than one percent of the waste sent to the Sites. To arrive at a PRP's volumetric share, the agency, using estimates, constituted a ratio between the volume of wastes that the PRP sent to the Sites and the total amount of wastes sent there.

The EPA sent notices of possible liability to some 671 PRPs, including generators and nongenerators. Administrative settlements were thereafter achieved with 300 generators (all de minimis PRPs). In short order, the United States and the two host states, Massachusetts and New Hampshire, brought suits in the United States District Court for the District of Massachusetts against 84 of the PRPs who had rejected, or were ineligible for, the administrative settlement. The suits sought recovery of previously incurred cleanup costs and declarations of liability for future remediation under the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA), 42 U.S.C. Secs. 9601-9675 (1987). The actions were consolidated.

With its complaint, the United States filed two proposed consent decrees. The first (the MP decree) embodied a contemplated settlement with 47 major PRPs, that is, responsible parties who were ineligible for membership in the DMC. This assemblage included certain generators whose volumetric shares exceeded the 1% cutoff point and certain nongenerators (like the owners of the Sites and hazardous waste transporters). The second consent decree (the DMC decree) embodied a contemplated settlement with 12 de minimis PRPs who had eschewed participation in the administrative settlement. As required by statute, notice of the decrees' proposed entry was published in the Federal Register. 53 Fed.Reg. 29,959 (Aug. 9, 1988). No comments were received.

The government thereupon moved to enter the decrees. Seven non-settling defendants objected.2 After considering written submissions and hearing arguments of counsel, the district court approved both consent decrees and dismissed all cross-claims against the settling defendants. United States v. Cannons Engineering Corp., 720 F.Supp. 1027, 1052-53 (D.Mass.1989). The court proceeded to certify the decrees as final under Fed.R.Civ.P. 54(b). Id. These appeals followed.

* We approach our task mindful that, on appeal, a district court's approval of a consent decree in CERCLA litigation is encased in a double layer of swaddling. In the first place, it is the policy of the law to encourage settlements. See, e.g., Donovan v. Robbins, 752 F.2d 1170, 1177 (7th Cir.1985); City of New York v. Exxon Corp., 697 F.Supp. 677, 692 (S.D.N.Y.1988). That policy has particular force where, as here, a government actor committed to the protection of the public interest has pulled the laboring oar in constructing the proposed settlement. See F.T.C. v. Standard Financial Management Corp., 830 F.2d 404, 408 (1st Cir.1987) (discussing need for judicial deference "to the agency's determination that the settlement is appropriate"); S.E.C. v. Randolph, 736 F.2d 525, 529 (9th Cir.1984) (similar). While "the true measure of the deference due depends on the persuasive power of the agency's proposal and rationale, given whatever practical considerations may impinge and the full panoply of the attendant circumstances," Standard Financial, 830 F.2d at 408, the district court must refrain from second-guessing the Executive Branch.

Respect for the agency's role is heightened in a situation where the cards have been dealt face up and a crew of sophisticated players, with sharply conflicting interests, sit at the table. That so many affected parties, themselves knowledgeable and represented by experienced lawyers, have hammered out an agreement at arm's length and advocate its embodiment in a judicial decree, itself deserves weight in the ensuing balance. See Exxon, 697 F.Supp. at 692. The relevant standard, after all, is not whether the settlement is one which the court itself might have fashioned, or considers as ideal, but whether the proposed decree is fair, reasonable, and faithful to the objectives of the governing statute. See Durrett v. Housing Authority, 896 F.2d 600, 603-04 (1st Cir.1990) (describing district court's task). Thus, the first layer of insulation implicates the trial court's deference to the agency's expertise and to the parties' agreement. While the district court should not mechanistically rubberstamp the agency's suggestions, neither should it approach the merits of the contemplated settlement de novo.

The second layer of swaddling derives from the nature of appellate review. Because approval of a consent decree is committed to the trial court's informed discretion, see id. 896 F.2d at 603-04; United States v. Hooker Chemical & Plastics Corp., 776 F.2d 410, 411 (2d Cir.1985); In re AWECO, Inc., 725 F.2d 293, 297 (5th Cir.), cert. denied, 469 U.S. 880, 105 S.Ct. 244, 83 L.Ed.2d 182 (1984), the court of appeals should be reluctant to disturb a reasoned exercise of that discretion. In this context, the test for abuse of discretion is itself a fairly deferential one. We recently addressed the point in the following terms:

Judicial discretion is necessarily broad--but it is not absolute. Abuse occurs when a material factor deserving significant weight is ignored, when an improper factor is relied upon, or when all proper and no improper factors are assessed, but the court makes a serious mistake in weighing them.

Independent Oil & Chemical Workers of Quincy, Inc. v. Procter & Gamble Mfg. Co., 864 F.2d 927, 929 (1st Cir.1988). Unless the objectors can demonstrate that the trier made a harmful error of law or has lapsed into "a meaningful error in judgment," Anderson v. Cryovac, Inc., 862 F.2d 910, 923 (1st Cir.1988), a reviewing tribunal must stay its hand. The doubly required deference--district court to agency and appellate court to district court--places a heavy burden on those who purpose to upset a trial judge's approval of a consent decree.II

With this introduction, we turn to our twice-swaddled assessment of the decrees here at issue. In beginning, we abjure an exegetic description of the decrees themselves or of the factual/legal background upon which they are superimposed, instead referring the motivated reader to the district court's comprehensive description of the governments' claims, Cannons, 720 F.Supp. at 1031-32; the administrative settlement, id. at 1033; the MP decree, id. at 1033-34; and the DMC decree, id. at 1034-35. We note only a few of the decrees' historical antecedents.

Originally, the EPA extended an open offer to all de minimis PRPs, including five of the six appellants,3 proposing an administrative settlement based on 160% of each PRP's volumetric share of the total projected response cost, that is, the price of remedial actions, past and anticipated. See id. at 1030 n. 1. The settlement figure included a 60% premium to cover unexpected costs and/or unforeseen conditions. Settling PRPs paid their shares in cash and were released outright from all liability. They were also exempted from suits for contribution, see 42 U.S.C. Sec. 9622(g)(5) (1987).

Following consummation of the administrative settlement, plaintiffs entered into negotiations with the remaining PRPs. These negotiations resulted in the proposed MP decree (accepted by 47 "major" defendants) and the DMC decree. The terms of the former have been memorialized in the opinion below, 720 F.Supp. at 1034, and do not bear repeating. The latter was modelled upon the administrative settlement, but featured an increased premium: rather than allowing de minimis PRPs to cash out at a 160% level, an eligible generator could resolve its liability only by agreeing to pay 260% of its volumetric share of the total projected response cost. The EPA justified the incremental 100% premium as being in the nature of delay damages.

With this admittedly sketchy background, we proceed with our consideration of the instant appeals, engaging in independent discussion of particular facts and decree provisions only to the extent required to afford needed perspective.

The lower court having made extensive, meticulously detailed findings in respect to the consent decrees, see Cannons, 720 F.Supp. at 1035-47, we see no point in repastinating well-ploughed terrain. We choose instead to set forth our general views as to the criteria that a district court should use in determining whether to approve a consent decree in the CERCLA context, explaining in the process why we believe the rulings below to be unimpugnable.

Our starting point is well defined. The Superfund Amendments and Reauthorization Act of 1986 (SARA), P.L. 99-499, Sec. 101 et seq., 100 Stat. 1613, authorized a variety of types of settlements which the EPA may utilize in CERCLA actions, including consent decrees providing for PRPs to contribute to cleanup costs and/or to undertake response activities themselves. See 42 U.S.C. Sec. 9622 (1987). SARA's legislative history makes pellucid that, when such consent decrees are forged, the trial court's review function is only to "satisfy itself that the settlement is reasonable, fair, and consistent with the purposes that CERCLA is intended to serve." H.R.Rep. No. 253, Pt. 3, 99th Cong., 1st Sess. 19 (1985), reprinted in 1986 U.S.Code Cong. & Admin.News 3038, 3042. Reasonableness, fairness, and fidelity to the statute are, therefore, the horses which district judges must ride.

That said, we are quick to concede that these three steeds are all mutable figures taking on different forms and shapes in different factual settings. Yet, the concepts' amorphous quality is no accident or quirk of fate. We believe that Congress intended, first, that the judiciary take a broad view of proposed settlements, leaving highly technical issues and relatively petty inequities to the discourse between parties; and second, that the district courts treat each case on its own merits, recognizing the wide range of potential problems and possible solutions. When a court considers approval of a consent decree in a CERCLA case, there can be no easy-to-apply check list of relevant factors.

A. Procedural Fairness.

We agree with the district court that fairness in the CERCLA settlement context has both procedural and substantive components. Cannons, 720 F.Supp. at 1039-40. To measure procedural fairness, a court should ordinarily look to the negotiation process and attempt to gauge its candor, openness, and bargaining balance. See, e.g., id. at 1040; United States v. Rohm & Haas Co., 721 F.Supp. 666, 680-81 (D.N.J.1989); Kelley v. Thomas Solvent Co., 717 F.Supp. 507, 517-18 (W.D.Mich.1989); In re Acushnet River & New Bedford Harbor, 712 F.Supp. 1019, 1031 (D.Mass.1989); Exxon, 697 F.Supp. at 693; State of New York v. Town of Oyster Bay, 696 F.Supp. 841, 844-45 (E.D.N.Y.1988); United States v. Hooker Chemicals & Plastics Corp., 540 F.Supp. 1067, 1080 (W.D.N.Y.1982).

In this instance, the district court found the proposed decrees to possess the requisite procedural integrity, Cannons, 720 F.Supp. at 1040-41, and appellants have produced no persuasive reason to alter this finding. It is clear the district court believed that the government conducted negotiations forthrightly and in good faith, and the record is replete with indications to that effect. Most of appellants' contrary intimations are vapid and merit summary rejection. But their flagship argument--that the procedural integrity of the settlement was ruptured because appellants were neither allowed to join the MP decree nor informed in advance that they would be excluded--requires comment.

Appellants claim that they were relatively close to the 1% cutoff point, and were thus arbitrarily excluded from the major party settlement, avails them naught. Congress intended to give the EPA broad discretion to structure classes of PRPs for settlement purposes. We cannot say that the government acted beyond the scope of that discretion in separating minor and major players in this instance, that is, in determining that generators who had sent less than 1% of the volume of hazardous waste to the Sites would comprise the DMC and those generators who were responsible for a greater percentage would be treated as major PRPs. While the dividing line was only one of many which the agency could have selected, it was well within the universe of plausibility. And it is true, if sometimes sad, that whenever and wherever government draws lines, some parties fall on what they may perceive as the "wrong" side. See Sprandel v. Secretary of HHS, 838 F.2d 23, 27 (1st Cir.1988) (per curiam). There was no cognizable unfairness in this respect. Moreover, having established separate categories for different PRPs, the agency had no obligation to let defendants flit from class to class, thus undermining the rationale and purpose for drawing lines in the first place.

Nor can we say that appellants were entitled to more advance warning of the EPA's negotiating strategy than they received. At the time de minimis PRPs were initially invited to participate in the administrative settlement, the EPA, by letter, informed all of them, including appellants, that:

The government is anxious to achieve a high degree of participation in this de minimis settlement. Accordingly, the terms contained in this settlement offer are the most favorable terms that the government intends to make available to parties eligible for de minimis settlement in this case.

Cannons, 720 F.Supp. at 1033. Appellants knew, early on, that they were within the DMC and could spurn the EPA's proposal only at the risk of paying more at a later time. Although appellants may have assumed that they could ride on the coattails of the major parties and join whatever MP decree emerged--the government had, on other occasions, allowed such cafeteria-style settlements--the agency was neither asked for, nor did it give, any such assurance in this instance. As a matter of law, we do not believe that Congress meant to handcuff government negotiators in CERCLA cases by insisting that the EPA allow polluters to pick and choose which settlements they might prefer to join. And as a matter of equity, we think that if appellants were misled at all, it was by their own wishful thinking.

The district court found the consent decrees to have been the product of fair play. Given that the decrees were negotiated at arm's length among experienced counsel, that appellants (except Crown, see supra note 3) had an opportunity to participate in the negotiations and to join both the first and the second de minimis settlements, and that the agency operated in good faith, the finding of procedural fairness is eminently supportable.

B. Substantive Fairness.

Substantive fairness introduces into the equation concepts of corrective justice and accountability: a party should bear the cost of the harm for which it is legally responsible. See generally Developments in the Law--Toxic Waste Litigation, 99 Harv.L.Rev. 1458, 1477 (1986). The logic behind these concepts dictates that settlement terms must be based upon, and roughly correlated with, some acceptable measure of comparative fault, apportioning liability among the settling parties according to rational (if necessarily imprecise) estimates of how much harm each PRP has done. Cf. Rohm & Haas, 721 F.Supp. at 685 (the most important aspect of judicial review is relationship of settlement figure to proportion of settlor's waste); Cannons, 720 F.Supp. at 1043 (charging more than proportionate liability must be justified in some way, as by unexpected costs or unknown conditions); Kelley, 717 F.Supp. at 517 (approving settlement because it was unlikely that settlor's comparative fault was less than percentage of cleanup costs it agreed to pay); United States v. Conservation Chemical Co., 628 F.Supp. 391, 401 (W.D.Mo.1985) (liability apportionment should be made on basis of comparative fault).

Even accepting substantive fairness as linked to comparative fault, an important issue still remains as to how comparative fault is to be measured. There is no universally correct approach. It appears very clear to us that what constitutes the best measure of comparative fault at a particular Superfund site under particular factual circumstances should be left largely to the EPA's expertise. Whatever formula or scheme EPA advances for measuring comparative fault and allocating liability should be upheld so long as the agency supplies a plausible explanation for it, welding some reasonable linkage between the factors it includes in its formula or scheme and the proportionate shares of the settling PRPs. See United States v. Akzo Coatings, 719 F.Supp. 571, 586-87 (E.D.Mich.1989); Acushnet, 712 F.Supp. at 1031: cf. Gardner & Greenberger, Judicial Review of Administrative Action and Responsible Government, 63 Geo.L.J. 7, 33 (1974) (courts must know why an agency has taken an action if they are to perform their review function adequately). Put in slightly different terms, the chosen measure of comparative fault should be upheld unless it is arbitrary, capricious, and devoid of a rational basis.4 See 42 U.S.C. Sec. 9613(j) (1987); Rohm & Haas, 721 F.Supp. at 681.

Not only must the EPA be given leeway to construct the barometer of comparative fault, but the agency must also be accorded flexibility to diverge from an apportionment formula in order to address special factors not conducive to regimented treatment. While the list of possible variables is virtually limitless, two frequently encountered reasons warranting departure from strict formulaic comparability are the uncertainty of future events and the timing of particular settlement decisions. Common sense suggests that a PRP's assumption of open-ended risks may merit a discount on comparative fault, while obtaining a complete release from uncertain future liability may call for a premium. See, e.g., Cannons, 720 F.Supp. at 1043; Superfund Settlements with De Minimis Waste Contributors: An Analysis of Key Issues by the Superfund Settlements Project, May 8, 1987, Vol. XIV Chem. Waste Lit. Rptr. 34, 46 (June 1987) [hereinafter Superfund Settlements ] (premium should be paid by PRP for benefit of being permitted to cash out). By the same token, the need to encourage (and suitably reward) early, cost-effective settlements, see, e.g., Acushnet, 712 F.Supp. at 1032 (quick settlement deserves recognition in terms of lowered settlement figure); United States v. Seymour Recycling Corp., 554 F.Supp. 1334, 1339 (S.D.Ind.1982) (similar), and to account inter alia for anticipated savings in transaction costs inuring from celeritous settlement, cf., e.g., Mathewson Corp. v. Allied Marine Indus., Inc., 827 F.2d 850, 855-56 (1st Cir.1987) (discussing range of considerations influencing private settlements), can affect the construct. Because we are confident that Congress intended EPA to have considerable flexibility in negotiating and structuring settlements, we think reviewing courts should permit the agency to depart from rigid adherence to formulae wherever the agency proffers a reasonable good-faith justification for departure.

We also believe that a district court should give the EPA's expertise the benefit of the doubt when weighing substantive fairness--particularly when the agency, and hence the court, has been confronted by ambiguous, incomplete, or inscrutable information. In settlement negotiations, particularly in the early phases of environmental litigation, precise data relevant to determining the total extent of harm caused and the role of each PRP is often unavailable. See Superfund Settlements, supra p. 16, at 43. Yet, it would disserve a principal end of the statute--achievement of prompt settlement and a concomitant head start on response activities--to leave matters in limbo until more precise information was amassed. As long as the data the EPA uses to apportion liability for purposes of a consent decree falls along the broad spectrum of plausible approximations, judicial intrusion is unwarranted--regardless of whether the court would have opted to employ the same data in the same way. See Rohm & Haas, 721 F.Supp. at 685-86 (reasonable relationship to some plausible estimate or range of estimates is standard of fairness).

In this instance, we agree with the court below that the consent decrees pass muster from a standpoint of substantive fairness. They adhere generally to principles of comparative fault according to a volumetric standard, determining the liability of each PRP according to volumetric contribution. And, to the extent they deviate from this formulaic approach, they do so on the basis of adequate justification. In particular, the premiums charged to de minimis PRPs in the administrative settlement, and the increased premium charged in the DMC decree, seem well warranted.

The argument that the EPA should have used relative toxicity as a determinant of proportionate liability for response costs, instead of a strictly volumetric ranking, is a stalking horse. Having selected a reasonable method of weighing comparative fault, the agency need not show that it is the best, or even the fairest, of all conceivable methods. The choice of the yardstick to be used for allocating liability must be left primarily to the expert discretion of the EPA, particularly when the PRPs involved are numerous and the situation is complex. See H.R.Rep. No. 99-962, 99th Cong., 2d Sess. at 253 (1986) ("[t]he President has the discretion to allocate the total response costs among potentially responsible parties as the President deems appropriate"). We cannot reverse the court below for refusing to second-guess the agency on this score.

Appellants' next asseveration--that the decrees favor major party PRPs over their less culpable counterparts--is a gross distortion. While the DMC and MP decrees differ to some extent in application of the volumetric share formula, requiring lower initial contributions under the latter, the good-faith justification for this divergence is readily apparent. In return for the premium paid, de minimis PRPs can cash out, thus obtaining two important benefits: reduced transaction costs and absolute finality with respect to the monetization of their overall liability. Cf. Superfund Settlements, supra p. 16, at 42-43. The major PRPs, on the other hand, retain an open-ended risk anent their liability at three of the Sites, see Cannons, 720 F.Supp. at 1042, making any comparison of proportionate contributions a dubious proposition. At the very least, assumption of this unquantifiable future liability under the MP decree warranted some discount--and the tradeoff crafted by the government's negotiators seems reasonable. Indeed, the acceptance of the first and second DMC settlement offers by so many of the de minimis PRPs is itself an indication of substantive fairness toward the class to which appellants belong. See Seymour, 554 F.Supp. at 1339. On this record, the district court did not misuse its discretion in ruling that the decrees sufficiently tracked the parties' comparative fault.

The last point which merits discussion under this rubric involves the fact that the agency upped the ante as the game continued, that is, the premium assessed as part of the administrative settlement was increased substantially for purposes of the later DMC decree. Like the district court, we see no unfairness in this approach. For one thing, litigation is expensive--and having called the tune by their refusal to subscribe to the administrative settlement, we think it not unfair that appellants, thereafter, would have to pay the piper. For another thing, rewarding PRPs who settle sooner rather than later is completely consonant with CERCLA's makeup.

Although appellants berate escalating settlement offers as discriminating among similarly situated PRPs, we think that the government's use of such a technique is fair and serves to promote the explicit statutory goal of expediting remedial measures for hazardous waste sites. See 42 U.S.C. Sec. 9622(a) (1987); see also Cannons, 720 F.Supp. at 1037 (emphasizing congressional interest in expedited cleanups); see generally, Note, Superfund Settlements: The Failed Promise of the 1986 Amendments, 74 Va.L.Rev. 123, 126 (1988) (chief congressional purpose of CERCLA was to provide immediate response to threat of uncontrolled hazardous waste). That the cost of purchasing peace may rise for a laglast is consistent with the method of the statute; indeed, if the government cannot offer such routine incentives, there will be little inducement on the part of any PRP to enter an administrative settlement. Of course, the extent of the differential must be reasonable and the graduation neither unconscionable nor unduly coercive, but these are familiar subjects for judicial review in a wide variety of analogous settings. Cf., e.g., United States v. Ven-Fuel, Inc., 758 F.2d 741, 763-64 (1st Cir.1985) (discussing standard of review anent imposition of civil penalty for oil import violation). We believe that the EPA is entitled to make use of a series of escalating settlement proposals in a CERCLA case and that, as the district court ruled, the serial settlements employed in this instance were substantively fair.

C. Reasonableness.

In the usual environmental litigation, the evaluation of a consent decree's reasonableness will be a multifaceted exercise. We comment briefly upon three such facets. The first is obvious: the decree's likely efficaciousness as a vehicle for cleansing the environment is of cardinal importance. See Cannons, 720 F.Supp. at 1038; Conservation Chemical, 628 F.Supp. at 402; Seymour, 554 F.Supp. at 1339. Except in cases which involve only recoupment of cleanup costs already spent, the reasonableness of the consent decree, for this purpose, will be basically a question of technical adequacy, primarily concerned with the probable effectiveness of proposed remedial responses.

A second important facet of reasonableness will depend upon whether the settlement satisfactorily compensates the public for the actual (and anticipated) costs of remedial and response measures. Like the question of technical adequacy, this aspect of the problem can be enormously complex. The actual cost of remedial measures is frequently uncertain at the time a consent decree is proposed. Thus, although the settlement's bottom line may be definite, the proportion of settlement dollars to total needed dollars is often debatable. Once again, the agency cannot realistically be held to a standard of mathematical precision. If the figures relied upon derive in a sensible way from a plausible interpretation of the record, the court should normally defer to the agency's expertise.

A third integer in the reasonableness equation relates to the relative strength of the parties' litigating positions. If the government's case is strong and solid, it should typically be expected to drive a harder bargain. On the other hand, if the case is less than robust, or the outcome problematic, a reasonable settlement will ordinarily mirror such factors. In a nutshell, the reasonableness of a proposed settlement must take into account foreseeable risks of loss. See Rohm & Haas, 721 F.Supp. at 680; Kelley, 717 F.Supp. at 517; Acushnet, 712 F.Supp. at 1028; Exxon, 697 F.Supp. at 692; Hooker, 540 F.Supp. at 1072. The same variable, we suggest, has a further dimension: even if the government's case is sturdy, it may take time and money to collect damages or to implement private remedial measures through litigatory success. To the extent that time is of essence or that transaction costs loom large, a settlement which nets less than full recovery of cleanup costs is nonetheless reasonable. See Rohm & Haas, 721 F.Supp. at 680 (interpreting "reason